Recycling of Lithium-Ion Batteries Libro

Sustainable Production, Life Cycle Engineering and Management
Series Editors: Christoph Herrmann, Sami Kara
Arno Kwade
Jan Diekmann Editors
Recycling of
Lithium-Ion
Batteries
The LithoRec Way
Sustainable Production, Life Cycle
Engineering and Management
Series editors
Christoph Herrmann, Braunschweig, Germany
Sami Kara, Sydney, Australia
Modern production enables a high standard of living worldwide through products
and services. Global responsibility requires a comprehensive integration of sustain-
able development fostered by new paradigms, innovative technologies, methods and
tools as well as business models. Minimizing material and energy usage, adapting
material and energy flows to better fit natural process capacities, and changing con-
sumption behaviour are important aspects of future production. A life cycle per-
spective and an integrated economic, ecological and social evaluation are essential
requirements in management and engineering. This series will focus on the issues
and latest developments towards sustainability in production based on life cycle
thinking.
More information about this series at http://www.springer.com/series/10615

Arno Kwade Jan Diekmann
•
Editors
Recycling of Lithium-Ion
Batteries
The LithoRec Way
123
Editors
Arno Kwade Jan Diekmann
Institute of Particle Technology, Battery Institute of Particle Technology, Battery
LabFactory Braunschweig LabFactory Braunschweig
Technische Universität Braunschweig Technische Universität Braunschweig
Braunschweig Braunschweig
Germany Germany
ISSN 2194-0541 ISSN 2194-055X (electronic)

Sustainable Production, Life Cycle Engineering and Management
ISBN 978-3-319-70571-2 ISBN 978-3-319-70572-9 (eBook)
https://doi.org/10.1007/978-3-319-70572-9
Library of Congress Control Number: 2017958613
© Springer International Publishing AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Series Editors’ Foreword
Private transportation is going through a dramatic technological transformation.

Increased uptake of electric mobility is expected to decrease the environmental
impacts caused by the tailpipe emissions from internal combustion vehicles sig-
nificantly. In addition, if powered with low carbon intensity energy sources, electric
vehicles might contribute to reducing the amount of greenhouse gas emissions
during their life cycle, therefore, diminishing their contribution to climate change.
This technology, however, involves the development of new automotive compo-
nents, which imply the emergence of new material supply chains. In this regard, the
environmental impact of road transportation has started to transform. Its environ-
mental hotspot increasingly shifts to the manufacturing phase of the vehicle and its
new components. The traction battery required for the vehicle’s operation is one
of the key components. Current commercial traction batteries are expensive, heavy
in weight, and linked with various potential environmental impacts. They contain
high amounts of key engineering metals such as copper, aluminum, nickel, and
cobalt. While the extraction processes of these materials cause numerous local
environmental impacts linked to mine tailing, their refining processes usually
demand large amounts of energy. Furthermore, some of the materials contained in
current lithium-ion traction batteries might face future supply risks due to the
geopolitical instability and potential market constraints caused by resources scarcity
and the low technical feasibility of the extraction processes.
In this context, recycling is a very attractive solution, as it is promoted to return
many of these materials back to their supply chains while preventing further
environmental and social implications. However, recycling is not without envi-
ronmental impact due to energy and resources use during the recovery processes.
Therefore, recycling processes for traction batteries should be designed with the
objective of compensating their life cycle environmental impacts. This means
recovering as much valuable and high-quality material as possible while optimizing
the consumption of energy, time, and resources during the recycling process.
The research within the LithoRec project aimed at developing recycling pro-
cesses, which are able to recover much of the materials used in a commercial
traction battery system while achieving a significant reduction of the energy
v
vi Series Editors’ Foreword
consumption compared to the current commercial pyrometallurgical processes. This

book is the result of more than 5 years of research. It gives a very detailed
description of the processes developed within the LithoRec project. The authors
provide the reader with highly valuable insights into understanding not merely the
theoretical aspects of recycling processes for traction batteries, but also present
analyses of all relevant technical and economic challenges emerging from its
implementation. Therefore, this book makes a significant contribution to our
understanding of many interactions among the technical, economic, and environ-
mental dimensions surrounding the complex process of a battery recycling process.
Christoph Herrmann
Technische Universität Braunschweig,
Braunschweig, Germany
Sami Kara
The University of New South Wales, Sydney
Australia
Preface
Lithium-ion batteries are increasingly applied in electrical vehicles, stationary

energy storage systems, and other consumer products like power tools. The
increasing usage of lithium-ion batteries requires a rise in their production capacity
and a minimization of their ecological impact, e.g., carbon dioxide footprint.
Moreover, the access to the raw materials has to be ensured and the material costs
have to be kept down, although production rates will rise dramatically in the next
years. In order to fulfill these tasks and goals, the spent lithium-ion battery systems,
especially those of the electric vehicles, have to be recycled and fed back into the
material cycle to close the loop. Today, the first recycling technologies have been
developed and are used for the recycling of lithium-ion batteries. The most common
way is the extraction of the most valuable components by disassembling followed
by pyro-metallurgical processes in a smelting furnace. This process recovers
components of high costs like cobalt, nickel, and copper, enabling synthesis of new
battery materials. However, several other materials including lithium are transferred
into slag and are therefore lost for further battery use.
In order to overcome these problems, a novel recycling process was developed
in the two LithoRec projects financially supported by the German Ministry for
Environment, Nature Conservation, Building, and Nuclear Safety. This book pre-
sents the results of the LithoRec II project. The LithoRec way incorporates the
battery system transport and the establishment of safety strategies for further battery
system handling, dismantling of the battery system, safe crushing of battery cells,
and separation of the different battery components, including electrolyte and active
materials. To gain new raw materials for the synthesis of active material, the
individual compounds were extracted from the coating materials and further sep-
arated by hydro-metallurgical processes. Special attention was paid to a safe pro-
cess design, the maximization of the recycling rate, and the ecological credits, as
well as the minimization of recycling costs. The knowledge was finally consoli-
dated in a pilot plant at the Technische Universität Braunschweig, where 1.4 tons of
battery systems were recycled. The different book chapters show that a sophisti-
cated technological strategy for the recycling of lithium-ion batteries exists already
today, which will enable a closed loop for battery materials in the near future.
vii
viii Preface
As the scientific speaker of the LithoRec projects, I would like to thank all
involved partners and collaborators for the intensive and highly motivated work
within the last years, and for the preparation of the different book chapters showing
the state of the art in recycling of lithium-ion batteries. In place of the collaborators
my special thanks go to Christian Hanisch, Jan Diekmann, and Martin Steinbild for
their continuous commitment. Our special thank goes to the German Ministry for
Environment, Nature Conservation, Building, and Nuclear Safety for the financial
support and to the project executing organization VDI/VDE Innovation + Technik
GmbH, especially to Dr. Randolf Schließer for the intensive supervision of the
project.
Braunschweig, Germany Arno Kwade

Scientific speaker of the LithoRec projects
Contents
1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Sergej Rothermel, Martin Winter and Sascha Nowak
2 The LithoRec Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Jan Diekmann, Sergej Rothermel, Sascha Nowak and Arno Kwade
3 Potential Dangers During the Handling of Lithium-Ion
Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Jan Diekmann, Martin Grützke, Thomas Loellhoeffel,
Matthias Petermann, Sergej Rothermel, Martin Winter,
Sascha Nowak and Arno Kwade
4 Overdischarging Lithium-Ion Batteries . . . . . . . . . . . . . . . . . . . . . . 53
Daniel Hauck and Michael Kurrat
5 Disassembly Planning and Assessment of Automation Potentials
for Lithium-Ion Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Felipe Cerdas, Roman Gerbers, Stefan Andrew, Jan Schmitt,
Franz Dietrich, Sebastian Thiede, Klaus Dröder
and Christoph Herrmann
6 Safe, Flexible and Productive Human-Robot-Collaboration for
Disassembly of Lithium-Ion Batteries . . . . . . . . . . . . . . . . . . . . . . . 99
Roman Gerbers, Kathrin Wegener, Franz Dietrich and Klaus Dröder
7 Crushing of Battery Modules and Cells . . . . . . . . . . . . . . . . . . . . . 127
Jan Diekmann, Steffen Sander, Guido Sellin, Matthias Petermann
and Arno Kwade
8 Separation of the Electrolyte—Thermal Drying . . . . . . . . . . . . . . . 139
Friederike Stehmann, Christian Bradtmöller and Stephan Scholl
ix
x Contents
9 Separation of the Electrolyte—Solvent Extraction . . . . . . . . . . . . . 155

Paul Haas, Stefan Pfeifer, Jannes Müller, Christian Bradtmöller
and Stephan Scholl
10 Electrolyte Extraction—Sub and Supercritical CO2 . . . . . . . . . . . . 177
Sergej Rothermel, Martin Grützke, Xaver Mönnighoff, Martin Winter
and Sascha Nowak
11 Off Gas Cleaning by Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Friederike Stehmann and Stephan Scholl
12 Material Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Jan Diekmann, Steffen Sander, Guido Sellin and Arno Kwade
13 Hydrometallurgical Processing and Thermal Treatment of Active
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Sergej Rothermel, Steffen Krüger, Martin Winter and Sascha Nowak
14 Realization in a Demonstration Plant . . . . . . . . . . . . . . . . . . . . . . . 247
Jan Diekmann, Thomas Loellhoeffel and Arno Kwade
15 Economic Assessment of the LithoRec Process . . . . . . . . . . . . . . . . 253
Christian Thies, Karsten Kieckhäfer, Claas Hoyer
and Thomas S. Spengler
16 Environmental Aspects of the Recycling of Lithium-Ion Traction
Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Felipe Cerdas, Stefan Andrew, Sebastian Thiede
Abbreviations
μCT Micro-computer tomography

3p4s 3parallel4series
A Adsorptive
AC Activated carbon
ACN Acetonitrile
ADP Abiotic depletion potential
AP Acidification potential
ARS Axial rotary shear
BET Brunauer–Emmett–Teller theory
BMU Battery management unit
BT Breakthrough
CC Constant current
CCD Charge-coupled device
CE Counter electrode
CED Cumulative energy demand
CHB Cyclo hexyl benzene
CID Current interrupt device
CMC Carboxymethyl cellulose
CP Constant power
CR Constant resistance
CV Constant voltage
DC Direct current
DEC Diethyl carbonate
DEDOHC Diethyl-2,5-dioxahexane dicarboxylate
DEFP Diethyl fluorophosphate
DMC Dimethyl carbonate
DMDOHC Dimethyl-2,5-dioxahexane dicarboxylate
DSC Differential scanning calorimetry
EC Ethylene carbonate
EDX Energy dispersive X-ray analysis
ELV End-of-life vehicle
xi
xii Abbreviations
EMC Ethyl methyl carbonate

EMDOHC Ethylmethyl-2,5-dioxahexane dicarboxylate
EMFP Ethylmethyl fluorophosphate
EOL End-of-life
EP Eutrophication potential
Epm Exergy primary material
Esm Exergy secondary material
EV Electric vehicles
FTIR Fourier transform infrared spectroscopy
GC Gas chromatography
GC–MS Gas chromatography–mass spectrometry
GHG Greenhouse gas
GWP Global warming potential
HEV Hybrid electric vehicle
HHPCO2 Helium head pressurized carbon dioxide
HSAL High surface area lithium
HTP Human toxicity potential
HV High voltage
IAST Ideal adsorbed solution theory
IC Ionic chromatography
IC/ESI-MS Ion chromatography–electrospray ionization-mass spectrometry
ICP-MS Inductively coupled plasma mass spectrometry
ICP-OES Inductively coupled plasma optical emission spectrometry
IDLH Immediately dangerous to life and health
LCA Life cycle assessment
LCO Lithium cobalt oxide LiCoO2
LEL Lower explosion limit
LFP Lithium iron phosphate
LIB Lithium-ion battery
LOD Limit of detection
LOQ Limit of quantification
MMU Module management unit
MOSFET Metal oxide semiconductor field effect transistor
NA Necessity to automate the corresponding disassembly operation
NCA Lithium nickel aluminum oxide
NCM Lithium nickel cobalt manganese oxide
NCM 111 Lithium nickel manganese oxide LiNi1/3Co1/3Mn1/3O2
NCM 532 Lithium nickel manganese oxide LiNi0.5Co0.3Mn0.2O2
NCM 622 Lithium nickel manganese oxide LiNi0.6Co0.2Mn0.2O2
NMP N methyl pyrrolidone
NPV Net present value
OEM Original equipment manufacturer
PA Polyamide
PC Propylene carbonate
PHEV Plug-in hybrid electric vehicle
Abbreviations xiii
PLC Programmable logic control

PMFP Particle matter formation potential
POCP Photochemical ozone creation potential
PTC Positive temperature coefficient
PTFE Polytetrafluoroethylene
PVDF Polyvinylidene fluoride
RE Reference electrode
rpm Rounds per minute
RRS Radial rotary shear
scCO2 Supercritical carbon dioxide
SEI Solid electrolyte interface
SEM Scanning electron microscopy
SFE Supercritical fluid extraction
SOC State of charge
SOH State of health
subCO2 Subcritical carbon dioxide
TAA Technical ability of a disassembly process to be automated
TAP Terrestrial eutrophication potential
TGA Thermogravimetric analysis
UEL Upper explosion limit
WE Working electrode
WEEE Waste electrical and electronic equipment
XRD X-ray powder diffraction
Symbols
V Volume (mL)
X Carbon loading (mmol/g)
m Mass (g)
Y Gas loading (g/L)
M Molar mass (g/mol)
q Density (kg/m³)
b Langmuir parameter (L/mmol)
b’ Langmuir parameter (–)
t Tóth Parameter (–)
u Saturation (–)
p Spreading pressure (–)
R Ideal gas constant (J/molK)
T Temperature (K)
y Concentration is gas phase (mol/mol)
x Concentration in adsorbed phase (mol/mol)
a Seperation factor (–)
A Area (m²)
t Time (h)
xv
List of Figures
Fig. 1.1 Overview of the global installed capacity of the wind and
solar power in the years 2005—2015 (Renewables 2015
Global Status Report 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2
Fig. 1.2 Simplified representation of the power grid situation of
tomorrow. The generated energy is comprised of the steady
growing share of renewables such as solar (a), wind power
(b) and conventional base load power generation (c). To
compensate fluctuations of wind and solar energy, the
mismatch between power generation (d) and load (f) is
managed by intermediately stored battery power (e). Inspired
by Hoelzle and Chang 2014 . . . . . . . . . . . . . . . . . . . . . . . . . .. 4
Fig. 1.3 Schematic illustration of the fundamental operating principle
of a LIB in the discharged (a) and in the charged state (b) . . .. 5
Fig. 1.4 Schematic illustration of a metal-based rechargeable lithium
battery showing the typical morphology of the lithium surface
before charge/discharge cycling (a) and after an undefined
number of cycles (b). The continuously growing dendrites
potentially penetrate the separator membrane and cause a
short circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7
Fig. 1.5 Schematic illustration of the graphite exfoliation process
initiated by the co-intercalation of solvated Li+ ion between
the graphene layers (above). Reductive decomposition of
intercalated solvent shell leads to gas formation and
delamination of individual graphene layers (exfoliation).
Protective function of the passivation layer (SEI), which
prevents the entry of the solvation shell and thus minimizes
the graphite exfoliation processes (below). Adapted from
Vetter et al. (2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Fig. 1.6 Cutaway view of a cylindrical 18650 cell . . . . . . . . . . . . . . . . . 11
Fig. 1.7 Cutaway view of a wound prismatic cell . . . . . . . . . . . . . . . . . . 13
Fig. 1.8 Cutaway view of a stacked pouch cell . . . . . . . . . . . . . . . . . . . . 14
xvii
xviii List of Figures
Fig. 1.9 Electrode fabrication process . . . . . . . . . . . . . . . . . . . . . . . . . .. 16

Fig. 1.10 Schematic illustration of the cylindrical cell fabrication
process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 17
Fig. 1.11 Schematic illustration of the wound prismatic cell fabrication
process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 19
Fig. 1.12 Schematic illustration of the stacked pouch cell fabrication
process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 20
Fig. 1.13 Major innovations in material technology. Past, existing and
expected battery chemistries
(Pavoni 2012; Andre et al. 2015) . . . . . . . . . . . . . . . . . . . . . .. 22
Fig. 1.14 Tetrahedral coordination of Co(II) (a) and octahedral
coordination of Ni(II) (b) by dialkylphosphonic acid (c). In
presence of water the octahedrally coordinated Ni(II) complex
undergoes a reversible ligand exchange increasing its
hydrophilic properties. The disintegration of the complex is
initiated by reducing the pH. Exemplarily demonstrated on the
Co(II) complex (c). Adapted from Kathryn (2008) . . . . . . . . .. 26
Fig. 2.1 Process flow chart of the investigated process steps and the
overall LithoRec process . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 36
Fig. 3.1 General decomposition pathways for the formation of
transesterifications products (a), oligocarbonate-based
products (b), organophosphate-based products (c) organic
fluorophosphate-based products and (d) hydrolysis products.
Reprinted with permission of (Nowak and Winter 2017) . . . . .. 42
Fig. 3.2 Pictures of the changes on the storage containers over storage
duration. Reprinted with permission of
(Grützke et al. 2015a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43
Fig. 3.3 Ionic decomposition and reaction products identified via
IC/ESI-MS. Reprinted with permission of
(Grützke et al. 2015a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43
Fig. 3.4 Decomposition and reaction products identified via GC-MS.
Reprinted with permission of (Grützke et al. 2015a) . . . . . . . .. 44
Fig. 3.5 Installation of the 18650 battery cell in a PTFE holder . . . . . .. 45
Fig. 3.6 Temperatures on the battery cells surfaces during nail
penetration tests with 18650 battery cells . . . . . . . . . . . . . . . .. 46
Fig. 3.7 Concentrations of carbon monoxide (CO) during nail
penetration tests of 18650 battery cells . . . . . . . . . . . . . . . . . .. 46
Fig. 3.8 Concentrations of hydrogen fluoride (HF) during nail
penetration tests of 18650 battery cells . . . . . . . . . . . . . . . . . .. 47
Fig. 3.9 Relative comparison of highest concentrations of the
identified gaseous products of nail penetration tests of 18650
battery cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 47
Fig. 3.10 μCT-images of the inner structure of the battery cells after
abuse (at top: NCM 2.2 Ah, at bottom: NCA 3.4 Ah, left:
structure including electrodes, right: rod and aluminum) . . . . .. 48
List of Figures xix
Fig. 3.11 Concentrations of hydrogen fluoride (HF) during nail

penetration test of 18650 battery cells, installed in a PTFE or
a steel holder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 49
Fig. 3.12 Schematic drawing of the hazards associated with lithium-ion
batteries and their interaction . . . . . . . . . . . . . . . . . . . . . . . . . .. 50
Fig. 4.1 Flowchart proposing how to deal with LIBs before and after
overdischarging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 55
Fig. 4.2 Definition of “overdischarging” and “pole reversal” . . . . . . . .. 57
Fig. 4.3 Different basic modes of electrical behaviour for
overdischarging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 58
Fig. 4.4 Devices used and their energy transformation concepts to
overdischarge cells, modules, and systems . . . . . . . . . . . . . . . .. 60
Fig. 4.5 Single cell resistor (variant A) . . . . . . . . . . . . . . . . . . . . . . . . .. 62
Fig. 4.6 Selectable resistor device (variant B) . . . . . . . . . . . . . . . . . . . .. 63
Fig. 4.7 Experimental setup to discharge battery cells in conductive
liquids (variant C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 64
Fig. 4.8 Device with MOSFETs to discharge cells and modules with
constant current (variant D) . . . . . . . . . . . . . . . . . . . . . . . . . . .. 65
Fig. 4.9 Electronically adjustable loads to discharge modules and
systems in CR, CC and CP modes [left: heat transformation
(variant E); right: energy recuperation into grid (variant F)] . .. 65
Fig. 4.10 Circuit diagram of basic setup for all discharge variants . . . . .. 66
Fig. 4.11 Overdischarge current/voltage characteristic curve for a 3.1
Ah cell with variant A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 68
Ah cell with variant B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 68
Ah cell with variant C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 69
Fig. 4.14 Experimental setup (left), cell poles (middle left), liquid
sample (middle right), brown liquid (right) . . . . . . . . . . . . . . .. 69
Ah cell with variant D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 72
Fig. 4.16 Overdischarge current/voltage characteristic curve for a VW
e-Golf battery system with variant E . . . . . . . . . . . . . . . . . . . .. 72
Fig. 4.17 Overdischarge and relaxation curves for 50 Ah NCM cells at
different temperatures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 74
Fig. 4.18 Relaxation curves for three 25 Ah and three 3.1 Ah cells for
different short circuit times . . . . . . . . . . . . . . . . . . . . . . . . . . .. 75
Fig. 4.19 Relaxation curves for three 3p4s modules for different short
circuit times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 76
Fig. 4.20 Short circuit current for a 2p4s module at 2.8 V
relaxation voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 77
xx List of Figures
Fig. 4.21 Overdischarge current/voltage characteristic curves for VW

e-Golf battery systems with 15 kW (variant E) and 3.5 kW
(variant F) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 78
Fig. 4.22 Heat distribution for a VW e-Golf battery system during
overdischarging at 9 °C ambient temperature . . . . . . . . . . . . .. 78
Fig. 4.23 Temperature rise and pole reversal for a 25 Ah NCM cell . . .. 79
Fig. 4.24 Cell pole reversal and heat rise while overdischarging a 2p4s
battery module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 80
Fig. 5.1 Methodical approach for the definition of a disassembly
system and the assessment of automation potentials . . . . . . . .. 88
Fig. 5.2 Setup of the battery system used in the hybrid Audi Q5 . . . . .. 89
Fig. 5.3 Main components of the Audi Q5 hybrid battery system
(closed system) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 89
Fig. 5.4 Main components of the Audi Q5 hybrid battery system (open
system) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 90
Fig. 5.5 Disassembly matrix for the Audi Q5 hybrid battery system
(Wegener et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 92
Fig. 5.6 Disassembly priority graph (Wegener et al. 2014) . . . . . . . . . .. 93
Fig. 5.7 Portfolio analysis (left) and bar chart for process oriented
visualisation (Herrmann et al. 2012) . . . . . . . . . . . . . . . . . . . .. 95
Fig. 6.1 Manual disassembly tests on battery systems
(Wegener et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Fig. 6.2 Disassembly workstation concept (Wegener et al. 2015) . . . . . . 110
Fig. 6.3 Cost-optimized robot tool (w/o protective cover) for the
disassembly of various screw connections
(Gerbers et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Fig. 6.4 Recording of the depth image camera with detected human
hand (left) compliance and force control for contact detection
(right) (Gerbers et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Fig. 6.5 Possibilities of intuitive programming and control
(Dröder et al. 2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Fig. 6.6 User interface for gesture control (Trentlage 2014) . . . . . . . . . . 114
Fig. 6.7 Functional principle of the direct positioning control
(Trentlage 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Fig. 6.8 Possibility of intuitive robot programming and control using a
handheld device (Dröder et al. 2017) . . . . . . . . . . . . . . . . . . . . . 116
Fig. 6.9 Template and screw database (Küthe 2015) . . . . . . . . . . . . . . . . 117
Fig. 6.10 Process steps for object recognition and localization
(Küthe 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Fig. 6.11 Process time in dependence on the chamfer diameter
(Gerbers et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Fig. 6.12 Experimental results of the tool validation
(Gerbers et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
List of Figures xxi
Fig. 6.13 Experimental accuracy analysis of direct positioning control

(Trentlage 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Fig. 6.14 Screw images for reliability testing . . . . . . . . . . . . . . . . . . . . . . 121
Fig. 6.15 False positives during screw detection (Küthe 2015) . . . . . . . . . 121
Fig. 6.16 Implementation of the hybrid disassembly workstation
(Dröder et al. 2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Fig. 7.1 Variations of module designs. . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Fig. 7.2 Composition of a generic battery module . . . . . . . . . . . . . . . . . 130
Fig. 7.3 Gas emissions during decomposition of lithium
hexafluorophosphate in the crushing product (measured via
Fourier transform infrared spectroscopy) . . . . . . . . . . . . . . . . . . 131
Fig. 7.4 Ageing influence on gas emissions of LIB cells (18650 battery
cells) during crushing (Diekmann et al. 2016a) . . . . . . . . . . . . . 133
Fig. 7.5 Gas emissions during crushing of LIB cells (PHEV 1 battery
cells) in a technical scale crusher. Reprinted with permission
of Diekmann et al. (2016b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fig. 7.6 Influence of different crushing mechanisms on the particle size
distribution of the crushing product (RRS: radial rotary shear,
ARS: axial rotary shear, M: mesh size); data obtained from
Mämpel (2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Fig. 7.7 Influence of mesh size of the discharge screen on the particle
size distribution of the crushing product (RRS: radial rotary
shear, M: mesh size); data obtained from Mämpel (2013) . . . . . 136
Fig. 7.8 Inert crusher designed for the recycling of lithium-ion battery
modules by Hosokawa Alpine Aktiengesellschaft . . . . . . . . . . . 136
Fig. 8.1 Vapor pressure (a) and boiling temperature (b) of remaining
liquid in the battery fragments . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Fig. 8.2 Mol fraction of carbonates in the residual liquid within the
material to be dried along the drying progress. Maximum
temperature is set to 120 °C at a pressure of 0.1 bar . . . . . . . . . 143
Fig. 8.3 Nitrogen demand dependent on the maximum temperature at
0.1 bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Fig. 8.4 Nitrogen demand for sufficient drying and fraction of
low-boiler in the gas after drying. Based on 1 battery
system = 220 kg and a maximum temperature of 60 °C
(Stehmann et al. 2017). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Fig. 8.5 Composition of off gas with condensation temperature of 20 °
C without extraction (a) and with E/Emin = 10 (b) . . . . . . . . . . 144
Fig. 8.6 Shovel dryer HTL-VT 10 (left) with drying chamber
including shovels (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Fig. 8.7 Experimental setup including dryer HTL-VT 10, two
condensers and vacuum pump (P1) . . . . . . . . . . . . . . . . . . . . . . 146
Fig. 8.8 Metal fragments sorted out prior to drying (left) and shovel
equipped with PTFE-plates (right) to prevent blocking . . . . . . . 146
xxii List of Figures
Fig. 8.9 Shovel dryer HTC-VT 140 (AVA-Huep GmbH & Co. KG,
Herrsching, Germany) at Solvay GmbH, Hanover and flow
diagram including the periphery . . . . . . . . . . . . . . . . . . . . . . . . 147
Fig. 8.10 Insight of the drying chamber with agitator shaft and
shovel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Fig. 8.11 Pressure and temperature course of all experiments . . . . . . . . . . 149
Fig. 8.12 Fouling at dryer exit and entry of vapor filter . . . . . . . . . . . . . . 150
Fig. 9.1 Preceding and following process steps of extraction . . . . . . . . . 156
Fig. 9.2 Büchi Glas Uster Versoclave 3 with anchor stirrer and filter at
the bottom, 0.5 L vessel on left side . . . . . . . . . . . . . . . . . . . . . 158
Fig. 9.3 Hexafluorophosphate concentration in extract with different
solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Fig. 9.4 Absolute extracted mass of lithium hexafluorophosphate for
different solvent to solid mass ratios . . . . . . . . . . . . . . . . . . . . . 161
Fig. 9.5 Temperature influence on extraction of lithium
hexafluorophosphate with PHEV 1 battery material from
Electrocycling GmbH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Fig. 9.6 Temperature influence on the extraction using Panasonic CGR
18650 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Fig. 9.7 Mass flow chart for the four stage extraction . . . . . . . . . . . . . . . 165
Fig. 9.8 Mass fraction of hexafluorophosphate and fluoride in extract
after extraction stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Fig. 9.9 Accumulated absolute mass of hexafluorophosphate in
extract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Fig. 9.10 Temperature influence onto extraction with water . . . . . . . . . . . 167
Fig. 9.11 Process sequence for an extraction sequence with DMC and
water each followed by a drying process . . . . . . . . . . . . . . . . . . 168
Fig. 9.12 Hexafluorophosphate and fluoride mass fraction in extract for
a combination of extraction with DMC and
water in 0.5 L scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Fig. 9.13 Optimized sequence of process steps for the extraction with
organic solvent and water with transfer of slurry . . . . . . . . . . . . 170
Fig. 9.14 Comparison of mass fraction lithium hexafluorophosphate in
extract for the experimental data and model based on dilution
of liquid residue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Fig. 9.15 Calculated mass fraction of LiPF6 in extract for different
solvent to solid mass ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Fig. 10.1 Time dependency of the amount of recovered electrolyte from
commercial 18650 cells after formation extracted with
supercritical (300 bar, 40 °C; red stars) and liquid (60 bar, 25 °
C; black squares) CO2 (Grutzke et al. 2015) . . . . . . . . . . . . . . . 180
List of Figures xxiii
Fig. 10.2 Compositions of the recovered electrolyte from commercial

18650 cells after formation extracted with supercritical a and
liquid b CO2 for the first three data points of Fig. 10.1
determined with GC-MS. Blue, top: EC; yellow, middle:
EMC; magenta, bottom: DMC (Grutzke et al. 2015) . . . . . . . . . 181
Fig. 10.3 Time dependency of the amount of recovered electrolyte from
commercial 18650 cells after formation extracted with liquid
CO2 and 0.5 mL/min additional solvents (black stars: ACN;
magenta triangles: ACN/PC (3:1); red circles: DEC) (Grutzke
et al. 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Fig. 10.4 Compositions and amounts (determined with GC-MS and IC)
of the recovered electrolytes from commercial 18650 cells
after formation extracted with liquid CO2 and additional
solvents for 30 min, with subsequent 20 min without
additional solvent. Red, top: LiPF6; blue, below: EC; yellow,
middle: EMC; magenta, bottom: DMC
(Grutzke et al. 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Fig. 10.5 Discharge capacities for aged (square, red: 20 °C, 70% SOH;
triangle, blue: 45 °C, 70% SOH) commercial 18650 cells
(Grützke et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Fig. 10.6 Chromatogram (GC-MS) of the extract from different aged
commercial 18650 cells (Grützke et al. 2014) . . . . . . . . . . . . . . 184
Fig. 11.1 Lab scale adsorber with model gas preparation: Nitrogen feed
(B01), heated DMC reservoir (B02), condenser (W01),
adsorber (B03), gas chromatography (GC) and vacuum pump
(P01) (Stehmann and Scholl 2015) . . . . . . . . . . . . . . . . . . . . . . 190
Fig. 11.2 Breakthrough curve of fixed bed adsorber with ideal
breakthrough time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Fig. 11.3 Adsorption equilibria of DMC onto activated carbon SC40.
Carbon loading dependent on (a) gas loading and
(b) saturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Fig. 11.4 Adsorption equilibrium of DMC and EMC onto activated
carbon SC40. (a) gas phase and adsorbed phase compositions
at a constant total gas loading of Y = 0.1 mmol L−1,
(b) selectivity at a constant gas composition of yEMC = 0.15
mol mol−1 (Stehmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . 197
Fig. 11.5 Tóth-parameter of two-component adsorption (DMC and
EMC) onto activated carbon SC40 at 30 °C with t = 0.3.
Dots: Parameter of experimental data. Lines: Parameter
calculated with IAST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Fig. 11.6 Breakthrough curve of model off gas loaded with DMC, fixed
bed is filled with carbon SC40 (Stehmann et al. 2017a) . . . . . . 198
xxiv List of Figures
Fig. 11.7 (a) Location of temperature peaks along the adsorption

process. (b) Parity chart of temperature and ideal
breakthrough time (Stehmann et al. 2017a) . . . . . . . . . . . . . . . . 199
Fig. 11.8 Calculated carbon loading corresponding to gas loading
(Stehmann et al. 2017a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Fig. 11.9 Carbon loading and conversion dependent on the equilibration
time for adsorption of DMC onto Norit RB4 (Stehmann et al.
2017b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Fig. 11.10 Conversion dependent on temperature for adsorption of DMC
onto Norit RB4 (Stehmann et al. 2017b) . . . . . . . . . . . . . . . . . . 201
Fig. 11.11 Decomposition rate and methanol concentration at steady
state occurring during a fixed bed adsorption of DMC onto
activated carbon SC40 at 30 °C . . . . . . . . . . . . . . . . . . . . . . . . . 202
Fig. 11.12 Exhaust gas cleaning of the mobile shredder (a) and analysis
of the resulting carbon loading and composition of electrolyte
adsorbed by carbon (b) (Stehmann et al. 2017a) . . . . . . . . . . . . 203
Fig. 11.13 Recommended setup for off gas cleaning with adsorption
including heating and purge gas, based on
(Mersmann et al. 2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Fig. 12.1 Composition of a generic battery system of an electric vehicle
(Diekmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Fig. 12.2 Scheme of the function principle of a zig-zag sifter (filled
dots: high density, unfilled dots: low density) . . . . . . . . . . . . . . 209
Fig. 12.3 Cumulated distribution of the separating parameter of
different materials of a crushed and dried battery module
(Diekmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Fig. 12.4 Composition and components of the heavy parts fraction . . . . . 211
Fig. 12.5 Change of the cumulated distribution of fragments of a battery
module after different process steps. The cumulated
distributions relate to the light product after classification and
to the residue after sieving (Diekmann et al. 2016) . . . . . . . . . . 212
Fig. 12.6 Yield of the coating materials with/without 2nd crushing and
mixing step and resulting impurities. Modified from
(Diekmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Fig. 12.7 Mass output of separator and aluminum in light and heavy
product of the 2nd air-classification without 2nd crushing step
(Diekmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Fig. 12.8 Mass output of separator and aluminum in light and heavy
product of the 2nd air-classification with 2nd crushing step
(Diekmann et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
Fig. 12.9 Product fractions of the separation processes and their
compositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
List of Figures xxv
Fig. 12.10 Mass output of aluminum and copper foil in light and heavy
product after separation via air-classification; data obtained
from (Mämpel 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Fig. 12.11 Mass output and content of Al and Cu foil in accept and reject
fraction of an optical separation (1st run); data obtained from
(Mämpel 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Fig. 12.12 Mass output and content of Al and Cu foil in accept and reject
fraction of an optical separation (2nd run); data obtained from
(Mämpel 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Fig. 13.1 Schematic illustration of the material process chain of the
proposed recycling concept starting from electrode processing
(Krüger et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Fig. 13.2 SEM-images of re-synthesized NCM materials from a pure
metal salts b rejects material and c cycled material as source.
The images were taken at an acceleration voltage of 3 kV
(Krüger et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Fig. 13.3 X-ray powder diffraction patterns of the reference, rejects and
cycled NCM materials in different stage of aging (Krüger
et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Fig. 13.4 Schematic cutaway illustration of a Swagelok® type T-cell
with a three-electrode setup in a fully assembled state. The
magnification shows the electrode stack placed between the
current collectors in the sequence from left to right: metallic
lithium/separator/graphite coating/NCM coating/aluminum
foil representing a half-cell setup. The reference electrode
equipped with metallic lithium is placed perpendicular to the
electrode stack, separated by a separator membrane
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Fig. 13.5 Cycling Performance of reference NCM material compared to
material with electrode rejects and cycled NCM as resource.
Cycling was performed at 1C in a 1 M LiPF6 EC:DMC 1:1
(wt/wt) electrolyte with lithium as reference and counter
electrode. The potential range was 3.0–4.2 V versus Li/Li+
(Krüger et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Fig. 13.6 C-rate test of the recycled materials originating from rejects
NCM and cycled NCM electrodes compared to the reference
material. The NCM/graphite full cells were investigated in an
LiPF6 EC:DMC 1:1 (wt/wt) electrolyte in a voltage region
from 3.0 to 4.2 V (Krüger et al. 2014) . . . . . . . . . . . . . . . . . . . 227
Fig. 13.7 Decomposition temperatures of specific surface oxygen
groups. Adapted from Ref. (Collins et al. 2015; Fuente et al.
2003; Figueiredo et al. 1999) . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
xxvi List of Figures
Fig. 13.8 Potential faradaic and electrostatic reactions of Li+ cation and
PF6+ anion with carboxilic, quinone and pyrone surface
oxygen groups. Adapted from Ref. (Collins et al. 2015) . . . . . . 229
Fig. 13.9 Electrochemical cycling of the Panasonic CGR18650CH
Li-ion MH12210 cell until 70% SOH. The charge/discharge
cycling was performed with a constant charge current of 0.44
A (1C) and a constant discharge current of 2.2 A (1C). The
charging and discharging cut-off voltages were set to 4.2 and
3.0 V respectively in accordance with the manufacturers
specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Fig. 13.10 TGA and DSC analysis of the untreated negative electrode
coating from a CGR18650CH Li-ion MH12210 cell after
electrochemical ageing (70% SOH) and as delivered (100%
SOH). The amount of samples: 27.143 ± 0.271 mg (70%
SOH), 27.640 ± 0.276 mg (100% SOH); inert atmosphere
(flow): argon (100.0 mL min−1); sample pan: alumina (90
μL); temperature ramp: 10.0 ± 0.1 °C min−1 (Rothermel et al.
2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Fig. 13.11 SEM image of the negative electrode originating from the
CGR18650CH Li-ion MH12210 cell (100% SOH) (Rothermel
et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Fig. 13.12 SEM image of the thermally treated graphite originating from
the CGR18650CH Li-ion MH12210 cell (100% SOH). White
colored (less electronically conductive) particles distributed
over the particle surface can be observed
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Fig. 13.13 Energy dispersive X-ray analysis (EDX) image of the
thermally treated graphite originating from the CGR18650CH
Li-ion MH12210 cell (100% SOH) shown in section a for the
elements oxygen b and phosphorus c of the scanned are
shown as well (Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . 233
Fig. 13.14 SEM image of graphite originating from the CGR18650CH
Li-ion MH12210 cell (100% SOH) after subcritical carbon
dioxide assisted electrolyte extraction with acetonitrile
addition as a co-solvent and subsequent thermal treatment.
The particle surface is free from crystallites observed in
Fig. 13.12 (Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . 233
Fig. 13.15 SEM image of the thermally treated graphite originating from
the CGR18650CH Li-ion MH12210 cell (100% SOH) after
supercritical carbon dioxide assisted electrolyte extraction and
subsequent thermal treatment. The particle surface is free from
crystallites observed in Fig. 13.12 (Rothermel et al. 2016) . . . . 234
List of Figures xxvii
Fig. 13.16 Raman spectrum of the thermally treated non-extracted

graphite originating from the CGR18650CH Li-ion MH12210
cell (100% SOH). Integrated peaks D and G are shaded in
grey. Laser wavelength: 532 nm; laser power: 15 mW;
diffraction grating: 330 gr mm−1; total acquisition time: 25 s
(5 image frames, 5 s per frame) (Rothermel et al. 2016) . . . . . . 236
Fig. 13.17 Raman spectra of thermally treated graphite without
electrolyte extraction (non-ex.), graphite after subcritical
carbon dioxide assisted electrolyte extraction with acetonitrile
addition as a co-solvent and subsequent thermal treatment
(subcCO2) and graphite after supercritical carbon dioxide
assisted electrolyte extraction and subsequent thermal
treatment originating from the CGR18650CH Li-ion
MH12210 cell (70% SOH/100% SOH). Laser wavelength:
532 nm; Laser power: 15 mW; Diffraction grating:
330 gr mm−1; Total acquisition time: 25 s (5 image frames,
5 s per frame) (Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . 237
Fig. 13.18 Constant current discharge capacities and Coulombic
efficiencies curves of thermally treated non-extracted graphite
originating from the CGR18650CH Li-ion MH12210 cell
(70% SOH/100% SOH) used as WE in a half cell
setup. Measurements were performed using a three-electrode
setup with metallic lithium as RE and CE. Electrolyte:
SelectiLyte™ LP47 (1 M Lithium hexafluorophosphate in
EC:DEC 3:7); WE: graphite/Super C65/Na-CMC (90/5/5);
discharging cut-off potential: 1.5 V versus Li/Li+; charging
cut-off potential: 0.02 V versus Li/Li+; specific current:
186 mA g−1, 0.5C (three initial formation cycles 37 mA g−1,
0.1C); temperature: 20 °C (Rothermel et al. 2016) . . . . . . . . . . 238
efficiencies of the graphite originating from the CGR18650CH
Li-ion MH12210 cell (70% SOH/100% SOH) after subcritical
addition as a co-solvent and subsequent thermal treatment
used as WE in a half cell setup. Measurements were
performed using a three-electrode setup with metallic lithium
as RE and CE. Electrolyte: SelectiLyte™ LP47 (1 M Lithium
hexafluorophosphate in EC:DEC 3:7); WE: graphite/Super
C65/Na-CMC (90/5/5); discharging cut-off potential: 1.5 V
versus Li/Li+; charging cut-off potential: 0.02 V versus
Li/Li+; specific current: 186 mA g−1, 0.5C (three initial
formation cycles 37 mA g−1, 0.1C); temperature: 20 °C
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
xxviii List of Figures

efficiencies of graphite originating from the CGR18650CH
Li-ion MH 12210 cell (70% SOH/100% SOH) after
supercritical carbon dioxide assisted electrolyte extraction and
subsequent thermal treatment used as WE in a half cell
setup. Measurements were performed using a three-electrode
setup with metallic lithium as RE and CE. Electrolyte:
SelectiLyte™ LP47 (1 M Lithium hexafluorophosphate in EC:
DEC 3:7); WE: graphite/Super C65/Na-CMC (90/5/5);
discharging cut-off potential: 1.5 V versus Li/Li+; charging
cut-off potential: 0.02 V versus Li/Li+; specific current:
186 mA g−1, 0.5C (three initial formation cycles 37 mA g−1,
0.1C); temperature: 20 °C (Rothermel et al. 2016) . . . . . . . . . . 241
Fig. 13.21 Constant current and constant potential discharge capacities
and Coulombic efficiencies of the recycled cathode material
used as WE. Measurements were performed using a
three-electrode setup with metallic lithium as the CE and RE.
Recycled LiNi1/3Co1/3Mn1/3O2 (NCM) is used as the WE.
Electrolyte: SelectiLyte™ LP47 (1 M Lithium
hexafluorophosphate in EC:DEC 3:7); WE:
NCM/PVdF/Super C65/SFG-6L (87/5/4/4); discharging
cut-off potential: 3.0 V versus Li/Li+; charging cut-off
potential: 4.3 V versus Li/Li+; specific current: 150 mA g−1, 1
C (three initial formation cycles 30 mA g−1, 0.2C); constant
potential charging step: 1 h; temperature: 20 °C (Krüger et al.
2014; Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
efficiencies of a full cell. Measurements are performed using a
three-electrode setup with metallic lithium as the RE.
Recycled graphite originating from the CGR18650CH Li-ion
MH12210 cell (70% SOH/100% SOH) after subcritical
addition as co-solvent and subsequent thermal treatment is
used as the CE. Recycled LiNi1/3Co1/3Mn1/3O2 is used as the
WE. Electrolyte: SelectiLyte™ LP47 (1 M LiPF6 in EC:DEC
3:7); counter electrode: graphite/Super C65/Na-CMC
(90/5/5); WE: NCM/PVdF/Super C65/SFG-6L (87/5/4/4);
discharging cut-off voltage: 4.25 V; charging cut-off voltage:
1.7 V; specific current: 150 mA g−1, 1C (three initial
formation cycles 30 mA g−1, 0.2C); temperature: 20 °C
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Fig. 14.1 Schematic drawing of the LithoRec demonstration plant . . . . . . 248
Fig. 14.2 3P-model for occupational safety in the recycling of
lithium-ion batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
List of Figures xxix
Fig. 15.1 Conceptual model with system boundaries for economic

assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Fig. 15.2 Past and future development of the electric vehicle stock in
Europe (EU28 + Norway) for three market scenarios . . . . . . . . 257
Fig. 15.3 NPV and payback period for different market scenarios and
price scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Fig. 15.4 Composition of NPV for realistic market scenario and realistic
price scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Fig. 15.5 a Negative and b positive contributions to NPV of input and
output factors in the realistic market scenario and realistic
price scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Fig. 15.6 Break-even analysis for discharge with energy recuperation
(compared to discharge without energy recuperation) with
regard to electricity price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Fig. 15.7 Break-even analysis for mechanical processing with
extraction (compared to mechanical processing with thermal
drying) with regard to price difference for electrode coating
powder with and without fluoride impurities in different
market scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
Fig. 15.8 Minimum required gate fees to make the process
economically feasible (NPV 0) for the case that value of
cathode coating powder is zero . . . . . . . . . . . . . . . . . . . . . . . . . 264
Fig. 16.1 Framework for the analysis of environmental effects of
recycling. Based on the work by Geyer et al. (2015). Dotted
red arrows represent the avoided material flow . . . . . . . . . . . . . 269
Fig. 16.2 Cradle to gate embodied energy for the production of one
kilogram battery pack. LFP: LiFePO4, NCM:
LiNi0.4Co0.2Mn0.4O2, LCO: LiCoO2, LMR-NCM:
0.5Li2MnO30.5LiNi0.44Co0.25Mn0.31O2, HT:
hydrothermal, SS: solid state. Data taken from
Dunn et al. (2015a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Fig. 16.3 Cradle to gate embodied energy for the production of different
cathodes materials. Data taken from Dunn et al. (2015a) . . . . . . 273
Fig. 16.4 Changes in exergy values for materials in different life cycle
stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
Fig. 16.5 Change in separation effort within the material separation
processes for lithium-ion batteries . . . . . . . . . . . . . . . . . . . . . . . 276
Fig. 16.6 Process chain developed in LithoRec. Based on Diekmann
et al. (2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Fig. 16.7 Modeled energy and material flows from the LithoRec
process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Fig. 16.8 Power operation profiles for the unit processes considered in
LithoRec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
xxx List of Figures
Fig. 16.9 Energy portfolio classification for the mechanical treatment

processes in LithoRec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
Fig. 16.10 Results of the Life Cycle Impact Assessment performed
within the LithoRec project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
List of Tables
Table 4.1 Battery cells, modules, systems investigated at the time

of the research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 60
Table 4.2 Most important battery data to be known for
overdischarging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 61
Table 4.3 Concentration analysis of the different samples . . . . . . . . . . .. 70
Table 4.4 Assessment of the discharge devices with four
crucial criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 73
Table 5.1 List of disassembly elements . . . . . . . . . . . . . . . . . . . . . . . . .. 91
Table 5.2 Disassembly steps and necessary tools
(Wegener et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 93
Table 5.3 Criteria catalogue (Herrmann et al. 2012) . . . . . . . . . . . . . . .. 94
Table 6.1 Experimental accuracy analysis of direct positioning control
(Trentlage 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Table 6.2 Positional deviations during object recognition
(Küthe 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Table 8.1 Model electrolyte composition that is used in the
simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Table 8.2 Concluded data of drying experiments . . . . . . . . . . . . . . . . . . 150
Table 8.3 Composition of condensate and model composition of
electrolyte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Table 9.1 Parameters for the IC analysis . . . . . . . . . . . . . . . . . . . . . . . . . 158
Table 9.2 Parameter for the solvent screening . . . . . . . . . . . . . . . . . . . . . 160
Table 9.3 Parameter for the influence of mass ratio of solvent
to solid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Table 9.4 Parameter for the temperature influence experiments with
PHEV 1 batteries from Electrocycling GmbH . . . . . . . . . . . . . 162
Table 9.5 Parameter for the temperature influence experiments with
Panasonic CGR 18650 batteries. . . . . . . . . . . . . . . . . . . . . . . . 163
Table 9.6 Parameter for four stage extraction . . . . . . . . . . . . . . . . . . . . . 164
Table 9.7 Parameters for the determination of the temperature
influence for aqueous extraction . . . . . . . . . . . . . . . . . . . . . . . 166
xxxi
xxxii List of Tables
Table 9.8 Mass fraction in extract and overall mass of fluoride for
different solid to solvent ratios . . . . . . . . . . . . . . . . . . . . . . . . 167
Table 9.9 Parameter for five stages of extraction with DMC . . . . . . . . . . 168
Table 9.10 Parameter for three stages of extraction with water . . . . . . . . . 169
Table 9.11 Parameter for extraction of PHEV batteries . . . . . . . . . . . . . . . 170
Table 9.12 Comparison of values from mass-balanced calculations and
experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Table 9.13 Necessary number of stages and amount of solvent for 40 g
Panasonic CGR 18650. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Table 10.1 Recovery rates and compositions for the scCO2 extraction of
LP30 (DMC/EC, 1 M LiPF6) from different separators
(Grützke et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Table 10.2 Recovery rates and compositions for the supercritical CO2
extraction of commercial 18650 cells after formation and
aging (Grützke et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Table 11.1 Parameters of Tóth Eq. (11.4) for different adsorptives onto
SC40 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Table 11.2 Parameters of Tóth Eq. (11.4) for different adsorptives onto
Norit RB4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Table 13.1 Measured concentrations of metals and impurities in the
mother-solutions used as source for the re-synthesized
materials. The concentrations were measured by ICP-OES
and listed in g L−1 (Krüger et al. 2014) . . . . . . . . . . . . . . . . . 222
Table 13.2 BET surface area of the reference, rejects and cycled
materials. Measurements were performed under nitrogen
and the used P/P0 pressure range was 0–0.3
(Krüger et al. 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Table 13.3 Anion and cation impurities of the resynthesized active
materials from cycled cells and electrode rejects compared to
the reference material. All values are listed in percent . . . . . . . 224
Table 13.4 Summary of the elemental ICP-MS analyses. All data are
given in ppb with respect to the previously digested amounts
of the graphite samples. Limit of detection (LOD) and limit
of quantification (LOQ) are shown. The LODs were
calculated according to the three-sigma criterion. For the
calculation of LOQ the nine-sigma criterion was applied
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
Table 13.5 Integrated areas of D and G peaks from Raman spectra
shown in Fig. 13.17, their ratio and the calculated in-plane
crystallite size La of each investigated graphite sample
(Rothermel et al. 2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
List of Tables xxxiii
Table 13.6 Overview of the discharge capacities and associated

Coulombic efficiencies obtained from constant current
charge/discharge cycling experiments according to data
demonstrated in Figs. 13.18, 13.19 and 13.20
(Rothermel et al. 2016). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Table 15.1 Assumptions for estimating the amount of spent batteries in
three market scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Table 15.2 Annual volume of spent battery packs for
recycling [EV-eq] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Table 15.3 Process modules with capacities, initial investments and
fixed operating expenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Table 15.4 Initial prices of selected materials in EUR2020. Negative
prices indicate that the material must be disposed . . . . . . . . . . 260
Table 15.5 Assumed prices for electrode coating powder in
EUR2020/kg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
Table 16.1 Components and material shares of a generic battery system
(Diekmann et al. 2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Table 16.2 Recycled material, recycling process and avoided primary
material. Data taken from Buchert and Sutter (2017) . . . . . . . 283
Chapter 1
Background
Sergej Rothermel, Martin Winter and Sascha Nowak
Abstract This chapter outlines a wide range of reasons for the wide dissemination
of lithium-ion batteries (LIBs) in our everyday lives and establishes a relationship
between renewable energy, its storage and changes in European regulations
regarding guidelines for a more efficient recycling of LIBs. To promote a clear
understanding of current challenges in technological development of appropriate
recycling strategies, the basic components and basic operation principles of a LIB
are explained. Next, the most widespread types of commercial LIBs and processes
for their manufacturing are presented in a more detailed way. Hereafter, two
existing and practically used recycling methods are described and analyzed from
the perspective of strengths and weaknesses.
1.1 Introduction
In view of the steadily growing demand for energy, the term “peak oil” has attracted
the focus of controversial discussions since the end of the last century (Nashawi
et al. 2010). The major concerns of the world’s society are summarized in the report
“Peaking of World Oil Production: Impacts, Mitigation, & Risk Management.”
published by the United States Department of Energy in 2005. It states: “The
peaking of world oil production presents the U.S. and the world with an
unprecedented risk management problem. As peaking is approached, liquid fuel
S. Rothermel M. Winter S. Nowak (&)

MEET Battery Research Center, University of Münster, Münster, Germany
e-mail: sascha.nowak@uni-muenster.de
M. Winter
Helmholtz-Institute Münster (HI MS), IEK-12, Forschungszentrum Jülich,
Münster, Germany
© Springer International Publishing AG 2018 1

A. Kwade and J. Diekmann (eds.), Recycling of Lithium-Ion Batteries,
Sustainable Production, Life Cycle Engineering and Management,
https://doi.org/10.1007/978-3-319-70572-9_1
2 S. Rothermel et al.
prices and price volatility will increase dramatically, and, without timely mitiga-
tion, the economic, social, and political costs will be unprecedented” (Hirsch et al.
2005).
Until today the peak oil has not occurred and the political, social and economic
developments in the world lead to an unsteady oil price development. This, of
course, does not exclude the eventuation of peak oil, but today the largely neglected
aspect of climate change increasingly attracts the public’s attention. In this context,
the recently negotiated Paris Agreement from the 21st United Nations Climate
Change Conference contains a binding declaration of the participating parties to
reduce the global annual emissions of greenhouse gases by 2020. The target is to
hold “the increase in the global average temperature to well below 2 °C above
preindustrial levels” (UNFCCC 2015).
The implementation of the previously negotiated agreements inevitably requires
the reduction of emissions derived from fossil power generation and a change
towards renewable energy sources. Due to this reason, the share of the globally
installed capacity of renewable power sources such as wind and solar power has
noticeable grown during the past decade. Wind power has seen the largest growth in
the last years and increasingly emerges in newly industrialized countries as it offers
a cost-efficient generation of renewable energy. The second main source of
renewable power is solar power, which has significantly increased in recent years as
a consequence of rapidly falling prices for solar modules and their subsidization.
The total amount of installed wind and solar power capacity is quantified with 433
and 227 GW, respectively, by the end of the year 2015 and is estimated to show
further growth. The annual increase of these power sources over the past decade is
demonstrated in Fig. 1.1 (Renewables 2015 Global Status Report 2015).
Since the existing classical power grid systems have been developed to distribute
energy produced by large, less responsive thermal generation plants, there is a
constant increase in complexity in their control through the addition of renewables.
Fig. 1.1 Overview of the global installed capacity of the wind and solar power in the years 2005
—2015 (Renewables 2015 Global Status Report 2015)
1 Background 3
Often, established power lines cannot transport the excess energy over long dis-
tances, which is associated with the necessity to reduce the energy input from
renewables into the grid. Therefore, there is a need for the storage of excess energy.
Due to this reason, the electrical energy system of tomorrow will require the
extended use of stationary energy storage devices for alignment of intermittent
renewable energy generation with load demands. In Fig. 1.2, the mismatch between
generated and consumed power is demonstrated by a typical constellation. It can be
seen that the excess demand can be compensated by stored excess energy.
To provide the maximum benefit, the grid energy storage must meet certain
economic, technical and safety targets. In this respect, lithium-ion batteries (LIBs)
become an increasingly attractive option, since they offer a good efficiency, long
cycle life, high power and high energy densities, which are particularly important
for grid management. To realize its full potential in stationary energy storage,
lithium-ion technology has to become cheaper and more environmentally friendly.
In general, battery storage power stations (unlike e.g. pumped hydroelectric
energy storage) can be dimensioned and located in accordance with the demand
profile of the associated grid section. The rising need for electrical energy storage
and falling prices per kWh battery capacity will stimulate the widespread imple-
mentation of battery storage power stations.
The second aspect of the measures for reducing the global carbon dioxide
emissions affects the transport sector, which is known as the second largest con-
tributor to greenhouse emissions. Here, the alternative to internal combustion
engines doubtless requires the change to electrically driven vehicles (EVs). When
charged with renewable power, the emissions of greenhouse gases are avoided
completely. Furthermore, EVs can be actively involved into the grid management
by remote controlled shifting of the charging periods by the grid operator. It is
generally expected that the market for EVs will continue to strongly increase
worldwide in the long term (Fortschrittsbericht 2014—Bilanz der
Marktvorbereitung 2014). Since the main part of the powertrain is the battery
module, the global LIB manufacturing capacities and thus the amount of produced
battery cells will increase extensively.
The widespread dissemination of EVs is a major challenge regarding their uti-
lization after reaching the end-of-life. To minimize the environmental impact, the
large amounts of spent battery modules need to be recycled in an effective way. The
quality of the recovered materials has to fulfill minimum requirements for LIB
manufacturing in order to complete the circle.
In this context, the development of new environmentally friendly and less
expensive energy storage devices for the extended use in battery storage power
stations and the development of a new, more efficient recycling strategy for LIBs
are the two main subjects of the present book.
Fig. 1.2 Simplified representation of the power grid situation of tomorrow. The generated energy
is comprised of the steady growing share of renewables such as solar (a), wind power (b) and
conventional base load power generation (c). To compensate fluctuations of wind and solar energy,
the mismatch between power generation (d) and load (f) is managed by intermediately stored
battery power (e). Inspired by Hoelzle and Chang 2014
1 Background 5
1.2 Introduction into Lithium-Ion Batteries
In the most common design, a LIB consists of a negative electrode made of graphite
and a positive electrode based on layered transition metal oxide such as LiCoO2 and
LiNiO2 or mixtures thereof, e.g. LiNi1/3Co1/3Mn1/3O2. A permeable membrane
soaked with an electrolyte, which is a mixture of organic carbonates as solvents and
the conductive salt LiPF6, separates the electrodes. During the charging process the
lithium-ions are de-inserted from the positive electrode, transported by the elec-
trolyte to the negative graphite electrode and intercalated therein. During dis-
charging, this process is reversed. The operation principle of a typical LIB is
demonstrated by Fig. 1.3 (Gores et al. 2011).
It is important to note that for reasons of simplicity the negative and positive
electrodes are often referred to as anodes and cathodes in literature, which is only
applicable to a battery in the discharging mode. In compliance with the generally
common practice, this simplified representation is followed in this thesis.
In the following subchapters, the role, properties and features of individual
battery components are described in more detail.
1.2.1 Liquid Nonaqueous Electrolytes
In general, the formulation of the electrolytes used in LIBs involves a mixture of

several solvents, additives and a conductive salt. Appropriate solvents include
cyclic carbonates such as ethylene carbonate (EC) or propylene carbonate (PC) as
Fig. 1.3 Schematic illustration of the fundamental operating principle of a LIB in the discharged
(a) and in the charged state (b)
well as open-chain carbonates such as dimethyl carbonate (DMC), ethyl methyl

carbonate (EMC) or diethyl carbonate (DEC). The combination of these solvents
provides the possibility to adjust certain physical characteristics such as liquid
range, dielectric permittivity or dipole moment.
Lithium hexafluorophosphate (LiPF6) is predominantly used as the conductive
salt. It moderately fulfills most of the essential requirements placed on the elec-
trolyte formulation, which can be in general summarized as followed (Gores et al.
2011):
• intrinsic thermal stability
• chemical inertness and electrochemical stability
• sufficient solubility in appropriate solvents
• high transference number of Li+ ions in electrolyte solutions
• low cost and molecular weight
• non-toxicity
Depending on the electrolyte, not all of the properties mentioned above are
fulfilled to the same extent. Albeit its excellent ionic conductivity, the limited
thermal and chemical stability as well as the sensitivity to moisture are considered
as major drawbacks of LiPF6 (Gores et al. 2011; Xu 2004).
In order to compensate the respective weaknesses of the electrolyte solvents and
the conductive salt, the addition of additives was found to be an effective and
economical alternative to modify certain functions of the electrolyte. Introduced in
small amounts, the additives are especially effective in terms of thermal electrolyte
stability, interfacial properties or overcharge protection of the cell (Xu 2004).
1.2.2 Negative Electrodes in Rechargeable Lithium

Batteries
Despite the widespread use of graphite as an anode material, its maximum theo-
retical discharge capacity of 372 mAh g−1 is significantly lower than that of
metallic lithium with 3860 mAh g−1. Since the energy density is the product of
discharge capacity and voltage, the lithium metal anode is considered as an
attractive option to increase the energy density of LIBs. Unfortunately, these anodes
could not be successfully commercialized in rechargeable cells since metallic
lithium tends to react with the most organic solvents forming a surface film, which
is also known as the solid electrolyte interphase (SEI). Although the presence of the
SEI is favorable to prevent continuous electrolyte decomposition by interruption of
the contact with metallic lithium, its inhomogeneous composition and thickness
leads to the formation of areas with preferred lithium deposition. During continued
charge/discharge cycling this effect leads to a significant enlargement of the elec-
trode surface resulting in a steadily growing SEI. Depending on the electrolyte
composition, this effect has two main negative consequences: the continued
1 Background 7
Fig. 1.4 Schematic illustration of a metal-based rechargeable lithium battery showing the typical
morphology of the lithium surface before charge/discharge cycling (a) and after an undefined
number of cycles (b). The continuously growing dendrites potentially penetrate the separator
membrane and cause a short circuit
electrolyte composition reduces the Coulombic efficiency and the deposited lithium
is mossy, grain or dendritic in shape. Especially the dendritic form of the high
surface area lithium (HSAL) is of particular concern, since the dendrites are
potentially able to create internal short circuits by penetration the separator. Such a
case is schematically illustrated in Fig. 1.4. Since metallic lithium is an indis-
pensable component of next generation battery technologies such as solid state,
lithium sulfur or lithium air, the scientific research addressing the above-mentioned
drawbacks is still ongoing (Xu et al. 2014; Yamaki and Tobishima 2011).
The issue of HSAL formation can be avoided if an intermetallic compound is
used as alternative. Such a compound formed electrochemically between lithium
and silicon for instance enables a theoretical discharge capacity of 3578 mAh g−1 if
combined in a Li15Si4 phase. The discharge potential of 0.4 V versus Li/Li+ as well
as the gravimetric discharge capacity are close to those of metallic lithium.
Furthermore, being the second most abundant element on earth, silicon is neither
expensive nor of environmental impact. Nevertheless, the electrochemical cycling
performance of silicon is associated with high volume expansions between pure
silicon and its lithiated phases which exceeds 300%. This effect results in insta-
bilities of the electrode/electrolyte interphases during cycling. A further issue is the
loss of electrical contact between particles and their following exclusion from
electrochemical reactions. Problems of comparable complexity also occur when
other alloys are in use, e.g. Al–Li or Sn-Li (Monconduit et al. 2015).
Taking into account the developments in the field of negative electrodes, it
becomes clear that to date lithiated carbon is considered to offer the best
compromise between energy density and cycling stability. Due to its layered nature,
graphite is able to incorporate Li + ions into a solid host according to the following
general reaction scheme:
discharge þ
LiX Cn
xLi þ xe þ Cn
charge
The associated volume expansion remains below 10%, thus providing a good
cycling stability. During the electrochemical reduction (or charge) of the carbon
host, Li+ ions intercalate between the graphene layers to form a graphite interca-
lation compound LixCn. Thereby, the formation of intercalate layers is subjected to
a certain regularity, which is called stage formation.
The maximum stoichiometry of lithiated graphite is reported to be LiC6, which
corresponds to a theoretical discharge capacity of 372 mAh g−1.
Since graphite is able to incorporate a variety of guest species, unfavorable
reactions may take place involving intercalated Li+ ions enclosed by a solvation
shell. This shell is formed by the electrolyte solvent and thus prevents the associ-
ation of ions to ion pairs, which strongly improves the ionic conductivity of the
electrolyte. Once intercalated, the solvated Li+ ion forms a ternary lithiated graphite
Lix(solv)yCn. The intercalation of the large solvated species is associated with a
high expansion of the graphite matrix potentially leading to irreversible structural
disintegration also known as exfoliation. Being intercalated, the solvation shell may
undergo several decomposition processes as the potential decreases below the
reductive stability limit of the electrolyte. The resulting gaseous decomposition
products may additionally support the exfoliation of graphite as schematically
illustrated in Fig. 1.5 (Winter and Besenhard 2011).
To prevent this unfavorable process, the composition of the electrolyte can be
optimized by addition of special additives. Their function is to form a protective
layer at the edge sites of graphite, which is selectively permeable for Li+ ions. This
protective layer is also known as the solid electrolyte interphase (SEI) and ensures
highly reversible lithiation/de-lithiation of graphite without exfoliation as shown in
Fig. 1.5. Because of its crucial significance for LIBs, the role, composition and
formation of the SEI are discussed in the following chapter (Winter and Besenhard
2011).
1.2.3 Positive Electrodes
Oxides with the general formula LiMO2 (where M = Co, Ni, or Mn) crystallize in a
layered face-centered rock-salt structure, in which the Li+ and M3+ ions occupy the
octahedral interstitial sites of the cubic oxygen array. This structure allows a
reversible two-dimensional insertion/extraction of Li+ ions between the covalently
bonded MO2 layers. Due to the good electronic conductivity deriving from the
1 Background 9
Fig. 1.5 Schematic illustration of the graphite exfoliation process initiated by the co-intercalation
of solvated Li+ ion between the graphene layers (above). Reductive decomposition of intercalated
solvent shell leads to gas formation and delamination of individual graphene layers (exfoliation).
Protective function of the passivation layer (SEI), which prevents the entry of the solvation shell
and thus minimizes the graphite exfoliation processes (below). Adapted from Vetter et al. (2005)
direct M-M interaction with a partially filled t2g band, the LiMO2 oxides have
become attractive cathode materials for LIBs (Manthiram and Muraliganth 2011).
The transition metal oxide LiCoO2 is one of the most used cathode in com-
mercial LIBs due to its high operating potential exceeding 4 V versus Li/Li+,
simple synthesis and good cycle life. The theoretically available discharge capacity
of this material is 274 mAh g−1 if the total amount of lithium would be extracted
during charge. However, due to structural and chemical instabilities only 50–60%
of lithium can be practically used, thus resulting in a discharge capacity of 140–
160 mAh g−1 (Manthiram and Muraliganth 2011). The limitation of LiCoO2 as a
cathode material is arising from a significant overlap of the Co3+/4+ t2g band with
the O2− 2p bands, which may result in oxygen evolution in the case of a deep
de-lithiation during its overcharging. In this case, strong oxidation of the electrolyte
may take place thus increasing the safety hazard (Zhang and Zhang 2015).
Despite the widespread use of LiCoO2 the safety concerns as well as the high
toxicity and cost of Co require an alternative chemistry for large-scale applications
such as electric vehicles or stationary energy storage applications (Manthiram and
Muraliganth 2011).
The exchange of Co by Ni results in an isostructural LiNiO2, although difficult to
synthesize and predicted to be less stable in the charged state than LiCoO2. The
solution to this problem was demonstrated by a partial substitution of Ni with Mn in
the mixed oxide LiNi0.5Mn0.5O2. This compound provides a higher discharge
capacity of around 200 mAh g−1 and significantly improved chemical stability.
Following the approach of mixed oxides the mixture of LiNi1/3Co1/3Mn1/3O2 (NCM

111) was found to be the best compromise between discharge capacity, structural
stability and stable cycling performance, and therefore increasingly gains the interest
in the industry. Beyond that, scientific efforts are being made to develop new cathode
materials based on Nickel rich stoichiometries such as LiNi0.5Co0.3Mn0.2O2 (NCM
532) or LiNi0.6Co0.2Mn0.2O2 (NCM 622) which offer the opportunity to higher
de-lithiation rates without structural drawbacks (Manthiram and Muraliganth 2011).
Another class of cathode materials are lithiated phospho-olivine structures. One
representative and extensivly investigated material is LiFePO4 (LFP). Because of its
environmental friendliness and the high abundance of the raw materials LFP attracts
an immense interest as a potential cathode material. But the relatively poor elec-
tronic conductivity complicates its application in LIBs. The introduction of carbon
coating as well as the minimization of the particle size is therefore necessary to
increase the electrochemical performance. Treated in such a way, a discharge
capacity of 160 mAh g−1 is reported. However, the redox potential of around 3.5 V
versus Li/Li+ decreases the corresponding energy density and limits the use of LFP
-based LIBs to stationary energy storage applications. Since considerable efforts in
nanosized synthesis and carbon coatings are necessary, the advantage of low cost
raw materials becomes relative (Manthiram and Muraliganth 2011).
With regard to the growing necessity of LIBs in new application areas, there is a
high interest in development of cathode materials providing a high energy and
power density, improved safety and simultaneously lower synthesis effort. In order
to achieve these objectives, more stable structures allowing a higher discharge
capacity at higher working potentials are in focus of research (Manthiram and
Muraliganth 2011).
1.3 Industrial Production of Lithium-Ion Cells

and Modules
1.3.1 Cell Design
To gain a better understanding of the challenges facing the recycling industry and
their strategies, it is a prerequisite to deal first with the diversity, structure and
manufacturing of batteries, the necessary raw materials and their requirements.
While the fundamental principles of cell design and operation are well known,
specific details, equipment and assembly techniques in commercial cell fabrication
are often kept undisclosed by the industry (Tagawa and Brodd 2009).
There are basically three common types of lithium-ion cells: cylindrical, pris-
matic and pouch. The organization for worldwide standardization in the electrical
and electronic engineering industry—International Electrotechnic Commission
(IEC)—established a common nomenclature for technical battery standards such as
form type and size. For cylindrical cells, the diameter is defined by the first two
1 Background 11
digits in millimeters and the next two digits indicate the cell height in tenths of a
millimeter. For instance, the indication R18650 is related to a cylindrical (or round)
cell which is 18 mm in diameter and 65 mm in height. The indication P366509
characterizes a prismatic cell, which is 36 mm wide, 65 mm long and 9 mm thick.
In addition, manufacturers may include other identifying information in combina-
tion with the nomenclature described above (Tagawa and Brodd 2009; Mikolajczak
et al. 2011).
The most widely produced cell is the 18650 cylindrical cell with nearly 660
million cells manufactured annually (Warner 2015; Anderman 2014). The reason
behind the high demand can be attributed to the fact that this small type of LIBs is
suitable for a variety of applications ranging from portable electronics to electric
vehicles. The inner structure of this cell type is illustrated in Fig. 1.6.
The electrode stack, wound to a jelly roll is placed in a cylindrical housing,
whereby the negative electrode tap is welded on the case bottom, so that the whole
case acts as the negative terminal. The positive electrode tab is connected to the cell
cap, which is electronically isolated from the housing by a plastic seal. The advantage
of the cylindrical cell is that the cell housing is made of steel and offers a
high-strength packaging, which protects the jellyroll from external mechanical
influences. Further, the cylindrical cell design incorporates additional safety devices
such as a current interrupt device (CID), which is a non-resettable fuse integrated into
the cell housing between the positive electrode tab of the jelly roll and the positive
electrode terminal. The CID disconnects the jellyroll from the terminal to stop the
current flow when the internal cell pressure overcomes a predetermined point.
Fig. 1.6 Cutaway view of a cylindrical 18650 cell

Usually, the occurrence of increased cell pressure is the result of overheating or

overcharge, whereby a significant amount of gas is generated in the cell interior prior
to entering a thermal runaway condition (Tagawa and Brodd 2009; Warner 2015).
Another type of safety feature is the positive temperature coefficient
(PTC) resistor, which is essentially a resettable thermal fuse placed internally
between the positive electrode terminal and the CID device. If the temperature of
the cell rises above a predetermined point, the polymer- containing conductive
carbon particles inside the PTC undergo an expansion and the conductive paths
break down, which increases its resistance. Consequently, the current flow is
minimized preventing the cell from operating at inadmissible temperatures (Warner
2015; Tagawa and Brodd 2009).
The production of a wound prismatic cell is similar to the production of cylin-
drical cells with the exception that a flat mandrel is used instead of a cylindrical one
during winding and the housing is different in shape and size. Typically, the neg-
ative current collector tab at the end of the wind is connected to the metallic case,
commonly used as the negative terminal. The positive current collector tab is used
to connect the positive terminal at the top of the cell (Xia 2011). Besides the wound
design, the electrodes may be present as an alternately stack of a number of indi-
vidual anodes and cathodes separated by a single z-folded separator, where the
anode and cathode electrode tabs are welded together respectively. A stacked cell
design is more suitable for high current applications compared to the wound type
since the current flow is distributed to a number of electrode tabs. Whereas
cylindrical cells are equipped with PTCs (except those designed for high power
applications because of the limiting impedance caused by the PTC), prismatic cells
usually need external PTCs. They still have a safety vent embedded in the housing
in order to prevent explosions if operating at inadmissible conditions with gas
evolution within the cell interior (Barsukov and Qian 2013). In contrast to the
cylindrical 18650 cell, the prismatic cell design enables thinner cell geometries and
its rectangular form can be easily dimensioned for each application (Guoxian and
Dean 2011).
While the cylindrical and prismatic cell designs use a hard casing, the pouch cell
case is made of a heat sealable aluminum laminated multilayer foil. This type of
enclosure enables light and slender designs, which cannot be realized in a hard case
format, resulting in improved gravimetric density compared to cylindrical and
prismatic versions (Mikolajczak et al. 2011; Guoxian and Dean 2011). Similar to
the prismatic cell, stacked but also wound electrode arrangements can be used in a
pouch cell. Since the pouch casing provides limited mechanical protection of the
electrodes and does not incorporate any safety devices, the cell needs to be
embedded in a respective device or external housing providing sufficient physical
stability, temperature and current control. The application focus of the pouch cell
design is on portable devices that require ultra-thin enclosures offering optimal
space usage in mobile phones or tablet computers (Yu et al. 2013). A wound
prismatic cell and a stacked pouch cell design is schematically illustrated in
Figs. 1.7 and 1.8.
1 Background 13
Fig. 1.7 Cutaway view of a wound prismatic cell
1.3.2 Electrode Fabrication
Regardless of the final design, the cell production begins with the preparation of the
electrodes. In this process, the electrode materials are mixed, coated onto a metal
current collector foil and then calendered. The electrode paste for the positive
electrode consists of active material such as LiCoO2 or LiNi1/3Mn1/3Co1/3O2, a
carbon based conductive agent and a binder like polyvinylidene difluoride (PVdF).
During the preparation, the active material and conductive carbon are thoroughly
mixed in e.g. a high speed thin-film mixer under dry solid-state conditions to ensure
a homogeneous distribution of the particles and disintegration of agglomerates. The
proportion of the active material is 90 wt%. After reaching the desired homo-
geneous distribution, the mixed solids are fed into the next planetary mixer along
with PVdF-containing N–methyl pyrrolidone (NMP) solution, and are then stirred.
Fig. 1.8 Cutaway view of a stacked pouch cell
In a planetary mixer the impeller shaft rotates about its axis while revolving around
the axis of the mixing vessel, pulling material from the bottom to the top. The
mixture of cathode material, conductive carbon and binder is first kneaded under
dry conditions to give a stiff paste. The desired viscosity is then adjusted by adding
the respective solvent. Upon reaching the desired homogeneity, the electrode paste
is coated onto a positive current collector. As the significantly smaller conductive
carbon particles fulfil the function of electronic connection between the single
particles, their uniform distribution is essential for battery performance (Friscic
2015; Tagawa and Brodd 2009).
The processing of the negative electrode is comparable, except that different
materials are applied. Graphite is used as active material and PVdF or car-
boxymethyl cellulose (CMC) serve as binder. Depending on the type of the binder
NMP or water is used as a solvent for PVdF or CMC respectively (Tagawa and
Brodd 2009).
Depending on the active material, the electrode slurry requires different current
collectors. For the positive electrode, Al foil with a thickness of 16 µm is nor-
mally used. The negative electrode is coated onto a Cu foil with 10 µm thickness
(Changhe et al. 2015).
Since there is an increasing technological trend in LIB manufacturing towards
automotive applications, the tolerance level for process related fluctuations
regarding the cell quality is constantly reduced. The batchwise slurry production
described above is no longer considered to be economically reasonable, since this
method requires an as longer processing time as more precisely the viscosity and
dispersion degree need to be adjusted. Moreover, the scale–up of the batch mixer is
difficult to do without any loss in quality. For this reason, the leading manufacturers
replace the conventional mixing methods with continuous dispersion mixing
devices (Ohata 2014).
1 Background 15
After mixing, the slurry is placed in sealed containers, where it is stirred and
circulated during the whole coating process to avoid formation of inhomogeneities
caused by precipitation or segregation. From this container, the slurry is fed to the
coating equipment. Common methods for coating operations are slot die, doctor
blade or back coating. In principle, all of these methods can be used for both
positive and negative electrode processing handling various viscosities and coating
speeds. Back coating is commonly used for discontinuous processing, whereby the
coating is interrupted by lifting the back rolls in regular intervals while the foil
continues to move. These coated segments either directly correspond to the elec-
trode dimensions of the final cell or additionally undergo a subsequent slitting to the
required dimensions. The coating thickness is electromechanically controlled across
and down the web, allowing a permissible deviation of less than 2 µm. Usually, a
second coating is applied on the opposite side of the current collector to increase
mass ratio of the active to inactive material. After coating, the foil passes through an
oven to evaporate the solvent (Changhe et al. 2015; Tagawa and Brodd 2009). The
dried electrode is compressed during the calendering process to form a continuous
thin foil and to increase the density of the electrode mass. The load and speed of the
calender are manufacturer-specific and vary with each product. After calendering,
the master roll is slit to the width specified by the cell dimensions and cut to the
specified length. Next, connection tabs need to be affixed to the electrodes to
provide electronic contact to the respective terminals in the subsequent step. An
aluminum tab is ultrasonically welded to the bare part of the positive electrode.
Likewise, nickel or nickel plated tabs are affixed to the negative electrode in the
same way (Tagawa and Brodd 2009). The electrode fabrication process is
schematically illustrated in Fig. 1.9.
1.3.3 Cylindrical Cell Fabrication
In the next step, the winding machine combines the electrode stack consisting of the
two electrodes and a separator to a jelly roll-like cell core by winding around a
mandrel as demonstrated in Fig. 1.10a, b. After winding, the cell core is taped to
keep it tightly wound and to prevent unwanted unwinding of the electrodes or their
displacement. The wound cell core is checked for presence of short-circuits by a
measurement of its impedance before being inserted into the cell housing (c). Next,
a small welding electrode is inserted through the hollow mandrel to weld together
the anode tab and the bottom of the cell case (d, e). After this procedure, a circular
insulator is placed on top of the jelly roll, whereby the positive electrode tab is
guided through the insulator (f). Before the cell is sealed by crimping, an annular
groove needs to be made at the top of the cell can to fix the jelly roll and to create a
socket for the sealing gasket and the positive terminal. This task is performed by a
grooving machine (g). Afterwards, the cap, which incorporates the sealing gasket,
the positive terminal, the CID and PTC safety devices is placed into the socket,
followed by ultrasonic welding of the positive electrode tab to the positive terminal
Fig. 1.9 Electrode fabrication process
(h). In the next step, the cell is filled with a predetermined amount of the electrolyte
using a precise vacuum injection device. During this step, the cell housing is set
under vacuum to ensure that the added electrolyte is evenly distributed in the
separator and electrode structures (i). Next, the cell is sealed (j, k). As any moisture
in the cell induces a negative effect on its operation, the cell assembly is carried out
under dry conditions. The finished cell (l) is washed with alcohol or acetone to
remove any adhering electrolyte from the cell surface and the sealing quality is
controlled by an electrolyte leakage test using an electronic sensing device that
detects the presence of carbonate vapors. The internal construction of the finished
1 Background 17
Fig. 1.10 Schematic illustration of the cylindrical cell fabrication process
cell is inspected by X-rays for proper top cap sealing, misalignment of the elec-
trodes and welding quality. Additionally, the voltage and impedance of each cell are
measured in order to detect and sort out any short-circuited cells (Tagawa and
Brodd 2009; Brodd and Tagawa 2002).
1.3.4 Prismatic and Pouch Cell Fabrication
The main difference between these two designs is that either a hard case made of
aluminum or plastic is used for prismatic cells or an aluminum laminated multilayer
foil for pouch cells. The electrode arrangement can be accomplished either in a
wound or stacked way.
In general, the assembly of wound prismatic cells is similar to that of cylindrical
cells until the step of enclosure integration (Fig. 1.11d–g). The process then
deviates due to characteristics of the cell size and geometry. Since no electrode tabs
are attached to the electrodes, the electrode terminals are directly connected to
them. For this reason, the electrode stack is asymmetrically arranged during
winding and uncoated areas extend beyond the edge of the stack (Fig. 1.11a, b).
After winding, the cell core is taped (c) and connected to the electrode terminals by
ultrasonic welding (d, e). If the cell case is made of aluminum, the interior is lined
with an insulating material (f) before the cell core is placed in the cell case (g). The
cap is then laser-welded to the cell case (h), followed by electrolyte filling using a
vacuum injection device through the vent provided in the cap (i). After filling, the
vent is sealed (j) and the finished cell (k) is subjected to final inspection procedures
as described before (Tagawa and Brodd 2009).
The pouch cell designs are essentially laminated constructions. Their manufac-
turing differs from one case to another as due to the great variety of forms and sizes
no general process can be established. Similar to prismatic cells, the electrode
arrangement can also be accomplished in a wound or stacked way. To give insights
into both processes, the manufacturing of the stacked pouch cell is shown in
Fig. 1.12. The corresponding electrodes are stacked alternately and separated by an
endless z-folded separator membrane (Fig. 1.12a). Once the stacking sequence is
complete, the stack is taped and the electrode tabs arranged on top of each other
are ultrasonically welded together to form the corresponding electrode terminals
(b, c, d). Afterwards, the electrode stack is inserted in a case, which is formed from
an aluminum laminated multilayer foil (e, f). The foil is sealed by means of heat in a
way that one edge remains unsealed (g). The assembly is then checked for internal
short circuits and the residual humidity is removed by heating in a vacuum dryer.
The remaining unsealed side is used for electrolyte filling in the next step (h). After
injection, the cell is vacuum-sealed in a two-chamber configuration aiming at a
better electrolyte penetration into the electrode stack (i). Since the pouch case
cannot provide mechanical stability, gaseous products released during the initial
charge may lead to an inflation of the cell followed by an ionic contact loss between
the electrodes. For this reason, the second chamber is used for gas collection during
the formation cycles (j). After this step, the cell is resealed and the gas filled
chamber is discarded (k, l). The finished cell (m) is subjected to leakage tests, X–ray
inspection, cleaning, labeling and storage (Zhang and Ramadass 2013).
1 Background 19
Fig. 1.11 Schematic illustration of the wound prismatic cell fabrication process
1.3.5 Cell Formation
Once the cell assembly is complete, within the specification cells are subjected to a
formation process in which they are charged for the first time. The conditions of the
first charge are important for the initial SEI formation on the anode and better
Fig. 1.12 Schematic illustration of the stacked pouch cell fabrication process
1 Background 21
electrolyte distribution between the electrodes. The initial charge is performed at

low currents followed by charging at slowly increasing currents according to the
specific characteristics of the cell for one or two more cycles after formation. After
this process, the cells are stored for a period between two weeks and one month.
The voltage of the cells is measured before and after this so-called aging process.
Deviations in voltage decrease indicate the presence of micro-short circuits. These
short-circuited cells are sorted out (Brodd and Tagawa 2002).
1.3.6 Battery System Manufacturing
On a module level, the single cells are connected in series and packaged to form
modules. A module includes a thermal management system, which is required to
regulate the module temperature and keep it within the desired operating range. The
use of air as a heat transfer medium is less complicated in the technical design, but
also less effective. Therefore, a thermal battery management based on a liquid
coolant such as ethylene glycol is often utilized in the automotive industry
(Dhameja 2002). Each module is equipped with an electronic module management
unit (MMU), which controls various functions such as temperatures, cell voltages
and currents and can therefore predict and actively prevent failures (Vezzini 2014).
On a battery pack level, the battery modules are connected in series and in
parallel to form a battery system with a nominal voltage typically ranging around
350 V. Nominal voltages up to 800 V are proposed by the industry in order to
increase the charging power at simultaneously reduced cable cross-sections and
thus weight and volume (Jung 2016). The values monitored by the MMUs are
transferred to the battery management unit (BMU). Based on these values, the
BMU develops strategies to protect the cells from their operation outside the
threshold conditions. Furthermore, it balances the state-of-charge of each single cell
to maintain equal voltage distribution over all cells preventing over-charge or -
discharge (Dhameja 2002; Vezzini 2014). Historical data collection can be included
into the functional range of the BMU, which can be used for condition assessment
of the battery system in case of its maintenance or utilization.
1.4 Recycling of Lithium-Ion Batteries
1.4.1 Introduction
The widespread use of LIBs in a variety of industrial and private applications leads
to the need for recycling and reutilization of their constituent components. On the
one hand, this necessity is economically driven by the price of Ni and Co. Both
metals are used in cathode materials. The respective prices are $10 per kg Ni and
Fig. 1.13 Major innovations in material technology. Past, existing and expected battery
chemistries (Pavoni 2012; Andre et al. 2015)
$28 per kg Co as traded in October 2016 (Current calendar year Monthly Average
Settlement Prices 2016). In theory, the value of these elements can be increased if
they are recovered as ready-to-use battery raw compounds without breaking down
to the pure elementary form for new synthesis. But the extremely high purity
requirements of the manufacturers make this approach difficult to realize.
Furthermore, Co is considered as a metal with rising scarcity resulting in an
increasing price. Although there is an economic value for its recycling, Co can be
substituted by Ni and Mn in the future (5 V spinel, Fig. 1.13), which are less scarce
in comparison. In contrast, Li has no foreseeable substitute and due to the con-
stantly growing need, its demand may exceed the global production in the early
2020s if no Li recycling is established until this point (Sonoc and Jeswiet 2014;
Reuter et al. 2013). Since the EU has only one Li occurrence in Portugal a strong
dependency on imported Li and its price may be the consequence (Perez et al. 2012;
Gruber et al. 2011). On the other hand, battery recycling is encouraged by legis-
lation, mainly because of environmental and health risks from hazardous battery
materials such as transition metals or electrolyte components.
With regard to these aspects, the European Parliament and the Council of the
European Union instituted the most advanced battery recycling legislation world-
wide as declared in the Battery Directive 2006/66/EC, which determines clear
collection and recycling targets. In accordance to the guidelines, the EU member
states should achieve a minimum collection rate of 45% and a minimum recycling
efficiency of 50 wt% by 26th of September 2016 with regard to batteries, which are
not lead-acid or nickel-cadmium (Directive 2006/66/EC of the European Parliament
and of the Council on Batteries and Accumulators and Waste Batteries and
Accumulators L 266/1 2006). Other relevant EU legislations affect the Waste of
1 Background 23
Electrical and Electronic Equipment (WEEE) 2012/19/EU and End of Life Vehicles
(ELV) 2000/53/EC directives, which target battery recycling of rechargeable bat-
teries in electronic products and electric vehicles. In compliance with these regu-
lations, the EU member states are obliged to collect at least 45% of the total weight
of electrical and electronic equipment waste starting in the year 2016. For
end-of-life vehicles, the reuse and recovery rates have to be increased to at least
85% with regard to the weight per vehicle and year (Directive 2012/19/EU of the
European Parliament and of the Council on on Waste Electrical and Electronic
Equipment (WEEE) L 197/38 2012; Directive 2006/66/EC of the European
Parliament and of the Council on Batteries and Accumulators and Waste Batteries
and Accumulators L 266/1 2006; Directive 2000/53/EC of the European Parliament
and of the Council on on End-of Life Vehicles L 269/34 2000; Reuter et al. 2013).
In the US and Canada, battery recycling is legislated by individual states and
provinces. As no efficient and environmentally-friendly recycling process is
available so far, the local recycling targets remain at low or even voluntary level.
However, in view of the rapidly growing LIB market, it is expected that these and
further countries will begin to establish or extend regulations on battery recycling
(Reuter et al. 2013).
1.4.2 Overview of Selected Lithium-Ion Battery Recycling

Technologies
The generic term “Lithium-Ion” to be found on labels of LIBs refers to a broad

range of different chemistries which have been adjusted for the specific applica-
tions. While in the beginning of the 2000s, when the LIB became increasingly
applied, most batteries used lithium cobalt oxide (LCO) compounds as the cathode
material. But the proportion of mixed transition metal oxides like nickel cobalt
manganese oxide (NCM) has been constantly increasing over the years. Anode
materials have been optimized as well. Besides established graphite electrodes,
silicon (Si) is expected to enter this segment in the future. The consequence is that
modern technology and scientific progress introduce ever new cell designs and new
chemistries and thus increase the complexity of LIB recycling. An overview of past,
current and expected battery chemistries in the market is given in Fig. 1.13.
In the following chapters the most important commercial LIB recycling pro-
cesses developed by Umicore AG & Co. KG and Toxco Inc. are presented.
Umicore process: Pyro-hydrometallurgical recycling process
The Umicore AG & Co. KG (hereafter referred to as Umicore) is one of the world
market leaders in recycling and refining of complex waste streams containing
precious and other non-ferrous metals. In September 2011, Umicore launched a
battery recycling plant in Hoboken (Belgium). Besides simultaneous treatment of
Lithium-Ion and NiMH batteries the process can handle various other types of
batteries, such as alkaline primary batteries. The fundamental technology is a
single-furnace pyrometallurgical process with subsequent hydrometallurgical

refinement of Ni and Co (Cheret and Santen 2005; Rothermel et al. 2016).
To begin the recycling process with a pyrometallurgical step is beneficial
because it eliminates the need for discharging each cell before its entering the
recycling process. After disassembling the EV battery modules at the Umicore
facility in Hanau (Germany), the battery cells are mixed together with small
amounts of coke, reducing agents, such as metallic Al and Zn, and slag formers.
The proportion of the cells is at most 50 wt%. Afterwards, this mixture is intro-
duced into a vertical shaft furnace entering the so-called “preheating zone” in which
the temperature of the mixture does not exceed 300 °C. A slow temperature
increase allows the electrolyte to be slowly evaporated, and thus reduces the risk of
explosions to a minimum (Cheret and Santen 2005; Tytgat 2011; Rothermel et al.
2016).
In the second zone, the “plastics pyrolising zone”, the temperature rises up to
700 °C, whereby the plastic components and binders of the cells are melted,
decomposed and partly evaporated. The released energy from partially incineration
of the before produced pyrolysis gases contributes to the heating process.
Afterwards, hot gases rise up into the “preheating zone”, cool down and are fed
together with the electrolyte vapors to a post combustion chamber in which they are
completely burned under high oxygen atmosphere. To prevent condensation, the
gas temperature at the top of the shaft must be kept high enough. This is realized by
the energy supply from a plasma torch. In front of the inlet to the post combustion
chamber, the temperature of the gases is increased above 1150 °C via a further
plasma torch. The effect of such a torch is an important increase in gas enthalpy
with a limited increase in gas volume. At this stage it is possible to capture halo-
genated products by injecting Ca- or Na-based products via the torch. After post
combustion, the gases are immediately cooled down below 300 °C in order to avoid
formation of dioxins and furans. Classical flue gas purification is installed for
further treatment (Cheret and Santen 2005; Rothermel et al. 2016).
In the third zone, “the smelting and reduction zone”, a regulated flow of hot and
optionally oxygen enriched air is injected via tuyères at the bottom of the shaft
furnace. The remaining battery waste is then transformed into two fractions: a slag
layer consisting of Al, Si, Ca and Fe and an alloy mainly containing Cu, Co and Ni
and residual Fe. These fractions are continuously or regularly tapped as a granulate
material at a temperature between 1200 and 1450 °C. Depending on the present
amount of aluminum in the charge, the oxidation energy of Al to Al2O3 might be
sufficient to melt all other products without additional heating making the process
self-sustaining. As Fe is not desirable in the subsequent refining process, its slightly
higher affinity to oxidation than Ni and Co is exploited by adjusting the redox
potential of the molten mass. The amount of the added reducing agent in the feed is
regulated. Additionally, SiO2-rich slag is helpful to capture high amounts of FeO
and thus to facilitate further oxidation of Fe. At this stage, the slag is leaving the
recycling process without any further refining as gravel substitute in the concrete
industry. In the process, Li also ends up in the slag as Li2O and gets lost (Cheret and
Santen 2005; Rothermel et al. 2016).
1 Background 25
The alloy fraction contains mainly valuable metals, such as Co, Ni, Cu and
residual Fe. For the hydrometallurgical refinement, this fraction is transported to a
refinery plant. The first stage of the refining process is the alloy dissolution in
sulfuric acid resulting in the oxidative formation of sulfate salts. Next, Cu is
removed from the solution by precipitation of CuS and Cu2S using SO2 at elevated
temperatures (Naboychenko et al. 2009; Tytgat 2011).
After the copper removal from the sulfate solution, residual iron can be pre-
cipitated by adjusting the pH of the solution. Afterwards, the iron-free solution is
pumped to the Ni/Co solvent extraction stage (Crundwell et al. 2011; Fischmann
et al. 2013).
Currently, the use of phosphonic acids is the most attractive option for solvent
extraction of cobalt from sulfate solutions. On an industrial scale the extractant bis
(2,4,4-trimethylpentyl)phosphonic acid diluted in an aliphatic solvent is used. Such
solutions are commercially distributed as CYANEX 272® or LIX 272® with a
typically extractant concentration of 10–25 vol% (Crundwell et al. 2011). The
separation principle is demonstrated in Fig. 1.14. “A” represents the extractant
which is a dimeric hydrogen-bonded ligand species (c). If the pH is between 3 and
5, the higher affinity of the extractant for Co leads to the formation of the tetra-
hedrally coordinated cobalt complex (a) which is more stable in the organic phase
than the octahedrally coordinated nickel complex (Crundwell et al. 2011; Kathryn
2008). While the cobalt complex is hydrophobic, the nickel complex can undergo
inner-sphere electron transfer reactions forming intermediates containing one or two
water molecules making the complex more hydrophilic (b) (Kathryn 2008).
In order to release cobalt from the organic phase, the pH of the aqueous phase
needs to be decreased to a value below 3 leading to a complex disintegration and
the following transfer of Co into the aqueous phase (Inc. 2008).
In the recycling process, the Co- and Ni- containing solution is pumped into a
mixer-settler, in which the organic extractant is added to the aqueous solution. The
phases are mixed together forming an emulsion followed by a quiescent settling
stage. At this stage the dispersed phase is separated into a Co-containing organic
and Ni-containing aqueous phases (Ingham et al. 2007). A second mixer-settler is
connected downstream in order to deplete the Co-rich organic phase by mixing with
a strong concentrated HCl solution. Co is transferred back to the aqueous phase as
CoCl2 and the Co free organic phase is used again in the extraction stage. The
CoCl2 solution is then injected into a calcination furnace, where it reacts with
oxygen to Co2O3 at high temperature. In the next step, Co2O3 and Li2CO3 are
introduced into a rotary kiln, in which they react to the final product LiCoO2. This
compound can be used for processing of new cathode materials. The Ni-containing
aqueous phase separated in the first mixer-settler and subsequently fed to a mixing
reactor. Here, the pH is increased by adding NaOH, whereby solid Ni(OH)2 is
precipitated and further reprocessed to new battery materials (Tytgat 2011).
In 2011, the German Federal Ministry for the Environment, Nature
Conservation, Building and Nuclear Safety funded a project called “LiBRi” led by a
consortium of Umicore AG & Co KG., Daimler AG, Öko–Institut (Institute for
Applied Ecology) and Clausthal University of Technology. The goal was to develop
Fig. 1.14 Tetrahedral coordination of Co(II) (a) and octahedral coordination of Ni(II) (b) by
dialkylphosphonic acid (c). In presence of water the octahedrally coordinated Ni(II) complex
undergoes a reversible ligand exchange increasing its hydrophilic properties. The disintegration of
the complex is initiated by reducing the pH. Exemplarily demonstrated on the Co(II) complex (c).
Adapted from Kathryn (2008)
an additional recycling route for lithium recycling from the slag and flue dust as it
accrues as a byproduct and waste in the Umicore process.
The developed hydrometallurgical process starts with rod milling of the slag to a
desired grain size of 150 µm. The crushed material is fed to a magnetic separator
which separates entrapped alloy particles from the slag. Afterwards, the material is
leached in 1–1.5 M sulfuric acid at 60 °C, whereby insoluble solids such as silicates
and gypsum are precipitated from the solution. Subsequently, the purified solution is
diluted in order to avoid precipitation of Li2CO3 in the following steps. Further, the
pH of the solution is increased to 7 by adding CaO which initiate precipitation of
compounds such as Al, Fe, Si, residual transition metals and gypsum. By addition
of further amount of CaO, the pH is raised to 12 in order to precipitate Mg and
residual metals. After solid-liquid separation, Ca is precipitated as CaCO3 by
addition of NaCO3. Afterwards, the solution is concentrated by vaporization and Li
1 Background 27
is precipitated as LiCO3 by further addition of NaCO3 at 100 °C. In the last step,
LiCO3 is washed with ethanol resulting in a purification degree of 98%. For reuse in
LIBs, the carbonate is refined by known techniques (Elwert et al. 2012). With regard
to the efforts for lithium extraction from the slag, Umicore will assumingly imple-
ment this procedure when it becomes commercially profitable, e.g. in case of
increasing lithium prices.
Toxco process: Mechanical-hydrometallurgical recycling process
Retriev Technologies Inc. (formerly TOXCO Inc.) in Trail (Canada) (hereafter
referred to as Toxco) operates a unique mechanical-hydrometallurgical recycling
process for LIBs. Before entering the process, the reactive potential of LIBs is
reduced by cryogenic cooling with liquid nitrogen making the following shredding
process safe as short circuits are avoided due to the frozen and thus non-conductive
electrolyte. After cooling down to about –160 °C, the LIBs are shredded into small
pieces and fed to a magnetic separator which removes the steel fraction. After
washing, this fraction is conveyed to further utilization. The remaining crushed
material is submerged into a high pH process solution consisting of lithium brine.
The brine is re-circulated from a subsequent process step. The liquid mixture of
shredded material and lithium brine are further wet-milled by a hummer mill in
order to separate the current collectors from the active materials, which are washed
out from the solution. After milling and washing, the shredded particles are
dewatered by a screw press, crushed again and fed to a shaker table, which sepa-
rates metals and plastics. The metal fraction is sent to further separation and
reutilization. The cold brine solution containing active materials is transferred into a
tank, in which it is heated and the active materials are neutralized. Afterwards, the
solution is filtered by a filter press, whereby the active materials are separated as a
mixture in form of a filter cake and cannot be recycled in this process (however, this
material could be fed into the Umicore process). The filtered lithium brine solution
is transferred to a downstream tank wherefrom it is partially re-circulated to the
previous crushing process or transferred to a following mixing tank in which
lithium is precipitated as Li2CO3 by addition of Na2CO3. The insoluble product is
dewatered in a further filter press (McLaughlin and Adams 1999; Thompson 2011).
For further refining of Li2CO3 a subsequent dissolution in sulfuric acid is carried
out. The filter cake is placed in a hybrid electrolytic cell, which is separated into an
acidic and a basic compartment by a Li+ conductive membrane. The acidic com-
partment is filled with mild sulfuric acid, whereas the basic compartment is filled
with LiOH. If Li2CO3 is dissolved in sulfuric acid, Li+ ions can pass through the
membrane while the anions are repelled. On the basic side, LiOH is formed, which
can be used for pH adjustment of the brine process solution in the previous steps.
The concentrated LiOH solution is either dewatered or converted to Li2CO3 by
addition of CO2 to the solution (McLaughlin and Adams 1999; Thompson 2011).
Summary and perspectives in battery recycling
As demonstrated above, there is a variety of industrial processes for the recycling of
LIBs following different approaches. Originally developed for recycling of primary
lithium batteries, the Toxco—process is considered to be inappropriate for recycling
of spent LIBs, since apart from lithium carbonate no further materials can be
recovered. Furthermore, the effort of deep freezing of LIBs prior to safe shredding
is neither economically nor ecologically applicable for large quantities.
The most advanced recycling process in commercial operation is the
pyro-hydrometallurgical process developed by Umicore. Based on the melting
furnace as the initial stage, the process offers a unique robustness towards changes
in cell chemistry of LIBs. This property is ideally suited to meet the market
requirements that are currently requiring the utilization of many different types of
small sized LIBs. The focus is on the recovery of the transition metals with suffi-
cient purity in order to reuse them in cathode synthesis. The process-related loss of
lithium and combustible components unfortunately decreases the recovery rate.
Overall, battery recycling has to follow the developments in the LIB market. The
increase in produced quantities as well as the change to large-sized battery modules
is expected to rise. At the same time, the legislative requirements on recycling
efficiency become stricter. With regard to these issues, more scientific and industrial
activities are expected in the field of LIB recycling in the future.
In the next chapter, another recycling process which is still in development—
LithoRec II is presented. It focuses on mechanical-hydrometallurgical treatment of
spent LIBs deriving from automotive applications. Here, the variation of the cell
chemistry is easy to manage due to well documented specification of each battery
module and their relatively limited diversification. Thus, the modules can be easier
sorted by the corresponding cell chemistry and utilized in the most appropriate way.
The process is designed to operate with discharged and partially automatically
disassembled modules, thus avoiding the need of deep freezing or smelting of the
individual cells. Almost every consistent component can be recycled by thorough
mechanical separation of the shredded materials, thereby reducing the subsequent
hydrometallurgical effort for extraction and refining of the precious metals.
References
Anderman M (2014) The Tesla Battery Report—Tesla Motors: Battery Technology, Analysis of
the Gigafactory, and the Automakers’ Perspectives. Total Battery Consulting Inc, Advanced
Automotive Batteries
Andre D, Kim S-J, Lamp P, Lux SF, Maglia F, Paschos O, Stiaszny B (2015) Future generations
of cathode materials: an automotive industry perspective. J Mater Chem A 3(13):6709–6732.
https://doi.org/10.1039/c5ta00361j
Barsukov Y, Qian J (2013) Battery safety and protections. In: Barsukov Y, Qian J (eds) Battery
power management for portable devices. Artech House
Brodd RJ, Tagawa K (2002) Lithium-Ion cell production processes. In: Schalkwijk WA,
Scrosati B (eds) Advances in lithium-ion batteries. Springer US, Boston, MA, pp 267–288. doi:
https://doi.org/10.1007/0-306-47508-1_10
Changhe C, Xiangyun Y, Shiyong Z, Yuping W, van Teunis R (2015) Assembly processes for
lithium-ion batteries. In: Yuping W (ed) Lithium-ion batteries: fundamentals and applications.
Electrochemical energy storage and conversion. CRC Press, pp 479–506. doi:https://doi.org/
10.1201/b18427-15
1 Background 29
Cheret D, Santen S (2005) Battery recycling. United States Patent

Crundwell FK, Moats MS, Ramachandran V, Robinson TG, Davenport WG (2011) Separation of
the Platinum-Group Metals from Base Metal Sulfides, and the Refining of Nickel, Copper and
Cobalt. In: Davenport WG, Crundwell FK, Moats MS, Ramachandran V, Robinson TG
(eds) Extractive Metallurgy of Nickel, Cobalt and Platinum Group Metals. Elsevier, Oxford,
pp 457-488. http://dx.doi.org/10.1016/B978-0-08-096809-4.10036-X
Current calendar year Monthly Average Settlement Prices (2016) The London metal exchange
limited. http://www.lme.com/*/media/Files/Market%20data/Historic%20Data/2016/Average
%20Prices/October%202016.xlsx. Accessed 22 Nov 2016
Dhameja S (2002) Electric vehicle battery performance. In: Electric vehicle battery systems.
Newnes, Woburn, pp 133–159. doi:http://dx.doi.org/10.1016/B978-075069916-7/50007-8
Directive 2000/53/EC of the European Parliament and of the Council on on End-of Life Vehicles L
269/34 (2000) vol EU Official Journal. Official Journal of the European Communities, Brussels
Directive 2006/66/EC of the European Parliament and of the Council on Batteries and
Accumulators and Waste Batteries and Accumulators L 266/1 (2006) vol EU Official
Journal. European Parliament and the Council, Brussels
Directive 2012/19/EU of the European Parliament and of the Council on on Waste Electrical and
Electronic Equipment (WEEE) L 197/38 (2012). vol EU Official Journal. European Parliament
and the Council, Brussels
Elwert T, Goldmann D, Schirmer T, Strauß K (2012) Recycling von Li-Ionen-Traktionsbatterien –
Das Projekt LiBRi. In: Thomé-Kozmiensky KJ, Goldmann D (eds) Recycling und Rohstoffe,
Band 5, vol Band 5. p 1004
Fischmann A, Wiggett S, Bednarski T, Cocalia V, Bourget C (2013) Hydrometallurgical nickel
laterite processing: a review of current SX flowsheets and industry trends. In: Battle T,
Moats M, Cocalia V et al. (eds) Ni–Co 2013. John Wiley & Sons, Inc., pp 391–403. doi:https://
doi.org/10.1002/9781118658826.ch32
Fortschrittsbericht 2014—Bilanz der Marktvorbereitung (2014). vol Nationale Plattform
Elektromobilität. Gemeinsame Geschäftsstelle Elektromobilität der Bundesregierung (GGEMO),
Berlin
Friscic T (2015) Ball-milling Mechanochemical synthesis of coordination bonds: discrete units,
polymers and porous materials. In: Stolle A, Ranu B (eds) Ball milling towards green
synthesis: applications, projects, challenges. The Royal Society of Chemistry, pp 151–189. doi:
https://doi.org/10.1039/9781782621980-00151
Gores HJ, Barthel J, Zugmann S, Moosbauer D, Amereller M, Hartl R, Maurer A (2011) Liquid
nonaqueous electrolytes. In: Daniel C, Besenhard JO (eds) Handbook of battery materials.
Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, pp 525–626. doi:https://doi.
org/10.1002/9783527637188.ch17
Gruber PW, Medina PA, Keoleian GA, Kesler SE, Everson MP, Wallington TJ (2011) Global
lithium availability. J Ind Ecol 15(5):760–775. https://doi.org/10.1111/j.1530-9290.2011.
00359.x
Guoxian L, Dean DM (2011) State-of-the-art production technology of cathode and anode
materials for lithium-ion batteries. In: Lithium-Ion batteries. Green chemistry and chemical
engineering. CRC Press, pp 327–394. doi:https://doi.org/10.1201/b11292-9
Hirsch RL, Bezdek R, Wendling R (2005) Peaking of world oil production: impacts, mitigation, &
risk management (trans: Energy USDo). Science Applications International Corporation
Hoelzle M, Chang P (2014) Advanced Battery Materials Technology for Cost Efficient Storage
Batteries. SEMI® Association
Inc. CI (2008) CYANEX® 272 Extractant brochure. Cytec Industries Inc
Ingham J, Dunn IJ, Heinzle E, Přenosil JE (2007) Modelling of stagewise processes. In: Ingham J,
Dunn IJ, Heinzle E, Přenosil JE (eds) Chemical engineering dynamics. Wiley-VCH Verlag
GmbH, pp 117–209. doi:https://doi.org/10.1002/9783527614103.ch3
Jung C (2016) The Future of EVs and fast charging at 800 V. In: Advanced Automotive Battery
Conference, Mainz, 28 January 2016. Advanced Automotive Batteries
Kathryn CS (2008) Solvent extraction in the hydrometallurgical processing and purification of

metals. In: Aguilar M, Cortina JL (eds) Solvent extraction and liquid membranes. CRC Press,
pp 141–200. doi:https://doi.org/10.1201/9781420014112.ch5
Manthiram A, Muraliganth T (2011) Lithium intercalation cathode materials for lithium-ion
batteries. In: Daniel C, Besenhard JO (eds) Handbook of battery materials. Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, Germany, pp 341–375. doi:https://doi.org/10.1002/
9783527637188.ch12
McLaughlin W, Adams TS (1999) Li reclamation process. United States Patent
Mikolajczak C, Kahn M, White K, Long RT (2011) Introduction to lithium-ion cells and batteries.
In: Mikolajczak C, Kahn M, White K, Long RT (eds) Lithium-ion batteries hazard and use
assessment. Springer US, Boston, MA, pp 1–24. doi:https://doi.org/10.1007/978-1-4614-3486-
3_1
Monconduit L, Croguennec L, Dedryvère R (2015) Negative electrodes. In: Monconduit L,
Croguennec L, Dedryvère R (eds) Electrodes for li-ion batteries. John Wiley & Sons, Inc.,
London, pp 1–28. doi:https://doi.org/10.1002/9781119007364.ch1
Naboychenko SS, Murashova IB, Neikov OD (2009) Production of nickel and nickel-alloy
powders. In: Neikov OD, Naboychenko SS, Murashova IV, Gopienko VG, Frishberg D
(eds) Handbook of non-ferrous metal powders. Elsevier, Oxford, pp 369–398. doi:http://dx.doi.
org/10.1016/B978-1-85617-422-0.00017-3
Nashawi IS, Malallah A, Al-Bisharah M (2010) Forecasting world crude oil production using
multicyclic hubbert model. Energy Fuels 24(3):1788–1800. https://doi.org/10.1021/ef901240p
Ohata T (2014) Innovations in lithium ion battery manufacturing—A continuous manufacturing
process applicable to nano-scale materials at a signficant cost reduction. PRIMIX Corporation,
Osaka
Pavoni F (2012) The Li-Ion Battery Value Chain. Roland Berger Strategy Consultants GmbH,
Istanbul
Perez AA, Safirova E, Anderson ST, Gurmendi AC, Newman HR, Soto-Viruet Y, Wallace GJ,
Wilburn DR (2012) The Mineral Industries of Europe and Central Eurasia. Geological Survey,
U.S
Renewables 2015 Global Status Report (2015). REN21 Paris
Reuter MA, Hudson C, van Schaik A, Heiskanen K, Meskers C, Hagelüken C (2013) UNEP
(2013) Metal recycling: opportunities, limits, infrastructure, a report of the working group on
the global metal flows to the international resource Panel
Rothermel S, Evertz M, Kasnatscheew J, Qi X, Grützke M, Winter M, Nowak S (2016) Graphite
recycling from Spent Lithium Ion Batteries. ChemSusChem [accepted]. doi:https://doi.org/10.
1002/cssc.201601062
Sonoc A, Jeswiet J (2014) A review of lithium supply and demand and a preliminary investigation
of a room temperature method to recycle lithium ion batteries to recover lithium and other
materials. Procedia CIRP 15:289–293. https://doi.org/10.1016/j.procir.2014.06.006
Tagawa K, Brodd RJ (2009) Production processes for fabrication of lithium-ion batteries. In:
Yoshio M, Brodd RJ, Kozawa A (eds) Lithium-ion batteries: science and technologies.
Springer, New York, pp 181–194. doi:https://doi.org/10.1007/978-0-387-34445-4_8
Thompson S Kinsbursky/Toxco overview. In: 4th US-China Electric Vehicle And Battery
Technology Workshop Argonne National Laboratory, 2011
Tytgat J (2011) Recycling of Li-ion and NiMH batteries from electric vehicles: technology and
impact on life cycle. Umicore Battery Recycling
UNFCCC (2015) Adoption of the Paris agreement. Proposal by the President. Paper presented at
the Paris Climate Change Conference, November 2015, COP 21, Paris, 12 December 2015
Vetter J, Novák P, Wagner MR, Veit C, Möller KC, Besenhard JO, Winter M, Wohlfahrt-Mehrens
M, Vogler C, Hammouche A (2005) Ageing Mechanisms in Lithium-Ion Batteries. J Power
Sources 147(1–2):269–281. https://doi.org/10.1016/j.jpowsour.2005.01.006
Vezzini A (2014) Lithium-ion battery management A2. In: Pistoia G (ed) lithium-ion batteries.
Elsevier, Amsterdam, pp 345–360. doi:http://dx.doi.org/10.1016/B978-0-444-59513-3.00015-7
1 Background 31
Warner JT (2015) Lithium-Ion and Other Cell Chemistries. In: Warner JT (ed) The Handbook of
lithium-ion battery pack design: chemistry. Elsevier Science, Components, Types and
Terminology
Winter M, Besenhard JO (2011) Lithiated carbons. In: Daniel C, Besenhard JO (eds) Handbook of
battery materials. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, pp 433–478.
doi:https://doi.org/10.1002/9783527637188.ch15
Xia D (2011) Lithium Ion rechargeable batteries. In: Liu R-S, Zhang L, Sun X, Liu H, Zhang J
(eds) Electrochemical technologies for energy storage and conversion. Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, Germany, pp 69–109. doi:https://doi.org/10.1002/978352763
9496.ch3
Xu K (2004) Nonaqueous liquid electrolytes for lithium-based rechargeable batteries. Chem Rev
104(10):4303–4417
Xu W, Wang J, Ding F, Chen X, Nasybulin E, Zhang Y, Zhang J-G (2014) Lithium metal anodes
for rechargeable batteries. Energy Environ Sci 7(2):513–537. https://doi.org/10.1039/
c3ee40795k
Yamaki J-I, Tobishima S-I (2011) Rechargeable lithium anodes. In: Daniel C, Besenhard JO
(eds) Handbook of battery materials. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim,
Germany, pp 377–404. doi:https://doi.org/10.1002/9783527637188.ch13
Yu A, Chabot V, Zhang J (2013) Electrochemical supercapacitor design, fabrication, and
operation. In: Yu A, Chabot V, Zhang J (eds) Electrochemical supercapacitors for energy
storage and delivery: fundamentals and applications. CRC Press, pp 203–246. doi:https://doi.
org/10.1201/b14671-6
Zhang Z, Zhang SS (2015) Challenges of key materials for rechargeable batteries. In: Zhang Z,
Zhang SS (eds) Rechargeable batteries: materials, technologies and new trends. Springer
International Publishing, Cham, pp 1–24. doi:https://doi.org/10.1007/978-3-319-15458-9_1
Zhang ZJ, Ramadass P (2013) Lithium-ion battery systems and technology. In: Brodd JR
(ed) Batteries for sustainability: selected entries from the encyclopedia of sustainability science
and technology. Springer, New York, pp 319–357. doi:https://doi.org/10.1007/978-1-4614-
5791-6_10
Chapter 2
The LithoRec Process
Jan Diekmann, Sergej Rothermel, Sascha Nowak and Arno Kwade
Abstract The LithoRec projects were funded by the German Federal Ministry of
the Environment, Nature Conservation Building and Nuclear Safety and VDI/VDE
Innovation+Technik GmbH. The projects aimed to develop a new recycling process
for lithium-ion batteries from electric and hybrid electric vehicles with a focus on
energy efficiency and a high material recycling rate. The developed process route
combines mechanical, mild thermal and hydrometallurgical treatment to regain
nearly all materials of a battery system.
2.1 Objectives and Results of LithoRec
The first LithoRec project lasted two years, starting in September 2009, and was
carried out by the scientific projects partners Technische Universität Braunschweig
and Westfälische Wilhelms-Universität Münster, as well as the industrial partners
Audi AG, Chemetall GmbH, Electrocyling GmbH, Evonik Litarion GmbH,
Fränkisches Recycling Zentrum, H.C. Starck GmbH, I+ME Actia GmbH, Recylex
GmbH, Süd-Chemie, and Volkswagen AG.
The consortium focused on the evaluation of different process routes and steps
for an efficient recycling of complete lithium-ion battery systems of electric
(EV) and hybrid electric vehicles (HEV). Nowadays, recycling processes mainly
focus on the pyro metallurgical recovery of nickel and cobalt as the most valuable
materials. For that reason, the special focus of LithoRec was the recovery of battery
grade lithium with the least possible effort, aiming to realize a new energy-efficient
J. Diekmann (&) A. Kwade

Institute of Particle Technology, Battery LabFactory Braunschweig,
Technische Universität Braunschweig, Braunschweig, Germany
e-mail: j.diekmann@tu-braunschweig.de
S. Rothermel S. Nowak

https://doi.org/10.1007/978-3-319-70572-9_2
34 J. Diekmann et al.
recycling process, abstain from high temperatures and trace mechanical and
hydrometallurgical process steps (Kwade and Bärwaldt 2012).
One of the main results of this project was a concept for the mechanical sepa-
ration of the components of a battery system in different fractions including the
coating materials. Therefore, manual and automated processes for the disassembly
of battery systems (Herrmann et al.) as well as different classifying and sorting
processes for the material separation were investigated in laboratory scale. The
recovered coating materials were treated hydrometallurgically afterwards.
The project partner Chemetall GmbH realized a pilot plant for this part of the
process in Langelsheim; combining leaching, filtration and different precipitation
steps. This kind of hydrometallurgical treatment regains the valuable materials
nickel, cobalt, manganese, and lithium hydroxide or lithium carbonate. It was found
that small impurities of aluminum have no negative impact on the electrochemical
performance of battery test cells with lithium nickel cobalt manganese oxide
(NCM) as a cathode material (Krüger et al. 2014).
Furthermore, the project partners investigated new ways for the recovery of
electrolyte solvents and the conducting salt. Processes like vacuum condensation
and extraction via supercritical carbon dioxide showed positive results.
Ecologic and economic assessments were carried out for the investigated con-
cept. While the ecological impact was high, the process is only economically
feasible at high throughputs and as such, very sensitive to the market for electric
driving systems (Hoyer et al. 2014).
2.2 Objectives and Project Progression of LithoRec II
The follow-on project LithoRec II started July 2012 and focused on the detailing of
the scientific and technical results, and the realization of the optimized process steps
in a pilot plant with a capacity of 100,000 kg battery cells per year. Besides the
same scientific partners, the industrial partners Actia I+ME GmbH, Audi AG,
Electrocycling GmbH, H.C. Starck GmbH, Hosokawa Alpine Aktiengesellschaft,
Rockwood Lithium GmbH (former Chemetall GmbH), Solvay Fluor GmbH, and
the associated partners Kuka GmbH, Lion Engineering GmbH, and Opel AG
participated in this project.
LithoRec II also foregrounded the recycling of battery systems from EV and
HEV. This focusing influenced the technical realization and, to an even greater
degree, the economic assessment. Investments in corresponding recycling equip-
ment could be profitable in most scenarios but there are also high uncertainties
regarding the main influencing factors: the future amount of spent battery systems
and the achievable prices for the recycling products. Because of these reasons, the
consortium decided not to realize a pilot plant in the projected scale, but to con-
stitute a smaller temporary demonstration plant in Braunschweig. The realized
process chain was composed of the process steps discharging, disassembly,
crushing, drying, separation, and sieving. During its operating time of four months,
2 The LithoRec Process 35
the process recycled 1.4 tons of battery systems and reached a material recycling
rate of 75–80% per battery system. The determined process data was used to update
the economic and ecologic assessments.
The investigated process steps and the overall developed LithoRec process are
described in the following.
2.3 The LithoRec Process Chain
The LithoRec process combines electrical, mechanical, mild thermal, and

hydrometallurgical treatment with the aim of recovering nearly all valuable mate-
rials of battery systems. A process flow chart of the process route for the preparation
of the different recyclates is shown in Fig. 2.1. Detailed descriptions of the process
steps and results of other thematic priorities of the LithoRec projects can be found
in the following chapters.
Descriptions of the overall process can also be found in other literature
(Diekmann et al. 2016a, b; Hanisch et al. 2015).
The first step is the discharging and short circuiting of the battery system to
lower the electrical hazard, regain the remaining energy, and enable safe disas-
sembly and crushing. The short circuit avoids relaxation and inherent voltage rise of
the battery system. Due to the many design deviations, an automation of this
process is not suitable at this state. The removed battery system peripheries such as
the housing, cables, and the battery management unit (BMU), controlling charge-
and discharge currents, voltage, and temperatures, is fed to established recycling
routes.
Due to the flammable electrolyte solvents, the dismantled battery modules are
crushed under inert atmosphere afterwards.
The following step aims to remove the remaining components of the electrolyte.
Nowadays, electrolytes are mixtures of linear carbonates, cyclic carbonates, and a
conducting salt. The removal of these components leads to an improved separation
process due to a lower adhesion between the crushing fragments and a minimized
hazard potential of intermediates and products. The project partners investigated
different technologies for the electrolyte removal during LithoRec II.
A possible process to not only regain the solvents but also the conducting salt, is
a solid-liquid extraction via dimethyl carbonate (Fig. 2.1, part I). The shredded
material is fed into a rotary kiln in which the conducting salt dilutes. The shredded
material is dried afterwards and dimethyl carbonate is recycled. For the recovery of
the electrolyte components, the solution is concentrated via evaporation.
Removal of the electrolyte solvents and a high amount of the conducting salt is
also possible via thermal drying (Fig. 2.1, part II). At temperatures between 80 and
140 °C and decreased pressure, the organic solvents of the electrolyte are vapor-
ized. The vaporized solvents can be recovered by condensation or combusted via a
thermal post-combustion process to regain the thermal energy. Additionally, the
conducting salt lithium hexafluorophosphate (LiPF6) is decomposed while
Fig. 2.1 Process flow chart of the investigated process steps and the overall LithoRec process
2 The LithoRec Process 37
generating hydrogen fluoride (HF) as a gaseous product. HF leads to the necessity

of a gas scrubber application.
The use of supercritical carbon dioxide (CO2) also strives to recover solvents and
conducting salt (Fig. 2.1, part III). At increased pressure of 120 bars, the electrolyte
components are extracted via CO2. After decompression to ambient pressure, the
electrolyte components, especially the solvents, are fluidized and can be recovered.
The shredded fragments remain completely dried.
The following mechanical separation is a combination of air-sifting, crushing,
and sieving processes and aims to separate the different materials. As electrodes are
made for long lasting adhesion, the separation of current collector foils and their
coating is the main difficulty. A possible combination is depicted in Fig. 2.1. The
dried battery fragments are firstly fed to a combination of a magnetic and a simple
air separator to regain iron and heavy parts such as modules and cell housings.
Afterwards, a homogenization of the fragments is carried out via a cutting mill. This
process step also increases the mechanical stress on the fragments, leading to a
higher yield of the coating materials; the so called black mass. Air classification via
zig-zag-sifting separates the fragments into two fractions: current collector foils and
separator/coating materials. The current collector foils, aluminum and copper,
can be separated afterwards via optical sorting. To recover the valuable coating
materials from anode and cathode a sieving process via a vibration sieve is carried
out.
These coating materials, the black mass, contain a transition metal oxide, for
example lithium nickel cobalt manganese oxide (NCM), and graphite. This powder
is processed hydrometallurgically to recover the metals. At first, the recovered
fraction is leached and solid graphite is removed via filtration. Metals like nickel
(Ni), cobalt (Co), or manganese (Mn) are precipitated via pH manipulation and a
precursor is prepared. The remaining lithium (Li) solution is cleaned by several
process steps. Lithium hydroxide or lithium carbonate is produced through crys-
tallization. The combination of lithium hydroxide or lithium carbonate and the
precursor can be used for new battery active material (Krüger et al. 2014). Two
aspects of the hydrometallurgical recovery of metals from the black mass of spent
lithium-ion batteries have to be considered. Firstly, the black mass has to be
unmixed, which means lithium iron phosphate (LFP) cannot be processed together
with NCM due to different precipitation limits. And secondly, the contamination
has to be taken into account because especially high impurities from the carbonates,
Cu, and Al will have a negative effect on the purity of the resynthesized battery
active material (Krüger et al. 2014).
Following a process route with a thermal drying process without recovery of the
electrolyte components, the LithoRec process reaches a material recycling rate of
75–80%. Graphite, separator, and the electrolyte components are not recycled yet.
References
Diekmann J, Hanisch C, Froböse L, Schälicke G, Loellhoeffel T, Fölster A-S, Kwade A (2016a)

Ecological recycling of lithium-ion batteries from electric vehicles with focus on mechanical
processes. J Electrochem Soc 164(1):A6184–A6191. https://doi.org/10.1149/2.0271701jes
Diekmann J, Hanisch C, Loellhoeffel T, Schälicke G, Kwade A (2016b) Ecologically friendly
recycling of lithium-ion batteries-the lithorec - process. ECS Trans The Electrochem Soc 73:9
Hanisch C, Diekmann J, Stieger A, Haselrieder W, Kwade A (2015) Recycling of lithium-ion
batteries. In: Yan J (ed) Handbook of clean energy systems. Wiley, UK
Herrmann C, Raatz A, Mennenga M, Schmitt J, Andrew S, Assessment of automation potentials
for the disassembly of automotive lithium ion battery systems. In: 19th CIRP international
conference on life cycle engineering. Berkeley
Hoyer C, Kieckhäfer K, Spengler TS (2014) Technology and capacity planning for the recycling of
lithium-ion electric vehicle batteries in Germany. J Bus Econ 85(5):505–544. https://doi.org/
10.1007/s11573-014-0744-2
Krüger S, Hanisch C, Kwade A, Winter M, Nowak S (2014) Effect of impurities caused by a
recycling process on the electrochemical performance of Li[Ni0.33Co0.33Mn0.33]O2.
J Electroanal Chem 726:91–96. https://doi.org/10.1016/j.jelechem.2014.05.017
Kwade A, Bärwaldt G (eds) (2012) Abschlussbericht “Recycling von Lithium-Ionen-Batterien”
Chapter 3
Potential Dangers During the Handling
of Lithium-Ion Batteries
Jan Diekmann, Martin Grützke, Thomas Loellhoeffel,

Matthias Petermann, Sergej Rothermel, Martin Winter,
Sascha Nowak and Arno Kwade
Abstract Due to their high voltage, high stored energy, and reactive components,
lithium-ion batteries present a specific and significant hazard potential. This espe-
cially comes into play during recycling because nearly every safety precaution of a
battery system and battery cell needs to be bypassed. Because the project partners of
LithoRec II spared a thermal pre-treatment step to deactivate the batteries, the
hazard potential and its handling played a major role. This chapter gives an over-
view of the hazards associated with lithium-ion batteries and describes their role in
every process step.
3.1 Hazard Potential of Lithium-Ion Batteries
The hazard potential of lithium-ion batteries has been the focus of many research
projects in recent years. For the design of safe process steps and the development of
safety concepts for the operation of a recycling plant, specific investigations were
necessary. Therefore, the project partners focused on specific safety issues. The
results of these investigations has led to a general understanding of these issues and
are presented in the following.

T. Loellhoeffel
Institute of Particle Technology, TU Braunschweig, Braunschweig, Germany
M. Grützke S. Rothermel M. Winter S. Nowak
M. Petermann
Volkswagen AG, Wolfsburg, Germany
M. Winter
Helmholtz-Institute Münster (HI MS), Forschungszentrum Jülich,
IEK-12, Münster, Germany

https://doi.org/10.1007/978-3-319-70572-9_3
The hazards of lithium-ion batteries can be roughly divided into three areas:
electrical hazard, fire and explosion hazard, and chemical hazard. Because these
areas are interacting, they cannot be observed separately (Diekmann et al. 2016).
3.1.1 Electrical Hazard
The electrical hazard is determined by the stored energy and high voltage. Battery
systems of electric vehicles contain around 16–90 kWh of electric energy and
operate between 300 and 400 V, and also 700 V in special cases. These facts come
into play when the battery system is handled outside the car, for example during
maintenance or recycling. The attached voltage requires a special qualification of
the executing workers.
One of the main hazards for humans is electric shock. It is important to take into
account that batteries emit direct current. An electric shock by direct current can
lead to paralysis of the musculature and electrolysis of the blood. Such an elec-
trolysis leads to decomposition and therefore gas release, and can also occur hours
or days after the accident. Nearly every electric shock provokes burn injuries at
entry and exit spots, depending on current path, exposure time, and voltage.
Short circuits of lithium-ion batteries lead to elevated temperatures due to joule
heating. This can be provoked externally by connection of the cells contacts or by
penetration of conducting solids, or internally due to cell failures. Elevated tem-
peratures from the inside or the outside of lithium-ion battery cells lead to further
response.
3.1.2 Fire- and Explosion Hazard
The fire and explosion hazard is determined by the applied solvents of the elec-
trolyte. Nowadays, electrolytes are a mixture of linear carbonates like dimethyl
carbonate (DMC), diethyl carbonate (DEC), or ethyl methyl carbonate (EMC), and
cyclic carbonates like ethylene carbonate (EC), or propylene carbonate (PC), as
well as the conducting salt lithium hexafluorophosphate (LiPF6) to establish ionic
conductivity. The carbonates are highly flammable and form an explosive atmo-
sphere with aerial oxygen.
Furthermore, the fire and explosion hazard arises from flammable, gaseous
reaction products due to chemical reactions. These reaction products could be
hydrogen, methane, ethane, carbon monoxide, and propene.
Generation of these and other gases also provokes a pressure rise of surrounded
and gastight aggregates. In the worst case scenario, this can lead to an explosion.
Another factor that has to be taken into account is the danger of dust explosion
induced by the fine particular coating materials after drying (compare particle size
distributions Chap. 7). This fraction has particle sizes smaller 200 µm, contains
3 Potential Dangers During the Handling of Lithium-Ion Batteries 41
carbon, and was examined on explosiveness by a certified testing institute. Testing

of three probes led to the result that the fraction is not explosive. This statement is
restricted to the tested battery chemistries and the applied process.
3.1.3 Chemical Hazard
The chemical hazard of lithium-ion batteries arises from their ingredients, but also
from the reactions of these ingredients during battery cell lifetime and under abuse
conditions.
In regards to toxicity, the cathode coating materials play a major role. These are
mainly lithium metal transition oxides containing nickel and/or cobalt oxide, for
example, lithium cobalt oxide (LCO), lithium nickel cobalt manganese oxide
(NCM), and lithium nickel aluminum oxide (NCA). Nickel and cobalt oxide are
classified as carcinogens and can also provoke severe intoxications (IFA Institut für
Abreitsschutz der Deutschen Gesetzlichen Unfallversicherung 2014).
Elevated temperatures provoked by joule heating due to external or internal short
circuits, over charging, or high ambient temperatures lead to chemical reactions
inside the battery cell. If a critical temperature is exceeded, exothermal reactions
take place, which again provoke other examples of such. An incident such as this is
called thermal runaway and leads to venting or even exploding of the battery cell.
Beside flammable gaseous products, also toxic gases are released during venting or
thermal runaway. Mainly these are carbon monoxide (CO) and hydrogen fluoride
(HF). Other toxic reaction products could be phosphorus pentafluoride (PF5) as the
thermal equilibrium of the decomposition of the conducting salt, and phosphor
oxyfluoride (POF3). HF and phosphorus compounds are mainly formed due to the
reactions of the conducting salt LiPF6 and the organic solvents. Another source of
HF is the endothermal decomposition of the binder polyvinylidene fluoride
(PVDF). This binder decomposes at temperatures around 300 °C, a temperature
that is easily achieved in battery cells during a thermal runaway.
The reactions of the conducting salt and the organic solvents not only take place
under abuse conditions, but also during ageing of lithium-ion batteries. Grützke
et al. (2015b) investigated the electrolyte decomposition products that can be found
in field tested battery cells of an HEV. Beside fluoro phosphates, they also found
(fluoro) alkyl phosphates as ageing products and proposed a reaction scheme
(Fig. 3.1).
Grützke et al. detected several degradation products. Hydrolysis of PF5 leads to
POF3 and HF, as well as phosphorus acid and fluorinated phosphates. The latter can
further react with the electrolyte solvents to form different alkyl phosphates under
release of carbon dioxide (CO2) and the corresponding alkanes.
The described reactions can also occur in the crushing product. After crushing,
“micro-short-circuits” between anode and cathode fragments of loaded battery cells
lead to joule heating and therefore to a temperature rise at hot spots. The increased
temperatures lead to the partly evaporation of the electrolyte solvents and the
(a)
(b)
(c)
(d)
Fig. 3.1 General decomposition pathways for the formation of transesterifications products (a),
oligocarbonate-based products (b), organophosphate-based products (c) organic
fluorophosphate-based products and (d) hydrolysis products. Reprinted with permission of
(Nowak and Winter 2017)
exposition of the conducting salt. The aforementioned reactions take place at

temperatures of around 60 °C, and in contact with aerial humidity (Diekmann et al.
2016). The results of a gas measurement during such an incident can be found in
Chap. 7.
The decomposition of the conducting salt and the reactions with the electrolyte
solvents were not only observed during spontaneous and fast events, but also in
experiments with long durations. To investigate the hazards during maintenance
and intermediate storage of the crushing product, the team of the University of
Münster carried out storage experiments. Fresh crushed battery cells fragments
were enclosed in tin cans. Gas samples and pictures were taken over the duration of
five months.
The changes to the tin cans during the experiments can be seen in Fig. 3.2.
After two months, the first patches of rust were observed. These probably
occurred because of the formation of HF and phosphorus acid due to the
Fig. 3.2 Pictures of the changes on the storage containers over storage duration. Reprinted with
permission of (Grützke et al. 2015a)
Fig. 3.3 Ionic decomposition and reaction products identified via IC/ESI-MS. Reprinted with
permission of (Grützke et al. 2015a)
decomposition of LiPF6. After storage for five months, holes due to rust were found
in the tin plate cans. After this storage duration, the experiments were aborted
because of the uncontrolled gas leakage.
The gas samples were analyzed by ion chromatography-electrospray
ionization-mass spectrometry (IC/ESI-MS) and gas chromatography-mass spec-
trometry (GC–MS) to detect and identify the decomposition and reaction products.
The ionic components in the gas product (Fig. 3.3) were identified as ethylene
phosphate (123 u), hydroxyethyl fluorophosphate (143 u), methyl fluorophosphate
(113 u), and ethyl fluorophosphate (127 u). These are intermediates corresponding
to the decomposition of the conducting salt.
Fig. 3.4 Decomposition and reaction products identified via GC-MS. Reprinted with permission
of (Grützke et al. 2015a)
The products created by the reaction of the conducting salt decomposition

intermediates and elctrolyte solvents were identified by GC-MS, as shown in
Fig. 3.1. Beside the linear and cyclic carbonates, these were dimethyl fluorophos-
phate (DMFP) at 98 mass-to-charge ratio (m/z), ethylmethyl fluorophosphate
(EMFP) at 115 m/z, and diethyl fluorophosphate (DEFP) at 113 m/z (Fig. 3.4).
Experts and researchers often compare alkyl phosphate, like DMFP and DEFP,
and the chemical warfare agent sarin, due to their similar chemical structures. It is
recommended to handle stored fragments, but also opened battery cells that came in
contact with water, cautiously. Inhalation, as well as contact with skin and mucosae
have to be avoided.
3.2 Interaction of Hazards
The before mentioned electrical, fire and explosion, and chemical hazards are
interacting. For example: a short circuit leads to elevated temperatures, which
trigger exothermal reactions leading to flammable and toxic reactions products.
To observe this interaction, TU Braunschweig developed and realized a battery
cell investigation chamber for nail penetration tests. During a mechanical abuse, a
high amount of flammable gases, as well as toxic gas components might be
released. Due to this fact, the battery testing unit basically is required to be a gas
tight chamber. Despite that, the chamber has to resist high pressure due to the
evolved gas and high temperatures. On this account, the testing unit is able to resist
an inner pressure up to 5 bars and is made of a non-corrosive material. The chamber

is equipped with a safety window and a camera to observe the internal space during
an abuse test. In this particular case, a nail made of non-corrosive steel with a
diameter of 3.0 mm and tip angle of 20° is used to puncture the battery cell and
provoke an internal short circuit by contacting the electrodes. The insertion depth
and velocity of the nail are controlled by a way-guided linear slide. To prevent short
circuit between the nail and the housing, the linear slide is electrically isolated from
the rest of the testing unit. The chamber enables temperature measurement on the
battery cell surface and gas analysis via Fourier transform infrared spectroscopy
(FTIR).
In the following, the test results of two different 18650 battery cells are com-
pared. One battery was a cell type with the cathode active material lithium nickel
manganese oxide (NCM) and a capacity of 2.2 Ah, and another 18650 battery cell
with the cathode active material lithium nickel aluminum oxide (NCA), containing
a higher capacity of 3.4 Ah. The battery cells were installed in a PTFE holder, as
shown in Fig. 3.5. During these tests the nail penetrates to the middle of the battery
with a velocity of 80 mm s−1. The tests were carried out under nitrogen
atmosphere.
During each test of the triple determination, it was observed that the rupture
disks of the NCM battery cells opened, while those of the NCA battery cells
remained closed. Further differences regarding surface temperatures and gas con-
centrations could be also observed. The resulting surface temperatures and their
time dependent course can be seen in Fig. 3.6.
Fig. 3.5 Installation of the

18650 battery cell in a PTFE
holder
700
600
cell surface temperature Ts [°C] Ts18650 NCA 3.4 Ah
500 Ts18650 NCM 2.2 Ah
400
300
200
100
0
0 2 4 6 8 10 12 14 16 18 20
time after penetration tap [min]
Fig. 3.6 Temperatures on the battery cells surfaces during nail penetration tests with 18650
battery cells
220000
concentration carbon monoxide cCO [mg*m-3]
200000
180000
160000
140000
cCO 18650 NCA 3.4 Ah
120000
cCO 18650 NCM 2.2 Ah
100000
80000
60000
40000
20000
0
0 2 4 6 8 10 12 14 16 18 20
Fig. 3.7 Concentrations of carbon monoxide (CO) during nail penetration tests of 18650 battery
cells
The surface temperature of the NCA battery cell is around 100 °C higher than
that of the NCM battery cells. Additionally, this temperature is reached earlier (after
0.7 min for the NCA battery cell and after 1 min for the NCM battery cell).
This trend can also be seen in the time dependent concentrations of CO and HF
in Figs. 3.7 and 3.8.
80
concentration hydrogen fluoride cHF [mg*m-3]

75
70
cHF 18650 NCA 3.4 Ah
65
cHF 18650 NCM 2.2 Ah
60
55
50
45
40
35
30
25
20
15
10
5
0
0 2 4 6 8 10 12 14 16 18 20
Fig. 3.8 Concentrations of hydrogen fluoride (HF) during nail penetration tests of 18650 battery
cells
Fig. 3.9 Relative comparison of highest concentrations of the identified gaseous products of nail
penetration tests of 18650 battery cells
The maximum concentrations of CO as well as HF are significantly higher for

the NCA battery cells. This can be explained by the higher surface temperatures, as
well as the different reactions between electrolyte components and active materials.
Differences of the gaseous products can also be seen in the relative comparison of
the highest concentration of the other measured gas components in Fig. 3.9.
A significant difference can be seen in post mortem studies of these two battery
cell types. The studies were carried out by micro-computer-tomography (µCT).
With this analysis technique, scans of cross sections can be used to build a three
dimensional model of the investigated probe. Figure 3.10 shows the models of the
Fig. 3.10 µCT-images of the inner structure of the battery cells after abuse (at top: NCM 2.2 Ah,
at bottom: NCA 3.4 Ah, left: structure including electrodes, right: rod and aluminum)
two battery cells after the test. The two pictures at the top show the inner structure
with electrodes and active material, as well as the aluminum rod of the electrode
wrap of the NCM battery cell. Comparison of the pictures of the NCM battery cell
and the NCA battery cell shows a higher amount of melted aluminum from the
NCA cathode. This fact can be explained by the closed rupture disk. During the
tests of the NCA battery cell, the materials of the electrodes, electrolyte compo-
nents, and reaction products can only leave the cell through the small hole created
by the nail. In particular, the electrode materials are sterically inhibited from this
way due to the wrapping of the electrodes. Because of this inhibition, the melted
aluminum stays in the battery cell.
The rupture disk of the NCM battery cells opens very fast and the melted
aluminum is set free, leading to nearly no accumulation inside the battery cell.
To prove the reaction of the NCA battery cell to a nail penetration test with
opening of the rupture disk, the PTFE holder was changed to an identically con-
structed holder made of stainless steel. This material has a significantly higher
Young’s modulus than PTFE. Therefore, the expansion of the battery cell during a
thermal runaway is inhibited and the pressure inside the battery will rise. This leads
to the opening of the rupture disk.
Change of the holder also leads to other results of the tests. As steel dissipates
the developed heat of the abused battery cells more efficiently, the temperatures
decrease faster. Also, the opening of the rupture disk leads to a faster increase of the
Fig. 3.11 Concentrations of hydrogen fluoride (HF) during nail penetration test of 18650 battery
cells, installed in a PTFE or a steel holder
measured gas components. Although the maximum concentration of HF stays the

same, it is reached about a minute earlier (Fig. 3.11).
These facts show that installation of the battery cells during testing can have a
significant influence on the test results.
Besides the gaseous components, the solid ingredients of the battery cells are
also released during a thermal runaway. Due to the melting of the aluminum foil,
these are also the cathode active materials containing nickel and cobalt. As men-
tioned before, these materials are carcinogenic.
The knowledge of the specific hazards of lithium-ion batteries and their inter-
action, as well as the gaseous and solid products and their properties under abuse
conditions lead to specific safety precautions. These can be found in the process
step chapters and Chap. 14.
3.3 Conclusion
The project partners evaluated the hazards along the process chain and roughly
divided them into three sections: electrical hazard, fire and explosion hazard, and
chemical hazard. As all three sections are intertwined, abuse tests of lithium-ion
battery cells were carried out.
Fig. 3.12 Schematic drawing of the hazards associated with lithium-ion batteries and their
interaction
The electrical hazards, like electric shock and short circuits, play a major role
during the process steps discharge and disassembly. Especially short circuits can be
a trigger to induce an interaction of the hazards and are provoked during crushing.
Short circuits can lead to increased temperatures inside the battery cells, which
provokes exothermal chemical reactions, resulting in a thermal runaway. The main
infrared active gases that are released during a nail penetration under nitrogen
atmosphere, like HF and CO, were identified and quantified via FTIR. During these
tests the influence of the installation of the battery cells, as well as the applied
cathode active material, became obvious.
Furthermore, the chemical hazard was determined by the decomposition of the
conducting salt LiPF6 and the carcinogenic cathode coating materials.
The gained knowledge of the hazard potential was used for the design of a safe
crusher and the development of a safety concept for the process realization in pilot
scale.
The hazards of lithium-ion batteries are condensed in the following scheme
(Fig. 3.12).
References
Diekmann J, Hanisch C, Froböse L, Schälicke G, Loellhoeffel T, Fölster A-S, Kwade A (2016)

Grützke M, Kraft V, Hoffmann B, Klamor S, Diekmann J, Kwade A, Winter M, Nowak S (2015a)
Aging investigations of a lithium-ion battery electrolyte from a field-tested hybrid electric
vehicle. J Power Sources 273:83–88. https://doi.org/10.1016/j.jpowsour.2014.09.064
Grützke M, Krüger S, Kraft V, Vortmann B, Rothermel S, Winter M, Nowak S (2015b)
Investigation of the storage behavior of shredded lithium-ion batteries from electric vehicles for
recycling purposes. Chemsuschem 8(20):3433–3438. https://doi.org/10.1002/cssc.201500920
IFA Institut für Abreitsschutz der Deutschen Gesetzlichen Unfallversicherung (2014) Dataset
nickel oxide
Nowak S, Winter M (2017) The role of sub- and supercritical CO2 as “Processing Solvent” for the
recycling and sample preparation of lithium ion battery electrolytes. Molecules 22(3). doi:10.
3390/molecules22030403
Chapter 4
Overdischarging Lithium-Ion Batteries
Daniel Hauck and Michael Kurrat
Abstract A main objective of the project LithoRec II is to study the overdischarge of

Li-Ion batteries for the purpose of recycling. For safety and functional reasons, the
batteries need to be overdischarged before undergoing the process steps of disas-
sembly and crushing. In this study, various devices for overdischarging are set up and
investigated together with corresponding battery behavior. Device properties such as
cost, safety, discharge time and discharge flexibility are assessed. Furthermore, the
study focused on parameters like battery heating during discharge and pole reversal,
relaxation amounts and heights as well as short circuit currents. The investigations
show, that it is most advisable to discharge whole battery systems by energy recu-
peration into grid with electronic adjustable loads, because of efficiency and safety
reasons. Overdischarging is not problematic if crucial battery parameters are
observed. It is recommended that overdischarging of automotive traction batteries
should only be done by high-voltage specialists.
4.1 Introduction
4.1.1 Overdischarging Batteries in the Literature
To date, relatively little has been reported on overdischarging lithium-ion batteries

(LIBs). This is possibly due to the fact that this field has been of little interest to
researchers and its importance has not even been questioned. LithoRec recycling as
well as electrical hazards are the key areas of focus forming the motivational
background to this project. A further aspect of studying overdischarging is cell
destruction or accelerated aging which can arise as a result of faulty programming in
D. Hauck (&) M. Kurrat

Institute of High Voltage Technologies and Electrical Power Systems, TU Braunschweig,
Brunswick, Germany
e-mail: d.hauck@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_4
54 D. Hauck and M. Kurrat
the battery management unit (BMU). Again, there is not much literature on LIBs,
whereas much is known about other battery technologies such as lead-acid batteries
in which undesired overdischarging may occur during stationary storage (Blank
2012). Similar investigations have however been conducted for NiCd and NiMH
batteries (Jossen et al. 2016). The chemical hazard of overdischarging compared to
overcharging, especially for LIBs, is much less which may be one of the reasons for
it not being the focus of much interest in the literature.
4.1.2 Motivation for Studying Overdischarging in LithoRec
For safety and operational capability reasons, overdischarging LIBs is a necessary

aspect of the LithoRec recycling chain, in particular for the later processes of
dismantling and shredding batteries.
Battery systems typically charged to around 400 V (DC) pose a potentially
life-threatening hazard to operators when they are dismantled. Generally, working
under high voltage (more than 60 V DC) is permitted only for high-voltage
(HV) specialists according to DGVU 200-005 (previously BGI 8686). Discharging
a battery system is preferable prior to dismantling it, as a discharged battery system
does not require HV specialists. Furthermore, the discharging of the system does
not entail much effort compared to discharging singular modules or cells. Therefore,
discharging at battery system level should be the preferred option.
Using overdischarged batteries is also very important for their subsequent
shredding. The mechanical pulp of charged batteries may cause local short circuits
which may then lead to local heating or sparking. The electrolyte-gas atmosphere in
a shredder is extremely flammable. As a consequence, only discharged batteries
should be shredded.
4.1.3 Individual Process Steps of Overdischarging

in LithoRec Recycling
Normally, batteries that have been overdischarged are no longer of any use, so that
their subsequent recycling is a reasonable last step in a battery’s life cycle. There is
an increasing effort to emphasize this aspect in industry, research and politics. The
concept of “prolonging or reusing automotive traction batteries after their first
application as second-life batteries” is becoming more widespread. From economic
and more importantly from environmental standpoints, there are significant argu-
ments for promoting this aspiration. To make a decision about which option is most
reasonable—recycling or a second life—the state of an old battery should be
assessed. Figure 4.1 shows a proposed scheme to find out how to meet this
challenge.
4 Overdischarging Lithium-Ion Batteries 55
Fig. 4.1 Flowchart proposing how to deal with LIBs before and after overdischarging
For the purpose of assessing the state of a battery, it is reasonable to classify the
battery into a system, modules and cells. Usually, the BMU which gives data on the
state of a cell, is designed according to master/slave architecture. In battery systems,
it is possible to recapture the lifetime data about the state of the BMU master if
there is access to the information and the respective reading equipment. In this way,
it is possible to undertake sorting and storage according to the charge level, energy
or cell chemistry. However, with modules and cells, such reading of data is not
possible, so that the characterization has to be done manually.1 Ultimately, all
batteries, even so-called second life batteries, need to go through the process of
recycling, meaning that overdischarging and short circuits are inevitable.
1
The manual characterization of the state will not be described here because overdischarging is the
focus of this paper.
4.1.4 Focus of the Research
In addressing the subject of “overdischarging LIBs”, several questions are raised,

for example:
• What does overdischarging mean and what happens at pole reversal?
• With which technique or equipment can a battery be overcharged or pole
reversed?
• How long does overdischarging need and how high is the voltage reversal?
• How safe is overdischarging and what risks are to be encountered?
• What is the cost of discharging and how much profit can be gained through
energy recovery?
• At what point of discharging is it safe to connect the battery to a short circuit?
The following sections attempt to answer these and further questions.
Section 4.2 explains the basics of overdischarging LIBs under electrical and elec-
trochemical aspects. Section 4.3 presents the applied battery systems, the applied or
developed instruments for discharging, as well as the measuring setups for
overdischarge investigations. Section 4.4 presents and discusses the results. The
key findings are summarized in Sect. 4.5 and an outlook on future challenges is also
given.
4.2 The Basics of Overdischarging Lithium-Ion Batteries
4.2.1 Definition of Overdischarging
Generally, overdischarging a battery means draining the current after the end-point
voltage, which leads to lasting damage (Bauer 2017). The following figure illus-
trates this situation.
In the example given in Fig. 4.2, overdischarging starts below 2.50 V.
Depending on the discharge current and the temperature, approx. 20% of additional
capacity can be discharged at up to 0 V. In the case that discharging will be
continued to the negative voltage range (below 0 V), a so-called pole reversal will
occur. In this case, the anode and cathode change polarity and thus the anode will
become the cathode and vice versa.
4.2.2 Electrochemical Basis
LIBs are electrochemical storages in which the lithium-ions—the charge carriers

migrate between the anode and the cathode. During discharging, the lithium-ions
are released from the anode and react with the electrons which flow via the external
Fig. 4.2 Definition of “overdischarging” and “pole reversal”
load circuit, thus forming a chemical compound at the cathode. While overdis-
charging, all lithium-ions finally migrate from the anode to the cathode. After this
point, forced overdischarging makes the battery try to continue delivering charged
ions. Thus, copper ions dissolve from the anode current collector and take over the
ion transport. This leads, on the one hand, to the formation of dendrites which
pierce the separator leading to short circuits. On the other hand, metallic copper is
deposited on the graphite coating in the separator. Additionally, traces of corrosion
can be formed on the graphite-coated copper current collector (Gellerich 2015).
The latter has to be dealt with during the subsequent processing steps (separa-
tion, leaching by treatment with bases, reprocessing). In addition, temperatures up
to 120 °C have been measured at the cell wall while overdischarging, which is
dangerous due to the possibility of starting an exothermic reaction of the cell
(Gellerich 2015).
In other overdischarging trials, the composition of the gases was measured
during overcharging and overdischarging processes. It was found that the cell
heated up by 29 °C; it expanded in volume and various gases were formed when
batteries were being overdischarged. The largest proportion of gas at 47% was
hydrogen, which is highly inflammable, and can form explosive oxyhydrogen with
air. Other detected gases were: carbon monoxide, carbon dioxide, methane, ethane
and ethylene (Mähliß 2012).
4.2.3 Electrical Basis of Overdischarging
In order to discharge a battery, it is necessary to connect an electric load to the main

poles. The electrons from the anode migrate via the external circuit and via the
electric consumer load to the cathode where they react with the lithium-ions. The
electric load can show different types of electrical behavior depending on the dis-
charge device/technique. Essentially, different modes of discharge behavior exist, as
qualitatively shown in Fig. 4.3.
Furthermore, different ways of controlling can also be undertaken successively
depending on the type of discharge equipment. The control types offer different
advantage/disadvantages and play a major part in the selection of the overdis-
charging process. The following section explains other important criteria for the
selection of the overdischarging process.
4.2.4 Criteria for Description and Selection

of Overdischarging Processes
There are a number of different technical ways (see Sect. 4.3.2) of discharging
electrical vehicle (EV) batteries. These options all have advantages and disadvan-
tages. Some of the important factors for selecting and defining the discharging
process are:
• the cost of the discharging device
– the acquisition cost
– the cost of (e.g. current, cooling liquid) or income from (reimbursement
through power recovery) the running process
– maintenance and repair costs
• other costs
– power lines for short circuit of cells, modules or systems
– storage areas for cells, modules or systems
– HV-qualified personnel
Fig. 4.3 Different basic modes of electrical behaviour for overdischarging

• speed of discharging
• safety of discharging
• flexibility in the formation of the discharge characteristic curve
• complexity and ease of handling of discharging device
• automation of the discharging device
• taking into account the material recycling share depending on the discharging
strategy ! discharging at high speed might lead—similarly to “fast discharg-
ing”—to a chemical deposition of Li or Cu which cannot be recovered in a good
yield.
• mobility of the discharging device
• sufficient power supply depending on the load (only in the case of power
regeneration).
For an optimal discharging strategy, the criteria listed above should be assessed
in a holistic approach.
4.3 Investigated Batteries and Devices

for Overdischarging
The devices and batteries for discharging, which were used and developed during
the project, will be explained in the following section. Batteries as well as devices
for discharging were studied during the research. The batteries can be divided into
battery cells, modules and systems. The devices for discharging can be differenti-
ated by energy transformation type. The following Fig. 4.4 illustrates this.
While the discharging of cells and modules was tested with devices that were
based on energy transformation into heat, the discharging of systems was addi-
tionally tested by energy transformation through power recovery. The following
section describes in detail both the batteries and devices investigated as well as the
experiments conducted.
4.3.1 Battery Cells, Modules, and Systems Tested
The tested batteries were provided by Volkswagen AG and AUDI AG who were
the project partners. The batteries were from recent automotive generations that
were produced at the time of the research. An overview of the cells, modules, and
systems used is given in Tables 4.1 and 4.2.
The tables above illustrate the different variations in cell and module connec-
tions, depending on whether the application is for electric vehicles (EVs), hybrid
electric vehicles (HEVs) or plug-in hybrid electric vehicles (PHEVs).
Whereas the system of the VW e-Golf consists of a triple parallel connection
(higher capacity or rather driving range), the HEV and PHEV are only connected in
Fig. 4.4 Devices used and their energy transformation concepts to overdischarge cells, modules,
and systems
Table 4.1 Battery cells, modules, systems investigated at the time of the research
BEV PHEV HEV
VW e-Golf VW Passat GTE AUDI Q5 Hybrid
System
Modules
Cells
series (higher voltage or rather power to accelerate/recuperate). This information is

helpful for calculating the maximum tolerable discharge current. It can also be
concluded that the systems exhibit nearly the same voltage, though the hybrid
variants possess much less energy.
Table 4.2 Most important battery data to be known for overdischarging
Property Unit BEV PHEV HEV
VW e-Gotf VW Passat GTE AUDI Q5 Hybrid
Cell Module System Cell Module System Cell Module System
Chemistry – NMC NCA NMC
Voltage Min. V 3 12 264 3 36 288 2.5 45 144
4 Overdischarging Lithium-Ion Batteries
Max. 4.1 16.4 360 4.1 49.2 394 4.3 73.8 309.6
Current Max. discharge A −116.6 −350 −350 −255 −255 −255 −1801 −1801 −1801
C-rate – −4.7 −4.7 −4 7 −9 1 −9.1 −9.1 −581 −581 −581
Nominal energy Wh 89 1.065 23 400 99 1.194 9 548 11.4 205 820
Capacity Ah 25 75 75 28 28 28 3.1 3.1 3.1
Weight Kg 0.72 10.5 320 0.72 10.5 155 <0.53 <9.5 38
Amount – 264 16 & 12 1 96 8 1 72 4 1
Connection – – 3p4s & 3p2s 3p88 s – 12s1p 96s1p – 18s1p 72s1p
1
Represents max. pulse discharge current for 10s
61
4.3.2 Applied Devices and Measuring Setup

for Overdischarging
To overdischarge the batteries, special overdischarging setups had to be developed

and constructed and then compared to commercially available loads.
Discharging with constant resistance
Obviously, the simplest way of discharging a battery is with constant resistance.
Those variants, A, B and C, indicated in Figs. 4.5, 4.6 and 4.7 involve the CR
control mode.
(A) Single cell resistor
The cells are connected directly to a resistor. The current flow can be calculated
according to Ohm’s law. The load resistance applied is 1.0 Ω and also 0.5 Ω, so that
the maximum current of 10 A is not exceeded. To conduct the heat away, the load
resistor is placed in a cooling body which is mounted on a cooling plate.
Fig. 4.5 Single cell resistor (variant A)

Fig. 4.6 Selectable resistor device (variant B)
(B) Selectable resistors

An extension of the singular resistance connection is the discharger with selectable
resistors. The device has three different resistance settings which, depending on the
battery voltage and the voltage limit set by programmable logic control (PLC), can
be switched on and off as required. Because the voltage drops during the dis-
charging process, the device has the advantage that the discharging current remains
high due to switching to low resistance values in contrast to variant A.
(C) Discharging in conductive liquids
Instead of a battery with a solid conductor for discharging, it is possible to use
conductive liquids. An obvious variant is to use salt water as the conductive liquid.
Figure 4.7 shows a schematic and a real experimental setup for cell discharge.
Fig. 4.7 Experimental setup to discharge battery cells in conductive liquids (variant C)
To measure the discharge current within the liquid, a positive cell pole is
insulated with silicone and passed over an ammeter outside the vessel serving as a
substitute pole, and then reintroduced close to the original pole.
This method falsifies the test results to some extent, but is however necessary to
obtain experimental data. For the setup, a solution of 300 g of NaCl in three liters of
water was used. Calculations based on the concentration, specific conductivity and
specific resistance give a resistance of 3.914 Ω for the shortest distance between the
minus and the plus pole substitute. The theoretical discharge current of a fully
charged cell is therefore approx. 1 A.
Discharging with constant current
(D) Discharging using a semiconductor
To discharge battery cells and modules with constant current, a circuit with
metal-oxide-semiconductor field-effect transistor (MOSFET) was develop as shown
in Fig. 4.8.
The device is conceived for batteries up to 80 V. Using the potentiometer, a
maximum discharge current of 50 A at a maximum allowed power of 300 W can be
set. A total of five MOSFETs were mounted on one common cooling body which,
in turn, was cooled using an air blower.
Fig. 4.8 Device with MOSFETs to discharge cells and modules with constant current (variant D)
Discharging with constant power

In order to discharge a battery in CP mode, commercially available electronically
adjustable loads were used as seen in Fig. 4.9.
(E) Discharging using energy transformation into heat
To discharge a battery at higher power, a TORKEL 860 from Megger GmbH
(Fig. 4.9 left) was used. It is limited to a maximum input voltage of 80 V with a
maximum current of 110 A or 15 kW. Three control modes of CC, CP and CR are
supported. Discharging occurs through resistors which are cooled down with eight
air blowers. The device can be controlled using a touch panel or the supported user
software.
(F) Discharging using energy recuperation into grid
To discharge a battery with energy recovery, the electronically adjustable load
EA-ELR-9500-30 supplied by EPS Stromversorgung GmbH (Fig. 4.9 right) was
Fig. 4.9 Electronically adjustable loads to discharge modules and systems in CR, CC and CP
modes [left: heat transformation (variant E); right: energy recuperation into grid (variant F)]
used. The maximum input voltage was 500 V and the maximum discharge current
was limited to 30 A or 3.5 kW. The four control modes of CC, CV, CP and CR are
supported. For power recovery, the DC energy introduced is converted into a power
synchronized sinusoidal current. The power can be introduced into local and public
power supplies. In this way, there is little heat development compared to other
discharging devices. The device can be regulated using a touch panel or propriety
software. The efficiency at maximum input voltage or power is 94.5%.
4.3.3 Experimental Setup and Research Questions
The investigations are related to the functionality and applicability of the dis-
charging device. Each discharging device has its technical specifications with
respect to voltage, current and power. The choice of which device is to be used for
battery discharge depends on these specifications. As in the previous section, the
devices and setups for overdischarging offer various features in terms of control
modes. As battery behavior when overdischarging is the focus of this research, no
detailed description of the experiments to investigate the behavior of the devices
will be given at this point. Instead, the main parameters relevant to overdischarging
and the experiments conducted to determine these parameters will be described.
The experiments will be divided according to battery size (cells, modules and
systems) and then described accordingly.
Discharging devices used and experimental setup
To investigate the discharging behavior of battery cells, a MOSFET device (variant
D) was used. It offers the advantage that the discharge occurs at constant current. To
discharge the module, the electronically adjustable load EA-ELR-9500-30 (variant
F) was used. To discharge the system, besides variant F, the TORKEL 860 battery
load tester (variant E) was also used as, with this device, a significantly higher
power can be achieved. The basic setup for all variants is shown in the following
circuit diagram in Fig. 4.10.
Fig. 4.10 Circuit diagram of basic setup for all discharge variants
To measure the current, the battery was connected to the load resistor (discharge
device) via a shunt resistor. An additional high voltage switch in series enabled the
safe switching of the battery to the discharge circuit. Parallel to the battery and to
the shunt, a short circuit was set up to enable the battery to be short circuited with
an additional high current switch.
Research issues
With the aid of an experimental setup, various questions are to be investigated
such as:
• How does the ambient/cell temperature influence discharge behavior?
• How much energy can be extracted in the overdischarging range and how long
does this take?
• How high are the short circuit currents depending on the electrical connection?
• How much does the voltage relaxation vary depending on the short circuit
duration?
• How much heat is developed while overdischarging?
• What effect does pole reversal have on the voltage relaxation?
4.4 Results and Discussion
Investigations related to the functional behavior and applicability of the discharging

devices (Sect. 4.4.1) as well as the behavior and characteristics of batteries when
they are overdischarged (Sect. 4.4.2) are reported in the next sections.
4.4.1 Investigations Related to the Functional Behavior

and Applicability of the Discharger
Discharging with constant resistance

The discharging device variants A–C all function following the control mode at
constant resistance. In Figs. 4.11 and 4.12 (representing variants A and B), the
current/voltage characteristic curves for the discharge of two 3.1 Ah cell are
illustrated.
The initial voltage of both the cells amounts to 3.67 V and a 50% state-of-charge
(SOC). The discharging process in variant A is conducted at a constant resistance
value of 0.5 Ω. The discharge current falls proportionally to the discharge voltage.
A voltage drop to 1 V is only observed after approx. 30 min.
With variant B, varying resistances for the discharge are used. After about
10 min, the voltage drops dramatically and the discharging resistance is switched to
a lower value, so that the discharge current remains relatively high. Due to the low
Fig. 4.11 Overdischarge current/voltage characteristic curve for a 3.1 Ah cell with variant A
Fig. 4.12 Overdischarge current/voltage characteristic curve for a 3.1 Ah cell with variant B
capacity in the low voltage range, a rapid drop to 0 V occurs, resulting in the
initiation of a short circuit of the cell. A direct comparison of both variants A and B
makes it clear that switching different resistances with a PLC leads to a significantly
faster discharge and is therefore more efficient.
Apart from these variants with a solid load resistor, there is variant C which uses
a liquid load resistor comprising a conductive salt/water solution. The discharge
curve for the current and voltage are shown in the following figure (Fig 4.13).
The data points were taken every second and with an accuracy of 10 mV/
10 mA. At the start of the measurements, the current drifted to a value of approx.
0.9 A which fell to approx. 0.6 A in the following 3 min. During the experiment
(until the 22nd minute), the current strength began to fluctuate and dropped
Fig. 4.13 Overdischarge current/voltage characteristic curve for a 3.1 Ah cell with variant C
continuously until it reached 0.1 A. Finally, using a metal bar, the cell was lifted out
of the liquid within the vessel a few times and immersed again (22nd to 25th
minute). This was done to remove potential resistance layers on the electrode and
thereby enable an increase in the current flow. Since the current did increase,
although only for a short period, this shows that both the layer and the gas buildup
on the electrode influence the discharge process negatively. The experiment was
therefore interrupted (25th to 28th minute) and the cell was removed from the
liquid. A strong dissolution of the original silver-colored coating of the substitute
pole (banana plug—the material is not known) was visible and a layer of black
coating was built up as shown in Fig. 4.14 (middle left).
In order to achieve better conductivity from the cathode side, the remnants of the
plug were wrapped in nickel foil. Subsequently, in a second trial, the battery cell
was reimmersed in the solution (37th minute). The current then reached a value of
0.8 A for 5 min (until the 42nd minute), before it started dropping continuously in
the following 6 min (until the 48th minute), reaching a value of approx. 0.6 A. In
the following 6 min (until the 54th minute), the current began to fluctuate and
reached its minimum at 0.1 A. The cell was then removed from the solution and the
data collection was stopped at the 68th minute. After the removal of the cell, it was
seen that the nickel foil was completely dissolved and the black covering layer on
the anode was even more distinct. Liquid samples were taken after each immersion
trial (Fig. 4.14 middle right) and sent to the battery research center MEET for
Fig. 4.14 Experimental setup (left), cell poles (middle left), liquid sample (middle right), brown
liquid (right)
Table 4.3 Concentration Concentration (ppm)

analysis of the different
Unshaken Shaken
samples
Ni Cu Ni Cu
Trial 1 370 4.498 1.252 15.838
Trial 2 436 217 18.183 7.072
Trial 3 212 99 4.233 1.577
analysis. To find out whether the battery would continue to discharge, a third trial
was conducted. The silicon insulation on the cathode was removed and the cell, tied
to a string, was immersed as shown in Fig. 4.14 (right). The open vessel was
allowed to stand and monitored for about 30 min. In Fig. 4.14 (right), it can clearly
be seen that the liquid turned brown and some material was found floating on the
surface. Apart from this, gas formation was observed. The loss of contact of the
electrode to the liquid may be the cause of the fluctuation of the current observed.
The cell was again removed from the vessel and a cell voltage of 3.78 V was
measured. Compared to the off-load voltage of 4.01 V after the second trial, a
significant drop in voltage of 0.23 mV was evident. This shows that in fact a further
and in this case a faster discharge has taken place. After another discharge (fourth
trial) and the subsequent removal of the cell, it was seen that the anode pole had
completely disconnected from the cell and the experiment was ended.
Table 4.3 shows an extract of the results of the MEET X-ray fluorescence
analysis of the samples from the first three trials of samples taken without and after
agitation.
Generally, a significant difference is observed between samples taken before and
after agitation. This can only be explained as being caused by precipitation or
floating materials (from Ni and Cu salts and hydroxides). In the agitated samples, a
high Ni content was found in trial 2, which can be explained as having been caused
by using Ni foil on the banana plug. The material used on the banana plug was not
investigated. With Cu, a continuous drop in concentration was evident. At the start
(trial 1), a large amount of Cu was probably dissolved as Cu salts in solution
(Eq. 4.1), which then led to the formation of Cu(I) oxide and Cu(II) oxide layers on
the electrode (Eq. 4.2a, b). This explains the dark coating on the electrode.
Cu þ 2H2 O þ 2NaCl ! CuCl2 þ 2NaOH þ H2 " ð4:1Þ
Cu þ H2 O ! CuO þ H2 " ð4:2aÞ
2Cu þ H2 O ! Cu2 O þ H2 " ð4:2bÞ
The visible dissolution of the banana plug (Ni electrode) led to an increase of
nickel ions in the solution. The cause of the dissolution is the formation of
hydrochloric acid (see Eqs. 4.3a, 4.3b to 4.4):
2NaCl ! Cl 2 þ 2Na þ þ 2e ð4:3aÞ
Cl2 þ H2 O ! HCl þ HOCl ð4:3bÞ
Ni þ 2HCl ! NiCl 2 þ H2 " ð4:4Þ
In the third and fourth trials, the insulation of the cathode was removed and
therefore the original cell potential made of aluminum was exposed. The reaction of
the cathode side can be represented in Eqs. 4.5 and 4.6.
Al ! Al3 þ þ 3e ð4:5Þ
3CuCl2 þ 2Al ! 3Cu þ 2AlCl3 ð4:6Þ
In analogy to Eq. 4.4, the dissolution of the aluminum cathode can be described
by Eq. 4.6. The experiment has shown that electrolysis at cell level strongly
influences the discharge characteristic curve and may even lead to the dissolution of
electrodes. Therefore, an opening of the cell and hence the formation of hydro-
fluoric acid (HF) cannot be excluded. The necessary extent of measures to neu-
tralize HF must be estimated. In addition, in order to guarantee safety, the amount
of hydrogen (H2) has to be analyzed to exclude the potential risk of oxyhydrogen
formation. At the module or system level, these effects are less likely because, on
the one hand, higher voltages are used and, on the other hand, there are larger
electrode surfaces. However, to allow a more stable discharging process at cell
level, it is necessary to test the use of other liquids and additives. One way of
minimizing the electrolysis process could be the application of a more resistant
coating on the electrode surfaces. Also, the use of media other than liquids, e.g.
conductive sands or polymers, should be considered. These substances would make
the processes easier to handle and would rule out danger from HF and
oxyhydrogen.
Discharging with constant current
In order to discharge a battery with constant current, the specially developed
MOSFET device (variant D) was used. A typical characteristic curve of the current/
voltage for the 3.1 Ah cell with this discharge mode is shown in Fig. 4.15.
The initial voltage of the cell is 3.67 V or 50% SOC. Due to the preset, there is a
higher current at the start when the MOSFET device is switched on. The current is
then set to a constant 6 A. After about 12 min, the lower voltage limit is reached
and after about 17 min, the cell is short circuited because it is close to 0 V.
If a higher discharge current were set, discharging would take place at a much
higher speed. Another advantage over variant A can be seen in the overdischarging
area, where the discharge current remains constant and leads to a fast decline in
voltage.
Fig. 4.15 Overdischarge current/voltage characteristic curve for a 3.1 Ah cell with variant D
Discharging with constant power

In order to discharge a battery in CP mode, the TORKEL 860 device (variant E) and
EA-ELR-9500-30 (variant F) are used. Figure 4.16 presents characteristic curves
showing the current, voltage and power of a VW e-Golf battery system with
TORKEL 860 at constant power.
The battery system is discharged at a constant power of 15 kW. After about
62 min, the power diminishes due to loss of voltage, and the maximum possible
discharging current through the load is reached (55 A). After 66 min, the current
also decreases because the battery system cannot deliver higher current due to the
Fig. 4.16 Overdischarge current/voltage characteristic curve for a VW e-Golf battery system with
variant E
lack of missing energy. From 90th minute onward, the system is short circuited and
the measurement stopped.
Assessment of discharging variants
The presented ways of controlling or discharging variants have different advantages
and disadvantages. Using the criteria from Sect. 4.2.4, the variants can be assessed.
In the following Table 4.4, four crucial criteria were used for assessment.
Those discharge devices using constant resistance (variants A, B and C) are low
cost and are considered safe due to the limited maximum current and the lower
power. Due to the proportionality of current and voltage, the discharge time is high
and the discharge curve inflexible. Discharging a cell in liquid is assessed as being
unfeasible, as concluded from the results of the test. However, at the system level,
project partners could achieve positive results.
The discharge via the developed MOSFET device (variant D) does not show any
significant disadvantages. It is slightly more expensive than the passive resistance
discharge devices, but due to it being controlled with constant current, it is much
more efficient with respect to the speed of discharge. Manually controlling the
power is possible by using the potentiometer and so this variant can be assessed as
being semiflexible. Safety is also rated as being mediocre as the discharge current
can be regulated, however there is still the danger of manually over loading with too
high current. By using a temperature controlled current input, it would be possible
to eliminate those deficits.
The discharge devices commercially available (variants E and F) are more
expensive than discharge devices which were developed in-house. However, they
have the advantage of allowing different ways of controlling (CR, CC and CP), and
they allow random discharge curves to be programmed. By monitoring the voltage,
Table 4.4 Assessment of the discharge devices with four crucial criteria
Discharge Discharge device Cost of Discharge Flexibility Safety
mode discharge time discharge during
device profile discharge
CR A Single cell Low High Inflexible Safe
resistor
B Selectable Middle Middle Middle Middle
resistors
C Conductive Low Middle Inflexible Middle
liquid
CC D MOSFET Middle Short Middle Middle
CR, CC, CP E Energy High Short Very Very safe
conversion flexible
into heat
F Energy High Short Very Very safe
recovery in flexible
grid
current and temperature, the discharging, with additional switch-off criteria can be
considered as being relatively safe.
4.4.2 Investigation of Battery Behavior in Cases

of Overdischarging, Pole Reversal, Short Circuit
and Voltage Relaxation
Influence of ambient/cell temperature on over discharge and relaxation

behavior
Temperature is known to have a strong impact on the inner resistance of a battery
and thus its discharge behavior. In order to assess the influence of temperature
during overdischarging or the relaxation of voltage after a short circuit, three 50
Ah NCM cells were discharged with 30 A at −20, 0 and 40 °C; they were then short
circuited and relaxed as shown in Fig. 4.17.
It can clearly be seen that at the beginning of the discharge process, the cell voltage
with the lowest temperature (−20 °C) drops the most. This is due to the increased
viscosity of the electrolyte along with its worsened ionic conductivity (Jossen and
Weydanz 2006). The inner cell resistance is about 26.5 mΩ after one second,
compared to 11.2 mΩ (0 °C) and 1.3 mΩ (40 °C). At the moment of the short
circuit (at 0.5 V), this cell was able to discharge only about 10.5 Ah, instead of 11.5
Ah (0 °C) and 12 Ah (40 °C). After stopping the short circuit after 1 h, it can
clearly be seen that cell voltages at the lowest temperature relax the fastest and
highest. After four hours of relaxation, the cell at −20 °C even increased above a
Fig. 4.17 Overdischarge and

relaxation curves for 50
Ah NCM cells at different
temperatures
2.5 V cell voltage, which means that it came back to “normal” voltage range. That
might be dangerous for the user because a defect-free battery is presumed, although
it might clearly already be overdischarged. The temperature-dependent relaxation
effect is again traced back to increased inner resistance. In contrast, the cell at the
lowest temperature still has the highest capacity due to the fact that less energy was
taken out in the former discharging step. This conclusion is generally valid and can
be transferred to other cell types or modules and systems.
Influence of short circuit duration, cell capacity and cell connection on
relaxation behavior
Apart from temperature, especially the preceding short circuit duration exerts a
direct effect on relaxation behavior. In the next trial, three 3.1 Ah cells and three 25
Ah cells were each discharged, short circuited and relaxed. The short circuit was
undertaken at 0.1 V for all cells and removed after different times (10 min, 1 h and
24 h). The relaxed voltage was then measured. The relaxation curves of the six cells
are plotted in Fig. 4.18.
The figure above demonstrates the dependence of relaxation behavior on the
length of the short circuit. After a short circuit lasting 10 min, the voltage of the
3.1 Ah cell as well as of the 25 Ah cell increased up to above 2.3 V. If the short
circuit time is longer, lower voltages will accordingly be achieved: 1–1.5 V at 1 h
and 0.1–0.2 V at 24 h. In order to ensure safe relaxation, cells should be short
circuited for at least 24 h before opening a short circuit again. Only in this way it
can be guaranteed that the cells will not become an ignition source during the
subsequent step of shredding.
It can also be seen that the 3.1 Ah cells exhibit a steeper increase in voltage than
at the beginning of the relaxation compared to the 25 Ah cells. Similar to tem-
perature, this can be traced back to the higher inner cell resistance of the 3.1 Ah
cells.
Fig. 4.18 Relaxation curves

for three 25 Ah and three 3.1
Ah cells for different short
circuit times
Apart from studying cell capacity, the goal was also to test the connection of the
cells with respect to relaxation behavior. The systems are usually made up of
several modules which themselves are connected of several cells. The modules can
be handled by humans with respect to their weight. They could therefore be con-
sidered as the smallest exchangeable unit in future generations of automotive
vehicles, because of the large effort required to dismantle a single cell.
In order to test the influence of the connection on the relaxation behavior, several
modules were overdischarged and their relaxation behavior was measured. Three
modules from a VW e-Golf with a 3p4s connection were tested again with different
short circuit times (10 min, 1 h and 24 h). The modules were discharged with the
MOSFET device with a current of 15 A starting with a voltage of 15 V. At a
module voltage of 0.5 V, the modules were short circuited. The relaxation curves
are shown in Fig. 4.19.
As expected and similar to the cell measurements, the relaxation voltage of the
modules grows increasingly as the short circuit time is reduced. What is notable is
the high relaxation to above 8 V (tendency still increasing) after four hours and
after a short circuit of 10 min. As with the cells, only the 24 h short circuit time
leads to a minimal relaxation voltage of the module.
Looking at the short circuit currents just before each short circuit, important
differences can also be found. After 10 min of relaxation, 2.22 A were measured
and at 60 min of short circuit time, there were 1.69 A. After 24 h, only 0.02 A of
short circuit current was measured.
High relaxation voltages might be dangerous to humans. To test the remaining
energy and the remaining current, one module in 2p4s configuration with a
relaxation voltage of 2.8 V was again short circuited. The current and voltage
curves can be seen in Fig. 4.20.
At the time of the short circuit, a current of 103 A was measured. After about
1.5 min of short circuit duration, a current of 3.1 A was still flowing. The relatively
high flows show that modules should also be short circuited for at least 24 h.
Fig. 4.19 Relaxation curves

for three 3p4s modules for
different short circuit times
Fig. 4.20 Short circuit current for a 2p4s module at 2.8 V relaxation voltage
The discharge of whole systems makes sense from a technical as well as from an
economic view. However, the system discharge carries the highest risk because
single cells without an active BMU cannot be controlled. The devices TORKEL
860 (variant E) and EA-ELR-9500-30 (variant F) were used to discharge the sys-
tems. The discharge curves with both variants are illustrated for two VW e-Golf
systems in Fig. 4.21.
Figure 4.21 shows that the discharge with a higher power of 15 kW (variant E)
occurs faster than with 3.5 kW (variant F). It is worth mentioning that in the case of
variant F, about 94% of the energy could be fed back into the power grid.
With variant E, a short circuit can already start after 1.5 h, whereas with variant
F, this is only the case after 3.5 h. In order to assess safety, a thermal imaging
camera was also used to measure temperature, as seen in Fig. 4.22.
The photo shows the back part of the battery system because this becomes
warmer than the front. The surrounding temperature is around 9 °C and the
warmest spots reach about 30 °C. The warming-up is clearly visible but is not
dangerous. In order to assess the relaxation behavior in the case of one system, the
short circuit was removed after about 10 min. After only a few seconds, the voltage
jumped to about 30 V, which would be a danger if the system were touched. After
altogether 25 min of relaxation, the voltage reached a value of 130 V which means
an even higher danger if touched. This clearly demonstrates that systems have to be
short circuited over a longer period of time (ideally several days) in order to avoid a
possibly dangerously high relaxation voltage above 60 V.
Influence of pole reversal on relaxation behavior
Pole reversal may cause extreme situations in overdischarging. To obtain a pole
reversal of individual cells, it is not sufficient to use an adjustable resistor for the
load. It is necessary to have an additional current source which displays a constant
discharge current independent of the battery voltage. The following Fig. 4.23
Fig. 4.21 Overdischarge current/voltage characteristic curves for VW e-Golf battery systems with
15 kW (variant E) and 3.5 kW (variant F)
Fig. 4.22 Heat distribution

for a VW e-Golf battery
system during
overdischarging at 9 °C
ambient temperature
Fig. 4.23 Temperature rise and pole reversal for a 25 Ah NCM cell
shows the discharge curves of cell voltage and temperature at a constant 30 A

discharge rate until a pole reversal occurs.
The figure shows that the cell reaches 0 V after 45 min. Subsequently, the cell is
discharged for a further 15 min and then the poles are reversed. The temperature
profiles show that the cell pole warms up significantly (50 °C) at reversal, so that a
longer pole reversal may be hazardous. After 60 min, the measurement was stop-
ped. It was noted that the voltage relaxes for a short moment and then remains at
0 V. In the case of a possible electrical hazard through a high relaxation voltage,
especially at system level, pole reversal may be of assistance.
During discharging modules or systems, a pole reversal of individual cells may
be inevitable. This is one reason why pole reversal needs further investigation.
A typical example is the discharge of cells connected in series, where a cell may
already exhibit lower capacity or a small state-of-health (SOH) value. The fol-
lowing figure demonstrates the discharge of a module (2p4s), where one cell
contains low power (Fig 4.24).
The figure clearly shows that the discharge profile of cell 7/8 drops sharply after
20 min and then reaches the overdischarging range, whereas the other cells remain
in the “normal” range. Since only four cells are connected in series, the voltage drop
observed is also visible in the module voltage profile which shows a profile similar
to the cell voltage profile. The temperature of cell 7/8 also increases significantly,
similar to that seen in the cell measurements, especially in the overdischarging
range. Due to the fact that every two cells are connected in parallel, only half the
current flows, namely 15 A, through each cell. Due to this, at the point of pole
reversal, a heat increase of only 15 K is observed in comparison to 25 K observed
in single cell measurements. In the course of pole reversal, the cell reaches −2 V
before the voltage returns in the direction of 0 V. The increase in cell voltage is
clearly visible in the module voltage profile. After 80 min, the other cells’ voltages
reach 0 V, at which point the discharge process can be terminated.
In this experimental case, pole reversal in combination with progressive dis-
charging is not critical and can, with sufficient expertise, be read from the module
voltage profile. The measurements show that both cell monitoring and temperature
monitoring of individual cells during overdischarging may be useful in preventing
Fig. 4.24 Cell pole reversal and heat rise while overdischarging a 2p4s battery module
critical situations. The behavior of the battery may differ and therefore pose other
hazards for higher discharge currents, other circuits or other cell types.
4.5 Summary and Outlook
When recycling was undertaken in the past, overdischarging did not have to be
carried out due to the pyrometallurgical recycling processes. Overdischarging was
not relevant for furnaces, as the residual voltage or energy remaining in a battery at
combustion were unimportant. For this reason, little research has been done about
overdischarging lithium-ion batteries, as they are more relevant to
non-pyrometallurgical recycling. Overdischarging LIBs is a necessary step in the
LithoRec recycling process. Factors such as safety, ease of handling and efficiency
are the key parameters that need to be considered in the realization of overdis-
charging. Inexpensive variants (A–D) which were developed by the project par-
ticipants as well as commercially available devices (E–F) were tested and validated.
The exploratory work in this project essentially led to the development of a
knowledge base about the behavior of cells, modules and systems when overdis-
charging in various situations. The influence of this behavior on cell properties as
well as the residual hazard potential were also studied. Important parameters in
these investigations on overdischarging behavior are the discharge times/capacities,
the short circuits that arise, the amount of the relaxation voltages, the influence of
the surrounding temperatures and the types of switching as well as behavior during
pole reversal.
All in all, it can be claimed that overdischarging poses no severe hazard, as long
as the discharge current and thus the battery temperatures allowed are not exceeded.
All discharge variants have their advantages and disadvantages and using them all
can therefore be justified.
The relevant questions for the operators of recycling plants are the costs, effi-
ciency and the degree of automation of the process steps. The recommendation here
is discharging a battery system using an electric load with energy recovery (variant
F) and high power, so that the remaining electrochemically stored energy in the
battery can also be “recycled”. The discharge in liquids at cell level has been proved
to be unviable, but as a system, this process showed promising results. The latter
variant, moreover, offers an interesting solution, especially under the aspect of
deactivating vehicles after accidents when the battery is damaged and contacting is
complicated.
Furthermore, it could be demonstrated that with pole reversal, little or no voltage
relaxation takes place. Especially from the aspect of HV safety due to electrical
hazards as well as the low process speed of recycling, the relaxation behavior of a
battery may be of importance. The influence of various discharge profiles or a pole
reversal on the recovery efficiency of individual raw material could not be inves-
tigated within this project. It is assumed that there might be an “ideal discharge
profile” for the optimum recovery of material.
To make discharging batteries more efficient in the future, the following is
recommended. Vehicle manufacturers could implement a release order entitled
“ready for recycling” in their BMUs, so that a monitoring system integrated into a
BMU and a connection to the system’s internal protection mechanism are possible.
Opening a battery system lid and installing HV circuits inside the opened system
may lead to the serious risk of an electric shock and should be avoided. Ideally,
there should be a standardized plug for HV contacts and BMUs as well as a
standardized software application for discharging, so that this process step can be
conducted safely and automatically. Due to the multiple types of battery systems
that are currently on the market, it is unfortunately not expected that this will be
realized in the near future.
References
Bauer S (2017) AkkuWelt‚ Erste Auflage. Würzburg: Vogel Business Media GmbH
Blank T (2012) Deep discharge behavior of lead-acid batteries and modeling of stationary battery
energy storage systems. 978-1-4673-1000-0/12/$31.00 ©2012 IEEE
Gellerich W (2015) Akkumulatoren - Grundlagen und Praxis, Zweite edn. Shaker Media, Aachen
Jossen A, Phan TB, Svoboda V (2016) Tiefentladung von Batterien - Ursachen, Mechanismen,
Lebensdauereinfluss. https://www.mikrocontroller.net/attachment/34643/Tiefentladung.pdf
Jossen A, Weydanz W (2006) Moderne Akkumulatoren richtig einsetzen. Reichard Verlag,
Untermeitingen
Mähliß J (2012) Aufbau, Funktionsweise und Gefährdungspotenzial von Li-Ionen-Zellen.
Elektronik - ecodesign. 2012
Chapter 5
Disassembly Planning and Assessment
of Automation Potentials for Lithium-Ion
Batteries
Felipe Cerdas, Roman Gerbers, Stefan Andrew, Jan Schmitt,

Franz Dietrich, Sebastian Thiede, Klaus Dröder
Abstract Traction batteries are composed of various materials that are both eco-
nomic valuable and environmentally relevant. Being able to recover these materials
while preserving its quality is not only economically attractive, but it can also
contribute to decrease the environmental impact of electric vehicles. Disassembly
can play in this regard a key role. On the one hand it might allow to separate
potential hazardous substances and avoid an uncontrolled distribution of these
substances into other material flows. One the other hand disassembly might pro-
mote improving the rate of material recovered while preserving its quality and
decreasing disassembly costs. In this chapter we present a methodology for the
estimation of disassembly sequences and for the estimation of automation potentials
for the disassembly of traction batteries. The methodology is illustrated with an
experimental case study.
5.1 Introduction
Electric vehicles (EV) are increasingly perceived as a promising strategy towards

reducing the environmental impacts of private transportation (Hawkins et al. 2013).
While the private transportation sector moves gradually from internal combustion
technologies towards full electric vehicles, new challenges arise concerning the
treatment of electric vehicles at the end of their useful life. A higher number of EV
on the roads means as well a higher number of spent batteries that must be treated in
F. Cerdas (&) S. Andrew S. Thiede C. Herrmann

Chair of Sustainable Manufacturing and Life Cycle Engineering, Institute of Machine Tools
and Production Technology, TU Braunschweig, Braunschweig, Germany
e-mail: juan.cerdas-marin@tu-braunschweig.de
R. Gerbers J. Schmitt F. Dietrich K. Dröder
Production Technology and Process Automation, Institute of Machine Tools and Production
Technology, TU Braunschweig, Braunschweig, Germany

https://doi.org/10.1007/978-3-319-70572-9_5
84 F. Cerdas et al.
the near future. Traction batteries are complex devices composed of highly valuable
materials. As the number of traction batteries rises and material costs increase, it
becomes economically attractive to recover the materials contained in these devices.
Being able to recover all or part of these materials is not only economically
attractive, but it is also an important strategy towards reducing the environmental
impact caused by the production of the battery pack in the first place (Dunn et al.
2014).
Historically, disassembly has been mostly executed manually due to the high
complexity in which the products are manufactured and therefore due to the large
number of variants at the end-of-life stages. As a consequence, the cost of disas-
sembly processes in countries with high labor costs has represented an obstacle
because the labor costs outweighs the potential revenues from the recovery of
usable parts and recyclable materials (Duflou et al. 2008). This causes products to
be entirely shredded which usually causes loses of material quality. As argued by
Duflou et al. (2008), disassembly planning is a key issue in defining the fate of the
remaining product’s embodied value. Disassembly can contribute to preserve the
quality of the material recovered, to recover subassemblies in working conditions
that might be remanufactured and re-built in a new product and can promote safer
recycling processes as hazardous components can be properly previously separated
(Duflou et al. 2008).
Thus, the planning of an optimal disassembly planning system for traction
batteries can improve the quality and amount of material recovered while reducing
processing times and costs. In this regard, automated and more flexible solutions for
the disassembly of traction batteries could not only improve the employee’s safety
but also it might contribute to reducing disassembly times as well as operation
costs. Within the overall scope of the LithoRec project, the goal of this chapter is to
determine experimentally the possible disassembly sequences, activities, required
tools and processing times of a disassembly planning for traction batteries. We
present first an overview of the state of the research on planning of disassembly
systems applied in general to electric and electronic devices and to traction bat-
teries. A methodology for the definition of disassembly sequences and the
assessment of automation potentials is next introduced. The application of the
methodology is illustrated through an experimental disassembly process applied to
a traction battery of the hybrid Audi Q5. At the end we present an approach for the
assessment of automation potentials in order to classify the required steps
depending on the degree of automation recommended.
5.2 State of the Art on Planning of Disassembly Systems
At the end-of-life or end-of-use stage different options exist: (direct) reuse,

remanufacturing, refurbishment, repair and material recycling. A reuse of lithium-
ion batteries (cells, modules, systems) can be defined as the direct continuation of
5 Disassembly Planning and Assessment of Automation Potentials … 85
its usage as battery, remanufacturing comprises any action necessary to have

as-new batteries using components taken from previously used battery systems as
well as new components. In distinction, refurbishment of batteries would aim at
batteries may be inferior to a new battery. And repair comprises all actions required
to correct faults in a battery system preventing its specified operation (United
Nations University—Step Initiative 2014). In addition, material recovery aims to
recover the materials of the battery to allow the production of new batteries
(closed-loop) or other products. For all options disassembly is often an important
process step, beside other such as cleaning, inspection, reprocessing, test,
reassembly and storage for instance in the case of remanufacturing (Sundin 2004).
Although there is numerous research done for each of these steps mentioned, much
of the research has been focused on the development of methodologies and tools for
the definition of the most economically attractive disassembly process. Several
question have been actively researched in this regard: (i) the definition of the
suitable object to be disassembled, (ii) the definition of the order, or sequence, in
which the different disassembly steps have to take place and (iii) the required or
recommended depth of disassembly based on economic and or environmental KPIs.
Particularly for the last two points, graph-theoretical approaches have been
widely applied. The research done by Scheuerer (1995), Johnson and Wang (1998),
Rudolph (1999) and Veerakamolmal (1999), present different sketches of possible
disassembly conditions for a particular range of products. These are afterwards
evaluated and classified regarding their costs and revenues. In a similar work,
Moore et al. (1998a, b) developed an approach for the planning of disassembly
procedures using Disassembly petri-nets automatically generated based on
disassembly-sequencing-matrixes. Using a heuristic together with a cost function
the number of possible solutions to a disassembly sequence is reduced. The applied
cost function considers changes in tools, directions, movement processes, variations
on the conditions of the products being dismantled as well as the existence of
dangerous substances inside the product. The algorithm selects for the part being
disassembled the process with the lowest disassembly cost. Zhou and Zussmann
(1998) presents an approach for the development and implementation of disas-
sembly programs and process also based on the use of petri-nets. Their approach
allows to find the disassembly system generating the highest revenues. In their
research they use Bayesian networks to deal with the uncertainties brought by the
interaction between the different conditions of the spent products being under
dismantling and the range of possible disassembly processes. In such a way, the
methodology developed by the authors enables the disassembly process to react and
adapt dynamically towards possible abrupt changes within the disassembly oper-
ations. The approach developed was tested using radios as an example.
Günger and colleagues (Güngör 1999; Güngör et al. 2000) present an approach
and a model for the analysis and improvement of disassembly processes and sys-
tems. Using CAD-Data, the algorithm developed generates geometry-based disas-
sembly sequence matrixes. These are afterwards used to sketch automatically a
disassembly process. An extension of the model including a Branch-and-Bound
86 F. Cerdas et al.
algorithm generates automatically a disassembly system under consideration of the

condition of the product being disassembled. For the balancing of the disassembly
line considering different assumptions and scenarios, a heuristic tool was developed
that contributes improving the material flow of the disassembly system. The
heuristic tool is based on a priority function that identifies and recommends a
suitable sequence of the required disassembly processes.
The approach presented by Mascle (2002) introduces a methodology to define
the optimal sequence of disassembly steps. The focus of the methodology lies on
enabling the reusability and recyclability of the components and the material
recovered. In this regard, the algorithm for disassembly planning developed
includes stability factors, geometries, parallel operations and irreversible operations.
Further researches by Kühn (2001) and Herrmann (2003), present methodologies
for the estimation of depth of dismantling and disassembly sequencing aiming at
supporting product development processes. The product models developed consist
on a product model composed of parts information such as materials, forms and
sizes, joining technologies and sequencing relationships. The model allows to
identify the optimal depth of disassembly as well as the most suitable sequencing of
steps in order to increase material recovering quotes and decrease endo-of-life
processing costs.
Kriwet (1995) and Lambert (2003) present extensive literature surveys on the
topic of disassembly sequencing for the identification and planning of suitable
disassembly processes and lines of electric and electronic products. The approaches
described define optimal disassembly sequences through the application of
heuristic, meta-heuristic or exact methods. By means of heuristic and meta-heuristic
methodologies disassembly sequences it is possible to identify suboptimal disas-
sembly sequences with a relatively high computational effort and time. Research by
Vongbunyong and Chen (2015) presents a detailed review of the different aspects
influencing the development of disassembly systems and its success. In their
research they review current available and emerging technologies that are expected
to ease the transition towards more flexible and automated disassembly systems.
Since lithium-ion battery systems contain a lot of valuable materials (e.g.
lithium, cobalt, copper, aluminium) several approaches have been developed to
recover these materials (Swain et al. 2007). However, these processes have been
developed for the recycling of small batteries from portable applications, such as
laptops, smartphones, or cameras (Lupi et al. 2004; Nan et al. 2005). Due to the
small size and low complexity of these batteries, the first step of the recycling
process was always the shredding of these batteries. Therefore, findings from these
studies are only partially transferable for large-scale automotive battery systems. In
particular, the size, weight and the complex design of automotive lithium-ion
batteries require additional disassembly and handling steps to realize an economic
recycling process. In contrast to directly shredding of the entire battery, disassembly
enables the extraction of the valuable lithium cells, hazardous substances as well as
reusable subassemblies e.g. the battery management Unit (BMU) (Herrmann 2003).
However, a high degree of cost-efficiency is required for an economic recycling process.
Since manual work steps are time and cost intensive a high degree of automation is
necessary to achieve a high cost-efficiency. Due to the high voltages of lithium-ion
batteries process automation can also improve the safety of employees. Due to high
development costs and technical restrictions, it is not necessarily reasonable to
automate all process steps. For the particular case of traction batteries, Herrmann
et al. (2014) present a scenario base methodology for the development of disas-
sembly processes for traction batteries. In their approach they make use of mor-
phological boxes to identify the best alternative for each disassembly scenario and
based on that the most suitable disassembly system and sketched. Wegener et al.
(2014) executed a complete disassembly of a traction battery system using as an
example a hybrid system from an Audi Q5. The disassembly sequences in their
research were derived from a priority matrix disassembly graphs.
5.3 Disassembly System for Li-Ion Traction Batteries:

Experimental Analysis, Planning and Assessment
of Automation Potentials
Within the project LithoRec,1 the methodology followed for the planning of dis-
assembly activities for traction batteries as well as for the assessment of the
automation potential consist of five steps as shown in Fig. 5.1. As a first step, a
product analysis is carried out to determine joining technologies, used materials and
disassembly priorities. In the next step, the obtained information is used to deter-
mine an optimal disassembly sequence. Afterwards the automation potential is
derived using a criteria checklist. Two key parameters, which identifies the
automation capability and the necessity of automation are used to calculate a score
of the automation potential for each disassembly operation. In the last step, the
calculated scores are used in a portfolio analysis to decide whether operations
should be automated, semi-automated or carried out manually and to prioritize
operations that should be automated first.
The disassembly of traction batteries is a complex task. A battery pack not only
might present very different designs and configurations, but also a battery system is
composed by a wide range of flexible components such as cables and dangerous
chemicals inside the battery cells that represent a safety issue when planning a
disassembly line for a traction battery. In this chapter we illustrate the application of
the approach shown in Fig. 5.1 for the case of disassembling completely the battery
system of an Audi Q5 as presented by Wegener et al. (2014).
1
This section is mostly based on the results presented in the research of Wegener et al. (2014)
which correspond to part of the work done within LithoRec.
88 F. Cerdas et al.
Fig. 5.1 Methodical approach for the definition of a disassembly system and the assessment of
automation potentials
5.3.1 Product Analysis
The basis for the product analysis are disassembly experiments carried out on
various battery systems. Standardized forms are used to document the results of
these experiments. During such a product analysis it was for example determined
that automotive lithium-ion battery systems have a similar set-up regardless of the
manufacturer. The information obtained in this step is therefore generic enough to
be used to estimate the automation potential for a wide variety of automotive
battery systems.
All of the disassembled systems in LithoRec, consisted of a battery management
system, battery modules and additional power electronics with similar joining
technologies. A sketch of the battery pack build in the Audi Q5 is shown in
Fig. 5.2. As seen, the battery pack consists of modules connected within the
electrical circuit. The modules contain in turn the battery cells connected in series.
Additionally, the battery system includes a Battery Management Unit BMU, a
cooling system, a set of elements conforming the power electronics.
Fig. 5.2 Setup of the battery system used in the hybrid Audi Q5
Fig. 5.3 Main components of the Audi Q5 hybrid battery system (closed system)
This battery pack has a nominal weight of 35 kg and its dimensions are
50 cm 70 cm 15 cm. The nominal capacity of the battery is 5 Ah and it has a
nominal voltage of 266 V. Each of the modules contain 18 cells with a nominal
voltage of 3.7 V. All the elements of the battery pack are protected by an insulated
casing system as shown in Fig. 5.3.
The modules in turn are packed in plastic casings and the battery cells are held
by a structural system which keep the cells fixed to each other. Flexible components
such as cables run through the cells in the modules connecting cells in series and
modules in parallel as can be appreciated in Fig. 5.4.
In both Figs. 5.3 and 5.4 the main disassembly elements identified for this case
are shown. These elements are labeled and classified in the next section. Together
with the data gathered regarding materials, weight, form and sizes, data including
information about on the disassembly tools used, the disassembly priorities, dis-
assembly time, as well as the necessary dexterity and forces for disassembly of the
battery system has to be considered.
90 F. Cerdas et al.
Fig. 5.4 Main components of the Audi Q5 hybrid battery system (open system)
5.3.2 Determination of Disassembly Sequences
A four steps procedure was followed in this part. In a first steps, the elements to be
disassembled are to be listed considering the information gathered in the previous
sections. As mentioned by Wegener et al. (2014), for the present case study we
considered a disassembly process starting a system (pack) level and ending at a
module level as the aim of the disassembly system is to obtain the modules.
Modules are afterwards crushed without being previously disassembled nor sepa-
rated due to the high economic costs that this would have otherwise implied. The
majority of the experimental disassembly was performed manually using hand
tools. The list of the disassembly elements is shown in Table 5.1.
Once the disassembly elements have been identified we proceed to identified the
sequence of the disassembly steps. To do this, a disassembly matrix was con-
structed in order to be able to identify precedence relationships between the dif-
ferent elements in the list. Figure 5.5 shows the resulting disassembly matrix for the
case under study. As explained by Wegener et al. (2014), a one to one comparison
of each of the elements is executed evaluating which of the elements needs to be
disassembled before to other one. The successor is given a minus one whereas the
predecessor is given a 1. In case it can not be determined which of the elements
needs to be disassembled first, then both of the elements are given a zero. So for
example when confronting the BMU (4) against the power electronics (5) the result
is a zero as it makes absolutely no difference to dismantled one element before the
other one and viceversa.
Once every element has been evaluated all the positive values in the rows are
summed so as to have an aggregated positive result for each of the disassembly
elements. As a final result, the higher the result of the summatory, the higher the
Table 5.1 List of Part number Part name

disassembly elements
1 Power electronics cover
2 System casing cover—side
3 Cable tie
4 Battery management unit (BMU)
5 Power electronics
6 System casing cover—top
7 Cable guiding
8 Gas venting
9 Cover of modules
10 Connector between modules
11 Modules fastener
12 System casing cover—bottom
13 Modules holder
14 Modules
92 F. Cerdas et al.
Successor
Priority matrix dismantling Total Rank
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 power electronics cover 1 1 1 1 1 1 1 1 1 1 1 1 1 13 1
2 system casing cover - side -1 1 1 1 0 1 1 1 1 1 1 1 1 11 2
3 cable tie 1 1 1 1 1 1 1 0 1 0 1 1 1 9 3
4 battery management unit (BMU) -1 -1 -1 0 1 1 1 1 1 1 1 1 1 9 3
5 power electronics -1 -1 -1 0 1 1 1 1 1 1 1 1 1 9 3
Predecessor
6 system casing cover - top -1 0 -1 -1 -1 1 1 1 1 1 1 1 1 8 6

7 cable guiding -1 -1 -1 -1 -1 -1 1 1 0 1 1 0 1 5 7
8 gas venting -1 -1 -1 -1 -1 -1 -1 0 0 1 1 1 1 4 8
9 cover of modules 1 1 0 1 1 1 1 0 0 0 1 1 1 3 9
10 connector between modules -1 -1 -1 0 -1 -1 0 0 0 0 1 1 1 3 9
11 modules fastener -1 -1 0 -1 -1 0 -1 -1 0 0 1 1 1 3 9
12 system casing cover - bottom -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 1 1 2 12
13 modules holder 1 1 1 1 1 0 1 1 1 1 1 1 1 1 13
14 modules 1 1 1 1 1 1 1 1 1 1 1 1 1 13 14
Fig. 5.5 Disassembly matrix for the Audi Q5 hybrid battery system (Wegener et al. 2014)
rank in the disassembly system and therefore the sooner will the element be dis-
assembled from the rest.
The following step to determine the disassembly sequence is to identify the
necessary operations in each of the steps. Considering that the objective of the
disassembly experiment is to gathered the battery modules, the 24 steps shown in
Table 5.2 were identified as necessary. As described by Wegener et al. (2014) the
numbers in brackets refer to the disassembly elements in Table 5.1 and in Fig. 5.4.
A quick look to the table shows that much of the disassembly operations can be
executed without the use of sophisticated tools as almost all of the elements of the
battery system are accessible by means of screw drivers or side cutters (for the case
of the cables holders).
In order to obtain a clearer picture of the sequence plan, a disassembly priority
graph is sketched as shown in Fig. 5.6. This kind of representation of the disas-
sembly process helps to understand the relationship between the different operations
to be executed with the predecessor operation. These operations are then combined
if applicable and ordered according to the disassembly matrix in Fig. 5.5. The graph
in Fig. 5.6 represents the information of the sequence of disassembly operations
recommended for the case of the Audi Q5.
As shown in Fig. 5.6, it is firstly recommended to release the screws on the top
and the bottom side of the battery casing. Following operations that take place are
the removal of the casing of the power electronics and the BMU. Once the top
casing and the electronic systems have been removed, it is recommended to proceed
removing all the cables, cable holders and the different plug connections between
the cell controllers and the BMU. Next, the gas venting as well as the different
connection systems and holders attached to the modules might be removed. Once
the module holders have been removed, the modules can be retired from the bottom
casing. As a final step, the modules fasteners can be unscrewed and the modules can
be removed.
Table 5.2 Disassembly steps and necessary tools (Wegener et al. 2014)
Disassembly Disassembly step Necessary tool
step number
I Unscrew covers (1), (6) and casing bottom (12) Screw driver
II Removal of the power electronics cover (1) Hand
III Checking of the state of charge Special
measuring device
IV Removal of the side covering (2) Hand
V Disassembly of the live lines from the modules/ Screw driver
stacks (14)
VI Cutting of the cable ties (3) Side cutters
VII Disassembly of the plug connection between the Hand
cell controllers and the BMU (4)
VIII Removal of the BMU (4) Hand
IX Removal of the power electronics (5) Hand
X Cutting of the bus for the thermo sensors (6) Side cutters
XI Disassembly of the system cover (6) Screw driver
XII Removal of the system cover (6) Hand
XIII Unscrew of the cable guiding (7) Screw driver
XIV Removal of the cable guiding (7) Hand
XV Removal of the gas venting (8) Hand
XVI Removal of the cover of stacks (9) Hand
XVII Disassembly of the connectors between the stacks (10) Screw driver
XVIII Removal of the connectors between the stacks (10) Hand
XIX Unscrew of the stacks holders (11) Screw driver
XX Removal of the stacks holders (11) Hand
XXI Removal of the casing bottom (12) Hand
XXII Unscrew of the stack fastener (13) Screw driver
XXIII Removal of stack fastener (13) Hand
XXIV Removal of stacks (14) Hand
Fig. 5.6 Disassembly priority graph (Wegener et al. 2014)

94 F. Cerdas et al.
As seen, and discussed by Wegener et al. (2014), removing cable guides, con-
nectors between the modules and the different fasteners holding the modules could
be combined into one disassembly operation step.
5.3.3 Assessment of Automation Potentials and Concepts

for Automated Disassembly
After the product analysis has been carried out and the disassembly sequences have
been derived the automation potential can be identified using a criteria catalogue.
Table 5.3 Criteria catalogue (Herrmann et al. 2012)

Description of disassembly operation Assignment to key
indicators
Process-related Number of disassembly motions NA
criteria Complexity of disassembly motion(s) TAA
Accessability of the join patch/the part to TAA
disassembly
Automation potential of the disassembly tool TAA
Joining type NA
Joining technique TAA
Loosening of the joint NA
Accuracy of the positioning of the disassembly tool TAA
[+/− mm]
Suspectance to failure TAA
Duration of manual disassembly NA
Safety- related Direct involvement of hazardous material NA
criteria Disconnecting of current carrying cables NA
Distance to an electrical potential NA
Legally required auxiliary material for manual NA
disassembly
Part-related Geometrical workpiece dimension (bounding NA
criteria geometry)
Weight NA
General characterisation TAA
Specific form elements (defined purchase for TAA
disassembly tool)
Number of specific form elements (defined TAA
purchase for disassembly tool)
Properties during force influence TAA
Part behaviour after disassembly TAA
econ.Criteria Economical priority of automation NA
Equipment costs (rough estimation) NA
Four groups of criteria (see Table 5.3) which identify different characteristics of the
disassembly operations are used to rate the automation potential. Each of the criteria
contributes to one of two key indicators the ‘technical ability of a disassembly
process to be automated’ (TAA) or the ‘necessity to automate the corresponding
disassembly operation’ (NA). The criteria are rated using a quantitative value or
qualitative attribute and the result is then multiplied by a weighting factor for each
criterion. A pairwise compression, carried out by an expert team, is used to define
the weighting factors. Finally, the results of each criterion is summed up into a final
score for NA and TAA.
5.3.4 Classification of Disassembly Sequences
The final scores for the automation capability (TAA) and the necessity of
automation (NA) for each disassembly operation are used in a portfolio analyses to
cluster the automation potentials (see Table 5.3). By means of the portfolio anal-
ysis, it can be identified whether a process can be automated with reasonable effort
and whether automation is necessary for an economical disassembly. For example,
it can be concluded from the example in Fig. 5.5, that the operations in field 2
should be automated with a high priority. However, the operations in field 2 should
be automated with a lower priority and all other operations should initially be
excluded from automation. Additionally, a bar chart is used to show the automation
potential for individual operations in relation to the disassembly sequences
(Fig. 5.7). With this process-oriented representation it is possible to identify
operations out of a disassembly sequences with high automation potential.
Fig. 5.7 Portfolio analysis (left) and bar chart for process oriented visualisation (Herrmann et al.
2012)
96 F. Cerdas et al.
5.4 Summary
In this chapter we presented a case study in which we experimentally disassembled

a traction battery. Using as an example the battery system of the Audi Q5, we
followed a methodology to determine possible disassembly sequences in order to
remove the battery modules from the battery pack. We first executed a product
description in which the most important disassembly elements were identified.
Next, through the application of sequence matrixes we ranked the different ele-
ments. Disassembly operations were afterwards described and ordered using a
disassembly priority graph. Finally, we described a method for the assessment of
automation potentials for disassembly of lithium-ion batteries. Two key indicators
are derived for each disassembly operation and presented in a portfolio analysis as
well as in a process-oriented diagram. In this context, recommendations for the
design of the automation system as well as for possible design changes of the
battery system can be made. The extended reference to the entire disassembly
process provides further information for disassembly to support the layout process.
The process orientation of the methodology complements the already established
planning processes of assembly/disassembly lines in relation to Lotter or REFA
(Hartel and Lotter 2006). The method can be used to provide additional information
for rough estimation of automation potentials. This supports the decision making
process, whether a complete or semi-automatic system should be realized.
References
Duflou JR, Seliger G, Kara S, Umeda Y, Ometto a, Willems B (2008) Efficiency and feasibility of
product disassembly: a case-based study. CIRP Ann Manuf Technol 57(2):583–600. https://
doi.org/10.1016/j.cirp.2008.09.009
Dunn JB, Gaines L, Kelly JC, James C, Gallagher KG (2014) The significance of li-ion batteries in
electric vehicle life-cycle energy and emissions and recycling’s role in its reduction. Energy
Environ Sci 8:158–68. Royal Society of Chemistry. doi:https://doi.org/10.1039/C4EE03029J
Güngör A (1999) Algorithmic models for analysis of disassembly processes and systems. PhD
thesis, Massachusetts
Güngör A, Gupta SM, Pochampally K, Kamarthi SV (2000) Complications in disassembly line
balancing. In: Proceedings of the SPIE international conference on environmentally conscious
manufacturing, vol 4193. pp 289–298
Hartel M, Lotter B (2006) Planung und Bewertung von Montagesystemen. In: Lotter B,
Wiendahl HP (eds) Montage in der industriellen Produktion (chap 13). Springer, Berlin,
pp 407–431
Hawkins Troy R, Singh Bhawna, Majeau-Bettez Guillaume, Strømman Anders Hammer (2013)
Comparative environmental life cycle assessment of conventional and electric vehicles. J Ind
Ecol 17(1):53–64. https://doi.org/10.1111/j.1530-9290.2012.00532.x
Herrmann C (2003) Unterstuetzung der Entwicklung recyclinggerechter Produkte. Vulkan-Verlag,
Essen
Herrmann C, Raatz A, Mennenga M, Schmitt J, Andrew S (2012) Assessment of automation
potentials for the disassembly of automotive lithium ion battery systems. In: Dornfeld D,
Linke B (eds) Leveraging technology for a sustainable world. Springer, Berlin, Heidelberg
Herrmann C, Raatz A, Andrew S, Schmitt J (2014) Scenario-based development of disassembly

systems for automotive lithium ion battery systems. Adv Mater Res 907:391–401. https://doi.
org/10.4028/www.scientific.net/AMR.907.391
Johnson M, Wang M (1998) Economical evaluation of disassembly operations for recycling,
remanufacturing and reuse. Int J Prod Res 36(12):3227–3252
Kriwet A (1995) Bewertungsmethodik für die recyclinggerechte Produktgestaltung. Dissertation,
Berlin, p 85ff
Kühn M (2001) Demontage- und recyclingorientierte Bewertung. Dissertation, TU Braunschweig,
Schriftenreihe des Instituts für Werkzeugmaschinen und Fertigungstechnik, Vulkan-Verlag,
Essen
Lambert AJD (2003) Disassembly sequencing: a survey. Int J Prod Res:37–41 (November 2014)
Lupi C, Pasquali M, Dell’Era A (2004) Nickel and cobalt recycling from lithium-ion batteries by
electrochemical processes. Waste Manag 25:215–220
Mascle C (2002) A system life-cycle model for disassembly-assembly line design. In: Proceedings
of the 15th triennial world congress of the international federation of automatic control,
Barcelona, 21–26 July 2002
Moore KE, Gungor A, Gupta SM (1998a) Disassembly process planning using petri nets. In:
Proceedings of the IEEE international symposium on electronics and the environment, Oak
Brook, pp 88–93
Moore KE, Gungor A, Gupta SM (1998b) Disassembly petri net generation in the presence of
XOR precedence relationships. In: Proceedings of the IEEE international conference on
systems, man, and cybernetics, San Diego, California, pp 13–18, 11–14 Oct 1998
Nan J, Han D, Zuo X (2005) Recovery of metal values from spent lithium-ion batteries with
chemical deposition and solvent extraction. J Power Sources 152:278–284
Rudolph A (1999) Altproduktentsorgung aus betriebswirtschaftlicher Sicht. Physica-Verlag,
Heidelberg
Scheuerer A (1995) Beiträge zur Steuerung des betrieblichen Recyclings unter besonderer
Berücksichtigung eines Informationssystems zur Unterstützung von Demontageprozessen.
Dissertation, Universität Erlangen-Nürnberg
Sundin E (2004) Product and process design for successful remanufacturing. Dissertation,
Linköpings Universitet
Swain B, Jeong J, Lee J-C, Lee G-H, Sohn J-S (2007) Hydrometallurgical process for recovery of
cobalt from waste cathodic active material generated during manufacturing of lithium ion
batteries. J Power Sources 167(2):536–544
United Nations University—Step Initiative (2014) “Solving the E-waste problem (step) white
paper: one global definition of E-waste, June, United Nations University. http://collections.unu.
edu/view/UNU:6120#.WcfUdFj4Kpw.mendeley
Veerakamolmal P (1999) Design and analysis of disassembly and remanufacturing systems in the
electronics industry. PhD thesis, Massachusetts
Vongbunyong S, Chen WH (2015) Disassembly automation. Springer International Publishing,
Cham. https://doi.org/10.1007/978-3-319-15183-0
Wegener K, Andrew S, Raatz A, Dröder K, Herrmann C (2014) Disassembly of electric vehicle
batteries using the example of the Audi Q5 hybrid system. Procedia CIRP 23(C):155–60. doi:
https://doi.org/10.1016/j.procir.2014.10.098
Zhou MC, Zussmann E (1998) Design and implementation of disassembly petri nets for intelligent
process planning. In: Technical report 0108-98, Multilifecycle Engineering Research Center,
NJIT, 20 July 1998
Chapter 6
Safe, Flexible and Productive
Human-Robot-Collaboration
for Disassembly of Lithium-Ion Batteries
Roman Gerbers, Kathrin Wegener, Franz Dietrich and Klaus Dröder
Abstract The rising number of electric vehicles will lead to an increase of EV

batteries reaching their end-of-life. Efforts are therefore being made to develop
technologies and processes for recycling, remanufacturing and reusing EV batteries.
One important and necessary step for the recycling process is the disassembly of
EOL EV batteries. Unpredictable lot sizes and volumes, as well as significant
variations in battery design between different car models challenges the disas-
sembly automation. Disassembly is therefore currently carried out manually. Fully-
automated disassembly would require high product specific investments which is
not economically feasible in changing production environments. Human robot
collaboration aims to overcome those problems with partial-automation by incor-
porating sensor integrated robotics in more fields of human activity. This chapter
presents the implementation of human robot collaboration for disassembly of
lithium-ion Batteries. While the human operator performs the more complex tasks,
the robot performs simple, repetitive tasks such as removing screws and bolts. An
intuitive programming environment, which does not require experience in robot
programming, is combined with cost efficient tooling and additional 3D safety
sensors to realize a safe, productive and ergonomic workspace.
6.1 Introduction
Advances in lithium-ion battery technology have increased the practicality and

attractiveness of electrically driven vehicles, leading to an increase in development
activities and production of electric vehicles (EV) and Li-ion batteries respectively.
Only recycling of these batteries can be a truly sustainable approach of dealing with
these batteries at their end-of-life (EOL). The realization of such a recycling pro-
cess, as proposed in “LithoRec II—Recycling of Lithium-Ion Batteries II”,
R. Gerbers (&) K. Wegener F. Dietrich K. Dröder

Chair of Production Technology and Process Automation, Institute of Machine Tools
and Production Technology, TU Braunschweig, Braunschweig, Germany
e-mail: r.gerbers@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_6
100 R. Gerbers et al.
requires the disassembly of batteries. In contrast to shredding the entire battery

directly, disassembly enables the separation of the valuable lithium cells from other
components. This ensures a high quality of the recovered raw materials and allows
intact subassemblies to be reused (Herrmann et al. 2012).
Due to small volumes and lot sizes, EV batteries are currently dismantled
manually. The projected volume considered, it can be foreseen that disassembly
must be rationalized in the future by increasing the degree of automation. This is a
critical factor for improved efficiency of the entire recycling process and thus for the
economic performance.
Previous results have shown that a fully automated disassembly is not feasible at
the moment (Herrmann et al. 2012), due to many product variants, non-existent
standards in battery design and the fact that the detailed designs of the batteries are
generally unavailable to the recycling company. Full automation will only be
recommendable if the standardization of the battery systems is well advanced and
the diversity of variants has been greatly reduced. Therefore, a partial automation
with the use of human robot collaboration was investigated for this project to
increase the efficiency compared to manual disassembly. Human robot collabora-
tion has gained a significant research interest in recent years. This is due to a large
variety of opportunities that come with the implementation of hybrid workspaces
where robots are not separated by safety barriers and are collaborating with human
operators. Such workstations allow the implementation of a flexible and reconfig-
urable production system, where humans carry out the tasks that require complex
sensing and motions and robots handle monotonous repeated and unergonomic
tasks. In this way a production system is achieved that can easily be adapted for
different product variants and batch sizes. The advantages of such a production
system contribute especially for industries with small scale productions and a wide
product variety. Therefore, human robot collaborations offer a great potential
especially for the disassembly of automotive battery systems to cover the wide
variety of different manufacturers and battery designs.
The aim of the presented project is to demonstrate methods to develop hybrid
workspaces, which leverage the advantages of partial automation with human robot
collaboration and their application for the disassembly of battery systems.
A prototypic implementation is presented to investigate the feasibility of the
developed concept. However, further industrialization of the demonstrator is nec-
essary for example, to increase the reliability and productivity or lower the costs for
hard- and software before it can be implemented in a series application.
The scope of this project is the development of concepts for increased flexibility
and productivity for human robot collaboration. Technologies for intuitive robot
programming and control are combined with suitable safety technology to develop
a concept that does not require conventional robot programming. This increases the
productivity and cost-efficiency of hybrid workspaces since conventional pro-
gramming of robots is a time-consuming and complex process for which trained
specialists are necessary. With these approaches, a disassembly workstation concept
has been developed which meets the requirements of a high diversity of product
variants combined with high productivity and profitability.
6 Safe, Flexible and Productive Human-Robot-Collaboration … 101
6.2 Background
This section provides an overview of technologies of relevance for partial

automation of disassembly of lithium-ion batteries using human robot collabora-
tion. This includes recent applications, safety technologies and hardware for human
robot collaboration as well as sensor technologies, algorithms and tools for auto-
mated disassembly.
6.2.1 Technologies and Applications for Safe Human Robot

Collaboration
One of the biggest challenges of realizing a hybrid workstation, where a human

operator works close to a moving robot, is to establish a certifiable safety tech-
nology without sacrificing productivity. Due to the high risk of collisions and the
associated injuries of moving robots, the safety of human operators has always been
the highest priority in human robot collaboration. For this reason, the productivity
has previously not been in the focus for the development of marketable
safety-critical systems. State of the art robotic systems enable a safe human robot
collaboration particularly through the use of sensitive force/torque sensors in the
drive units of the robot axes. The torque of every axis is constantly measured during
robot motion to detect collisions. A difference between the measured torque and a
variable with the acceptable value stored in the control system directly triggers the
robot’s safety circuit (Luca and Flacco 2012). This technology requires the con-
sideration of the robots braking distance and braking time in order to prevent
injuries of the human operator. Therefore, only low velocities can be permitted and
the productivity is decreased significantly compared to conventional robotic
workspaces. In addition, robots can be found on the market, that include vision
sensors, 3D area sensors (Fanuc 2015) or tactile sensor skins (Bdiwi 2014) which
can be used for the detection of parts, for visual servoing (Cherubini et al. 2015) or
as additional safety feature. A recent development, which has not yet been
implemented in industrial robots, are safety features based on capacitive sensors
(Hoffmann et al. 2015). This safety feature enables the detection of a collision at an
earlier stage. Thus, faster movements can be allowed in collaborative workspaces.
Not all sensors that have been developed for human robot collaboration are directly
integrated into the robot. For some applications, additional vision sensors are
integrated into the workstation to enhance the scope of application. Such systems
can be used to detect humans and obstacles in a larger area of the workspace and to
adapt the robot’s velocity dynamically according to the distance of the object.
Additionally, such sensors can be used to implement online path planning to avoid
collisions in advance by adapting the robots motion sequence (Flacco et al. 2014;
Wang et al. 2013). These technologies demonstrate that a safe and certifiable human
robot collaboration is already feasible. However, the technologies available on the

market are associated with a loss of productivity, which indicates the need for
further research to develop a certifiable safety technology without sacrificing
productivity.
In recent industrial applications, robots are mainly used to assist workers in
assembly operations. For example a MRK-Systeme GmbH “KR5SI” robot that is
used at Audi AG for handling coolant expansion tanks (Audi AG 2015) or UR 5
robots that are used at the Volkswagen plant in Salzgitter for cylinder head
assembly (Heyn and Krüger 2015), UR 10 robots that are used at BMW for col-
laborative door sealing (Knight 2014) or Bosch „APAS“ robots used for handling
actuators and sensors at Volkswagen Sachsen (Heyn and Krüger 2015). These
examples show that human robot collaboration is already being used on an
industrial scale, but that the applications are limited to simple repetitive tasks. This
is due to the required experience for robot programming since flexible applications
would require a frequent reprogramming of the application. Further research is
therefore necessary for reduced effort of robot programming to enable human robot
collaboration for applications that require a higher flexibility.
The demonstrator developed in this project for disassembly of battery systems
incorporates 3D-Vision technology combined with an online image processing as a
safety system concept for increased productivity. Additionally, the 3D-Vision
systems is used to implement an intuitive human machine interface that does not
require conventional robot programming. Thus, a concept is demonstrated that can
meet the requirements for high productivity and flexibility.
6.2.2 Technologies for Automated Disassembly
One essential aspect for the realization of a disassembly workstation is the auto-
mated disassembly of different joining types (e.g. adhesive, click or screw con-
nections). Screw and bolt connections are standardized components and are
therefore found in a multitude of battery systems. Thus, the realization of an
automated unscrewing is highly relevant for the disassembly of lithium-ion bat-
teries. Nave for example developed a robotic system for unscrewing, where the
inaccurate visual localization of screws was compensated using a compliance
mechanism (Nave and Jünemann 2003). Büker et al. (2001) developed a system for
unbolting car wheels, whereby an active stereo vision system was combined with a
force-torque sensor and a task-planning module for disassembly of the screws.
A contour analysis and generalized hough transformation was used to identify the
position of the bolts followed by template matching for increased precision of the
detection algorithm The detection algorithm yielded detection rate of 98%. Gil et al.
(2007) developed a multi-sensor robotic system for unscrewing and removing of
electronic components from consumer products. A combination of visual servoing
and force control was used to detect screws from a camera image and remove them
with a robot. The detection of the screws in the camera images was achieved using
the Douglas-Peucker algorithm and template matching. Cruz-Ramírez et al. (2011)
developed a vision based system for robotic disassembly of ceiling beams.
A template matching algorithm was used to localize the screw position in combi-
nation with a support vector machine that integrated information from multiple
images. This method resulted in a reduced number of false positives. However, a
desired high reliability could still not be achieved. Vongbunyong et al. (2013)
approached the problem of a low reliability of camera-based detection with the Haar
Cascade method. This method utilizes a decision tree that classifies each region
based on features that can be mathematically calculated with low effort (Viola and
Jones 2001). The approach achieved a promising detection rate of 98% in a testing
environment but significantly decreased to 64% after application in an actual dis-
assembly scenario. The greatest issue was the influence of light reflections, which
led to an increased occurrence of false positives. Further approaches with increased
reliability for automated disassembly, rely on the availability of product information
or a combination of product data and computer vision (Gil et al. 2007; Büker et al.
2001; ElSayed et al. 2012). However, product structure data is rarely available or
takes considerable effort to create. Further research is therefore necessary in order to
develop a system that can deal with the low reliability of computer vision and that
does not require product structure data.
The automated disassembly workstation developed in this project incorporates
manual support from a human operator to overcome the problem of unreliable
computer vision without the need for product structure data. The operator can verify
the computer vision process, check for false positives and has the opportunity to
directly teach new positions to the robot. This can be done intuitively by unskilled
workers in a short time on the factory floor, has reduced software requirements and
can be easily integrated into an existing manual disassembly setup.
6.3 Requirements for the Development of Human

Robot-Collaboration Systems
This section elaborates the requirements for successful implementation of appli-

cations for human robot collaboration. This poses the foundation for the develop-
ment of the automated disassembly application for this project. The development of
workstations for human robot collaboration requires the contemplation of aspects
and constraints that go beyond the considerations for conventional robotic work-
stations. The proximity of the human operator to the robot leads to problems and
hazards that must be considered in order to implement an application economically,
ergonomically and safely. In comparison to conventional applications, this adds
effort that must be evaluated while planning potential applications. In particular, the
considerations for robot acceptance, safety requirements and the control architec-
ture as well as the associated risks and economic influences are discussed.
6.3.1 Robot Acceptance
The integration of collaborative robot into existing manual workstations always

involves the placement of a robot close to the human operator. This proximity of a
moving and working robot to a human operator can lead to fear and discomfort
resulting in acceptance problems of the application. The operator can be afraid of
injuries or he can fear the possibility that his workplace could be completely
automated in the future. Such fears can lead to acceptance problems and provoke an
avoidance of the robotic functions or even lead to situations where the human
operator manipulates the robots functionality. The operator’s acceptance for the
robot is therefore a key factor for a successful implementation of hybrid worksta-
tions. The individual characteristics of the operator, such as age, sex, previous
experience as well as the cultural attitude mainly influences the robot acceptance
(Wu et al. 2014). For example, there is a skeptical attitude towards robots in
Europe, whereas in Japan there is a high degree of acceptance, which is caused in
particular by educational influences (Kaplan 2004). Special attention must therefore
be given to the particular characteristics of the employees in terms of communi-
cation, training and support (Wu et al. 2014). Technical and organizational factors
have been defined which have a positive or negative impact on robot acceptance.
There is a negative impact on process security, legal certainty with regard to data
protection and liability as well as the subjectively perceived occupational safety
(Ezer et al. 2009). The negative effects can be counteracted by the individual benefit
of the new system component. If the integration of the robot leads to a workload
reduction and to an increase in flexibility and spontaneity for the operators, their
individual benefit and thus the acceptance will be high. At the same time, a
reduction of the abilities of the employee should be avoided and a good usability
should be ensured. A positive factor can also be achieved by the humanoid design
of the robots. Such a design can support familiarity, well-being and therefore result
in positive reactions of the operator. For those reasons, human factors can make a
decisive contribution to a successful implementation of a desired application for
human robot collaboration and should therefore be considered during the entire
development process. In this project, the robot acceptance was taken into account
by the implementation of an appropriate movement planning (e.g. no spontaneous
movements) and by the development of a human machine interface based on natural
human to human communication (e.g. teaching by demonstration).
6.3.2 Safety Requirements and Risk Assessment
In order to operate a human robot collaboration workstation on an industrial scale, a

certification is required where various safety-critical requirements must be con-
sidered. The certification must be carried out in accordance with the corresponding
standards such as, (ISO 10218-1, 2011) and (ISO/TS 15066, 2016).
The requirements for the safety system depend on the application respectively the
mode of cooperation between the robot and the human operator. Further, the
requirements differ depending on the distance of the operator from the robot or the
average length of stay in the workspace. Four different safety principles are dis-
tinguished in the standards and guidelines. During the “safety-rated monitored
stop”, the robot stops while the operator enters the collaborative workspace and
continues when the operator has left the collaborative workspace. In “hand guid-
ing”, the operator controls the robot motion by touching the robot. During “speed
and separation monitoring”, the distance between robot and operator is constantly
measured. The robot speed is adjusted dynamically and the robot is stopped when
the minimum distance is exceeded. In “power and force limiting”, the contact force
between the robot and the operator is limited to a safe level. The respective safety
principle must be determined during the risk assessment and the appropriate
operational scenario. Additionally, the risk assessment must define the necessary
technical safety requirements. However, the safety distance, the biomechanical
limits and design of component have a great influence on the productivity and
flexibility of the later system as well and will be discussed in more detail below.
Safety distance
To reduce the risk potential of a collision to a minimum a certain safety distance
must be respected. If the minimum safety distance is exceeded, the system must
switch to a safety monitored stop or to a collaborative operating mode. The change
between an autonomous and collaborative operating mode must be in such a way
that the robot cannot endanger the human operator. In addition, the operating state
of the robot must be recognizable from the outside of the collaboration workspace.
The minimum safety distance according to (ISO/TS 15066, 2016) depends on the
following factors: The motion speed of the operator and the robot, the robot’s
braking distance and braking time, the potential intrusion distance as well as the
position accuracy of the measuring system and the robot. Due to these factors, a
large safety distance must be selected in current applications. However, a large
safety distance leads to a frequent change of the robot to a less-productive mode,
thereby reducing the overall productivity. This circumstance must be considered
when designing a safety system for human robot collaboration and has been taken
into account for the safety concepts developed in this project.
Biomechanical limits
A collision between a collaborative robot and a human may lead to injuries of
the affected parts of the human body, while the components of the robot or the tool
are usually not deformed. Therefore, it is necessary to verify that the tool can be
used in the worst-case scenario of a collision. The potential degree of injury
depends on the type of contact (transient/static), the effective robot mass and speed,
the duration of impact, the collision surface area and the affected body part. The
resulting forces must not exceed the maximum permissible body compression for
minor injuries. To comply with this requirement, medical-biomechanical limits
have been specified according to (ISO/TS 15066, 2016). This standard defines
values for the maximum permissible force (shock and clamping/squeezing) and the
admissible surface pressure. Additionally, compression constants were identified
for each body area, which are used as a characteristic value for the maximum
compression displacement until the defined limits are reached. By using mathe-
matical representations of these limits, the maximum overall external force can be
determined for a collision surface area. This force is to be determined for the worst
case with the smallest collision surface and for the most vulnerable body part that
may be affected. The maximum transient force is to be compared with the adjus-
table range of the force sensors in the robot controller. An implementation of the
planned collaboration concept is only possible if this force can be detected. All
calculations at this stage must to be carried out with respect to the entire work-
station design and should be utilized to coordinate requirements for a speed and
separation monitoring.
Component design specification
The probability of a collision between humans and robots at collaborative
workplaces must to be minimized by appropriate measures. Nevertheless, design
requirements for the dynamic and static components of the workstation must be
respected for an unlikely event of a collision. The standards and guidelines define
the specification, which must be respected while designing components for col-
laborative workstations. In case of a collision, no sharp, pointed, cutting edges or
rough surfaces must be in the contact area and no surface load is allowed that
exceeds the limits specified for the maximum total force and surface pressure. The
workspace must be designed in such a way that the operator is able to avoid
clamping situations. For this purpose, appropriate housing, covers and separation
planes must be used to minimize the risk of injury. The determination of geometric
values is still being discussed. So far the guidelines recommend a distance between
two opposite points of the outline of the collision surface of greater than 5 mm (BG/
BGIA 2011) and a minimum radius of the edges of 2.5 mm (ISO/TS 15066, 2016).
The requirements listed here have been taken into account for the tools developed in
this project.
6.3.3 Economic Efficiency
Besides the challenges regarding certification of safety functions and operator

acceptance, there are also requirements for economic operability. This is due to
investments for hardware, which can be very expensive in case of human collab-
oration robots and engineering. There are still no standardized application scenar-
ios; workstation designs must be developed individually and costs are increased. An
important factor to advance human robot collaboration is therefore the development
of robotic tools with standardized interfaces that enable a fast and cost efficient
adaptability of workstations for different tasks and product variants. Since human
collaboration robots usually entail a wide field of integrated sensors, they enable the
opportunity to develop standardized tools that take advantage of those sensors. By
using those integrated sensors, tools can be designed, which are technically much
simpler than conventional tools for automation and thus tool costs can be reduced.
For the development of simplified tools which should be used in human robot
collaboration, two approaches have been proven to be suitable (Gerbers et al. 2016).
Since a lot of robotic tools have already been developed for fully autonomous
production, these tools can be used as a basis for simplification. Automatic
screwdriving systems, for example, are equipped with sophisticated vision and
force based sensor technology (Tsujimura and Yabuta 1991). Costs can be reduced
by omitting this hardware leveraging the integrated sensors of the robot. Additional
costs can be reduced by the abandonment of a tool specific controller, by integrating
the algorithms for operation into the robot controller. Another approach for
development of simple tools is the contemplation of tools used for manual opera-
tion. For the example of the screwdriving tool, automation could be carried out by
adapting an electric powered screwdriver for manual handling as robotic tool.
Depending on the desired degree of automation, additional functions may have to
be integrated into the robotic tool. Following the example of the screwdriving tool,
the screw could be fed by the operator or by additional actuators integrated into the
tool. Based on the analysis of conventional tools morphological approaches and
utility analysis should be used to determine the optimal solution for a tool for a
specific application. These considerations have been incorporated into the devel-
opment of the tools used in this project.
6.3.4 Control Architecture and Interoperability
For the development of applications for human robot collaboration, the focus must
set be on the interoperability for further application. This must include in particular
the contemplation of the development of tools and algorithms. This is important in
order to prevent insular solutions that can only be used within a specific hard- and
software environment. In the future, an application may require a robot form a
different manufacturer (e.g. higher load capacity), which may not be compatible
with the already developed tools or algorithms, without additional effort. The key
aspect is therefore the robot controller, as its interfaces define the programming
language, the human machine interfaces as well as the integration of sensors and
actuators. Currently, most industrial robot controllers have proprietary interfaces
that prevent interoperable solutions. Robotic companies would have to standardize
their interfaces and open up their controllers for software package that can be
executed in the controller’s real-time core for sensor guided tasks. Since this is
currently not the case, proprietary or open source higher-level controllers can be an
appropriate alternative. First implementations of such an open-source control
solution are currently under investigation in research projects (Awad et al. 2016).
For the realization of this project the development of a control solution was chosen,
which offers the possibility to implement applications independently from the robot
manufacturer for increased reusability. Due to the far-reaching influence of the
choice of robot control, this decision should be defined at an early stage in the
development process.
6.4 Human Robot Collaboration for Disassembly

of Lithium-Ion Battery Systems
This section is concerned with the development of a human-robot-collaboration

demonstrator for disassembly of lithium-ion batteries. This includes an analysis of
the disassembly steps, the derivation of a workstation concept and the development
of necessary tools and control algorithms. Furthermore, the potential of intuitive
robot programming is investigated and possible technological implementations are
presented.
6.4.1 Disassembly Steps and Requirements for Human

Robot Collaboration
The first step for the realization of the hybrid workstation was the identification of
disassembly sequences, which are suitable to design efficient, ergonomic and safe
workstations for the desired application. A systematic planning of the material and
information flow as well as the work sequences was necessary to enable the
co-operation of manual work and robot-supported activities. Various concepts were
developed and evaluated. The individual work contents of the processes were
investigated in manual disassembly tests as shown in Fig. 6.1 (Wegener et al.
2014). Based on these findings, the required procedures and tools were compared
with the performance of potentially applicable automation technologies. The
analysis of the disassembly steps showed that many handling and unscrewing
operations are necessary, especially with regard to the disassembly of the modules.
The handling operations, in particular the separation of components, often require
high maneuverability and varying forces and techniques. Since the required tech-
niques are highly dependent on the design of the product, the automation of these
tasks was deemed to be prohibitively difficult and costly. Conversely, screws and
bolts are standardized components and can be removed using the same technique.
As stated before, unscrewing is a relatively simple technique that has been auto-
mated previously and has widespread application in the disassembly of battery
systems. Therefore, unscrewing was identified as a suitable application for human
robot collaboration.
Fig. 6.1 Manual disassembly tests on battery systems (Wegener et al. 2014)
The findings from the disassembly analysis are the work sequences, the division
of manual and automated activities and the choice of tools. This information was
used to derive first drafts for the workstation layout, which offer the greatest
potential in terms of flexibility and short cycle times.
6.4.2 Layout Concept
As next step, the developed concepts for the hybrid workstation were detailed
according to ergonomic and safety aspects. For example, a moving robot in the field
of view of the operator may cause mental pressure to the worker. To avoid such
mental pressure, the movements of the robot must be programmed in such a way
that they are predictable for the operator. Rapid movements of the robot must be
avoided, especially in the facial central of the worker. Furthermore, suitable safety
technology must be used in order to protect the human operator from a collision
with the robot. For this purpose, the advantages and disadvantages of different
sensoric and actoric safety mechanisms and their impact on the workplace of the
human operator were examined and evaluated. The findings of these investigations
were used to adapt the concepts in such a way that the requirements of existing
standards and guidelines for safety as well as the requirements on ergonomics are
met. These insights lay the foundation for a decision support system to enable a
methodological planning of hybrid workstations for future applications. This
decision support system facilitates the layout planning, the selection of suitable
automation components and safety technology as well as the allocation of auto-
mated and manual activities.
Based on this work, the workstation concept for robot-assisted disassembly of

the EV batteries, as shown in Fig. 6.2 was elaborated and chosen for realization.
Both the human and the robot require access to the disassembly object, i.e. the
battery. The chosen concept contains shared workspaces, which minimizes the
system complexity and the time required to frequently transport the disassembly
object between separate workspaces. Furthermore, the human and robot also require
access to their own disassembly tools. For the human this may be a variety of tools
including pliers, screwdrivers, a hammer and cutting tools. For the robot, this
consists of different kinds of socket wrench bits for its unscrewing tool (Fig. 6.3).
The human carries out more complex tasks such as prying apart components joined
with snap fits or (to a limited extent) glue, and pulling out or cutting cables, while
Fig. 6.2 Disassembly

workstation concept
(Wegener et al. 2015)
Fig. 6.3 Cost-optimized robot tool (w/o protective cover) for the disassembly of various screw
connections (Gerbers et al. 2016)
Fig. 6.4 Recording of the depth image camera with detected human hand (left) compliance and
force control for contact detection (right) (Gerbers et al. 2016)
the robot unfastens all screws and bolts. The location of the screws and bolts can be
either taught manually or detected via a 3D camera. In Fig. 6.4, this is indicated by
a sketch of a camera with its field of view on the battery. In addition, the camera is
intended to monitor the distance of the human operator and the robot. This infor-
mation can be used to prevent collisions while adjusting the robots speed depending
on the distance to the human operator. Therefore, the operating speed of the robot
and thus the productivity of the workstation can be increased. With this layout, an
optimal combination of productivity and ergonomics is achieved while complying
with the required safety guidelines.
6.4.3 Tool Development
One main aspect of methodical planning is the design of cost-optimized,

function-integrated robot tools for human robot collaboration. In order to meet the
requirements for an economical and flexible production system, hybrid worksta-
tions must be adapted to new tasks as cost-effectively as possible. The methods
elaborated in this work support the development process for tools with high
demands on cost efficiency (see Sect. 6.3.3). In order to demonstrate the saving
potential, the robot tool shown in Fig. 6.3 has been developed for the automated
disassembly of screw connections. This robotic tool is based on a cost-effective
cordless screwdriver for manual handling and is equipped with an additional
single-finger gripper for the removal of screws. For safe collaboration with human
operators, a housing for the tool was designed in accordance with the specifications
described in Sect. 6.3.2. The hardware only requires simple electronic motion
control units that can be driven from the robot controller. Using the internal sensors
and the controller of the robot an additional compensation mechanism could be
waived. Therefore costs are significantly reduced compared to conventional
unscrewing systems with an additional active compensation (Apley et al. 1998).
The total hardware cost of the prototype is about 160 € (80 € pneumatic compo-
nents, 60 € motion control electronics and drive, 20 € additional components). The
functional integration of a gripper further reduces the hardware cost. Additionally,
the developed algorithms can be reused for other joining operations reducing the
development effort for future applications.
6.4.4 Sensor Technology and Control Algorithms
By leveraging the sensors of the collaborative robots, additional sensors or control

hardware can be waived for the development of tools and components for the
workstation. This reduces hardware and development costs for the tools, but also
requires additional effort for programming control algorithms for the robot con-
troller. Therefore, tool-independent algorithms were developed, which minimize the
programming effort for the development of future applications. The developed
algorithms are based on compliance and force control for process automation. Such
an algorithm is shown in Fig. 6.4, right. This algorithm uses the compliance and
force control to compensate for the inaccuracies in the contact search during
unscrewing between the screw and tool. This principle of contact search can be
transferred to other fields of application and thus reduces the development effort for
future applications.
Another aspect, which has a significant impact on the efficiency and productivity
of hybrid workstations, is the safety technology. In current industrial applications,
the safety of hybrid workstations is achieved by a virtual, spatial separation in
which the speed of the robot is considerably reduced (Luca and Flacco 2012). In the
event of human intervention, this results in a drop in productivity. In order to

increase the productivity of hybrid workstations, safety principles have been
developed, which allow for adjusting the robots speed in dependence of the distance
to the human operator. The increase in productivity is achieved with a safety
concept using 3D measuring technology for constantly monitoring the distance
between the robot and the human operator. For this purpose image processing
algorithms were used, which enable the detection and localization of human
obstacles. The successful recognition of a human hand in the workspace of the
robot is shown in Fig. 6.4, left. Based on these functionalities, an algorithm has
been developed which determines the minimum distance between a human operator
and the robot and thus adapts the speed of the robot or stops the robot when a
predetermined safety distance is reached. This 3D measurement based safety
principle does not yet meet all the requirements of the current standards and
guidelines. These concepts, however, provide the basis for further developments of
future safety systems, which will enable a robot to be driven with a higher average
speed and thus increase the overall system productivity.
6.4.5 Intuitive Human-Machine Interfaces
The development of the hybrid workstation has shown that a further core aspect for
the economic realization at a high product variety is the simple and efficient pro-
gramming of the robot system. This is because conventional programming of robots
is a time-consuming and complex process for which trained specialists are neces-
sary. Frequent programming and retooling of robotic systems to new applications or
components is correspondingly cost-intensive, in particular, since the programming
must take place at the system itself during the primary processing time. This is
especially true for the disassembly of battery systems, since a large variety of
variants can be expected due to the lack of standardization. An economic imple-
mentation of a hybrid workstation can therefore only be realized, if the program-
ming of the robots is made possible with minimal effort. Hence, methods have been
developed, which enable intuitive programming of the hybrid workstation without
prior knowledge in robot programming. Therefore, the use of handheld devices with
intuitive user interfaces as well as other communication channels, inspired by
natural human-to-human communication were investigated. These include both
finger gestures and hand gestures as well as demonstrating movement sequences
(see Fig. 6.5). These simple and intuitive human-machine interfaces can save sig-
nificant personnel costs for hybrid workstations.
Programming by Demonstration and Gesture Control
The first type of programming, which was investigated for this project, is pro-
gramming by gestures or by demonstration. This type of programming is inspired
by natural human-to-human communication and is therefore very simple to
Fig. 6.5 Possibilities of intuitive programming and control (Dröder et al. 2017)
Fig. 6.6 User interface for gesture control (Trentlage 2014)
understand and intuitive to use. The gesture control is intended to enable an intu-
itive, completely contactless adjustment of the robot position. This is particularly
useful when the operator is wearing work gloves that would make it difficult to
operate a conventional control unit. The gesture control was implemented in the
hybrid workstation using the same 3D-camera technology and similar vision
algorithms developed for the safety system (see Sect. 6.4.4). The 3D measuring
technology is used to monitor the workspace of the human operator and to detect
and localize human obstacles. A visual user interface (see Fig. 6.6) was designed to
give the operator direct feedback on his actions. The interface is used to visualize
the possible actions of the operator and to provide additional information on the
status of the controller and the robot.
While the robot is idling, the main menu of the gesture controller is active. From
this menu basic commands, such as moving the robot to its home position (NULL),
activating the gravitational compensation mode (Grav) or selecting the control type
(!) can be called. In this way, four different control types can be activated. The
“Direct Joint Control” allows the positioning of the tool center point via the direct
control of the individual joints. The actuation can be performed using a push
gesture or a resting hand on the user interface element. If the motion is approved,
the selected joint can be controlled directly using the other control elements. Each
joint can be turned in positive and negative directions (−/+) while the robot velocity
can be adjusted to slow (S), medium (M) or fast (F). The “Direct Cartesian Control”
allows the positioning of the tool center point in the coordinate system of the Robot.
Fig. 6.7 Functional principle

of the direct positioning
control (Trentlage 2014)
A hand motion in the corresponding direction of the user interface element will
move the tool center point to the desired direction. The “Direct Positioning Control”
allows the operator to specify a robot position by pointing a finger, as shown in
Fig. 8.7. This position can be saved and used in program sequences. The
“Programming by Demonstration mode” is activated when the gravitational com-
pensation mode is turned on. In this mode, the force and torques in the robots joints
are constantly measured to compensate the weight of the robot tool while enabling
hand guiding by a human operator. The operator can move the robot to a desired
position (e.g. unscrewing position) and save the position for later use in program
sequences.
Especially the last two control types provide a very efficient way of program-
ming, which does not require any experience with specific robots. Compared to
conventional robot programming, where the robot is moved to the desired position
using a teach pendant, considerably time and therefore cost can be saved (Fig. 6.7).
Programming with intuitive user interfaces
The intuitive programming methods presented in the previous section offer the
possibility to program a robot by teaching position for position without the
necessity for expert knowledge. This enables the reduction of labor costs. However,
the need to teach each position separately limits the potential of this type of pro-
gramming. In order to further improve the efficiency of intuitive robot program-
ming, the development of a new visualization and programming concept was
necessary. For this purpose, a human-machine interface concept with an intuitive
user interface for programming devices (e.g. Handhelds) was developed. The
chosen visualization concept is based on the superposition of a real scene with
computer-generated additional information or virtual objects. The environment of
the hybrid workstation is recorded with a camera and displayed on the operator’s
device, superimposed by additional information. The robot can then be pro-
grammed via the interaction with those virtual objects. These functions were
Fig. 6.8 Possibility of intuitive robot programming and control using a handheld device (Dröder
et al. 2017)
implemented using image processing methods and a database system (see

Sect. 6.4.5). The segmentation of recordings from the real scene allows a pattern
matching combined with a database search to detect object and assign operations. In
order to demonstrate the feasibility of this programming concept, the described
functionalities were integrated into the hybrid workstation for disassembly of bat-
tery systems. In the demonstrator, it is possible to identify and locate components or
screws and automatically assign these objects to suitable standard robotic functions,
which are stored in the database (e.g., gripping, screwing, etc.). Once these func-
tions are assigned to the detected objects, they are displayed on the operator’s
programming device, for example a tablet, as recommendation for action (see
Fig. 6.8). The operator can choose from these recommendations and thus program
the disassembly task with several screw points simultaneously. Accordingly, the
user is provided only with functions, which are necessary for the current task. This
considerably reduces the complexity and expenditure of time of the robot pro-
gramming for disassembly tasks.
Object recognition and localization
The intuitive programming environment presented in the previous section
requires the automatic identification and localization of various objects in the hybrid
workstation. Therefore, a flexible concept for object recognition and localization
was developed based on a database and image processing algorithms. The database
is required to store known components and robotic tools or functions necessary for
the processes to be performed. The requirements for the database were a high
degree of flexibility and an efficient data processing. Therefore, a database based on
the XML data format was chosen to ensure simple expandability and processing by
algorithms. Components are stored as templates consisting of bitmaps, which
contain the binary edge representations of the component. In order to minimize the
complexity of the database, the use of a complete edge image was omitted and a
combination of an outer and an inner edge representation was used instead. The
outer edge images can be reused multiple times for similar components and the size
of the database can be minimized. Also the detection speed can be increased, since
the image processing algorithm first uses the outer edge renderings for a template
matching in a rough recognition before the component type is identified.
To further increase the reliability and efficiency of the recognition process, a
function was implemented to identify known objects in addition to individual
components. Such objects can contain multiple components (e.g. screws) and the
associated positions in the objects coordinate system. This enables the detection of
several elements of an object within a common coordinate system, thereby reducing
the overall system error. An object is identified using an algorithm that measures the
distances between the transformed object points and their nearest neighbors to the
component points already detected in the camera image. A hierarchical search tree
is used to ensure a quick location of the next neighboring point for several object
points in the camera image. An object is evaluated as found, if more than 50% of
the object points fall below a distance of one component diameter to their nearest
neighbor. The transformation of the object points is performed by a brute-force
approach, in which the yaw angle is increased in 1° steps. Furthermore, a tool
database has been implemented in the detection process, which contains the defi-
nition of a tool, as well as its corresponding component type. When a disassembly
task is requested on the user interface, an algorithm is used to determine the tool
which corresponds best to the detected component type (e.g. screw type). The
database also contains the position of the tool changing station, so that an auto-
mated tool change can be triggered. In order to demonstrate the feasibility of the
developed concept, the object recognition has been implemented in the demon-
strator for the automated disassembly of screw connections. The database was used
to store the inner and outer screw template to allow for detection of different screw
types (see Fig. 6.9). Additional information such as the screw class, a description
text and a thumbnail for the graphical representation in the selection menu of the
user interface were also stored in the XML file.
The individual stages of the recognition process are shown exemplarily in
Fig. 6.10. Here an incomplete component search result was corrected using the
Fig. 6.9 Template and screw database (Küthe 2015)

Fig. 6.10 Process steps for object recognition and localization (Küthe 2015)
object recognition by adding a missing screw and dismissing the screw, which is
not part of the object but recognized as component. This type of data processing
enables a very efficient and reliable recognition process and also enables the pos-
sibility to discard false positives from the component detection. The flexible design
allows the developed concept to be transferred to various different applications (e.g.
joining) in future.
6.5 Results and Discussion
This section presents the results of the experimental validation of the key tech-
nologies as well as the combination of the individual technologies into a functional
demonstrator. These results are intended to demonstrate the feasibility of the rel-
evant individual technologies and their integration into the developed hybrid
workstation concept for disassembly of battery systems. The experimental valida-
tion focuses on the reliability and the achievable accuracy of the individual tech-
nologies in order to assess possible fields of application, to identify problems and to
highlight development potential.
6.5.1 Robot Based Unscrewing
In order to assess the applicability of the developed unscrewing system, the

achievable process times and the compensable positional deviations were deter-
mined (Gerbers et al. 2016). It was found, that the process time decreases in
accordance with an increasing thread diameter. This also leads to a larger chamfer
of fitting which describes the local lateral tolerance compensation (Fig. 6.11). Thus,
compared to existing tools, which are dimensioned for the smallest expected
chamfer, the process time can be reduced. As a result, a tight fit for less than one
Fig. 6.11 Process time in

dependence on the chamfer
diameter (Gerbers et al. 2016)
second could be achieved for screws larger than M8. For smaller screws such as
M4, the time increases up to 5 s. However, no smaller screws than M6 are used in
the investigated battery systems and therefore an efficient disassembly could be
guaranteed for the developed demonstrator.
A series of measurements of the functionality of the unscrewing system for
different screw types and positional deviation between the tool and the screw has
been conducted and the results are shown in Fig. 6.12. These results demonstrate
the functionality of the implemented process, for certain offsets between the tool
and the screw, with a high reliability. However, the maximum compensable tol-
erance is limited and depends on the diameter and type of the screw head or nut.
Therefore, it can be concluded that a localization accuracy of at least 5 mm (M6
Screw) is required in this project for a reliable unscrewing process in the demon-
strator for automated disassembly.
Influence of lateral offset as a function of the shaft

diameter at external hexagon drive Influence of lateral offset depending on the drive type
100 100
M4 Torx
Tight fit probability
Tight fit probability
80 80 filister head
60 60 M4 Phillips
filister head
40 40
M4 Phillips
20 20 flat head
0 0 M4 External
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 5,5 6 6,5 7 7,5 8 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 hexagon
lateral offset in mm lateral offset in mm
M4 M6 M10
Fig. 6.12 Experimental results of the tool validation (Gerbers et al. 2016)
Table 6.1 Experimental accuracy analysis of direct positioning control (Trentlage 2014)
Position Δx (mm) Δy (mm) Δz (mm) Displacement (mm)
DPC1 41.7 −6.1 −40.9 58.8
DPC2 39.6 −32.6 40 65.1
DPC3 30.1 −15.6 9 35.1
DPC4 42.8 −23.3 18.6 52.2
DPC5 40.9 −59.1 39.8 82.2

Fig. 6.13 Experimental

accuracy analysis of direct
positioning control (Trentlage
2014)
6.5.2 Gesture Control
An accuracy analysis has been carried out for the developed gesture control in order
to assess possible applications. To determine the accuracy of the gesture control a
measuring tip was placed with a fixed position in the field of view of the used
camera system. As next step the position of the measuring tip was determined
relatively to the robots coordinate system. The gesture control was then used to
point the position of the measuring tip to the robot. The displacement of the tool
center point was determined after the robot’s movement to the desired position. The
resulting data for the displacements for various positions is shown in Table 6.1. The
actual position of the measuring tip (red) and the tool center points
(TCP) approached by the robot (blue) is shown in Fig. 6.13. It was found, that the
deviation from the desired position and the TCP-position are in the range of cen-
timeters. Thus, the use of the gesture control in combination with the automatic
unscrewing is not possible. However, this control concept can be used in many
further applications such as pointing objects or moving the robot from one area of
the workspace to another. An accuracy increase of at least tenfolds would be
required to use the gesture control with the developed search algorithm for robotic
unscrewing. Since the camera used for this experiments has a low resolution of
640 480 pixels for the depth map a higher accuracy could be achieved using a
camera with a higher depth resolution.
6.5.3 Object Recognition and Localization
The reliability and accuracy of the object recognition and localization is decisive for
the efficient use of a fully automatic system for disassembly of screw connections
and was therefore examined more closely. The reliability of the object detection
results from the probability of unrecognized screws, false-positives and false
detections of the component type. A few examples of screw connections, which
have been used to investigate the reliability, are presented in Fig. 6.14 This figure
shows screw connections, which are increasingly difficult to identify from left to
right. These images indicate that a general assessment of the reliability is not
possible. While a close to 100% reliability can be achieved for images as shown on
the left side of the figure only a low reliability of 10% can be obtained for images of
screws as shown on the right side of the figure.
While the recognition algorithm is reliable for good images, several false
detections occurred during the actual application (see Fig. 6.15). This is particularly
due to the poor image quality of the chosen hardware. The RGB image quality of
the used ASUS Xtion Pro Live is not satisfactory despite the selected maximum
resolution of 1280 1024 pixels. It can be concluded, that the use of higher quality
cameras would improve the detection performance. However, the achieved accu-
racy is sufficient for the developed concept for automated disassembly, since
incorrect recognized screws are intended to be double-checked by a human operator
during programming of the robot (see Sect. 6.4.5).
In addition to the reliability, the achievable positioning accuracy of the object
detection is relevant for the disassembly process. Several experiments were carried
out in order to determine the accuracy of the object localization. Therefore, the
position of the components relative to the coordinate system of the workstation was
measured manually and compared with the measurement results of the object
recognition. A sample of the results is shown in Table 6.2. For the experiments, the
Fig. 6.14 Screw images for reliability testing
Fig. 6.15 False positives

during screw detection (Küthe
2015)
Table 6.2 Positional Position Δx Δy Δz Displacement

deviations during object (mm) (mm) (mm) (mm)
recognition (Küthe 2015)
SL1 2.42 −1.33 −2.17 3.51
SL2 1.54 0.21 −4.03 4.32
SL3 3.13 0.70 −2.96 4.36
SL4 −0.16 −1.74 2.24 2.84
SL5 0.31 −1.06 0.51 1.22
SL6 1.68 0.70 −1.08 2.12

camera has been aligned at a distance of approx. 90 cm and at an angle of approx.

35° to the surface of the test object. It can therefore be assumed, that the deviations
are reduced once the camera is positioned vertically.
The respective deviations from the ideal point of the screw are shown in the last
column of the table. In general, the accuracy is comparable to the results from
Vongbunyong et al. (Vongbunyong and Chen 2015), which indicated a deviation of
up to 6 mm. The accuracy is thereby mainly influenced by the resolution of the
screw image. This also explains the range of the deviations. Screws that are closer
to the camera can be detected with greater accuracy. A maximum inaccuracy of
4.5 mm could be determined in the tests carried out in the demonstrator (M6
screws). Since the developed unscrewing process can compensated deviations up to
5 mm (see Sect. 6.5.3) the reached accuracy is sufficient to enable a reliable dis-
assembly process,
6.5.4 Demonstrator
The concept of the hybrid workstation, which has been implemented as demon-
strator is shown in Fig. 6.16. The demonstrator is equipped with a Robot, 3D
camera technology, a screwdriving tool and a tool changing station. The robot for
such a workstation has to be lightweight to reduce the risk associated with colli-
sions. Additionally, the robot should have a sufficient load capacity and should be
able to compensate for forces and torques generated from handling an electric
screwdriver and loosening screws. One robot, that fulfils these criteria and that was
therefore chosen for realization of the demonstrator is the KUKA Lightweight
Robot (LWR). The robot programming in the demonstrator is carried out using the
developed interactive and intuitive user interface in which the operator can select
the necessary disassembly operations. The trajectories for the robot are computed
from the 3D camera data and transferred to the robots controller. The robotic
disassembly process is executed in parallel to the manual disassembly tasks. During
disassembly, the operator can intervene in the robot process by gestures at any time
to interrupt or adept the process. The speed of the robot is continuously adapted
Fig. 6.16 Implementation of

the hybrid disassembly
workstation (Dröder et al.
2017)
with the developed safety system. This is done in accordance to the distance
between the operator and the robot, thus ensuring that the robot always works at the
maximum efficiency that is safe to operate.
Disassembly experiments have been carried out using this demonstrator and the
results have shown that a partial-automated disassembly of battery systems is
feasible. The intuitive operating concept significantly reduces the programming
time of the robot and surveys among unskilled users of the demonstrator indicate a
high degree of acceptance for the workstation concept. However, there is still the
need for optimization. Currently the cycle times for the hybrid disassembly are
relatively high compared to manual disassembly. Especially errors in the image
processing result in the necessity for too many manual interventions of the operator,
which decreases the productivity. Additionally, the compensation mechanism used
in the unscrewing process is not efficient enough and the disassembly takes more
time compared to manual unscrewing.
6.6 Conclusion and Outlook
The aim of this project was to develop a concept for increased flexibility, pro-
ductivity and safety in the disassembly of battery systems. Since a fully automated
system is not feasibly at the moment, due to unpredictable lot sizes and the lack of
standardization of the battery designs, the decision was made to develop a
partial-automated disassembly system. Therefore, the use of human robot collab-
oration has been investigated, a concept where a human operator carries out the
tasks that requires complex sensing and motions and a robot handles monotonous,
hazardous and unergonomic tasks. In order to be able to develop such a system,
requirements for the disassembly of batteries have been elaborated and layout
concept as well as concepts for the implementation of the individual subtech-
nologies (e.g. HMI) have been derived. The individual concepts have then been
implemented prototypically and incorporated into a demonstrator for the disas-

sembly of battery systems. With this realization of a hybrid workstation as well as
with the implementation of the developed machine interfaces, tools and algorithms
an efficient, partial automated disassembly system for batteries could be demon-
strated successfully. Disassembly experiments with this demonstrator have shown
that human robot collaboration is a promising concept for economic rationalization
of disassembly tasks and that the requirements for variant flexibility can be fulfilled.
In order to implement this concept on an industrial scale there is still the need for
further development to improve the reliability and therefore the disassembly pro-
ductivity. In addition, the safety concept must be subjected to certification, for
which further extensions (e.g. a second safety circuit) are necessary. Since only
simple hardware was used for the demonstration of the developed concepts and no
extensive optimization of the process parameters has been carried, it can be
assumed that the productivity can be significantly improved. Thus, a concept is
available for the future to increase the degree of automation for disassembly of
battery systems when an increasing demand of recycling capacity requires it.
The realization of an economical plant for large scale recycling of battery systems is
therefore possible even for high-wage countries. Furthermore, the principles and
methods elaborated in this project can be used for the development of hybrid
workstations in a variety of related applications.
References
Apley DW, Seliger G, Voit L, Shi J (1998) Diagnostics in disassembly unscrewing operations.
Int J Flex Manuf Syst 10(2):111–128. https://doi.org/10.1023/A:1008089230047
Audi AG (2015) New human-robot cooperation in Audi’s production processes. http://www.audi.
com/aola/brand/en_lc/tools/news/pool/2015/02/New_human-robot_cooperation_in_Audi_s_
production.html. Accessed 22 Nov 2015
Awad R, Heppner G, Roennau A, Bordignon M (2016) ROS engineering workbench based on
semantically enriched app models for improved reusability. In: 2016 IEEE 21st international
conference on emerging technologies and factory automation (ETFA), pp 1–9. https://doi.org/
10.1109/ETFA.2016.7733581
Bdiwi M (2014) Integrated sensors system for human safety during cooperating with industrial
robots for handing-over and assembling tasks. Procedia CIRP 23:65–70. https://doi.org/10.
1016/j.procir.2014.10.099
BG/BGIA (2011) BG/BGIA risk assessment recommendations according to machinery directive.
BGIA—institute for occupational safety and health of the german social accident insurance
Büker U, Drüe S, Götze N, Hartmann G, Kalkreuter B, Stemmer R, Trapp R (2001) Vision-based
control of an autonomous disassembly station. Robot Auton Syst 35(3–4):179–189. https://doi.
org/10.1016/S0921-8890(01)00121-X
Cherubini A, Passama R, Fraisse P, Crosnier A (2015) A unified multimodal control framework
for human–robot interaction. Robot Auton Syst 70:106–115. https://doi.org/10.1016/j.robot.
2015.03.002
Cruz-Ramírez SR, Mae Y, Arai T, Takubo T, Ohara K (2011) Vision-based hierarchical
recognition for dismantling robot applied to interior renewal of buildings. Comput Aided Civ
Infrastruct Eng 26(5):336–355. https://doi.org/10.1111/j.1467-8667.2010.00689.x
Dröder K, Dietrich F, Tornow A, Löchte C, Wonneberg B, Gerbers R, Bobka P (2017)

Transfersysteme. Handbuch Industrie 4.0(1):429–451. ISBN: 978-3-446-44642-7
ElSayed A, Kongar E, Gupta SM, Sobh T (2012) A robotic-driven disassembly sequence generator
for end-of-life electronic products. J Intell Robot Syst 68(1):43–52. https://doi.org/10.1007/
s10846-012-9667-8
Ezer N, Fisk AD, Rogers WA (2009) Attitudinal and intentional acceptance of domestic robots by
younger and older adults. Univers Access Hum Comput Interact 5615:39–48. https://doi.org/
10.1007/978-3-642-02710-9_5
Fanuc (2015) Robot finder LR Mate arc welding CR-35iA Paint Robots M-10iA M-20 M-410
M-710iC M-900 M-2000iA R-1000iA R-2000 M-1iA M-2iA M-3iA range overview
accessories collaborative robot CR-35iA. http://www.fanuc.eu/pt/en/robots/robot-filter-page/
collaborative-cr35ia. Accessed on 24 Nov 2015
Flacco F, Kroeger T, Luca A de, Khatib O (2014) A depth space approach for evaluating distance
to objects. J Intell Robot Syst. https://doi.org/10.1007/s10846-014-0146-2
Gerbers R, Mücke M, Dietrich F, Dröder K (2016) Simplifying robot tools by taking advantage of
sensor integration in human collaboration robots. Procedia CIRP 44:287–292. https://doi.org/
10.1016/j.procir.2016.02.135
Gil P, Pomares J, Diaz SvPC, Candelas F, Torres F (2007) Flexible multi-sensorial system for
automatic disassembly using cooperative robots. Int J Comput Integr Manuf 20(8):757–772.
https://doi.org/10.1080/09511920601143169
Herrmann C, Raatz A, Mennenga M, Schmitt J, Andrew S (2012) Assessment of automation
potentials for the disassembly of automotive lithium Ion battery systems. In: Dornfeld DA,
Linke BS (eds) Leveraging technology for a sustainable world: Proceedings of the 19th CIRP
conference on life cycle engineering. University of California at Berkeley, Berkeley, USA,
23–25 May 2012. Springer, Berlin Heidelberg, pp 149–154
Heyn H, Krüger T (2015) Towards human-robot cooperation—systematic approach and
applications within the Volkswagen group. International Conference on Intelligent Robots
and Systems, Hamburg
Hoffmann A, Schierl A, Angerer A, Stüben M, Vistein M, Reif W (2015) Robot collision
avoidance using an environment model for capacitive sensors. In: Planning, control, and
sensing for safe human-robot interaction. IEEE International Conference on Robotics and
Automation (ICRA), Seattle, USA
ISO 10218-1 (2011) Robots and robotic devices—safety requirements for industrial robots—
Part 1: Robots
ISO/TS 15066 (2016) Robots and robotic devices—collaborative robots
Kaplan F (2004) Investigating cultural differences in the acceptance of robots. Int J Human Robot
01(03):465–480. https://doi.org/10.1142/S0219843604000289
Knight W (2014) How human-robot teamwork will upend manufacturing. http://www.
technologyreview.com/news/530696/how-human-robot-teamwork-will-upend-manufacturing/.
Accessed on 22 Nov 2015
Küthe D (2015) Robotergestütztes Lösen von Schraubverbindungen mit Hilfe eines
Bildverarbeitungssystems
Luca A de, Flacco F (2012) Integrated control for pHRI: Collision avoidance, detection, reaction
and collaboration. In: 4th IEEE RAS & EMBS ìnternational conference on biomedical robotics
and biomechatronics (BioRob 2012), pp 288–295
Nave M, Jünemann R (2003) Beitrag zur automatisierten Demontage durch Optimierung des
Trennprozesses von Schraubenverbindungen, 1. Aufl. Logistik für die Praxis. Praxiswissen
GmbH Gesellschaft für innovative Qualifizierung, Dortmund
Trentlage C (2014) Entwicklung einer Gestensteuerung für einen KUKA Leichtbauroboter unter
Einsatz der Tiefenbildkamera ASUS XtionPRO LIVE
Tsujimura T, Yabuta T (1991) Adaptive force control of screwdriving with a
positioning-controlled manipulator. Robot Auton Syst 7(1):57–65. https://doi.org/10.1016/
0921-8890(91)90027-I
Viola P, Jones M (2001) Rapid object detection using a boosted cascade of simple features. In:
IEEE computer society conference on computer vision and pattern recognition. CVPR 2001,
I-511-I-518
Vongbunyong S, Chen WH (2015) Disassembly automation: automated systems with cognitive
abilities. Sustainable production, life cycle engineering and management. Springer, Cham
Vongbunyong S, Kara S, Pagnucco M (2013) Application of cognitive robotics in disassembly of
products. CIRP annals—manufacturing technology 62(1):31–34. https://doi.org/10.1016/j.cirp.
2013.03.037
Wang L, Schmidt B, Nee AY (2013) Vision-guided active collision avoidance for human-robot
collaborations. Manuf Lett 1(1):5–8. https://doi.org/10.1016/j.mfglet.2013.08.001
batteries using the example of the Audi Q5 hybrid system. Procedia CIRP 23:155–160. https://
doi.org/10.1016/j.procir.2014.10.098
Wegener K, Chen WH, Dietrich F, Dröder K, Kara S (2015) Robot assisted disassembly for the
recycling of electric vehicle batteries. Procedia CIRP 29:716–721. https://doi.org/10.1016/j.
procir.2015.02.051
Wu Y-H, Wrobel J, Cornuet M, Kerhervé H, Damnée S, Rigaud A-S (2014) Acceptance of an
assistive robot in older adults: a mixed-method study of human-robot interaction over a
1-month period in the living lab setting. Clin Interv Aging 9:801–811. https://doi.org/10.2147/
CIA.S5
Chapter 7
Crushing of Battery Modules and Cells
Jan Diekmann, Steffen Sander, Guido Sellin, Matthias Petermann

and Arno Kwade
Abstract Crushing is a substantial process step for the following separation, as it

transfers the battery cells or modules to a storable and conveyable bulk material.
Crushing also leads to the opening of the battery cells and release of valuable
materials. Because of the deliberate destruction of the battery cells, this process has
a high hazard potential due to the remaining attached energy of the battery cells and
the flammable electrolyte components. The project partners evaluated requirements
for the design of a crusher and drafted a concept for the realization of a safe and
efficient crushing process.
7.1 Purpose of Crushing and Basic Technologies
The first crushing process has a considerable impact on the following process route.
During this treatment, the battery modules are transformed into a conveyable and
storable bulk material by opening the battery cell housing. Moreover, the valuable
materials are set free and become accessible for the following material separation.
The deliberate destruction of lithium-ion battery (LIB) cells leads to hazards due
to the flammable electrolyte components or exothermal reactions caused by short
circuits by the crushing tools and consequently toxic gas emissions. To deal with
these hazards, different basic crushing technologies are described in the literature or
realized in industrial recycling processes.

S. Sander
Hosokawa Alpine Aktiengesellschaft, Augsburg, Germany
G. Sellin
Electrocycling GmbH, Goslar, Germany
M. Petermann
Volkswagen AG, Wolfsburg, Germany

https://doi.org/10.1007/978-3-319-70572-9_7
After different pre-treatments to minimize the hazard potential of LIB, such as

discharging, thermal pre-treatment, and freezing, two general crushing technologies
can be distinguished: wet and dry crushing.
The usage of a thermal pre-treatment enables safe dry crushing. At temperatures
around 300 °C in a rotary kiln, exothermal reactions take place and the solvents
evaporate, leading to an opening of the cells and combustion of the organic com-
ponents. A thermal pre-treatment inactivates the battery cells (Georgi-Maschler
et al. 2012). Drawbacks of this technology are the energy consumption of the
pre-treatment, as well as the high amount of the fine particular coating materials
which are set free during crushing.
Retriev Technology Inc., formerly Toxco Inc., uses a wet crushing process after
the deactivation of lithium batteries by freezing. Crushing is carried out by a
hammer mill under a brine solution to neutralize harmful reaction products and to
minimize the fire and explosion hazard (Vezzini 2014).
Zhang et al. compared wet and dry crushing processes for the recycling of LIB
(Zhang et al. 2013). They investigated the crushing of batteries from consumer
devices and concluded that selective crushing was presented in both methods. They
also showed that electrode active materials are fully shed from the current collector
foils and concentrate in fine fraction with less impurities when dry crushing is
carried out.
The high amount of waste water that has to be cleaned during wet crushing, the
high energy consumption of a thermal pre-treatment, as well as the higher purity
that can be reached lead to the decision to use a dry crushing under an inert
atmosphere in LithoRec II.
7.2 Input Material
One important parameter for the design of a safe crushing process as well as for a
suitable crusher is the knowledge about the input material. These are the discharged
and short circuited battery modules. One of the main problems for the design of a
suitable crusher is the variety of different designs and constructions. Figure 7.1
shows a selection of different module designs from current EV and HEV.
These design variations show the differences in used materials, battery cell type,
number of battery cells, connections, as well as mass and dimensions. The masses
range from 1.5 to more than 30 kg. The width can vary from 10 to 450 mm. This
data is most important for the dimensioning of material feed, rotor drive, and the
construction of the rotor itself.
Figure 7.2 shows the composition of a generic battery module with battery cells
containing lithium nickel cobalt manganese oxide (NCM) with a mass ratio of nickel,
cobalt, and manganese of 1:1:1 as the cathode active material. This composition is
based on an the analysis of different modules and estimations of experts (Buchert and
Sutter 2015). The data shows that around 89 wt% are related to the battery cells. The
module periphery consists of steel (4.1 wt%), aluminum (4.8 wt%), and plastics
7 Crushing of Battery Modules and Cells 129
Fig. 7.1 Variations of module designs
(1.9 wt%). Electronic parts can also be found. These are slave battery management
units (BMU), which control charge and discharge current, voltage, and temperatures
of one or more modules.
When it comes to crushing of the module, the periphery and cell housings play a
major role. These materials are mostly ductile with contacts and module housing
components being especially massive (steel with thicknesses around 1 mm or more)
and posing challenges to a crusher.
Next to the mechanical challenges, the ingredients of LIB play a major role in
safe crushing. The corresponding hazards and requirements are described in the
following. A more general description of the hazards along the process route can be
found in Chap. 3.
7.3 Requirements on Safe Crushing
Although the hazards of LIB and therefore health and safety are important for the
entire process chain, these issues play a special role during crushing. This process
represents a willful destruction of the battery cells. Consequently, all safety
mechanisms are by-passed.
Fig. 7.2 Composition of a generic battery module
7.3.1 Hazards Generated by Chemical Reactions
Chapter 3 presents a rough division of the hazards of LIB in electrical, fire and
explosion, and chemical hazards. Also the interaction of these three areas, leading
to chemical reactions is described. This interaction is determined by the stored
energy of the battery cells. A short circuit of a LIB leads to high temperatures due to
joule heating and therefore to exothermal and endothermal reactions. If a critical
temperature is reached, the so called thermal runaway takes place. Such an incident
can be caused by internal or external shorts, but also by overcharging or over-
heating. A deeper look at the thermal runaway can be found in Chap. 3.
In the case of crushing, the tools cause an external or internal short due to
electrically conductive materials or contact of the electrodes. Discharging and short
circuiting of the battery systems before disassembly and short circuiting of every
module before crushing eliminates the stored energy and avoids joule heating.
The interaction of joule heating and chemical reactions can be observed in the
crushing product. After crushing, “micro-short-circuits” between anode and cathode
fragments of loaded battery cells lead to joule heating and therefore an increase the
temperature at hot spots. The increased temperatures lead to the evaporation of the
electrolyte solvents and the exposition of the conducting salt lithium
hexafluorophosphate (LiPF6). At temperatures of around 60 °C, and in contact with
aerial humidity, LiPF6 decomposes and forms hydrogen fluoride (HF) and further
gaseous and liquid products (Diekmann et al. 2016a). Figure 7.3 shows the com-
position of the gaseous product and highest concentrations of the substances during
such a decomposition. In this case, four PHEV 1 battery cells (25 Ah) were crushed at
a State-of-Charge (SOC) of 80% and then brought into contact with aerial humidity.
It can be seen that next to the evaporated solvents also the additive cyclo hexyl
benzene and HF are released. Although the concentration of HF seems to be very
small, it is around 17 times higher than the IDLH (immediately dangerous to life
and health) of 30 mg * m−3.
A possibility to eliminate this reaction seemed to be cooling, as rise of the
temperature is one of the triggers. Therefore, battery cells (PHEV 1, 25 Ah) were
cooled with dry ice and acetone to core temperatures of around 65 °C. This
concentration of gaseous components cgas [mg*m-3]
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
e
ne
e
e
id
at
at
at
id
id
an
ze
ox
ox
on
on
on
or
ex
en
flu
di
on
rb
rb
oh
ar
lb
ca
ca
on
en
cl
lc
xy
Cy
rb
on
yl
e
og
hy
e
en
h
oh
Ca
rb
dr
et
et
yl
cl
Ca
Hy
Cy
Et
Di
yl
h
Et
Fig. 7.3 Gas emissions during decomposition of lithium hexafluorophosphate in the crushing
product (measured via Fourier transform infrared spectroscopy)
treatment enabled safe crushing of loaded battery cells, but only shifts the
decomposition of the conducting salt.
In order to decrease the time needed for discharging and to set limits for crushing
of battery cells which cannot be discharged due to internal cell failures, the project
partners varied the SOC. Depending on the battery chemistry and capacity, a SOC
of 30% for safe crushing was discovered. As expected, battery cells with the
cathode active material lithium nickel aluminum oxide (NCA) show a higher
reactivity.
7.3.2 Gas Release During Crushing
In addition to the gas emissions of loaded battery cells, the gas release of discharged
and short circuited battery cells is important for the crushing process, which will be
described in the following.
To investigate the influence of ageing on the recycling of lithium-ion batteries, a
simple recycling process in laboratory scale was used, combining discharging and
short circuiting, crushing, drying, and sieving. The experiments were carried out
with Panasonic CGR18650CH 18650 battery cells with a nominal voltage of 3.6 V
and capacity of 2250 mAh in a fourfold determination. While four cells were aged
at 1 C to a State-of-Health (SOH) of 80% and short circuited after discharging to
1 V, the others were directly discharged and short circuited.
During drying (65 °C, 5 h), only small differences in the mass loss could be
observed. Also, no significant distinctions were made regarding yield and com-
position of the recovered coating materials. While only small changes in the solid
materials were investigated, ageing shows an impact on the gaseous products
during crushing (Diekmann et al. 2016a).
The discontinuous crushing of the battery cells was carried out with a converted
cutting mill under an inert atmosphere. The released infrared active gas components
were identified and quantified via Fourier transform infrared spectroscopy (FTIR).
In Fig. 7.4, the masses of the released organic solvents ethyl methyl carbonate
(EMC) and dimethyl carbonate (DMC), as well as carbon dioxide (CO2) are shown.
It becomes obvious that the overall gas release of aged battery cells is smaller
(around 18 wt%) than of non-aged, simply formatted, cells. During crushing, these
battery cells only lost 3.6 and 4.5 wt% of their overall volatile components as gases
respectively.
The organic carbonates EMC and DMC are well known to be reduced during
ageing. Their reactions with lithium form components of the solid electrolyte
interface (SEI), for example CH3OCO2Li (Vetter et al. 2005; Aurbach et al. 2002).
This reduction leads to a smaller released mass of the organic solvents.
One of the gaseous components that is produced during formation of the SEI is
CO2 and therefore it is an ageing product. During prolonged cycling, CO2 reacts with
traces of water to form solid ageing products (Aurbach et al. 2002). This effect leads
to the lower released masses of CO2 of aged battery cells (Diekmann et al. 2016a).
110 10
100 9
mass organic solvents mos [mg]
90 8
mass CO2 mCO2 [mg]

80
7
70
6
60
5
50
4
40
3
30
20 2
10 1
0 0
uncycled cycled uncycled cycled uncycled cycled
80%, 1C 80%, 1C 80%, 1C
EMC DMC CO2
Fig. 7.4 Ageing influence on gas emissions of LIB cells (18650 battery cells) during crushing
(Diekmann et al. 2016a)
The highest concentration of DMC during the crushing of the 18650 battery cells
is 0.2 vol%, which is far away from its lower explosion limit of 4.22 vol%.
Crushing of battery cells from EV and HEV have other requirements due to their
higher weight and slightly higher content of volatile components.
Crushing of such battery cells with a capacity of 25 Ah leads to higher con-
centrations of the organic solvents in the atmosphere of the inner space of a
technical scale crusher (Fig. 7.5). For the used crusher and battery type, the lower
explosion limit of DMC is reached for a cell mass of three kilograms. This is equal
to four or five battery cells from EV.
When it comes to crushing battery cells or modules from EV or HEV, the
investigations show that operating under a dry and inert atmosphere is highly
recommended.
7.3.3 Influence of Different Crushing Mechanisms
Besides the control of the hazard potential, the selective crushing and therefore
preparation of the material separation is most important for the design of a crusher
for lithium-ion battery modules.
concentrations organic solvents cos [Vol.-%]

4,5
4,0
3,5
3,0
2,5
2,0
1,5
1,0
0,5
0,0
C
C
C
EM
M
EM
D
D
Cell Mass 1,5 kg Cell Mass 3 kg
Fig. 7.5 Gas emissions during crushing of LIB cells (PHEV 1 battery cells) in a technical scale
crusher. Reprinted with permission of Diekmann et al. (2016b)
To evaluate the different mechanisms, crushing experiments were carried out

with EV battery cells. As the different inert crushers were nonexistent, this eval-
uation was carried out with battery cells containing no electrolyte and lithium iron
phosphate (LFP) as the cathode active material. Because LFP is not carcinogenic,
these experiments had a minimized hazard potential.
Lithium-ion battery cells and modules need to be crushed with low deformation
and compression of the fragments to avoid inclusions, and therefore loss of valuable
materials. Due to the hazard potential regarding electrolyte and partly carcinogenic
coating materials, the process steps and conveyors have to be surrounded and
gastight. To ensure the proper operation by avoiding sticking of the fragments, the
upper particle size has to be limited. Rotary shears are the best to fulfil these tasks.
Five different rotary shears were evaluated. These were radial rotary shears
(RRS) and axial rotary shears (ARS). Axial rotary shears crush materials with
interdigitating crushing tools. Normally, these rotary shears have at least two rotors,
although one of them can be fixed, and could be equipped with discharge screens.
Radial rotary shears crush with a geared rotor and a stator. Generally, these
aggregates have only one rotor and are always equipped with discharge screens.
Furthermore, the feed material is transported hydraulically or pneumatically.
Figure 7.6 shows the cumulated distributions of the five tested crushers and mesh
sizes.
The radial rotary shears (RRS Lindner and RRS Erdwich) show the best results
regarding the amount of smaller fragment sizes, meaning better isolation of the
1,1
1,0 RRS Lindner, M 50

RRS Lindner, M 35
cumulated distribution Q3 (x)
0,9 RRS Erdwich, M 30
ARS Erdwich, 2 shafts
0,8 ARS Erdwich, 2 shafts, 2 times
ARS Erdwich, 3 shafts, M 40
0,7
ARS UZ
0,6
0,5
0,4
0,3
0,2
0,1
0,0
0,1 1 10 100
particle size x [mm]
Fig. 7.6 Influence of different crushing mechanisms on the particle size distribution of the
crushing product (RRS: radial rotary shear, ARS: axial rotary shear, M: mesh size); data obtained
from Mämpel (2013)
different components, and limitation of the upper fragment sizes. The axial rotary
shear UZ (“Universalzerkleinerer”—universal crusher), which is a rotary shear with
one shaft and a stator between every gear of the rotor, shows the lowest amount of
smaller fragment sizes. The axial rotary shear type Erdwich, with three shafts and a
discharge screen mesh size of 40 mm, also leads to a good isolation of the frag-
ments, but limitation of the upper fragment size was not carried out sensitively.
Generally, the crushing products of the axial rotary shears contain strips of elec-
trodes or the separator. These strips cannot be separated from each other by con-
ventional mechanical separation techniques.
Crushing of lithium-ion batteries via the radial rotary shear type Lindner is very
sensitive to the mesh size of the discharge screen, which can be seen in Fig. 7.7.
Variation of these mesh sizes lead to the expected changes of the amount of
smaller fragment sizes and the limitation of the upper fragment size. Furthermore,
no difference can be seen between two serial crushing processes with a first mesh
size of 50 mm and a second mesh size of 20 mm, and one simple crushing process
with a mesh size of 20 mm.
Due to the sensitivity to the mesh sizes, crushing and the following separation
processes are very adaptable.
In further experiments it was shown that application of a liquid electrolyte
substitute to the battery cells has no influence on the results of the crushing process
in regards to particle size distribution (Mämpel 2013).
1,1
1,0
RRS Lindner, M 50
cumulated distribution Q3 (x)
0,9
RRS Lindner, M 35
0,8 RRS Lindner, M 20
RRS Lindner, M 50 + M 20
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0,0
0,1 1 10 100
Fig. 7.7 Influence of mesh size of the discharge screen on the particle size distribution of the
crushing product (RRS: radial rotary shear, M: mesh size); data obtained from Mämpel (2013)
Fig. 7.8 Inert crusher designed for the recycling of lithium-ion battery modules by Hosokawa
Alpine Aktiengesellschaft
7.4 Design of a Crusher for Lithium-Ion Battery Modules
The aforementioned requirements, determined by the hazard potential and the

mechanical isolation of the different components, lead to a new design of a crusher
for lithium-ion battery modules and cells which operates batch wise. This crusher
and its periphery was drafted and planned by Hosokawa Alpine Aktiengesellschaft.
A scheme of this draft can be seen in Fig. 7.8.
The design is based on the radial rotary shear type Lindner and is equipped with
a replaceable discharge screen. The following drying process is connected by a
screw conveyor and functions as intermediate storage. During the batch wise
crushing process, the dryer is filled. Before the drying process starts, the connection
between the crusher and the dryer is shut down. To establish an inert atmosphere,
the entire aggregate is filled with nitrogen. Therefore, every single aggregate is
gastight and protected against over pressure.
The battery cells and modules are fed into the crusher via an airlock which is
nitrogen flooded after every feeding. The slide of the airlock only opens if the
oxygen (O2) concentration is less or equal to 3%. Afterwards, a hydraulic slide
pushes the battery cells or modules against the rotor, leading to a slow abrasion of
material. The resulting fragments are transported to the dryer by the screw
conveyor.
Detection of a thermal runaway or the decomposition of the conducting salt is
carried out by the measurement of temperature, pressure, and carbon monoxide
(CO). In the case that all three measurements are beyond the defined values, the
nitrogen inflow, as well as the cleaned gas outflow are increased to remove the toxic
and corrosive gases. Crushing will be continued after the reactions stop. The
exhaust gas treatment for the crusher and dryer are carried out via condensation and
a gas scrubber. For maintenance, the entire rotor can be removed from the crusher
which leads to very good ventilation during these tasks.
Because the project partners decided not to constitute a pilot plant in large scale,
this crusher design has not been realized and tested yet.
References
Aurbach D, Markowsky B, Rodkin A, Cojocaru M, Levi E, Kim H (2002) An analysis of

rechargeable lithium-ion batteries after prolonged cycling. Electrochim Acta 00:13
Buchert M, Sutter J (2015) Ökobilanzen zum Recyclingverfahren LithoRec II für
Lithium-Ionen-Batterien
Diekmann J, Hanisch C, Froböse L, Schälicke G, Loellhoeffel T, Fölster A-S, Kwade A (2016a)
Diekmann J, Hanisch C, Loellhoeffel T, Schälicke G, Kwade A (2016b) Ecologically friendly
recycling of lithium-ion batteries—the lithorec-process. ECS Trans Electrochem Soc 73:9
Georgi-Maschler T, Friedrich B, Weyhe R, Heegn H, Rutz M (2012) Development of a recycling
process for Li-ion batteries. J Power Sources 207:173–182. https://doi.org/10.1016/j.jpowsour.
2012.01.152
Mämpel C (2013) Recycling von Lithium-Ionen-Batterien aus Elektromobilen. Bachelor thesis,
Technische Universität Bergakademie Freiberg
Vetter J, Novák P, Wagner MR, Veit C, Möller KC, Besenhard JO, Winter M, Wohlfahrt-Mehrens
M, Vogler C, Hammouche A (2005) Ageing mechanisms in lithium-ion batteries. J Power
Sources 147(1–2):269–281. https://doi.org/10.1016/j.jpowsour.2005.01.006
Vezzini A (2014) Manufacturers, materials and recycling technologies, pp 529–551. https://doi.

org/10.1016/b978-0-444-59513-3.00023-6
Zhang T, He Y, Ge L, Fu R, Zhang X, Huang Y (2013) Characteristics of wet and dry crushing
methods in the recycling process of spent lithium-ion batteries. J Power Sources 240:766–771.
https://doi.org/10.1016/j.jpowsour.2013.05.009
Chapter 8
Separation of the Electrolyte—Thermal
Drying
Friederike Stehmann, Christian Bradtmöller and Stephan Scholl
Abstract The recycled lithium-ion batteries are shredded to access value compo-
nents for further processing. The material that has been shredded before must be
dried. This is important not only because of the simplified separation of dry
material, but also due to safety issues. A material loaded with electrolyte will be
surrounded by a gaseous atmosphere loaded with electrolyte. Such a mixture of
electrolyte and air has an environmental impact and can build an explosive atmo-
sphere. Therefore, undried material requires gas tight as well as explosion protected
equipment. Drying can be achieved by high temperature, low pressure and rinsing
with inertization gas. To support process design and optimization, the influence of
each parameter and advantageous parameter combinations can be determined with a
flow sheet simulation. However, experimental investigations are required to verify
simulation data and to identify critical aspects regarding handling of the material.
This chapter discusses application and results of a flow sheet simulation as well as
experimental investigations on drying.
8.1 Introduction
Dry battery fragments ease the handling in all recycling steps. Wet material sticks to
the equipment walls increasing the cleaning effort and the carry-over between
recycling batches. Regarding the separation process, wet materials stick together
and hinder the separation of different materials. Besides these handling issues, also
F. Stehmann C. Bradtmöller S. Scholl (&)

Institute for Chemical and Thermal Process Engineering, TU Braunschweig,
Braunschweig, Germany
e-mail: s.scholl@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_8
140 F. Stehmann et al.
Table 8.1 Model electrolyte composition that is used in the simulations

Component Formula Electrolytea/ Boiling temperature/ Heating BLEL/
gi g−1
total °C value/ vol%
kJ mol−1
DMC C3H6O3 0.398 90 1373 3.13
EMC C4H8O3 0.250 107 1985 2.17
EC C3H4O3 0.341 248 1169 3.68
CHB C12H16 0.011 240 5920 0.73
a
Composition according model electrolyte considered in the LithoRec process without conducting
salt
safety constraints need to be considered. Wet material can form an explosive

atmosphere resulting in increased equipment costs for explosion protection. Lower
and upper explosion limits (LEL and UEL) identify the region of potentially
explosive gas loadings. The LEL can be calculated according to Nitsche (2015).
The critical fuel content BLEL is reached if the LEL reaches 100% and can be
calculated with Eq. (8.1).
kJ
43007 kmol
BLEL ðVol %Þ ¼ 100 % ð8:1Þ
kJ
H kmol
H represents the heating value of the fuel and can be calculated for the elec-
trolyte components with Eq. (8.2).
g
kJ
H ¼ 4:1876 M ð81:3 C þ 243 H 23:05 OÞ ð8:2Þ
kmol mol
C, H and O represent the wt% of each atom. For instance, the weight fraction of
C in the molecule DMC is calculated according to Eq. (8.3).
~C g
nC M
g
C¼ mol 100 % ¼ 3 12 mol 100 % ¼ 40 % ð8:3Þ
~ DMC g
M 90 mol
g
mol
The resulting heating values and calculated LELs for a model electrolyte are
summarized in Table 8.1. All simulations are based on a model electrolyte with the
volatile components DMC, EMC, EC and CHB. The conducting salt is not con-
sidered. The table also contains the composition of the model electrolyte.
Operating conditions should be kept outside these explosion limits to ensure a
save process. However, a complete removal of the electrolyte is favorable regarding
the solids handling.
8 Separation of the Electrolyte—Thermal Drying 141
8.2 Flow Sheet Simulation of the Drying Process
A flow sheet simulation was used to establish consistent overall material and energy
balances based on the process steps and connectivity. In addition, it is a powerful
tool to identify the sensitivity of single process parameters on the material flows and
to identify promising combinations of process parameters. However, a simulation
must always be verified by experiments. This chapter presents parameter studies
based on the LithoRec process. Due to the variety of process set-ups, parameters
and influenced material flows the presented data can only be a small excerpt. The
processes of shredding and drying are analyzed for their off gas composition and
changes in the composition over time. Furthermore, the demand of nitrogen as
inertization gas and maximum temperature is presented.
8.2.1 Setup of the Simulation
A flow sheet simulation is based on physical properties and phase equilibria as well
as selection and combination of all relevant process functionalities. This simulation
was conducted for stationary process conditions using the commercial flow sheet
simulation tool CHEMCAD. The existing database was supplemented with
self-determined substance data. A continuous process can easily be depicted in this
software. However, the steps in the LithoRec process were set up in batch mode. To
depict the time dependency of a batch process the simulations were set up in
discretized sections, see Stehmann et al. (2017). The composition of the liquid
electrolyte in the shredded battery fragments entering the drying process was set
according to the composition in Table 8.1.
8.2.2 Sensitivity Study of Parameters Related

to the Drying Process
The evaporation of a mixture, consisting of low-boiling (DMC and EMC) and

high-boiling components (EC and CHB), will lead to a change in composition of
liquid as well as gas phase over time throughout the drying progress. Low-boilers
will accumulate in the gas phase, while high-boilers accumulate in the liquid phase.
The resulting change of composition leads to an increase of the temperature that is
required for drying or to a higher demand of nitrogen after a maximum temperature
is reached. These time dependencies are displayed in the following sections.
Figure 8.1 shows the vapor pressure and boiling temperature of the liquid that is
present in the shredded material depending on the drying progress. The drying
progress is defined as the mass fraction of the already removed carbonates. Along
the drying process, first low-boiling components are removed, while high-boiling
components accumulate. This leads to an increasing boiling temperature of the

electrolyte mixture remaining in the solid material. The boiling temperature can be
reduced if the overall pressure of the system is lowered. The boiling temperature at
a pressure of 0.1 mbar is about 60–80 K lower than at ambient pressure. If the
drying process is only driven by temperature, meaning an evaporation without the
use of an inert gas, the boiling temperature is required for a sufficient drying.
However, the required temperatures can also be decreased by decreasing the partial
pressure of the evaporated components. This can be achieved by decreasing the
overall pressure or the partial pressure itself keeping the overall pressure at ambient
conditions by applying an inert gas flow.
Figure 8.1a also depicts the vapor pressure of the residual mixture at a given
drying progress at 25 and 35 °C. The vapor pressure decreases with the drying
progress due to a higher fraction of high-boiling components. Nitrogen demand will
increase with decreasing vapor pressure of the residual liquid. Therefore, it is
beneficial, if the liquid has a considerable vapor pressure. The vapor pressure of the
system approaches a very low value of 0.1 mbar at a drying progress of 70% at 25 °
C as well as 35 °C. This shows that a drying at ambient temperature might not be
beneficial as the nitrogen demand increases exponentially. Even though, if the
vapor pressure drops significantly, the LEL is kept and the system is not explosive.
Figure 8.2 depicts the change of composition of the remaining liquid in the
material to be dried. At the beginning of the drying process, the mol fraction of
high-boiler in the remaining liquid increases, since initially only low-boilers DMC
and EMC are removed. The increasing fraction of high-boilers increases the
required temperature to enable the evaporation of liquid. Eventually, the tempera-
ture reaches the defined maximum temperature of 120 °C. From this point on,
nitrogen has to be fed to the system to remove the high boiling components.
Alternatively, a reduction of the total pressure to decrease the vapor pressure of the
mixture is possible.
(a)
100
(b)
250
boiling temperature / °C
vapor pressure / mbar
10 200
150
1
100
0.1
T = 25 °C 50 p = 1 bar
T = 35 °C p = 0.1 bar
0.01 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
drying progress / gevaporated gtotal-1 drying progress / gevaporated gtotal-1
Fig. 8.1 Vapor pressure (a) and boiling temperature (b) of remaining liquid in the battery
fragments
composition of remaining 1.0 140 DMC

EMC
liquid / moli moltotal-1
0.8 120 EC
temperature / °C
CHB
temperature
0.6 100
0.4 80
0.2 60
0.0 40
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
drying progress / gevaporated gtotal-1
Fig. 8.2 Mol fraction of carbonates in the residual liquid within the material to be dried along the
drying progress. Maximum temperature is set to 120 °C at a pressure of 0.1 bar
Fig. 8.3 Nitrogen demand 180 2

nitrogen demand with Tmax=120°C
nitrogen demand / gN2 gLiquid-1

dependent on the maximum
temperature at 0.1 bar Tmax=160°C
temperature / °C
160 1.5
Tmax=140°C
Tmax=140°C
140 1
Tmax=120°C
120 0.5
Tmax=160°C
100 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
drying progress / gevaporated gtotal-1
Figure 8.3 depicts the influence of the maximum temperature on the nitrogen
demand and temperature profile for a drying process at 0.1 bar. As shown before,
the temperature must be increased during the drying process to enable the removal
of electrolyte. After the defined maximum temperature is reached, a nitrogen flow is
applied to allow further drying. For lower maximum temperatures, the demand of
nitrogen is significantly higher.
As shown above, the composition of the residual liquid determines drying
conditions: temperature, total pressure or nitrogen demand. The LithoRec process
includes a solvent extraction prior to drying. The extraction is used to remove the
conducting salt, but also has the chance of a reduced high-boiling content in the
material depending on the choice of the extraction agent. High excess of the solvent
used for extraction will lead to a dilution of the electrolyte with solvent. Dimethyl
carbonate is preferably used as an extraction solvent, since it is already present in
the process. Additionally, DMC has the advantage to be low boiling. Hence, it can
be easily removed in the drying process. Increasing the amount of extraction sol-
vent will decrease the content of high boiling components in the feed material for
fraction of low-boiler / mol mol -1

0.8
nitrogen demand / t
fraction of low-boiler in
gas after drying
0.6
nitrogen demand
0.4
0.2
0
0 5 10
solvent excess E Emin-1 / mol mol-1
Fig. 8.4 Nitrogen demand for sufficient drying and fraction of low-boiler in the gas after drying.
Based on 1 battery system = 220 kg and a maximum temperature of 60 °C (Stehmann et al. 2017)
drying. The minimum solvent demand Emin is given by the liquid hold up of the
solid material. Pretests determined a ratio of Emin = 1 kgsolvent kg−1
solids. Figure 8.4
shows that the nitrogen demand during drying decreases with increasing solvent
excess E E−1
min.
The extraction also influences the composition of the off gas. Considering a
subsequent condensation at 20 °C, the resulting off gas composition is depicted in
Fig. 8.5. These results were obtained with a model electrolyte only consisting of
DMC, EMC, EC and CHB. Without extraction (Graph a) the concentration course
shows the analogue behavior as before. At the beginning of the drying progress,
mostly low-boilers are present in the off gas. Only after 60% of the drying progress,
the concentration of the high-boiling component EC increases. The concentration
courses smoothen if the shredder-output is extracted with DMC prior to drying.
1E+01 1E+01
DMC
gas loading / mmol L -1
gas loading / mmol L -1
DMC
1E-01 EMC 1E-01
1E-03 EC 1E-03 EMC

E Emin-1=0 E Emin-1=10
A B
1E-05 Tc= 20 °C 1E-05 Tc= 20 °C
CHB
1E-07 CHB 1E-07
EC
1E-09 1E-09
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
drying progess / mol mol -1 drying progess / mol mol-1
Fig. 8.5 Composition of off gas with condensation temperature of 20 °C without extraction
(a) and with E/Emin = 10 (b)
This leads to a constant DMC concentration in the off gas and a significant decrease
of EMC and EC. A constant off gas concentration is favorable for the design of a
subsequent off gas cleaning.
8.3 Experimental Investigation on Thermal Drying

of Battery Fragments
Drying experiments were conducted by Solvay GmbH, Hannover. The experiments

were carried out with shredded batteries that were prepared in the mobile shredder;
see the previous chapter “Crushing of Battery Modules and Cells”. The material to
be dried was stored in a PE-bag and a PE-barrel to avoid contact to the ambient as
well as evaporation of the carbonates prior to drying.
8.3.1 Small Scale Experiments
Small-scale experiments were carried out in a 10 L shovel dryer (AVA-Huep

GmbH & Co. KG, Herrsching, Germany), see Fig. 8.6. The dryer was equipped
with two condensers and a vacuum pump (P1), see Fig. 8.7. The first condenser was
operated with tap water to recover high boiling components such as PC and EC.
The second condenser was operated with a cryostat at 5 °C to recover most of the
low-boiling components. A vacuum pump was installed between both condensers
to evacuate the drying chamber. The off gas was cleaned by absorption.
Fig. 8.6 Shovel dryer HTL-VT 10 (left) with drying chamber including shovels (right)
Fig. 8.7 Experimental setup including dryer HTL-VT 10, two condensers and vacuum pump (P1)
The shredded battery fragments were manually presorted prior to the experi-
ments to remove large fragments such as screws that would block the shovels of the
dryer. The presorting was carried out at ambient conditions. In this experiment,
shredded stacks consisting of small Sanyo Cells (*450 g/cell) were dried.
The material was stirred at 62 rpm and the drying chamber evacuated down to
40 mbara. After 45 min of operation, the temperature of the heating fluid was
increased from ambient temperature to 80 °C. At this temperature, low amounts
(some mL) of colorless liquid appeared in the first condenser. After the volume of
condensate did not visibly increase any further, the temperature was increased to
120 °C.
The experiment was repeated three times. However, at some point in each
experiment the stirrer was blocked by some massive metal parts und the experiment
could not be finished. Therefore, the shovels were equipped with PTFE-plates
(Fig. 8.8), to prevent blocking. With this system, the operation time could be
Fig. 8.8 Metal fragments sorted out prior to drying (left) and shovel equipped with PTFE-plates
(right) to prevent blocking
extended up to 5–6 h but at some point, each experiment had to be stopped because
of blocking problems.
8.3.2 Pilot Scale
The experiments were repeated in a dryer with a volume of 140 L (HTC-VT 140,
AVA-Huep GmbH & Co. KG, Herrsching, Germany). This drier has the advantage
of a larger distance between shovels and wall and a more powerful propulsion.
Furthermore, the larger filling volume enables a more detailed and representative
investigation of the drying process.
The horizontal dryer is equipped with four shovels driven with a shaft beared on
both sides. A vapor filter is positioned above the drying chamber for evacuation and
to remove the exiting vapor. The drying chamber is equipped with a double shell
that can be kept at a constant temperature using an oil thermostat. Two cold traps
located downstream the vapor filter ensure the removal of organic carbonates. The
cooling traps are filled with a mixture of isopropanol and dry ice to achieve a
temperature of −80 °C. The plant was evacuated with a scroll pump (Edwards XDS
10). The exiting off gas was once again led through a gas washer containing dry ice.
A flow diagram of the dryer and its periphery is shown in Fig. 8.9. The dryer can be
filled via a door that also carries the vapor filter, see Fig. 8.10.
While filling the material into the dryer, the bag and environment were aspirated
due to safety issues. Afterwards the door was tightly closed. To start the drying
process, the agitator was set to 50 rpm, followed by evacuation and heating. The oil
thermostat was operated with a heating ramp from 20 °C up to 70 °C, 100 °C and
finally 120 °C. The temperature inside the drying chamber was monitored with a
thermocouple. The mass of the inserted and the dried battery fragments as well as
the condensate were determined gravimetrically. Three experiments were carried
out and are described hereafter. The pressure and temperature courses of all
Fig. 8.9 Shovel dryer HTC-VT 140 (AVA-Huep GmbH & Co. KG, Herrsching, Germany) at
Solvay GmbH, Hanover and flow diagram including the periphery
Fig. 8.10 Insight of the drying chamber with agitator shaft and shovel
experiments are depicted in Fig. 8.11. Additionally, key data of each experiment
are summarized in Table 8.2.
The drying process was operated for 5–8 h. In the first experiment a mixture of
Sanyo and JCS cells were dried, while experiment 2 and 3 were operated with only
Sanyo and/or JCS cells, respectively. The pressure was set as low as possible,
reaching 200 mbar in the first and 200–300 mbara in the third experiment. During
the second experiment leaking problems occurred, leading to an unstable and higher
pressure. Each drying process was paused several times to empty the cooling traps
resulting in a short increase of pressure as it can be observed in every experiment.
The temperature in the drying chamber was approximately 10–20 K below the oil
temperature.
In the first minutes of the experiments, flaky crystals were observed in the
cooling traps followed by liquid condensate. Each time no further condensation was
observed, the temperature was increased, leading to a temporary increase of con-
densation. After the drying process, the dried material was transferred into a
PE-barrel. The dried material showed differences to the humid material. The foils
appeared to be shredded and the active material, that was humid and attached to the
foils prior to drying, appeared to be a loose dry powder.
The leaking problems during the second experiment resulted in a non-sufficient
drying. The product still contained a high amount of humidity after drying. The
higher pressure also led to a crystalline fouling at the dryer exit, where the vapor
filter is attached (Fig. 8.12). High-boilers, such as EC may have crystallized here,
since this part was not actively heated.
The highest weight loss (20.1%) could be achieved in the first experiment. The
low weight loss in experiment 2 is a result of the not-effective drying. In addition,
the weight loss in experiment 3 is lower (13.8%). However, since the product
appeared to be dry, it is likely, that the starting material contained less volatile
electrolyte.
During the first experiment, 55% of the weight loss could be recovered in the
condensate. The recovery ratio increases for each experiment up to 90% in
1200 120
No. 1
1000 100
temperature / °C
pressure / mbara
800 80
600 60
400 40
200 20
0 0
0 1 2 3 4 5 6
time / h
1200 120
No. 2
1000 100
temperature / °C
pressure / mbara
800 80
600 60
400 40
200 20
0 0
0 1 2 3 4 5 6 7
time / h
1200 120
No. 3
1000 100
temperature / °C
pressure / mbara
800 80
600 60
400 40
200 20
0 0
0 1 2 3 4 5 6 7 8
time / h
pressure oil temperature product temperature
Fig. 8.11 Pressure and temperature course of all experiments
experiment 3. It is likely, especially between experiment 2 and 3, that a certain

amount of electrolyte remained in the equipment and was carried over to the next
experiment, falsifying the mass balance.
Table 8.2 Concluded data of drying experiments

No. 1 2 3
Material Sanyo-cells/ Sanyo-cells JCS-cells
JCS-cells
Mass ina/kg 14.62 16.38 11.46
Mass outa/kg 11.68 14.56 9.88
Weight loss/kg/% 2.94 1.82 1.58
20.1 11.1 13.8
Mass condensatea/kg 1.63 1.13 1.42
Recovery ratiob/% 55 62 90
Throughputc/kg h−1 m−3 19.7 17.5 10.5
a
Determined gravimetrically
b
(mass condensate) (weight loss) −1
c
(mass in) (operation time)−1 (dryer volume)−1
Fig. 8.12 Fouling at dryer

exit and entry of vapor filter
The collected condensate fractions were combined and analyzed with gas chro-
matography without calibration, see Table 8.3. The analysis therefore only provides
information about the composition and not the total mass of the components. It is
likely that more components are present that are not detectable with this analysis
method. The cooling traps were kept at a temperature of −80 °C, therefore, it can be
assumed, that only a negligible part of the electrolyte passed the cooling traps
without condensation. Besides common electrolyte components, also decomposition
products methanol, ethanol and DEC are detected with a low percentage. However,
the main components are the electrolyte constituents DMC, EMC, EC and PC. The
content of the low boiling components in the condensate is similar to the model
electrolyte in the LithoRec process. The high-boiling component EC was detected
with only an area percentage of 12.1% in the condensate. The model electrolyte
Table 8.3 Composition of Component Condensate/Area % Electrolytea/wt%

condensate and model
MeOH 0.96 –
composition of electrolyte
EtOH 0.36 –
DMC 41.96 39.8
EMC 29.58 25.0
DEC 0.25 –
EC 12.18 34.1
PC 9.12 –
CHB 2.83 1.1
a
Composition according model electrolyte considered in the
LithoRec process without conducting salt
contains 34.1 wt% of EC. This loss may result from an uncomplete drying. PC was
not included in the model electrolyte. The loss of high-boiler agrees with the weight
loss in Table 8.2 and the course of drying, determined in the simulation. These
simulations identified that the high-boiling components will be removed only at the
end of the process and may already appear to be solid, if all low-boiling components
are removed. An uncompleted drying process will therefore lead to a reduced
recovery of high-boiling components in the condensate.
8.4 Conclusions
8.4.1 Transferability of Simulation Results on Experiments
The simulated results were obtained assuming a thermodynamic equilibrium that

may not be present in a real process. Therefore, deviations between simulation and
experiments are likely. However, the simulation is a powerful tool to show main
influences on the material flow and promising parameters for optimization. The
identified influences of the parameters are summarized in the following.
Drying temperature
Higher temperatures enhance drying with respect to lower humidity at the end of
the drying process and a higher drying rate. Higher temperatures are needed at the
end of the drying process to remove high boiling components. However, high
temperatures can lead to decomposition of conducting salt and increase energy
demand.
Drying pressure
A reduction of pressure decreases the boiling temperature of the electrolyte present
in the solid material and therefore reduces the required drying temperature.
However, the equipment must be suitable for vacuum operation. Leaking can cause
major fouling problems, since the electrolyte may crystallize due to low tempera-
tures if the pressure is not kept at a constant low level.
Inert gas flow
Purging with inert gas such as nitrogen lowers the partial pressure of the electrolyte
components and therefore enhances the drying rate. However, the off gas is diluted
with inert gas impeding the subsequent condensation.
Mass of extraction solvent
An extraction solvent containing only low boiling components decreases the high
boiler composition and therefore enhances drying. A higher mass of extraction
solvent also increases the solvent recycling costs and the required extractor volume.
8.4.2 Recommendations for the Design of a Drying Process
The results show, that a shovel dryer is a suitable device to dry battery fragments.
Based on the results, it can be assumed, that even at optimum conditions one batch
will need one-day shift to be dried. The experiments had a throughput of 10.5 to
19.7 kg h−1 m−3 dryer volume. In accordance to that, the drying process in an
industrial plant needs to be scaled up.
Additionally, several critical parameters were observed in the experiments:
• Blocking of the shovels: Massive metal fragments can cause a blocking of the
shovels. The blocking can be reduced by an appropriate shovel design and a
powerful propulsion.
• Leakages: Insufficient vacuum during the drying process inhibits the drying of
high boiling components. Even longer process times will not compensate the
insufficient vacuum.
• Fouling: Sections of the drying apparatus, that are not actively heated, may have
a lower temperature and cause the condensation of the vapor. High-boiling
components may as well crystallize in these areas building a solid fouling layer.
8.4.3 Further Research Needs and Outlook
As shown in the simulations, the concentrations of the electrolyte components in

the off gas change along the drying progress. A fractioning of the gained condensate
will ease the purification of single components. The feasibility needs to be analyzed
experimentally.
Drying experiments were only conducted at 120 °C and without extraction and
without nitrogen flow. Optimization of the drying process will require a temperature
reduction and nitrogen purge flow.
The presented simulation only covers the specified LithoRec process with the
model electrolyte composition. The main influences of the parameter might be
generally valid. However, to receive absolute values, the set-up of the simulation
needs to be adjusted individually to the considered process.
References
Nitsche M (2015) Abluft-Fibel: Reinigung lösemittelhaltiger Abgase. Springer Vieweg, Berlin

Stehmann F, Jahnke S, Balmforth-Slater C, Scholl S (2017) Adsorptive exhaust gas cleaning for
recycling of Li-Ion-batteries. J. Electrochem, Soc, p 164
Chapter 9
Separation of the Electrolyte—Solvent
Extraction
Paul Haas, Stefan Pfeifer, Jannes Müller, Christian Bradtmöller

and Stephan Scholl
Abstract The extraction of electrolyte from lithium-ion batteries is a possibility to

remove the high boiling organic components and the conducting salt from the
battery material in the recycling of lithium-ion batteries. In these studies, dimethyl
carbonate was employed as organic solvent. The influence of temperature and
solvent to solid mass ratio have been tested with Panasonic CGR 18650 batteries in
a stirred vessel. Although the conducting salt was successfully extracted with a
crossflow extraction with four stages, the remaining fluoride loading of the battery
material was too high for further processing. Therefore, a second series of
extractions with water was used to remove fluoride. The combination of four
extraction stages with dimethyl carbonate and six stages with water for the pro-
cessing of hybrid electric vehicle batteries resulted in a fluoride loading of
173.3 mg per kg fine fraction of the raffinate. Suggestions for the design of an
extraction apparatus based on the experiments were worked out.
9.1 Introduction
The aim of the solvent extraction is the recovery of the electrolyte components and
to ensure the requirements of the subsequent steps in the processing of the active
material in respect to the fluoride content. In the solvent extraction the extraction
agent is mixed with grinded battery material and the conducting salt and organic
carbonates are transferred to the liquid phase. The solid material is dried afterwards
to evaporate the remaining light boiling extraction agent. Thus, only small con-
centrations of the electrolyte remain in the battery material which is transferred to
the hydro and pyro metallurgic processes to recover the metallic value compounds
P. Haas S. Pfeifer J. Müller C. Bradtmöller S. Scholl (&)

Institute for Chemical and Thermal Process Engineering,
TU Braunschweig, Braunschweig, Germany
e-mail: s.scholl@tubraunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_9
156 P. Haas et al.
Fig. 9.1 Preceding and

following process steps of
extraction
(lithium, manganese and cobalt) from the anodes and cathodes. The extraction of
the electrolyte with solvents which are present in lithium-ion batteries enables
following processes to be conducted in a safer way due to removing toxic con-
ducting salt and its decomposition products. The removal of organic solvents is
performed to avoid formation of an explosive atmosphere in subsequent process
steps. Furthermore, the separation of the organic solvents and conducting salts
allows to recover the value compounds from the electrolyte and the coatings of the
electrodes with the intended process pathways. In Fig. 9.1 the planned process
sequence at the beginning of the project is displayed.
The integration of the extraction step allows for the removal of the conducting
salt and high boiling compounds. Thus, the latter steps can be performed with less
hazardous material and less effort in terms of the drying. The recovery of the
conducting salt and organic electrolyte components is possible due to the transfer
into liquid phase and phase separation from solid phase.
9.2 Basics for the Extraction of Electrolyte Compounds
9.2.1 Battery Material and Electrolyte Composition
Decomposition of electrolyte compounds

The conducting salt lithium hexafluorophosphate dissociates to lithium fluoride and
phosphorus pentafluoride in non-aqueous solvents.
LiPF6 ðsolÞ
LiF ðsÞ þ PF5 ðsolÞ ð9:1Þ
Lithium hexafluorophosphate reacts with water and hydrofluoric acid and phos-
phoryl fluoride are the result.
LiPF6 ðsolÞ þ H2 O
LiF ðsÞ þ 2HF ðsolÞ þ POF3 ðsolÞ ð9:2Þ
The resulting products are toxic and acidic and are reacting further to phosphoric
acid (Xu 2004; Grützke et al. 2015; Terborg 2011; Korthauer 2013; Kawamura
et al. 2006). Further information related to the degradation of the electrolyte can be
found in (Nowak and Winter 2015).
9 Separation of the Electrolyte—Solvent Extraction 157
Organic carbonates
The batteries contain dimethyl carbonate (DMC), ethyl methyl carbonate (EMC),
ethylene carbonate (EC), propylene carbonate (PC) and cyclohexyl benzene (CHB).
DMC and EMC are light boiling components with the boiling temperatures 90 °C
respectively 107 °C at ambient pressure. The high boiling solvents are 243–244 °C
(EC), 240 °C (PC) and 239–240 °C (CHB).
9.2.2 Experimental Setup
Extraction apparatus
The stirring vessel for the experiments was a Versoclave 3 (Büchi Glas Uster, Uster,
Switzerland) with a 0.5 and 5 L vessel. The temperature was set by a CC 405 (Peter
Huber Kältetechnik GmbH, Offenburg, Germany) and controlled by an internal
temperature probe. The whole vessel was made from stainless steel. The apparatus
was located in a chamber flushed with nitrogen and the vessel was filled with argon
prior to the start of the process. Most of the experiments were conducted at pressure
of 2 bar absolute or slightly higher. For the extraction kinetics a pneumatic sample
system was used to take samples during the experiment from the center of the vessel
using internal pressure. For the other experiments the samples were taken through a
valve with a 2 µm filter at the bottom of the vessel. For later experiments the filter
and the stirrer were modified. The stirrer shaft was shortened and the anchor stirrer
was replaced by a propeller stirrer. The 2 µm stainless steel filter matrix was fixed
by a bracket at the bottom of the vessel. The outlet was controlled by an external
valve. Thus, the blocking of the small filter area because of the accumulation of fine
particles was reduced and the outlet time for the liquid phase was reduced. It was
necessary to increase the pressure to enable the outlet. The version with 2 µm filter
at the outside of the stirred vessels bottom is shown Fig. 9.2. The extractions with
water have been conducted with argon in the vessel but without nitrogen. For the
determination of the temperature influence in Sect. 9.3.3 the experiments were
performed in a glass beaker because the inert gas atmosphere was not necessary.
Furthermore, the handling was improved.
Analytical setup
The focus of analysis was the determination of hexafluorophosphate and decom-
position products. For this reason the ion chromatograph 881 IC compact in
combination with 919 IC Autosampler (Deutsche Metrohm GmbH & Co. KG,
Filderstadt, Germany) was used. The ion chromatography and accessory was
controlled by using MagIC Net software by Metrohm. All columns were supplied
by Metrohm (Deutsche Metrohm GmbH & Co. KG, Filderstadt, Germany) as
shown in Table 9.1. A linear matrix elimination was integrated with an external
6-port-valve (VICI Vlaco Instruments Co. Inc., Schenkon, Switzerland). The flow
rate of eluent was 0.7 mL/min and a temperature of 55 °C was set in column oven.
The composition of eluent was 0.7 L ultrapure water, 0.3 L acetonitrile and 10 mL
158 P. Haas et al.
Fig. 9.2 Büchi Glas Uster Versoclave 3 with anchor stirrer and filter at the bottom, 0.5 L vessel
on left side
Table 9.1 Parameters for the Parameter Setting

IC analysis
Flow rate eluent 0.7 mL/min
Oven temperature 55 °C
Sample volume 20 µL
Run time 15 min
Trap column Metrosep A Trap 1–100/4
Guard column Metrosep RP 2 Guard/3.5
Pre-column Metrosep A Supp 4/5 Guard 4.0
Column Metrosep A Supp 5 100/4.0
Ghost column Metrosep A PCC 1 HC/4.0
1 M sodium carbonate and 2 mL 1 M sodium hydrogen carbonate.1 The eluent was

degassed before use. For the calibration fluoride and phosphate standards (Merck
KGaA, Darmstadt, Germany) and lithium hexafluorophosphate salt 1 M dissolved
in a mixture of DMC and EC supplied by the MEET Battery Research Center
(MEET Battery Research Center, University of Münster, Münster, Germany). The
samples for the extract analysis were diluted with DMC and samples for determi-
nation of fluoride concentration were diluted with ultrapure water. From the anion
chromatography result for hexafluorophosphate the mass of lithium hexafluoro-
phosphate was determined. The extract samples were prepared by filtration with a
0.45 µm pre-syringe filter and diluted with DMC.
1
The basic parameters were derived from a method note by Deutsche Metrohm GmbH &
Co. KG, Filderstadt, Germany.
For the determination of fluoride and hexafluorophosphate in the solid phase a part
of the fine fraction of the dried raffinate was dispersed in ultrapure water. A sample
of the water was taken and filtered with a 0.45 µm filter, diluted with ultrapure
water and measured with ion chromatography.
Battery material
For most of the experiments Panasonic CGR 18650 battery cells were used. They
were ground in the “Battery analysis mill”, a modified SM 200 cutting mill (Retsch,
Haan, Germany), in nitrogen atmosphere after discharging and packaged immedi-
ately in inert gas atmosphere. The grinding apparatus is described by detail in
Chap. 7 of this book. The battery weight of a single cell was 42.64 g on average
before the grinding and 37.79 g afterwards. The electrolyte was 15–20 wt% in the
investigated batteries. The organic carbonates were analyzed and their mass fraction
in the electrolyte was assumed to be 40 wt% DMC, 20 wt% EMC, 10 wt% PC and
30 wt% EC. The conducting salt concentration was 1.18 mol LiPF6/L electrolyte.2
In the first experiments and latter multi-stage extractions PHEV 1-format bat-
teries with 25 Ah were used. The battery material for the initial experiments was
shredded at the site of Electro Cycling GmbH (Goslar, Germany). The grinding was
performed in inert gas atmosphere, but the batches have not been sealed to maintain
the nitrogen atmosphere to avoid further degradation. For the second experiment in
Sect. 9.3.4 the material was grinded by Volkswagen AG (Wolfsburg, Germany).
These batteries are used in hybrid electro vehicles and a single cell weight is about
1 kg. They were from hybrid electric vehicle systems and consisted mainly of
batteries from the same kind as used for latter experiments.
9.3 Experimental Progress and Transfer to Process Design
9.3.1 Single-Stage Extractions
Comparison of different extraction agents

Acetonitrile, ethanol and acetone were tested in comparison to DMC, the lowest
boiling organic component from the battery. The parameters for the experiments are
listed in Table 9.2. The experiments have been performed in nitrogen atmosphere in
a glovebox. The samples were immediately analyzed subsequently after the
extraction and after 48 h storage at 8 °C. The samples have been sealed to avoid
degradation due to other mechanisms than influence of the solvent.
The hexafluorophosphate concentration in different solvents is displayed in
Fig. 9.3. DMC showed the highest concentration in extract after the extraction and
also after the 48 h storing. The increase of hexafluorophosphate in ethanol after
storage could only be explained by evaporation of the solvent or errors in the
2
The content of LiPF6 in the Panasonic CGR 18650 batteries was determined by the MEET
Münster (WWU Münster, Münster, Germany).
160 P. Haas et al.
Table 9.2 Parameter for the Parameter Values

solvent screening
Battery material Panasonic CGR 18650
Mass of the solid 2.5 g
Mass of solvent 25 g
Extraction time 5 min
Runs per solvent 2
Fig. 9.3 Hexafluorophosphate concentration in extract with different solvents
dilution of the samples. For latter analysis there was no evidence of fluoride or other
traces which indicated degradation. DMC is preferred as it is already present in the
battery and therefore no additional substance which has to be separated in the
downstream processing. Hence, DMC was selected for the further experiments.
Analysis of the grinded battery material
The PHEV battery material grinded by Electrocycling GmbH was used in first tests.
The necessary amount of solvent for the wetting of the battery material was
determined by pouring 60 mL DMC over 50 g of battery material in a beaker glass.
The remaining liquid was separated from the solid material and by calculation in
reference to the solid mass the liquid retention or the necessary amount of DMC for
the wetting of the battery material was calculated to be (0.98 ± 0.09) kg solvent/kg
solid material. The bulk density of the ground battery material is 150 kg/m3.
Mass ratio of solvent to solid
Table 9.3 shows the parameters for the determination of the influence of the solvent
to solid mass ration S/L.
In Fig. 9.4 the calculated absolute mass of Lithium hexafluorophosphate in
extract over time for three different mass ratios of solvent to battery material L/S (5,
7.5 and 10) is displayed. The largest absolute mass of lithium hexafluorophosphate
in the extract was obtained with the highest solvent to solid mass ratio of L/S = 10.
The amount of lithium hexafluorophosphate for L/S = 5 is higher than for L/
S = 7.5. This unexpected result is probably caused by a deviation in the probe
material due to different storing times. The delay in the increase of salt
Table 9.3 Parameter for the influence of mass ratio of solvent to solid
Parameter Values
Battery material PHEV 1 (Electrocycling)
Mass of the solid 30 g for ratio of 10 and 50 g for ratio of 5 and 7.5
Mass of DMC 250 g (ratio of 5); 375 g (ratio of 7.5) and 300 g (ratio of 10)
Sample volume 3 mL
Stirrer speed 100 rpm
Temperature 30 °C
Fig. 9.4 Absolute extracted mass of lithium hexafluorophosphate for different solvent to solid
mass ratios
concentration for the solvent to solid ratio of L/S = 10 is also unexpected and the
reason could be within the sample preparation system. For this set of experiments,
the highest concentration was achieved after 30 min. Afterwards the concentrations
decrease because of the degradation of the conducting salt during extraction or
between sampling and measurement. Therefore, the extraction time has to be
limited to reduce losses and the samples must be sealed. Furthermore, the grinding
and storing of battery material has to be performed with inertization to avoid the
degradation of lithium hexafluorophosphate.
Temperature
The temperature has a major impact onto the decomposition of LiPF6 but also on
the dissolving of conducting salt in solvent. Therefore, the extraction should be
performed at the highest possible temperature with an acceptable degree of
decomposition.
The concentrations of the samples for the extraction of lithium
hexafluorophosphate in the extract for temperatures of 10, 20, 30 and 40 °C are
displayed Fig. 9.5 over the extraction time with the maximum time of 120 min. The
parameters for the experiments are listed in Table 9.4. The concentration for all
162 P. Haas et al.
Fig. 9.5 Temperature influence on extraction of lithium hexafluorophosphate with PHEV 1

battery material from Electrocycling GmbH
Table 9.4 Parameter for the temperature influence experiments with PHEV 1 batteries from
Electrocycling GmbH
Parameter Values
Battery material PHEV 1 (Electrocycling)
Mass of the solid 30 g
Mass of DMC 300 g
Sample volume 3 mL
Solvent DMC
Times of sampling [min] 0, 1, 2, 3, 5, 10, 15, 30, 60, 120
Working in nitrogen atmosphere No
temperatures beside 40 °C is increasing over time, while for 40 °C the concen-

tration declines after 5 min. This indicates the decomposition of the conducting salt.
The highest concentration of 1.24 g PF6−/kg extract after 60 min was achieved by
choosing 30 °C. The experiments were repeated with the model battery system, the
Panasonic CGR 18650 cells. The parameters are listed in Table 9.5.
Figure 9.6 shows concentrations of hexafluorophosphate over time for different
temperatures. The experiments have been performed with single Panasonic 18650
cells and the extraction time was limited to 30 min. For this experiment the highest
values are achieved at 20 °C with 6.3 g PF6−/kg extract while for the higher
temperatures and 10 °C the values are lower. For 30 and 40 °C the value declines
after 15 min which indicates the decomposition of conducting salt. For 10 °C the
concentrations are slightly increasing with a maximum PF6− concentration in
extract of 3.1 g PF6−/kg extract.
Table 9.5 Parameter for the temperature influence experiments with Panasonic CGR 18650
batteries
Parameter Values
Mass of the solid (10, 20, 30 and 40 °C) [g] 36.50, 37.21, 37.79 and 37.15
Mass ratio of solvent to solid for the first stage 7.5
Times of sampling [min] 1, 3, 5, 10, 15, 30
Working in nitrogen atmosphere Yes
Fig. 9.6 Temperature influence on the extraction using Panasonic CGR 18650
The results of these two sets of experiments differ from each other. In both
experiments are the low concentrations were achieved at 10 °C and degradation of
the conducting salt was seen at 40 °C. For 20 °C and 30 °C the results differ for both
sets. The reasons are within test material, the processing and the sample preparation.
For the latter set, the 18650 batteries were prepared individually for the experiments
and were grinded and sealed in nitrogen atmosphere. The filling of the vessel and
sample preparation were performed in nitrogen atmosphere to avoid degradation. As
mentioned above the batteries from EC have been grinded in nitrogen atmosphere,
but storage was at ambient atmosphere. The material for the experiments was taken
from one batch. The storage times are different for each experiment. The generally
higher concentrations in the latter set of experiments show the efficient prevention of
degradation. This shows the necessity of subsequent handling and working in inert
gas atmosphere. The following multi-stage extractions were performed at 20 °C due
to longer process times to avoid degradation.
164 P. Haas et al.
9.3.2 Multi-Stage Extractions
The specification for hydrometallurgical processes of 10 mg fluoride per kg active

material was not met with the single-stage extractions. The remaining fluoride,
including the hexafluorophosphate concentration, was still too high. Thus,
multi-stage extractions were tested to determine the necessary amount of solvent
and the necessary number of extraction stages to remove the hexafluorophosphate
as far as possible.
The multi-stage extractions have been performed in crossflow configuration.
Therefore, the solid material was filled into the stirred vessel and remained in the
vessel after each extraction step, while the fluid exited the vessel through a filter at
the bottom after extraction time. After each stage fresh solvent was added. The
extraction time started when all phases were in the vessel and when the time was
over the drainage of the liquid was started. However, in reality the contact time of
solid and liquid phase is longer because of the time required for the draining.
Four stage extraction
The experimental procedure and sampling was tested in up to three stage experi-
ments. The first stage was performed with a solvent to solid mass ratio of 10, while
the following stages have been performed a ratio of 7.5 (Table 9.6, Fig. 9.7).
The additional solvent in the first stage was added corresponding to the retention
of liquid after draining in the first stage.
In Fig. 9.8 the mass fraction of hexafluorophosphate and fluoride in the extract
after each stage are displayed at a logarithmic scale. After the first stage fluoride
was detected while no fluoride was present in extracts of following stages. This
indicates the absence of degradation processes of hexafluorophosphate to fluoride
during extraction but decomposition has happened during usage, discharging and
the grinding of batteries. This must be taken into account for the process design.
The mass fraction of hexafluorophosphate in the extract decreases significantly after
consecutive stages.
Table 9.6 Parameter for four stage extraction

Parameter Values
Mass ratio solvent to solid 10 (first stage), 7.5 (other stages)
Sample volume 3 mL
No. of samples first stage 6
No. of samples for each following stages 4
Solvent DMC
Temperature 20 °C
Extraction time per stage 30 min
Fig. 9.7 Mass flow chart for the four stage extraction
Fig. 9.8 Mass fraction of

hexafluorophosphate and
fluoride in extract after
extraction stages
Fig. 9.9 Accumulated

absolute mass of
hexafluorophosphate in
extract
The absolute mass of extracted lithium hexafluorophosphate in the extract is

displayed in Fig. 9.9. The overall extracted mass is increasing by 0.025 g lithium
hexafluorophosphate for the third extraction stage and 0.006 g lithium
hexafluorophosphate for the fourth stage. The difference between the stages beyond
stage 3 is minimal. After four stages 1.31 g LiPF6 have been extracted overall. The
calculated mass of LiPF6 in this battery type from analysis was 1.27 g. For the
following process was the fluoride and hexafluorophosphate loading in the fine
166 P. Haas et al.
fraction are crucial parameters. A fraction of raffinate of 56.56 g was dried at 105 °C
for three days to evaporate the organic solvents and sieved with Alpine e200 LS
(Hosokawa Alpine AG, Augsburg, Germany) with 200 µm sieve mesh. The loading
of fine fraction was 3558 mg fluoride and 233 mg hexafluorophosphate per kilogram
fine fraction. From Fig. 9.8 the degradation is visible from the fluoride concentration
in the extract. Therefore, it is not possible to completely extract lithium
hexafluorophosphate with DMC. A major part of the residue is fluoride as result of
the degradation, but also lithium hexafluorophosphate. The solubility of lithium
fluoride, the product of the decomposition of lithium hexafluorophosphate, in DMC
is very low in comparison to the solubility in water or mixtures of water and alkyl
carbonates. The solubility of LiF in EC is higher than for PC or DMC (Jones et al.
2009). The extract from first stage consists mainly of DMC but it still contains traces
of the other organic carbonates. Therefore, the solubility of LiF in the extract from
the first stage was slightly increased in comparison to the following stages.
9.3.3 Single-Stage Extractions with Water
The necessity to reduce the fluoride content in the solid material and the limitations
in the extraction of fluoride with DMC revealed the need for a solvent for fluoride.
Water has been used for the analytic determination of fluoride and
hexafluorophosphate and is therefore a suitable solvent for fluoride extraction.
Furthermore it was possible to transfer the slurry to the following hydro metallurgic
processes without the necessity of drying. The optimum temperature for the
extraction of fluoride was determined in a set of experiments with 40, 60 and 80 °C
(Table 9.7).
Figure 9.10 displays the fluoride content in the extract for the aqueous extrac-
tions for different temperatures. The concentrations are increasing over time and
values increase with rising temperatures. The maximum value in the experiment is
reached after 10 min for 80 °C with 0.47 g fluoride/kg extract. The solubility in
water is reported with 1.07 g/L at 293.15 K. (Jones et al. 2009) the solubility is not
the limiting factor.
Table 9.7 Parameters for the determination of the temperature influence for aqueous extraction
Parameter Values
Battery material Raffinate from Panasonic CGR 18650 extractions
Mass ratio solvent to solid 7.5
Sample volume 3 mL
Solvent Water
Extraction time per stage 30 min
Fig. 9.10 Temperature

influence onto extraction with
water
Table 9.8 Mass fraction in extract and overall mass of fluoride for different solid to solvent ratios
Extraction time Mass fraction of fluoride in Extracted overall mass
[min] extract [g/kg] fluoride [g]
10 mL 30 0.483 0.0048
60 0.345 0.0035
12.5 mL 30 0.384 0.0048
60 0.429 0.0054
15 mL 30 0.344 0.0052
60 0.357 0.0054
20 mL 30 0.309 0.0062
60 0.288 0.0058
25 mL 30 0.263 0.0066
60 0.264 0.0066
In a second set of experiments 2 g of the raffinate fine fraction from the DMC
extraction were dissolved in ultrapure water at 30 °C. The mass of water was 10,
12.5, 15, 20 and 25 g. Samples were taken after 30 and 60 min. The fluoride mass
fraction as determined from ion chromatography and calculated extracted fluoride
mass are shown in Table 9.8.
The mass fractions of fluoride in the extract show no clear trend. For higher
solvent to solid ratios the mass fractions of fluoride decrease while the overall
extracted mass increases as expected. Overall, the extraction at the highest tem-
perature and a high solvent to solid ratio extracts the largest amount of fluoride. An
overall process optimization will have to consider the effort to recover fluoride from
the extract as it gets more diluted with a higher solvent to solid ratio. Hence, the
optimal processing for the aqueous extract has to be investigated.
168 P. Haas et al.
9.3.4 Multi-Stage Extractions with DMC and Water
Extraction of whole battery material with DMC and water

For further reduction of fluoride in the fine fraction of the raffinate, multi-stage
crossflow extractions with water have been performed. Thereby, a sufficiently low
level of fluoride for the next process steps is to be obtained. Figure 9.11 displays the
changed sequence of process steps for the minimum remaining fluoride in the solid
battery material.
The process sequence in Fig. 9.11 was experimentally tested to evaluate the nec-
essary number of extraction stages and to collect data for the extraction with water.
The parameters for DMC extraction are listed in Table 9.9 and for extraction with
water the parameters are listed in Table 9.10. The solvent to solid mass ratio for the
DMC extraction was limited by the volume of the stirred vessel.
Figure 9.12 depicts the mass fractions of fluoride and hexafluorophosphate in the
extract for a five stage crossflow extraction with DMC and water for three following
extraction stages. After the extraction with DMC the solid material was dried at
105 °C and ambient pressure and was returned into the vessel. The DMC extraction
was performed at 20 °C to reduce the decomposition of the conducting salt and the
temperature of the aqueous extraction was set to 80 °C to increase the solubility of
the fluoride and decompose the remaining conducting salt. During extraction with
DMC fluoride was only detectable in extract of the first stage like shown for the
four stage extraction in Sect. 9.3.2 while after the drying fluoride was detected at
Fig. 9.11 Process sequence for an extraction sequence with DMC and water each followed by a
drying process
Table 9.9 Parameter for five Parameter Values

stages of extraction with
DMC Battery material Panasonic CGR 18650
Mass of DMC 400 g
Solvent to solid mass ratio DMC 5 for all stages
Temperature DMC extraction 20 °C
Extraction time for each stage 30 min
Table 9.10 Parameter for three stages of extraction with water

Parameter Values
Battery material Panasonic CGR 18650 from previous DMC extraction
Mass of water 300 g
Solvent to solid mass ratio water 10 for all stages
Temperature water extraction 80 °C
Extraction time for each stage 30 min
Fig. 9.12 Hexafluorophosphate and fluoride mass fraction in extract for a combination of
extraction with DMC and water in 0.5 L scale
every aqueous extraction stage with decreasing concentration in extract. It was

unexpected to detect hexafluorophosphate in extract after drying and aqueous
extraction due to expected degradation. The loading was 1092.8 mg fluoride per kg
solid material and 218.4 mg hexafluorophosphate per kg solid material after the
extraction with DMC and drying. After the extraction with water and the subse-
quent drying, hexafluorophosphate was not detected and the loading of the solid
phase with fluoride was 166.4 mg per kg solid material.
Extraction of whole battery material with DMC and only fine fraction with
water
In an alternative process concept only the fine fraction was treated with DMC and
water consecutively, while for the coarse fraction only DMC extraction is performed.
This allows the performance of separation by particle size after first drying and
transfer of the coarse fraction to recovery steps, while the fine fraction is treated by
extraction with water. This also allowed the slurry to be transferred without further
drying required. The coarse fraction consists mainly of the housing, the foils of
copper, aluminum and plastics. The value compounds are mostly present in the fine
170 P. Haas et al.
Fig. 9.13 Optimized sequence of process steps for the extraction with organic solvent and water
with transfer of slurry
fraction smaller than 200 µm. The ratio of the fine fraction to overall mass after
drying was 0.5. Thus, the amount of solid mass for the second extraction step could
be reduced and the solid phase would be more homogenous due to the absence of
larger parts from foils. It is easier to handle the slurry than the graphite powder for the
transfer to next process steps and the energy demand for heating and condensation is
reduced without the additional drying process after the extraction with water.
The process sequence is displayed in Fig. 9.13. The process was performed with
a four stage crossflow extraction with DMC and a six stage crossflow extraction
with water in a 5 L stirred vessel. The battery material has been grinded in the
battery shredder of Volkswagen and packaged in inert gas atmosphere. The
parameters are listed in Table 9.11.
Table 9.11 Parameter for Parameter Values

extraction of PHEV batteries
Battery material PHEV 1
(Volkswagen)
Mass of the solid 0.69 kg
Mass of DMC 4.35 kg
Sample volume 10 mL
Solvent to solid mass ratio DMC 6.2
Solvent to solid mass ratio water 10.2
Temperature DMC extraction 20 °C
Temperature water extraction 80 °C
No. of extraction stages with DMC 4
No. of extraction stages with water 5
Vessel volume for DMC extraction 5L
Volume of water extraction 0.5 L
Extraction time for each DMC stage 60 min
Extraction time for each water stage 30 min
The solvent to solid mass ratio for the DMC extraction was defined by the
maximum loading of stirred vessel and is 6.2. The drying was performed at 105 °C in
the stirred vessel. The product was separated by sieving with an air jet sieve e200 LS
(Hosokawa Alpine AG, Augsburg, Germany) with a 200 µm mesh size. 147.5 g of
the fine fraction were transferred to the aqueous extraction. The solvent to solid mass
ratio was 10.2. The fluoride in the remaining solid phase after the final extraction
stage was 173.29 mg per kg solid material. The result is in contrast to the previous
findings and deviates from the predicted result based on mass balances. There a
fluoride loading of 10 mg per kg solid material was targeted. Hexafluorophosphate
has not been detected after the process while it was still present in the previous
experiment. The additional extraction stages with water resulted in the complete
degradation of the conducting salt. The reasons for the deviation could be the aging of
the battery and therefore the changes in composition in comparison to the 18650
model cell with a simplified discharging cycle. Within the investigations it could not
be verified whether a loading of 10 mg fluoride per kg solid material can be reached
by the combination of the extractions. As a fallback position the following recovery
processes of precious materials could be enabled to handle slightly higher levels of
fluoride in the active material. However, there is a significant potential for the
reduction of the solvent demand and number of stages by using multi-stage extraction
in countercurrent mode. Therefore, it may be more efficient to modify the following
process to reduce the necessary amount of solvent and the number of stages.
9.3.5 Degradation of Fluoride-Containing Compounds
The polyvinylidene fluoride binder at the anode or other fluoride containing com-
pounds showed no effect in terms of increasing the fluoride content after the pro-
cessing including extraction and drying. Samples of the experiment for the
extraction of PHEV 1 batteries with DMC and water shown in Sect. 9.3.4 have also
been analyzed at the MEET Münster without detection of further fluoride com-
pounds. Therefore, it is assumed to neglect these compounds in terms of the
fluoride content of the solid fine fraction.
9.3.6 Extraction of Organic Carbonates
The high boiling organic carbonates have been extracted in the experiments. Gas
chromatography measurements using a SCION 456-GC (Bruker Corporation,
Billerica, USA) showed low concentrations of high boiling EC in the extract from
multi-stage extraction of Panasonic CGR 18650 batteries with DMC and water
shown in Sect. 9.3.4. Up from the third stage no EC was detected anymore. The
raffinate was analyzed with a GC headspace method and after the DMC extraction
only DMC and EMC were detected.
172 P. Haas et al.
9.3.7 Modelling of the Multi-Stage Extractions with DMC
Influence of the liquid retention

The influence of the retention of liquid in the raffinate was determined by modelling
based on the experiment described in the analysis of the grinded material in
Sect. 9.3.1. The mass fraction of the anion of conducting salt in the extract was
calculated based on the measurement from the first stage of extraction. An equal
distribution and complete availability in of the conducting salt in the solution was
assumed as well as a complete equilibration of the liquid phase of the raffinate and
the composition of extract. The value for the second stage is calculated based on
dilution of the liquid phase in the vessel due to the adding of the new solvent. The
values for further stages were calculated based on the value for the second stage and
mass balancing.
Figure 9.14 shows a comparison of experimental and calculated values for the mass
fraction of the lithium hexafluorophosphate in the extract after each extraction
stage. While the calculated value for the second stage is slightly higher than the
value form the experiment, for later stages the measured values are higher than the
calculated ones. However, a very good qualitative match is achieved.
Modelling based on mass balances with known battery composition
The solid-liquid extraction crossflow mode was modelled based on (Mersmann
et al. 2005). For the further modelling based on mass balances the composition of
the Panasonic CGR 18650 batteries was determined or estimated. The concentration
of LiPF6 in electrolyte was determined to be 1.18 mol/L by the MEET Münster
(WWU Münster, Münster, Germany). The density of the electrolyte was determined
to be 1100 kg/m3 and the mass fraction of the electrolyte in the battery was 175 g/
kg battery. Therefore, the mass fraction of LiPF6 in the battery was calculated to be
29 g/kg battery. Based on the liquid retention and mass balances for extract and
Fig. 9.14 Comparison of mass fraction lithium hexafluorophosphate in extract for the experi-
mental data and model based on dilution of liquid residue
Table 9.12 Comparison of values from mass-balanced calculations and experiment

Extraction Mass fraction of PF6− in Mass fraction of PF6− in Deviation
stage extract from balance-based extract from experimental [%]
model [g/kg] data [g/kg]
1. 3.106 2.852 8.19
2. 0.415 0.335 19.16
3. 0.078 0.079 1.20
4. 0.021 0.019 9.40
raffinate the mass fractions were calculated for the four stage extraction and eval-
uated with the latter presented five stage extraction with DMC in Sect. 9.3.4. For
other kinds of batteries it is necessary to determine the composition in regard to the
LIPF6 and solid content. The values from experimental data and the balance-based
model are listed in Table 9.12 with deviations. These deviations are calculated by
the following equation.
!
xPF6 ;exp xPF6 ;cal
deviation ¼ abs ð9:3Þ
xPF6 ;exp
For the experiment the mass fractions are slightly higher, only the value for the third
extraction stage is higher for the calculation. The largest deviation is found for the
second stage with 19%, while for the other stages it is less than 10%. The reason for
deviation could be resulting from loss of battery material in the mill during grinding
or differences in the battery composition of the commercial available batteries.
Overall a very good qualitative and quantitative match is achieved.
9.3.8 Number of Necessary Extractions Stages and Specific

Solvent Demand
The number of necessary extraction stages and the specific solvent demand were
calculated for a 40 g Panasonic CGR 18650 battery for crossflow extractions with a
starting solvent to solid mass ratio of 5 and 10 based on the modelling based on
mass balances in Sect. 9.3.7. To keep the overall amount of liquid in the extractor
constant the added solvent in the following stages was reduced by 40 g after the
first stage. The aim was to reduce the amount of LiPF6 in raffinate to 0.007 g per kg
raffinate. The correlated mass fractions of LiPF6 in extract and raffinate for both
solvent to solid mass ratios have been calculated based on the experimental data for
the retention of liquid and mass balances.
The mass fractions of LiPF6 for the initial solvent to solid mass ratio of 5 and 10
are displayed in Fig. 9.15. For both mass ratios the reduction of LiPF6 content in
the extract with increasing number of stages is visible. The reduction for the higher
174 P. Haas et al.
Fig. 9.15 Calculated mass fraction of LiPF6 in extract for different solvent to solid mass ratios
Table 9.13 Necessary number of stages and amount of solvent for 40 g Panasonic CGR 18650
Initial solvent to solid Necessary Necessary mass Specific amount of
mass ratio [kg solvent/kg number of of DMC overall DMC [kg DMC/kg
solid] stages [-] [g] battery]
5 7 1160 29
10 4 1480 37
amount of solvent is larger. The necessary number of stages, overall mass of DMC
and specific amount of DMC for the two solvent to solid mass ratios are listed in
Table 9.13.
While the necessary number of extraction stages decreases with higher mass
ratios the amount of required solvent increases as expected. The degradation of
LiPF6 in the batteries is not included in these calculations and would decrease the
available amount of conducting salt. This effect has to be regarded in further
studies. Other possibilities to reduce the necessary amount of extraction stages and
required amount of solvent are the increase of tolerated fluorides in the active
material of the batteries as mentioned above and the operation in countercurrent
extraction mode.
9.4 Conclusions
9.4.1 Summary
Comprehensive experimental and theoretical investigations have been conducted on

the extraction of electrolyte and conducting salt LiPF6 in the recycling process of
lithium-ion batteries. The experiments have shown the influence of the
decomposition of the conducting salt and the influence of the temperature. DMC was
chosen as solvent because of the low boiling temperature, the high solubility of
lithium hexafluorophosphate and its presence in the batteries. A multi-stage cross-
flow extraction was effective in the reduction of conducting salt content, whereas the
combination with an aqueous extraction was necessary to reduce the fluoride loading
of the solid material. The experiments have been performed with commercially
available 18650 cells and also with hybrid electric vehicles batteries. The reduction
of the fluoride content employing solvent extraction was shown to be successful in
lab experiments, but the parameters for the following processing were not met and
the integration in the demonstration plant has not been performed. The extraction
was modelled based on data from extract and mass balances. A very good qualitative
and good quantitative agreement with experimental data was achieved.
9.4.2 Remarks for the Design of an Extraction Plant
The remarks for the equipment design are based on the properties of battery
material including chemical composition, results of the experiments and the models
from the extraction experiments. According to the requirements of following pro-
cesses, the fluoride concentration was set to a desired level and the experiments
were designed to achieve this level. It was mandatory to work in inert gas atmo-
sphere to avoid further decomposition of the conducting salt and to reduce the
explosion hazard (Grützke et al. 2015).
The results in this chapter were achieved in experiments in 0.5 and 5 L scale
with grinded battery cells. However the industrial applications should be able to
handle 1000 kg batteries per hour, which is roughly equivalent to 7 m3 of solid
material per hour. The composition of the material in regard to the proportion of fine
and coarse fraction depends on the disassembly whether the cell or the module is
the target for the grinding. Therefore, further testing is necessary for the scale-up to
larger apparatuses to perform the detail engineering and choose the right kind of
equipment. It is expected beneficial to switch from crossflow operation to coun-
tercurrent flow mode to minimize the amount of solvent. However, this will go
along with an increase in the required number of stages so that a change in the
extractor equipment technology should be considered. Belt extractors or screw
extractors are frequently employed for multi-stage countercurrent solid-liquid
extractions. This would go along with a change of the extraction mode from
maceration to percolation in large-scale extractions because of the possibility of
avoiding the stirring of the solid material.
The processing of the extract and its recycling has not been tested extensively in
experimental conditions while flowsheet simulations based on phase equilibrium
176 P. Haas et al.
data have been performed. Without the recycling of the major fraction of the extract
the disposal of the special chemicals and gas cleaning increase the costs for the
inclusion of the extraction. Hence, the extract has to be processed and reused as
solvent to minimize the need for fresh solvent. At the same time, the handling and
processing of the extract also requires sensitive care due to the light boiling com-
ponents and the decomposition of the conducting salt. The differences between
batteries at the end of the lifecycle in contrast to new batteries have been analyzed
to compare the influence of the aging products with field-tested vehicles but only
with specific batteries. For the design of an universal recycling facility for all kinds
of lithium-ion batteries further investigations on the influences of these ageing
compounds have to be performed. The changes in battery composition have to be
considered.
References
Grützke M, Kraft V, Hoffmann B, Klamor S, Diekmann J, Kwade A, Winter M, Nowak S (2015)

Aging investigations of a lithium-ion battery electrolyte from a field-tested hybrid electric
vehicle. J Power Sources 273:83–88. https://doi.org/10.1016/j.jpowsour.2014.09.064
Jones J, Anouti M, Caillon-Caravanier M, Willmann P, Lemordant D (2009) Thermodynamic of
LiF dissolution in alkylcarbonates and some of their mixtures with water. Fluid Phase Equilibr
285(1–2):62–68. https://doi.org/10.1016/j.fluid.2009.07.020
Kawamura T, Okada S, J-i Yamaki (2006) Decomposition reaction of LiPF6-based electrolytes for
lithium ion cells. J Power Sources 156(2):547–554. https://doi.org/10.1016/j.jpowsour.2005.
05.084
Korthauer R (2013) Handbuch Lithium-Ionen-Batterien. SpringerLink: Bücher. Imprint: Springer
Vieweg, Berlin, Heidelberg
Mersmann A, Kind M, Stichlmair J (2005) Thermische Verfahrenstechnik: Grundlagen und
Methoden, 2., wesentlich erw. und aktualisierte Aufl. Chemische Technik Verfahrenstechnik.
Springer, Berlin
Nowak S, Winter M (2015) Review—chemical analysis for a better understand-ing of aging and
degradation mechanisms of non-aqueous electrolytes for lithium ion batteries: method
development, application and lessons learned. J Electrochem Soc 162(14):A2500–A2508.
https://doi.org/10.1149/2.0121514jes
Terborg L (2011) Analytical methods for the determination of aging products in lithium-ion
batteries. Dissertation, Wilhelms-Universität Münster
Xu K (2004) Nonaqueous liquid electrolytes for lithium-based rechargeable batteries. Chem Rev
104(10):4303–4418. https://doi.org/10.1021/cr030203g
Chapter 10
Electrolyte Extraction—Sub
and Supercritical CO2
Sergej Rothermel, Martin Grützke, Xaver Mönnighoff,

Martin Winter and Sascha Nowak
Abstract This chapter reports on experiments aimed at investigating the capability

of pressurized carbon dioxide to extract the electrolyte from commercial available
LIBs on a laboratory scale. Two different phase conditions of carbon dioxide
(subcritical and supercritical) and two different extraction (static and dynamic) have
been considered and analyzed for their strengths and weaknesses. Furthermore, the
addition of co-solvents is examined with regard to their contribution to higher
recovery rates. After reporting the optimized extraction method, the extracted
electrolyte was analyzed by gas and ionic chromatography methods for potential
de-composition products and their relative amount.
10.1 Introduction
Another electrolyte extraction method investigated within the LithoRec II project is

the use of sub- and supercritical CO2 (scCO2) as extractant. The motivation behind
this approach is a principle often taught in organic chemistry: same with the same
dissolves. Indeed, the main constituents of the electrolyte are either cyclic or linear
carbonates having low molecular weights and non-polar fractions. Besides its
gaseous, liquid and solid phases, carbon dioxide is easy to transfer into its 4th
accessible supercritical phase which is the most applicated medium in supercritical
fluid extraction (SFE). The most common example is the decaffeination process of
coffee. The advantage of scCO2 as extractant is that its dissolving properties can be
adjusted by pressure according to the substance to be extracted. Especially in
combination with additional solvents, many organic substances can be extracted
from complex matrices.
S. Rothermel M. Grützke X. Mönnighoff M. Winter S. Nowak (&)

MEET Battery Research Center, University of Münster, Muenster, Germany
M. Winter
Helmholtz-Institute Münster (HI MS), Forschungszentrum Jülich,
IEK-12, Muenster, Germany

https://doi.org/10.1007/978-3-319-70572-9_10
In this context, a static and dynamic (flow-through) method were developed at a

laboratory scale to extract the electrolyte with sub- and supercritical carbon dioxide
(sub and scCO2). The extraction performance in combination with different
co-solvents was studied. The investigated methods were analyzed in view of the
recovery rate and duration. Furthermore, the achieved extracts were analyzed with
gas and ion chromatography for their qualitative and quantitative composition.
Commercial available Panasonic CGR18650CH Li-ion MH12210 (2.2 Ah,
LiNi1/3Co1/3Mn1/3O2 (NCM)/graphite) 18650 cells as well as in-house-made pouch
cells (4.8 Ah), both based on LiNi1/3Co1/3Mn1/3O2 (NCM)/graphite electrode
chemistry were used as study objects.
10.2 Static Electrolyte Extraction by Supercritical CO2
To provide the proof of principle that supercritical CO2 is applicable as extractant

for the electrolyte extraction from LIBs, two different types of the separators
drenched with 4 mL of the electrolyte each were selected as substrates for the initial
studies. Two electrolytes, namely: LP30 (DMC/EC, 1 mol/L LiPF6) and LP50
(EMC/EC, 1 mol/L LiPF6) were used to investigate the influence of their different
compositions on the recovery rate. For this reason, two equal sized separators
(6 6 cm2): polyethylene fleece (Freudenberg 2190) and porous glass fiber
(Whatman® GF/D) were drenched with 4 mL of electrolyte and placed in a
stainless-steel autoclave (Berghof BR-100) with a reactor volume of 200 mL and a
Teflon® lined interior. The reactor was filled with helium head pressurized CO2
(HHPCO2) from a gas cylinder to an inner pressure of 90 bar. Applying additional
heating to the reactor, the inner pressure increased to 120 bar (at 40 °C). After a
period of one 90 min, the autoclave was snap-frozen by liquid nitrogen which
immediately caused a substantial loss of pressure to 50 bar. Afterwards the
enclosed CO2 was released via an external valve. The remaining extract of frozen
electrolyte was collected from the bottom of the Teflon® inlet (Grützke et al. 2014).
The resulted recovery rates are summarized in Table 10.1. Furthermore, the
composition of the extract substances determined by Gas Chromatography—Mass
Spectrometry (GC–MS) experiments and Ionic Chromatography (IC) is given
(Grützke et al. 2014).
Table 10.1 Recovery rates and compositions for the scCO2 extraction of LP30 (DMC/EC, 1 M
LiPF6) from different separators (Grützke et al. 2014)
Analytical Substance/recovery Reference Freudenberg Whatman
method rate LP30 2190 GF/D
GC DMC [wt%] 50.2 ± 1.5 43.0 ± 1.3 81.8 ± 1.9
EC [wt%] 49.8 ± 1.5 57.0 ± 1.3 18.2 ± 1.9
IC LiPF6 [mol/L] 1.07 ± 0.01 1.15 ± 0.04 –
Total recovery rate [wt%] 73.5 ± 3.6 36.7 ± 1.6
10 Electrolyte Extraction—Sub and Supercritical CO2 179
Overall, (73.5 ± 3.6) wt% could be recovered from the polyethylene fleece and
(36.7 ± 1.6) wt% from the porous glass fiber. The relatively low recovery rates can
be partly explained by losses during CO2 release which allowed vapors of the
electrolyte to escape. Another part of remained/turned back into the separator,
which was confirmed by weighing the separator before and after experiment
Substantial difference in extraction efficiency between the different separator
types can be explained by the higher polarity of EC (dipole moment 4.61 D)
compared to linear organic carbonates (DMC: 0.76 D; EMC: 0.89 D), thus a higher
adsorption capacity on the polar surface of the glass fiber separator is assumingly
the main reason for this difference. The polar nature of LiPF6 is also attributable to
the fact, that no detectable amount was recovered from the Whatman® separator
The deficiencies of this method motivated the development of a dynamic
flow-through extraction method, which is considered separately in the next section.
10.3 Dynamic Electrolyte Extraction

by Sub- and Supercritical CO2
The electrolyte extraction experiments began with the development of an appro-

priate extraction method and settings adjustment. Similar to the investigations made
before, polyethylene fleece separator (Freudenberg 2190) was drenched with bat-
tery grade electrolyte (BASF Selectilyte™) and introduced into an flow-through
extraction chamber (made of stainless steel and of 50 mL volume) and subjected to
a continuous supercritical CO2 flow. The extraction conditions were set to 40 °C,
the pressure to 80 bar and one hour duration. It has become necessary to heat the
release valve to 55 °C to avoid electrolyte freezing as a result of the
Joule-Thompson-Effect. The electrolyte extract was collected in a cryogenic trap
(−78 °C) installed downstream of the heated release valve. Alternatively studied
method to separate the electrolyte from the extractant by using washing solutions
(e.g. acetonitrile) has not succeeded (Grützke et al. 2015).
After specifying the operating mode, further experiments were performed using
2.2 Ah 18650 cells. For this reason, Panasonic CGR18650CH Li-ion MH12210
cells were chosen as a basis. To simulate the real End of Life conditions, the cells
were electrochemically cycled for 1000 cycles. The targeted ageing was performed
in a controlled environment at 20 C. Electrochemical experiments were carried out
using a Maccor Series 4000 Battery Tester 20 A. The cells were charged with a
constant current of 2.25 A (1C) to a charging cut-off voltage of 4.2 V with a
subsequent constant voltage step of 30 min. The discharge procedure was per-
formed with equal current rate of 2.25 A with 3.0 V as lower cut-off voltage. In
addition to the cycling procedure an extended constant current step was added to
ensure their safe handling afterwards. Prior to the dismantling of the cell enclosure
in a protecting Helium atmosphere, the discharged cells were frozen to −18 °C to

reduce electrolyte evaporation and avoid biased results. For this purpose, an
in-house designed lathe machine was used. Clamped into the lathe, the cell rotates
around its own axis, whereby a cutting tool positioned just underneath the sealing
gasket moved forward to complete a radial cut. After this step, the upper part of the
cell could be opened and the jelly roll contained therein was released quickly
unfolded and transferred into the pressure chamber (made of stainless steel and
50 mL volume) of the extraction equipment (Grützke et al. 2015).
The unwound electrodes were densely packed into the extraction chamber. As
the extraction chamber was inserted into the Applied Separations Speed SFE
System (from ERC), the temperature was adjusted to 40 °C and the extraction
repeated varying CO2 pressure between 80 and 300 bar. The changes in pressure
did not remarkably influence the recovery rate. Whereas variated flow rates at
constant pressure and extraction time demonstrated a higher yield of the electrolyte.
The electrolyte extract was collected in a cryogenic trap (−78 °C) installed
downstream of the release valve. To counteract the Joule-Thomson-Effect and
prevent freezing and blocking of the gas outlet, the release valve was continuously
heated to 55 °C—the upper limit of the thermal electrolyte stability. At the same
time, the difficulty of electrolyte freezing limited the throughput of the extractant.
Therefore, the extractant flow was limited to 15 ± 3 L−1 at the outlet as a com-
promise between the flow rate and sublimation of CO2 (Grützke et al. 2015).
Using the previously collected experience in flow adjustment, two different
conditions of CO2 with fixed temperature, pressure and flow settings were inves-
tigated in dependence on time: subcritical (25 °C, 60 bar) and supercritical (40 °C,
300 bar). Figure 10.1 summarizes the recovery rates achieved during systematically
extended extraction durations starting from 2 to 20 min in fixed time intervals of
2 min. It can be clearly seen that the most amount of electrolyte is eluted during the
first minutes by each method, whereby the subcritical CO2 seems to be more
effective (Grützke et al. 2015).
Fig. 10.1 Time dependency

of the amount of recovered
electrolyte from commercial
18650 cells after formation
extracted with supercritical
(300 bar, 40 °C; red stars)
and liquid (60 bar, 25 °C;
black squares) CO2 (Grützke
et al. 2015)
Fig. 10.2 Compositions of the recovered electrolyte from commercial 18650 cells after formation
extracted with supercritical a and liquid b CO2 for the first three data points of Fig. 10.1
determined with GC-MS. Blue, top: EC; yellow, middle: EMC; magenta, bottom: DMC (Grützke
et al. 2015)
As mentioned afore, the dissolving properties of the carbon dioxide are influenced
by the conditions prevailing in the extraction chamber. In order to investigate the
impact of these different conditions on the qualitative extract composition, gas
chromatography-mass spectrometry (GC-MS) investigations were performed.
Analyses reveal that linear carbonates are preferably eluted by subcritical CO2,
whereby ethylene carbonate (EC) is more efficiently extracted by supercritical CO2.
This is in line with the assumption that with increasing pressure the dissolution
character of CO2 becomes of rather non-polar nature. Therefore, the less polar EC is
recovered in higher amounts. The non-polarity of the extractant is also the reason for
the negligible extraction of the conductive salt lithium hexafluorophoasphate (LiPF6).
Its traces in the extractant were confirmed by ionic chromatography experiments as
in both extractants (sub- and supercritical setup) (Grützke et al. 2015) (Fig. 10.2).
To indicate the efficiency of each method, the amount of the electrolyte initially
contained in the 18650 cell needed to be determined. This was done by simply
drying an opened cell at 120 °C for 10 h. The weight reduction was found to be
3.76 g (Grützke et al. 2015).
10.4 Supporting Electrolyte Extraction Additives
Under extraction conditions applied so far, it would be challenging to increase the

extraction yield of EC and LiPF6.
To overcome these limitations and long extraction times extraction experiments
with subcritical CO2 and additional polar co-solvents were realized.
The time-resolved for the recovery rate gained using acetonitrile (ACN), a
mixture of acetonitrile and propylene carbonate (ACN/PC) and diethyl carbonate
(DEC) are demonstrated in Fig. 10.3.
Fig. 10.3 Time dependency of the amount of recovered electrolyte from commercial 18650 cells
after formation extracted with liquid CO2 and 0.5 mL/min additional solvents (black stars: ACN;
magenta triangles: ACN/PC (3:1); red circles: DEC) (Grutzke et al. 2015)
Fig. 10.4 Compositions and amounts (determined with GC-MS and IC) of the recovered
electrolytes from commercial 18650 cells after formation extracted with liquid CO2 and additional
solvents for 30 min, with subsequent 20 min without additional solvent. Red, top: LiPF6; blue,
below: EC; yellow, middle: EMC; magenta, bottom: DMC (Grützke et al. 2015)
The additional solvent was added to the CO2-stream with a flow rate of
0.5 mL min−1 my means of a Knauer Platin Blue UHPLC pump. The experiments
were repeated by five different duration intervals: 2, 10, 20, 30 and 50 min. Overall,
the observed recovery rates follow an exponential growth. The experiments with
ACN and DEC show a comparable trend with high recovery rates in the initial
phase compared to the ACN/PC mixture, which shows a relatively low performance
in the beginning but outperforms the other co-solvents at extraction times beyond
20 min. Thus, the conclusion can be made, that the addition strongly polar solvents
reasonable to improve the extraction result (Grützke et al. 2015).
Figure 10.4 summarizes the extractant compositions achieved by means of
different co-solvents. In addition, the composition of a reference is shown, which is
based on the quantitative investigation of an electrolyte sample (10 mL) precipi-
tated by centrifugation of another dismantled jelly roll. The composition observed is
normalized to the amount of electrolyte per 18650 cell (Grützke et al. 2015).
It can be seen, that ACN effectively extracts linear carbonates as well as EC. But
the recovery of LiPF6 is still strongly limited. The use of DEC as a co-solvent
allows the extraction of all constituent electrolyte components including LiPF6,
although at strongly reduced efficiency in EC recovery. The best recovery efficiency
of 89.1 ± 3.4 wt% is demonstrated if the ACN/PC (3:1) mixture is added to the
extractant. Furthermore, the conductive salt recovery is approximately complete,
although at the expense of the linear carbonates (Grützke et al. 2015).
10.5 Analysis of the Extracted Electrolytes
Analogously to the static extraction method described in one of the previous

chapters, similar experiments were performed with commercial available 18650
cells with the aim to determine the potential presence of decomposition products
therein. To depict different scenarios of the possible end of line conditions, one
18650 cell—as delivered from the factory—was used as a source of the non-aged
electrolyte. Another cell was electrochemically aged to 70% SOH at 20 °C and a
third exemplar was electrochemically aged to 70% SOH at 45 °C. The electro-
chemical cycling data is shown in Fig. 10.5. Higher temperatures initially improve
the cycling performance of the cell, which, however, quickly decreases beyond the
350th cycle. The performance at 20 °C is rather uniform and steadily decreasing
All three cells were dismantled and the uncased jelly rolls were separately
subjected to the static electrolyte extraction by using supercritical CO2 and an
autoclave. Unwinding of the extracted electrodes revealed significant differences in
adhesion strength of the coating and current collectors (Grützke et al. 2014).
Fig. 10.5 Discharge

capacities for aged (square,
red: 20 °C, 70% SOH;
triangle, blue: 45 °C, 70%
SOH) commercial 18650 cells
(Grützke et al. 2014)
The extract of the non-aged electrolyte consisted of (41.9 ± 0.1)% DMC,

(48.2 ± 0.2)% EMC and (4.6 ± 0.2)% EC which were identified with GC MS and
quantified with Gas Chromatography—Flame Ionization Detector (GC FID).
Similar to the extraction experiments with electrolyte drenched separator, the
conductive salt LiPF6 could be recovered in traces (Grützke et al. 2014).
Besides well-known basic electrolyte compounds EMC, DMC and EC, the GC
investigations revealed also substances based on transesterification of the carbonates
EMC and DMC, which are only observable in electrochemically aged samples. The
substances are dimethyl-2,5-dioxahexane dicarboxylate (DMDOHC), ethylmethyl-
2,5-dioxahexane dicarboxylate (EMDOHC) and diethyl-2,5-dioxahexane dicar-
boxylate (DEDOHC) identified as reaction products of linear carbonates. The
extracted amounts of the electrolytes, their percentage compositions and concentra-
tions of the dicarboxylates are summarized in Table 10.2. The evolution of the con-
centrations from non-aged via 20 to 45 °C samples reveals that the concentration of
Table 10.2 Recovery rates and compositions for the supercritical CO2 extraction of commercial
18650 cells after formation and aging (Grützke et al. 2014)
After formation 20 °C, 70% SOH 45 °C, 70% SOH
Total Recovery [g] 0.35 0.25 0.21
DMC [wt%] 41.9 ± 0.1 40.1 ± 1.1 39.4 ± 0.4
EMC [wt%] 48.2 ± 0.2 42.2 ± 1.1 42.1 ± 0.3
EC [wt%] 4.6 ± 0.2 7.0 ± 0.3 7.1 ± 0.2
DEC [wt%] LOD < c < LOQ 4.5 ± 0.3 3.6 ± 0.4
DMDOHC LOD < c < LOQ 2.9 ± 0.1 2.0 ± 0.1
EMDOHC ✓ ✓ ✓
DEDOHC – 0.7 ± 0.1 0.5 ± 0.1
Fig. 10.6 Chromatogram (GC-MS) of the extract from different aged commercial 18650 cells
(Grützke et al. 2014)
DMC and EMC is decreasing, while the relative EC concentration is increasing. This
finding is in confidence with the assumption that EC is not primarily involved into the
formation of dicarboxylates (Grützke et al. 2014) (Fig. 10.6).
10.6 Conclusion
In this chapter, the suitability of high pressurized CO2 as extractant for the elec-
trolyte in LIBs is investigated. Initially, the extraction capability of supercritical
CO2 was assessed on predefined substrates in a static experiment. It was demon-
strated that the chosen approach is applicable in principle, albeit with two limita-
tions, namely the inability of LiPF6 recovery and losses of organic carbonates
during equalization the inner gas pressure in the autoclave.
Based on this experience, modifications had been made to improve the extrac-
tion result. First, the static experimental setup was changed to a dynamic
flow-through method which minimizes losses of the carbonates. Second, the opti-
mized pressure conditions were evaluated to reduce the technical effort. Third, the
addition of co-solvents was investigated with the aim to enable simultaneous
extraction of the conductive salt LiPF6. Finally, commercial available 18650 cells
of different state-of-health were utilized as substrates to reflect real conditions. The
results of this investigations finally demonstrated that the use of subcritical CO2 in
combination with acetonitrile and dynamic extraction setup could achieve a
recovery rate of almost 89 wt% including LiPF6.
Further analyzes of the extracted electrolyte revealed the presence of decompo-
sition products in electrochemically aged cells, namely dimethyl-2,5-dioxahe-
xane dicarboxylate (DMDOHC), ethylmethyl-2,5-dioxahexane dicarboxylate
(EMDOHC) and diethyl-2,5-dioxahexane dicarboxylate (DEDOHC).
The demonstrated method offers a unique opportunity for electrolyte removal
during the recycling process of LIBs. On the one hand, the extracted electrolyte can
be easily purified and reutilized in LIB manufacturing. On the other, the shredded
material is free from hazardous, flammable and humidity-sensitive substances thus
offering the advantage of lower demands on personnel and equipment.
In view of these results, CO2 assisted electrolyte extraction is considered to an
alternative in recycling LIBs.
References
Grützke M, Kraft V, Weber W, Wendt C, Friesen A, Klamor S, Winter M, Nowak S (2014)

Supercritical carbon dioxide extraction of lithium-ion battery electrolytes. J Supercrit Fluids 94
(1):216–222. https://doi.org/10.1016/j.supflu.2014.07.014
Grützke M, Monnighoff X, Horsthemke F, Kraft V, Winter M, Nowak S (2015) Extraction of
lithium-ion battery electrolytes with liquid and supercritical carbon dioxide and additional
solvents. RSC Adv 5(54):43209–43217. https://doi.org/10.1039/c5ra04451k
Chapter 11
Off Gas Cleaning by Adsorption
Friederike Stehmann and Stephan Scholl
Abstract Some process steps in the recycling process containing a significant

amount of volatile organic compounds, such as shredding and drying, need to be
inertised by flushing with dry nitrogen. Consequently, the off gas of this inertization
is loaded with electrolyte or extraction solvents. Due to environmental restrictions,
this off gas has to be cleaned before it is released. This chapter addresses the off gas
cleaning with adsorption onto activated carbon and gives a basic understanding of
the operation mode of a fixed bed adsorber and the database to design an adsorptive
off gas cleaning. Relevant data of adsorption equilibria of single as well as binary
mixtures of electrolyte components onto activated carbon at temperatures from 20
to 60 °C are presented. Additionally, binary equilibria were calculated with the
Ideal Adsorbed Solution Theory and compared to experimental results. Finally,
results of the adsorption in a fixed bed are discussed. Breakthrough behavior and
temperature profiles of the adsorption process are presented and compared to the
adsorption equilibria. Another focus is set onto decomposed electrolyte components
that may occur during the lifetime of a battery, the recycling process and the
adsorption itself. These decomposition products can affect the adsorption equilib-
rium and therefore have to be considered.
11.1 Introduction
Several recycling steps of the LithoRecII-process produce an off gas. The shredder
must be flushed constantly with nitrogen to prevent an explosive atmosphere. Also
during drying all volatile components are vaporized at elevated temperature, low
pressure and purged with nitrogen. According to German environmental restrictions
F. Stehmann S. Scholl (&)

Institute for Chemical and Thermal Process Engineering,
e-mail: s.scholl@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_11
188 F. Stehmann and S. Scholl
the mass flow or the mass concentration must not exceed 0.5 kgC h−1 or
50 mgC m−3, respectively (Bundesministerium für Umwelt 2002). Both limits are
based on the total carbon mass. For dimethyl carbonate with a total carbon mass of
−1
0.4 gC g1
DMC this results in a maximum flow of 1.25 kgDMC h or a maximum
−3
concentration of 125 mgDMC m .
The off gas cleaning can be designed as a combination of condensation and
adsorption. Cooling of the off gas will result in the condensation of high-boiling
components as well as most of the low-boiling components, for details see
Stehmann et al. (2017a). However, to achieve a sufficient cleaning, very low
temperatures are required resulting in an energy consuming process. Adsorption has
proven to be an advantageous process for the final cleaning step. This cleaning
procedure operates at ambient temperature and—depending on the chosen des-
orption method—does not produce any wastewater. After the adsorbent is fully
loaded, it can be desorbed and the retrieved desorbate containing previously
adsorbed electrolyte can be recovered. Alternatively, if the adsorbent is supposed to
be replaced and not recycled, it can be thermally disposed.
This chapter provides a database of adsorption equilibria of dimethyl carbonate
(DMC) and ethyl methyl carbonate (EMC) onto activated carbon. Single compo-
nent equilibria as well as two-component equilibria are presented for temperatures
of 20–60 °C. Single-component isotherms are described with the model of Tóth and
model parameters are presented. The Ideal Adsorbed Solution Theory (IAST) is
employed to calculate binary adsorption equilibria based on single component
isotherms.
Additionally, the adsorption in a fixed bed is discussed. The operation mode of a
fixed bed adsorber is presented including break through behavior and temperature
profiles. Retrieved data can be used to identify advantageous monitoring parameters
for estimation of the adsorber lifetime while the adsorber is operating.
Finally, this chapter addresses decomposition during adsorption. Decomposition
can already occur during the lifetime of a battery or the recycling process, but also
appears to be catalyzed by activated carbon. Hence, DMC and EMC decompose to
methanol and ethanol during adsorption. Since these decomposition products are
also adsorbed by the activated carbon, their presence has to be considered for the
design of the adsorptive off gas cleaning. Therefore, this chapter also addresses the
main influences on the decomposition by adsorption and quantifies the decomposed
amount.
11.2 Experimental Procedure
11.2.1 Materials
Dimethyl carbonate (Merck KGaA, Darmstadt, Germany) and ethyl methyl car-
bonate (BASF SE, Ludwigshafen, Germany) as well as ethanol (VWR Chemicals,
11 Off Gas Cleaning by Adsorption 189
Darmstadt, Germany) and methanol (VWR Chemicals, Darmstadt, Germany) were

used as adsorptives. As adsorbent different activated carbons, namely SC40
(Silcarbon GmbH, Kirchhundern, Germany) and Norit® RB4 (CABOT Norit
Nederland B.V., Zaandam) were tested.
11.2.2 Preparation of Batch Samples
Samples were prepared in gas tight vials with a volume of VVial = 20 mL. Activated
carbon was inserted and heated at 120 °C for 24 h. After cooling down in a des-
iccator, the adsorptive was added and the vial closed. Masses of carbon mAC and
adsorptive mA were determined gravimetrically. The prepared samples were stored
for a defined time at a set temperature. After equilibration, the gas loading Y GC was
analyzed by gas chromatography. Hereafter the carbon loading X can be calculated
with a mass balance according to Eq. (11.1) (Stehmann et al. 2016).

m
A
M YGC V
A
X¼ ð11:1Þ
mAC
The volume V describes the free volume in the vial according to Eq. (11.2).
m
V ¼ VVial VAC ¼ VVial AC ð11:2Þ
qAC
The volume of the carbon VAC is calculated with the buoyancy density of the
carbon, which is derived by buoyancy measurements in a magnetic suspension
balance and therefore represents the carbon volume displacing the present air.
11.2.3 Fixed Bed Experiments
The breakthrough behavior was investigated in a lab-scale adsorber with a volume

of 300 mL, an inner diameter of 4.25 cm and a bed height of 21 cm. Eleven
thermocouples monitor the temperature profile. A gas loading unit consisting of a
saturator and condenser provides the model off gas with nitrogen as carrier gas
loaded with DMC, see Fig. 11.1.
Before starting the adsorption process, the inlet gas stream was analyzed and
only DMC was detected. The exiting gas stream of the adsorber was analyzed
during the operation time of the experiments (Stehmann et al. 2017a).
Fig. 11.1 Lab scale adsorber with model gas preparation: Nitrogen feed (B01), heated DMC
reservoir (B02), condenser (W01), adsorber (B03), gas chromatography (GC) and vacuum pump
(P01) (Stehmann and Scholl 2015)
11.3 Calculation Methods
11.3.1 Single Component Isotherms
Adsorption equilibria are commonly described by adsorption isotherms. For off

gases containing only one non-inert component a multitude of single component
models is available in literature. In the present study models of Langmuir (1918)
and Tóth (1971) in Eqs. (11.3) and (11.4) were used, respectively.
Xmax b Y
X¼ ð11:3Þ
1þb Y
Xmax b Y
X¼ 1 ð11:4Þ
1 þ ðb Y Þt t
Here, Xmax is the maximum carbon loading, b is a coefficient describing the ratio
of ad- and desorption rate and Y the equilibrium gas loading. The Tóth parameter
t represents the heterogeneity of the adsorbent. For t = 1, the Tóth equation equals
Langmuir, where the adsorbent surface is assumed to be totally homogeneous. For
t = 0 the surface shows maximum heterogeneity. The parameter t only depends on
the adsorbent, while the parameters b and Xmax also vary with adsorptive and
temperature. Equations (11.3) and (11.4) describe the carbon loading depending on
the gas loading Y. However, the same equations can be used with the gas saturation
u instead of the gas loading and b’ instead of b. Both parameters can be derived
with the saturation gas loading Ysat according to Eqs. (11.5) and (11.6).
Y
u¼ ð11:5Þ
Ysat
b0 ¼ b Ysat ð11:6Þ
11.3.2 Multi-component Isotherms
The experimental determination of multi-component isotherms is time-consuming.

Hence, the prediction of those equilibria based on single-component isotherms is
favorable. Common models to describe multi component equilibria are the model of
Markham and Benton (1931) or the Ideal Adsorbed Solution Theory (IAST) (Myers
and Prausnitz 1965). The model of Markham and Benton is based on the Langmuir
equation and can be solved analytically, see Eq. (11.7). However, this model is not
predictive in nature but may serve as an easy-to-handle way to describe influences
of temperature and pressure on multi-component equilibria.
Xmax;i bi Yi
X¼ P ð11:7Þ
1 þ nk¼1 bk Yk
The IAST developed by Myers and Prausnitz (1965) is set up analogue to the
ideal gas law and predicts multi-component adsorption equilibria based on any
single-component model represented by X 0i ðY Þ, see Eq. (11.8).
p0i Aspec. YZi Xi0 ðY Þ

0
¼ dY ¼ const. ð11:8Þ
RT 0
Y
The parameters on the left hand side including the spreading pressure p0i , the
specific surface of the adsorbent Aspec: , the ideal gas constant R and the temperature
T are identical for all adsorptive components in a certain equilibrium state. Hence,
the right hand side must match this for every component as well. The fictional
partial gas loading Y 0i consists of the overall gas loading Y as well as the fraction of
the component i in the gas phase yi and adsorbed phase xi , see Eq. (11.9).
yi
Yi0 ¼ Y ð11:9Þ
xi
The overall carbon loading is calculated with the hypothetical carbon loading
X 1 , if only component i was present, and the mole fraction xi of component i in the
adsorbed phase. For a two component system this results into Eq. (11.10).
1 x1 x2
¼ þ ð11:10Þ
X X1 X2
Any model of single-component isotherms may be used in Eq. (11.8) to express

the single component adsorbent loading Xi0 ðY Þ. For the Langmuir model, the
integral can even be solved analytically. However, solving the overall set of
equations for all components requires numerical iteration.
The composition in gas and adsorbed phase will differ due to the selectivity of
the carbon and different single-component equilibria. This selectivity can be
expressed with Eq. (11.11).
yEMC =xEMC yEMC 1 xEMC

a¼ ¼ ð11:11Þ
yDMC =xDMC 1 yEMC xEMC
11.3.3 Breakthrough Time
A fixed bed can be described by three characteristic zones. In the first zone, the
adsorbent is fully loaded with the equilibrium loading corresponding to the entering
off gas concentration. This zone is followed by the mass transfer zone (MTZ), in
which the actual adsorption takes places. The last part of the adsorber is the
unloaded zone, where only the upstream cleaned off gas passes. During the
adsorption process, the MTZ is moving from the entrance to the exit of
the adsorber. As soon as the MTZ is reaching the adsorber exit, the off gas is not
sufficiently cleaned anymore leading to an increase of gas loading. This break-
through time indicates when the adsorbent must be exchanged or recycled.
The ideal breakthrough time tBT,ideal describes the time at which the break-
through would occur if a vertical breakthrough curve was present, see Fig. 11.2,
Fig. 11.2 Breakthrough

curve of fixed bed adsorber
with ideal breakthrough time
and is calculated with Eq. (11.12), where tmax represents the total operation time of
the experiment and Yti the gas loading at time ti.
Ptmax
Yin tmax ti ¼0 Yti ðti ti1 Þ
tBT;ideal ¼ ð11:12Þ
Yin
The carbon loading after each experiment was calculated with Eq. (11.13). V_
and mAC represent the volumetric flow through the adsorber and the mass of carbon
in the adsorber, respectively.
Yin V_ tBT,ideal
X¼ ð11:13Þ
mAC
11.3.4 Decomposition
Several authors describe methanol as a minor decomposition product in not cat-

alyzed reactions based on the hydrolysis of DMC (Yoshida et al. 1997; Tundo
2001; Fu et al. 2005). The stoichiometry of this hydrolysis can be expressed with
Eqs. (11.14) and (11.15) (Stehmann et al. 2017b). The positive reaction enthalpy
indicates an endothermic reaction.
DMC þ H2 O ! 2MeOH þ CO2 ðDhR ¼ 16:43 kJ=molÞ ð11:14Þ
DMC þ H2 O ! 2MeOH þ CO þ 1=2O2 ðDhR ¼ 299:36 kJ=molÞ ð11:15Þ
The decomposition can be depicted with the conversion that is detected in the
gas phase Cgas calculated with the gas loading of methanol YMeOH, the free air
volume (VVial – VAC) and the added mass of DMC nDMC, see Eq. (11.16)
(Stehmann et al. 2017b).
0:5 YMeOH ðVVial VAC Þ

Cgas ¼ ð11:16Þ
nDMC
11.4 Adsorption Equilibria of Electrolyte Components

on Activated Carbon
Adsorption equilibria are used to determine the carbon loading that will be reached
in presence of a certain gas loading. The equilibrium is often described as an
adsorption isotherm with equations depending on adsorptive, adsorbent and tem-
perature. These equilibria need to be determined experimentally and subsequently
can be used to calculate the carbon mass required to clean a certain off gas. The
following chapter presents experimentally investigated adsorption equilibria of the
electrolyte components dimethyl carbonate and ethyl methyl carbonate such as
possible decomposition products methanol and ethanol.
11.4.1 Single-Component Adsorption Equilibria of DMC,

EMC, Methanol and Ethanol
Single-component adsorption equilibria were determined for DMC and EMC as

well as for their decomposition products methanol and ethanol on the activated
carbons SC40 and Norit RB4. Figure 11.3 shows the adsorption equilibria of DMC
on SC40. The experimental data were fitted with the model of Tóth, see Eq. (11.2).
The adsorption equilibria are depicted as isotherms showing the corresponding
carbon and gas loading of an equilibrium state. The carbon loading is increasing
with higher gas loading, eventually reaching a maximum carbon loading. For a
specific gas loading the carbon loading is decreasing with higher temperature due to
an exothermic adsorption process. However, for a specific gas saturation the tem-
perature has a minor influence on the carbon loading, see Fig. 11.3b. Therefore, the
influence of temperature onto the adsorption capacity is primarily based on the
vapor/liquid-equilibria and not the adsorption process itself. The data of all material
combinations were summarized with the corresponding parameters of the Tóth
(a) 10 (b) 10
carbon loading X / mmol g-1
carbon loading X / mmol g-1
1 1
0.1 0.1
0.001 0.01 0.1 1 0.0001 0.001 0.01 0.1 1
gas loading Y / mmol L-1 gas saturation φ / -
20 30 40 50 60 °C
20 30 40 50 60 °C
Fig. 11.3 Adsorption equilibria of DMC onto activated carbon SC40. Carbon loading dependent
on (a) gas loading and (b) saturation
Table 11.1 Parameters of Tóth Eq. (11.4) for different adsorptives onto SC40
T Ysat Xmax Xu = 1 b t R2
/°C /mmol L−1 /mmol g−1 /mmol g−1 /L mmol−1 /– /–
MeOH 20 5.29 27.62 18.47 0.38 1.00 0.947
30 8.64 27.62 18.47 0.23 0.925
40 13.60 28.25 18.47 0.14 0.956
50 20.70 28.25 18.47 0.09 0.864
60 30.58 28.25 18.47 0.06 0.868
EtOH 20 2.42 14.15 10.78 7.26 0.60 0.979
30 4.18 14.69 10.78 3.33 0.979
40 6.91 15.24 10.78 1.64 0.985
50 11.02 15.71 10.78 0.88 0.987
60 16.98 15.71 10.78 0.57 0.963
DMC 20 2.29 12.47 7.87 500 0.28 0.991
30 3.68 12.13 7.52 270 0.994
40 5.72 11.73 7.11 147 0.990
50 8.59 11.29 6.85 98 0.988
60 12.53 10.64 6.45 67 0.997
EMC 20 1.04 7.09 5.54 2458 0.32 0.982
30 1.73 6.70 5.24 1471 0.955
40 2.78 6.70 5.24 916 0.985
50 4.31 6.70 5.24 591 0.989
60 6.48 6.99 5.24 234 0.967
equation in Tables 11.1 and 11.2. These tables contain the saturation gas loading
Ysat, the parameter Xmax, b and t according to Eq. (11.4) and the carbon loading at a
gas saturation of u = 1. The parameters were calculated with the thermodynamic
constraints that Xu = 1 and b are decreasing with temperature.
11.4.2 Two Component Equilibria
The off gas of a real process will contain multiple components. High boiling
components such as EC can be easily condensed; therefore, the main constituents of
the off gas will be the low boiling components such as DMC and EMC and possibly
their decomposition products that occurred during the lifetime of the battery or
previous recycling steps. With two and more components present in the off gas a
competitive adsorption occurs between those components and the adsorption
equilibria will be a result of the ternary interaction of the two adsorptives and the
adsorbent. The investigation and evaluation of binary adsorption equilibria in batch
experiments is challenging, since only the total mass and composition of the added
adsorptive and the mass of carbon can be influenced during the sample preparation.
Table 11.2 Parameters of Tóth Eq. (11.4) for different adsorptives onto Norit RB4
T Ysat Xmax Xu = 1 b t R2
/°C /mmol L−1 /mmol g−1 / mmol g−1 /L mmol−1 /– /–
MeOH 20 5.29 13.66 11.36 0.93 1.00 0.965
40 13.6 13.66 11.36 0.36 0.988
60 30.58 13.66 11.36 0.16 0.953
EtOH 20 2.42 8.16 7.23 12.80 0.71 0.978
40 6.91 8.16 7.23 4.48 0.971
60 16.98 8.16 7.23 1.83 0.966
DMC 20 2.29 6.45 5.23 1051 0.34 0.968
40 5.72 6.57 5.23 324 0.932
60 12.53 6.45 5.02 111 0.993
EMC 20 1.04 4.87 4.28 9763 0.34 0.916
40 2.78 4.74 4.15 3433 0.947
60 6.48 4.63 4.01 1192 0.978
The resulting gas loading and composition of the gas phase is appearing only after
reaching equilibrium and therefore cannot be adjusted while sample preparation.
However, the composition of the adsorbed phase is assumed to equal the compo-
sition of the added adsorptive, since the adsorbed mass highly exceeds the not
adsorbed mass in the gas phase. Therefore, the results have been evaluated for each
constant composition of the adsorbed phase represented by the adsorbed mole
fraction of EMC xEMC. There are two main characteristics of multi-component
equilibria that need to be addressed separately. First, the selectivity of the adsorbent
material for a specific component can be depicted as the dependency of the
adsorbed phase composition on the gas phase composition. Secondly, the total
carbon loading can be depicted as an isotherm dependent on the total gas loading
for a specific adsorbed phase composition.
The compositions of adsorbed and gas phase are presented in Fig. 11.4. Graph a
shows the relation between the composition of the gas phase and the adsorbed
phase for a constant gas loading and compares experimental results with IAST
calculations. Graph b depicts the selectivity calculated using Eq. (11.11) based on
IAST data.
Graph a shows, that the mol fraction of EMC in the adsorbed phase is higher
than in the gas phase. Therefore, EMC is adsorbed preferentially. This is also
indicated by the selectivity a, which is higher than 1. Furthermore, the experimental
results do not show a significant difference in selectivity depending on the tem-
perature, while the IAST calculations predict a higher selectivity at high tempera-
tures. Graph b shows, that higher gas loadings decrease the selectivity eventually
reaching a = 1, indicating that both DMC and EMC are adsorbing equally. Even
though the IAST calculation represents the selectivity correctly for a total gas
loading of Y = 0.1 mmol L−1 and a temperature of 50 °C, the selectivity at a
temperature of 30 °C is underestimated.
(a)
xEMC / molEMC molEMC+DMC-1 1 (b)
3
0.8
0.6
2
0.4 Y = 0.1 mmol L-1

50 °C (exp)
1
50°C (IAST)
0.2 yEMC = 0.15 mol mol-1
30°C (exp)
50 °C
30°C (IAST) 30 °C
0 0
0 0.2 0.4 0.6 0.8 1 0.001 0.01 0.1 1
yEMC / molEMC molEMC+DMC-1 Y / mmol L-1
Fig. 11.4 Adsorption equilibrium of DMC and EMC onto activated carbon SC40. (a) gas phase
and adsorbed phase compositions at a constant total gas loading of Y = 0.1 mmol L−1,
(b) selectivity at a constant gas composition of yEMC = 0.15 mol mol−1 (Stehmann et al. 2016)
Fig. 11.5 Tóth-parameter of 12 1750

two-component adsorption
(DMC and EMC) onto 11 1500
activated carbon SC40 at
10 1250
Xmax / mmol g-1
b / L mmol-1
30 °C with t = 0.3. Dots:
Parameter of experimental 9 1000
data. Lines: Parameter
calculated with IAST 8 750
7 500
6 250
5 0
0 0.2 0.4 0.6 0.8 1
xEMC / mol mol-1
Xmax b
b (IAST) Xmax (IAST)
The results can also be presented as adsorption isotherms analogue to

Sect. 11.4.1. Figure 11.5 shows the parameter of a Tóth fit, Eq. (11.4), for different
adsorbed compositions. In this graph the values at xEMC = 0 molEMC mol total-1 and
xEMC = 1 molEMC moltotal-1 are equal to the single component data of DMC and
EMC, respectively, which were already presented in Table 11.1. The parameter in
between were calculated from equilibria data of the binary system DMC and EMC
for each constant adsorbed composition xEMC.
The maximum carbon loading Xmax is linearly dependent on the composition of
the adsorbed phase. This shows that the capacity of the adsorbent in only dependent
on the space requirement of the present molecules. This trend is also represented in
the IAST calculations. In contrast to that, the parameter b is increasing exponen-
tially with mole fraction of EMC in the adsorbed phase. The parameter b describes
the ratio of adsorption and desorption rate and represents the slope of the isotherm
at low gas loadings. This course of the parameter b indicates, that the adsorption
equilibria at low gas loadings is mainly influenced by the component with the
highest concentration. The experimental data is constantly underestimated by IAST
calculation. Concluding, IAST calculations are not sufficient to precisely calculate
the carbon loading at low gas loadings, since neither the selectivity nor the carbon
loading at low gas loadings can be described correctly. However, since the
adsorption of EMC and DMC is very similar, the resulting total carbon loading of
certain off gas mixture can be estimated with IAST.
11.4.3 Off Gas Cleaning with Fixed Bed Adsorption
The off gas cleaning in the battery recycling can be done with fixed bed adsorption.
Using this method, the off gas is led through a packing of adsorbent, in this case
activated carbon. The contaminants are adsorbed by activated carbon leading to a
reduction of gas loading. The following pages describe typical breakthrough curves
and methods to monitor and predict the breakthrough progress.
The fixed bed adsorption can be depicted with a breakthrough curve, see
Fig. 11.6. This figure depicts the DMC and MeOH gas loading of the off gas exiting
the adsorber bed as a function of the operation time. Additionally, the temperature
at 19.5 cm downstream the inlet of the adsorber (1.5 cm upstream the adsorber exit)
is presented. The gas loadings of DMC and MeOH are slowly increasing after 30 h
indicating that the MTZ has reached the adsorber exit. While the gas loading of
33 1.6
Temperature at 1
19.5 cm 1.4
DMC
C
temperature / °C
32 1.2
MeO
OH
Yi / mmol L-1
1
31 0.8
0.6
30 0.4
0.2
29 0
0 20 40 60
time / h
Fig. 11.6 Breakthrough curve of model off gas loaded with DMC, fixed bed is filled with carbon
SC40 (Stehmann et al. 2017a)
(a) 20 (b) 60
ideal breakthrough time / h

location of ΔT = 1K / cm
15
40
10
5
y = 0.4102x + 1.8563 y = 1.1191x - 1.7379
R² = 0.9985 R² = 0.9884
0 20
0 20 40 20 40 60
time / h temperature breakthrough time / h
Fig. 11.7 (a) Location of temperature peaks along the adsorption process. (b) Parity chart of
temperature and ideal breakthrough time (Stehmann et al. 2017a)
methanol is decreasing again, the gas loading of DMC is sharply increasing after
45 h until a maximum value of appx. 1.4 mmol L−1 is reached. The ideal break-
through time of this experiment is 49.7 h, see Eq. (11.12).
The adsorber is held at a constant temperature of 30 °C. However, after the MTZ
reaches the adsorber exit, the temperature increases up to 32 °C and hereafter
decreases again. This increase is explained by the exothermic adsorption process,
leading to an energy release in the MTZ. Figure 11.7a depicts the location of the
temperature increase along the adsorber. The temperature front and therefore the
mass transfer zone migrate through the adsorber with a constant velocity.
The breakthrough time of the temperature front can be approximated with the linear
fit. This was done for several experiments and depicted in Fig. 11.7b.
The ideal breakthrough time is higher than the temperature breakthrough time
and therefore identifies the temperature increase at the adsorber exit as a promising
parameter to monitor the adsorption process.
Beyond monitoring of an operating process, the prediction of the breakthrough
time is favorable. This can be done with short-cut calculations using previously
determined adsorption equilibria. These equilibria can be easily determined in batch
experiments, but may also be calculated from breakthrough experiments with
Eq. (11.13), see Fig. 11.8.
The results of breakthrough experiments at 30 °C show the carbon loading
increasing linearly with the gas loading. Additionally, the temperature influence as
observed in batch experiments is visible, meaning that the carbon loading increases
at lower temperatures. However, the results obtained in batch experiments show a
high deviation. The reason of this deviation was not determined, yet. Possible
reasons may be an influence of the gas flow or the decomposition as addressed in
the following chapter.
14
carbon loading / mmol g-1 12 Batch, 30°C

10 25 °C
30 °C
8
40 °C
6 Fit (30°C)
4 y = 7.6485x
R² = 0.9948
2
0
0 0.5 1 1.5 2
gas loading / mmol L-1
Fig. 11.8 Calculated carbon loading corresponding to gas loading (Stehmann et al. 2017a)
11.5 Decomposition of Electrolyte Components

on Activated Carbon
Dimethyl carbonate and ethyl methyl carbonate are decomposing. This may already
happen during the ageing of the battery and the recycling process itself. As it is
presented in the chapter “Separation of the Electrolyte—Thermal Drying”, the
condensed off gas produced in a real drying process contains the decomposition
products ethanol and methanol. These components were also detected in the
adsorption experiments, where only the pure components DMC and EMC were
used. The decomposition products may influence the adsorption behavior and the
recycling strategy regarding the removed electrolyte. Therefore, major influences
on the decomposition were identified in batch experiments and the decomposing
behavior in fixed bed adsorption was investigated.
11.5.1 Main Influences on Decomposition
Figures 11.9 and 11.10 show the influence of equilibration time and temperature.
Gas phase adsorption has a fast kinetic, therefore, the adsorption equilibrium itself
is barely influenced by equilibration times over 6 h (see Fig. 11.9a). However, the
conversion increases with time and up to 23 h of equilibration time no steady state
could be detected (see Fig. 11.9b). The total amount of decomposed material is low,
so that the divergent equilibrium of methanol is not affecting the total carbon
loading. The conversion increases with higher gas loadings as a result of the
asymptotic adsorption isotherm.
Moreover, higher temperatures accelerate decomposition (see Fig. 11.10). This
underlines the endothermic character of the reaction. Higher temperatures will
occur during desorption. The carbon has to be heated to remove the previously
Fig. 11.9 Carbon loading and conversion dependent on the equilibration time for adsorption of
DMC onto Norit RB4 (Stehmann et al. 2017b)
Fig. 11.10 Conversion

dependent on temperature for
adsorption of DMC onto
Norit RB4 (Stehmann et al.
2017b)
adsorbed molecules. The gained desorbate may be recycled and the contamination
with decomposition products can lower the value.
11.5.2 Decomposition During Fixed Bed Adsorption
As presented in Sect. 11.5, decomposition also occurs during fixed bed adsorption.
Due to a different residence time and ratio of adsorbent to adsorptive, the results are
not fully comparable to the results of batch experiments. Figure 11.11 shows the
rate of decomposition and the gas loading of methanol at steady state as a function
of the inlet gas loading.
decomposition / mmolMeOH h-1 0.7 0.025
YMeOH,steadystate / mmol L-1

0.6
0.020
0.5
0.4 0.015
0.3 0.010
0.2
0.005
0.1
0 0.000
0 0.5 1 1.5 2 0 0.5 1 1.5 2
gas loading Yin / mmol L-1 gas loading Yin / mmol L-1
Fig. 11.11 Decomposition rate and methanol concentration at steady state occurring during a
fixed bed adsorption of DMC onto activated carbon SC40 at 30 °C
Both parameters are increasing with gas loading, indicating that the decompo-
sition is dependent on the amount of DMC passing the system.
Decomposition was also observed in the activated carbon filters of the mobile
shredder in the demonstration plant. The exiting off gas of the shredder was led
through two consecutive adsorbers filled with activated carbon SC40. After the off
gas exiting the second filter appeared to be not cleaned sufficiently anymore, carbon
samples were taken from the top and the bottom of each filter and analyzed for their
total loading and the composition of the corresponding gas phase. The composition
of the corresponding gas phase was determined by analyzing the gas phase that was
built after filling appx. 3–4 g of a sample into a gas tight vial and storing it for 6 h
at 60 °C. The total loading was determined by the weight loss that appeared after
storing the open vials for 24 h at 120 °C.
The results are depicted in Fig. 11.12. Graph (a) shows the setup of the carbon
filters, while graph (b) depicts the carbon loading and the composition of the
headspace above the carbon. The carbon loading is decreasing with the length of the
adsorber. At the entrance of the first adsorber, an uptake of 25 wt% was observed.
This value decreases down to 5 wt% at the exit of the second adsorber. Therefore, it
is likely, that the exiting off gas was not cleaned sufficiently anymore. Also, the
mass transfer zone in this system appears to be not as sharp as shown in the lab
scale experiments in Sect. 11.5. This may result from the low gas flow and therefore
high operation time of several months. This long operation time may have also
caused the high concentration of methanol. As shown before, a long contact time
between carbon and electrolyte increases decomposition. Such high values of
decomposition as in these off gas filters have been observed in lab scale experi-
ments neither with carbon nor in drying experiments with real battery fragments.
Fig. 11.12 Exhaust gas cleaning of the mobile shredder (a) and analysis of the resulting carbon
loading and composition of electrolyte adsorbed by carbon (b) (Stehmann et al. 2017a)
11.6 Conclusions
This chapter presented adsorption equilibria determined in batch experiments as

well as the breakthrough behavior in fixed bed experiments. In addition, the
decomposition of electrolyte components present at adsorption onto activated car-
bon was investigated.
The adsorption equilibria give a database for the electrolyte components DMC
and EMC as well as their decomposition products methanol and ethanol. Lower
temperatures enhance the adsorption caused by the temperature dependent vapor/
liquid-equilibrium. However, the maximum carbon loading for a saturated gas is
mainly constant for temperatures from 20 to 60 °C. Regarding binary equilibria,
temperature could not be identified to influence the adsorption selectivity of the
components DMC and EMC. At low gas loadings, EMC is adsorbed preferentially.
With increasing gas loading, the adsorption selectivity decreases, eventually leading
to an equal adsorption of DMC and EMC. The experimental data of two component
equilibria was compared with IAST calculations based on single component data.
The maximum carbon loading Xmax can be sufficiently estimated with the IAST
model representing the adsorption at higher gas loadings, while the parameter b that
is mostly relevant for low gas loadings was underestimated.
Fixed bed experiments identify the temperature increase at the adsorber outlet as
a parameter to easily monitor the breakthrough progress. The calculated carbon
loading from the breakthrough experiments increases linearly with the gas loading.
Therefore, this adsorption equilibrium is not consistent with the equilibrium
determined in batch experiments.
Decomposition was detected in batch experiments as well as fixed bed experi-
ments with model off gas only containing DMC and real off gas of the
demonstration plant. It increases with temperature and residence time. The total
decomposition using model off gas increases linearly with gas loading. The gas
loading of methanol peaks at the beginning breakthrough of DMC. After the peak,
the gas loading of methanol is decreasing again eventually reaching a low constant
value. This plateau value is determined by the decomposition rate of DMC under
fixed bed adsorption conditions. The analysis of the adsorbers of the demonstration
plant shows a very high amount of decomposition products caused by the long
operation time and therefore the long contact time of electrolyte components and
carbon.
11.6.1 Recommendations for Designing the Adsorption

Process
The adsorption process should be setup with two adsorber, according to Fig. 11.13.
Two adsorbers enable a continuous off gas cleaning in one adsorber and parallel
Fig. 11.13 Recommended

setup for off gas cleaning with
adsorption including heating
and purge gas, based on
(Mersmann et al. 2005)
regeneration by heating and/or purging in the second adsorber. The mass flow of the
purge gas should be kept as low as possible to result in a desorption gas with a high
gas loading. This off gas can be condensed in a separate process step or even be fed
upstream the already existing condensation to regain the electrolyte. Alternatively,
the loaded carbon can be thermally disposed. In this case, no heating or purging is
required. The carbon in the loaded adsorber can be exchanged, while the second
adsorber ensures the continuous off gas cleaning.
Special attention should be paid to the time dependency of the off gas loading.
As presented, the accessible carbon loading is increasing with increasing gas
loading. Vice versa, decreasing gas loading of the off gas will wash out previously
adsorbed electrolyte. Therefore, the design of the adsorptive off gas cleaning should
consider the lowest gas loading that is likely to appear during the operation time of
the adsorber. In any case, a comprehensive recognition of the different operating
situations is required in the design of a suitable and robust off gas cleaning process.
Further research works should address the influence of decomposition on adsorp-

tion equilibrium and the fixed bed adsorption. Furthermore, the reason for the
difference of adsorption equilibria between batch and breakthrough experiments
should be investigated.
Within the project LithoRec II, desorption of the loaded carbon was proven to be
possible with temperatures of 60 and 90 °C and reduced pressure. However, these
parameters should be adjusted with respect to desorption rate, energy consumption
and decomposition.
References
Bundesministerium für Umwelt (2002) Naturschutz und Reaktorsicherheit. Erste Allgemeine

Verwaltungsvorschrift zum Bundes-Immisionsschutzgesetz, Technische Anleitung zur
Reinhaltung der Luft - TA Luft
Fu Y, Zhu H, Shen J (2005) Thermal decomposition of dimethoxymethane and dimethyl carbonate
catalyzed by solid acids and bases. Thermochim Acta 434(1–2):88–92. https://doi.org/10.1016/
j.tca.2005.01.021
Langmuir I (1918) The adsorption of gases on plane surfaces of glass, mica and platinum. J Am
Chem Soc 40(9):1361–1403. https://doi.org/10.1021/ja02242a004
Markham EC, Benton AF (1931) The adsorption of gas mixtures by silica. J Am Chem Soc 53
(2):497–507. https://doi.org/10.1021/ja01353a013
Mersmann A, Kind M, Stichlmair J (2005) Thermische Verfahrenstechnik: Grundlagen und
Methoden, 2., wesentlich erw. und aktualisierte Aufl. Chemische Technik Verfahrenstechnik.
Springer, Berlin
Myers AL, Prausnitz JM (1965) Thermodynamics of mixed-gas adsorption. AIChE 11(1):121–126
Stehmann F, Scholl S (2015) Adsorption von Batterieelektrolytkomponenten aus einem Inertgas

an Aktivkohle. Chem Ing Tech 87(8):1037. https://doi.org/10.1002/cite.201550013
Stehmann F, Prziwara P, Bradtmöller C, Schoenitz M, Scholl S (2016) Adsorption equilibria of
dimethyl carbonate and ethyl methyl carbonate onto activated carbon. Chem Ing Tech 88
(3):327–335. https://doi.org/10.1002/cite.201500130
Stehmann F, Jahnke S, Balmforth-Slater C, Scholl S (2017a) Adsorptive exhaust gas cleaning for
recycling of Li-ion-batteries. J. Electrochem, Soc, p 164
Stehmann F, Wiegmann E, Scholl S (2017b) Decomposition of dimethyl carbonate caused by
adsorption onto activated carbon. Adsorption 23. doi:10.1007/s10450-016-9858
Tóth J (1971) State equations of the solid-gas interface layers. Acta Chim Hung Tomus 69(3):
311–328
Tundo P (2001) New developments in dimethyl carbonate chemistry. Pure Appl Chem 73
(7):1117–1124
Yoshida H, Fukunaga T, Hazama T, Terasaki M, Mizutani M, Yamachi M (1997) Degradation
mechanism of alkyl carbonate solvents used in lithium-ion cells during initial charging.
J Power Sources 68(2):311–315. https://doi.org/10.1016/S0378-7753(97)02635-9
Chapter 12
Material Separation
Jan Diekmann, Steffen Sander, Guido Sellin and Arno Kwade
Abstract The material separation processes combine all mechanical treatments

after crushing and drying, which aim to separate the different materials of
lithium-ion battery modules. In the following text, the input material for these
processes is characterized before the multistage separation and different process
influences are described. The resulting process is a combination of mixing,
zig-zag-sifting, crushing, and sieving processes, leading to an overall material
recycling rate of 75–80% of an electric vehicle (EV) battery system.
12.1 Introduction
The composition of a generic battery system of an electric vehicle (EV) can be seen
in Fig. 12.1. The main components are the battery system periphery, combining
housing materials, cables, battery management system, module periphery, and the
battery cells. These cells consist of cathodes, anodes, electrolyte, and separator, as
well as the battery cell housing. The battery cells represent around 55% of the
overall weight of an EV.
While the different materials of the battery system periphery can be recovered
via disassembly, the battery modules, and simultaneously the battery cells are
crushed and dried. The resulting fragments and powders have to be separated
efficiently and with a purity that enables selling of the material fractions or further
processing. As the volatile components are removed via drying, the main materials
that have to be separated afterwards are copper foil from the anode, aluminum foil

S. Sander
Hosokawa Alpine Aktiengesellschaft, Augsburg, Germany
G. Sellin
Electrocycling GmbH, Goslar, Germany

https://doi.org/10.1007/978-3-319-70572-9_12
Fig. 12.1 Composition of a generic battery system of an electric vehicle (Diekmann et al. 2016)
from the cathode, the housing materials of the module (aluminum, steel, plastics)
and the battery cell housings (aluminum), the separator (plastics), as well as the
coating materials of the electrodes. While the anode coating mainly consists of
graphite, the coating material of the cathode contains a lithium transition metal
oxide, in this case lithium nickel cobalt manganese oxide (NCM). These materials
should be processed hydrometallurgically or pyrometallurgically after separation to
recover lithium, nickel, cobalt, and manganese.
12.2 Characterization of the Input Material
To characterize the input material of the separation processes after crushing and
drying, sieving analysis and manual material sorting of each fraction were carried
out. Because the aim was to mainly use mechanical separation processes, the so
called separating factor was used instead of the particle size. This separating factor
is used to characterize separation via air-classification. To explain the meaning of
this, it is necessary to take a look at the function principle of a zig-zag-sifter, which
is shown in Fig. 12.2.
A zig-zag-sifter is a tube with a cascade of cross-flow sifting processes, repre-
sented by the zig-zag and an air flow from the bottom. For particles with the same
size, the separation is carried out by the different densities. Particles with higher
12 Material Separation 209
light product
heavy product
Fig. 12.2 Scheme of the function principle of a zig-zag sifter (filled dots: high density, unfilled
dots: low density)
Coating Materials
1.1
Separator
1.0 Plastic foil
Al current collector foil
) [-]
0.9 Cu current collector foil

Plastics
0.8
cumulated distribution Q (x
Cell housing (Al)

0.7 Module housing (Al)
Steel
0.6 Electric conductor (Cu)
0.5
0.4
0.3
0.2
0.1
0.0
0.01 0.1 1 10
separating parameter x [mm (g cm-3)]
Fig. 12.3 Cumulated distribution of the separating parameter of different materials of a crushed
and dried battery module (Diekmann et al. 2016)
density are going to the heavy product, while particles with lower density form the
light product. For fragments or particles with different sizes and densities, the
separating factor comes into play. It combines the two crucial factors of a separation
via air-classification: particle or fragment size and therefore uplift force, as well as
density and therefore sedimentation velocity. Similar separating parameters have a
negative influence on the selectivity of an air-classification (Diekmann et al. 2016).
Figure 12.3 shows the cumulated distribution connected to the separating

parameter for the different material fractions of a crushed and dried EV battery
module. The used battery modules were field tested and had an average weight of
4.95 kg. They contained six 25 Ah battery cells with NCM as cathode and graphite
as anode active materials.
According to the separating parameter, separation of the current collector foils
(Al and Cu), separator, and plastic foils would be unselective. Separation of the
materials with higher densities, like module housing, cell housing, plastics, and
steel from the other materials would be possible via zig-zag-sifting. Due to the
smaller particle size, the coating materials of anode and cathode can be recovered
by sieving. Under the applied drying conditions, hardly any coating materials
remain on the current collector foils.
To recover nearly all materials in high purity, a multistage separation is applied.
12.3 Recovery of the Heavy Parts (1st Air-Classification)
The first separating process step is an air classification via zig-zag-sifting to recover
the so called heavy parts. This classification is carried out with a mass load of
109 g kg−1 and an air velocity of 3.34 m s−1 (Diekmann et al. 2016). While the
separator, current collector foils, plastic foils, and coating materials are located in
the light product, the materials with high density are found in the heavy product.
The components of the heavy product can be seen in Fig. 12.4.
The heavy parts mainly consist of aluminum of the module and cell housing
(74.3%), but also steel, copper, plastics, and conclusions. These conclusions are
compounds of separator, current collector foils, and black mass. They represent the
first loss of the process and originate from the first crushing step. The removal of
the heavy parts is essential for the protection of the following process steps and
leads to a change of the cumulated distribution of the remaining fragments
(Fig. 12.5).
12.4 Recovery of the Coating Materials and Process

Influences
The recovery of the coating materials is carried out via sieving with a vibration
sieve at a mesh size of 500 µm. It can be executed before or after the separation of
current collector foils and separator. In this case, a second crushing step is applied
before sieving, exerting a cutting stress on the fragments and increasing the overall
mechanical stress. Therefore, a cutting mill with a discharge screen of 10 mm was
used. To show the influence of this second crushing step, two different process
routes were compared: one with a second crushing step and another without.
Fig. 12.4 Composition and

components of the heavy
parts fraction
Al module Copper
26.6 % 5.0 %
Al housing Plastics
47.7 % 4.0 %
Steel Inclusions
13.8 % 2.9 %
The change in the cumulated distribution of the fragments after different process
steps can be seen in Fig. 12.5.
The first air-classification already transforms the cumulated distribution to a
higher share of smaller fragment sizes. The application of a second crushing step
further increases the mass of smaller fragment and particle sizes. Significantly, this
change can be seen for particle sizes smaller than 500 µm.
After the following sieving, this influence becomes obvious in an increased yield
of the coating materials (Fig. 12.6). The yield is increased from 60 to 75% due to
the comminution of the coating material fragments as well as the decomposition of
the remaining conclusions. Furthermore, the increased yield only results in slightly
increased impurities. Cu impurities are increased from 1.7 to 1.8 wt%, while the
concentration of Al rises from 0.5 to 0.6 wt% (Diekmann et al. 2016). Obviously,
the “mild” cutting stress only slightly influences the impurities of the coating
materials.
Noticeably, a supplemental process influence can be shown by adding a mixing
step directly after crushing and drying. This additional process step increases the
1,1
1,0
cumulated distribution Q (x) 0,9
0,8
0,7
0,6
0,5
after 1st crushing/ drying
0,4
after 1st classification
0,3 after 1st classification/ sieving
after 1st classification/ 2nd crushing
0,2 after 1st classification/ 2nd crushing/ sieving
0 2 4 6 8 10 12 14 16 18 20 22 24
Fig. 12.5 Change of the cumulated distribution of fragments of a battery module after different
process steps. The cumulated distributions relate to the light product after classification and to the
residue after sieving (Diekmann et al. 2016)
Fig. 12.6 Yield of the 100

mass output black mass mBM [%]
Coating materials
coating materials with/ 90 Fe impurities
without 2nd crushing and Al impurities
mixing step and resulting 80 Cu impurities
impurities. Modified from 70
(Diekmann et al. 2016)
60
0
with with mixing
2nd crushing step and
nd
2 crushing step
yield of coating materials to more than 90% (Fig. 12.6). This mixing step was
carried out at 300 rpm. At this rotation velocity, the components of the module and
cell housing exert a shear stress to the fragments and especially the coating material
fragment. Although the impurity concentrations are also increased, (Cu: 1.9 wt%,
Al 0.8 wt%, Fe: 0.3 wt%) this fraction can be treated pyrometallurgically or
hydrometallurgically afterwards to recover the valuable metals.
12.5 Separation of Current Collector Foils and Separator

(2nd Air–Classification)
The applied second crushing step also influences the separation of the current
collector foils and the separator. This process step aims to remove the valueless
separator, which nowadays is mostly polyethylene. The main challenge herein is the
separation of the aluminum foil and the separator, as these materials have a very
similar density. The experiments were carried out via zig-zag-sifting with a mass
load of 25 g kg−1 air and varied air velocities (Diekmann et al. 2016). The results of
this second air-classification step without preceding a second crushing step are
shown in Fig. 12.7.
Generally, an increasing air velocity leads to a higher share of the overall alu-
minum and separator in the light product. For an acceptable amount of 1 wt% of the
overall aluminum in the light product, around 40 wt% of the entire separator
remains in the heavy product.
The positive influence of adding a second crushing step to the separation per-
formance is shown in Fig. 12.8. This figure shows the share of the overall alu-
minum and the overall separator in the light and heavy products of the second
air-classification with a preceding second crushing step.
For an aluminum loss of around 1 wt%, the share of the overall separator in the
heavy product can be minimized to 23.7 wt%. Also, the selectivity of this process is
enhanced via the second crushing. The reason for this increased selectivity can be
found in the change of the cumulated distributions in Fig. 12.5. Second crushing
limits the upper fragment size and decomposes the inclusions. Following sieving
limits the lower particle size. These two process steps lead to a closer particle or
fragment size distribution, and therefore the different densities of the materials
becoming more important. The homogenized distribution enables a more precise
air-classification.
Fig. 12.7 Mass output of 110

vS 1.08 m s-1
separator and aluminum in 100 99.44 98.98 98.55
vS 1.23 m s-1
light and heavy product of the 90
mass output Al foil and
vS 1.32 m s-1
2nd air-classification without
separator m Al, Sep [%]
80
2nd crushing step (Diekmann
70
et al. 2016) 60.26 60.09
60
51.39
50 48.61
39.91 39.74
40
30
20
10
0.56 1.02 1.45
0
light product heavy product light product heavy product
aluminum foil separator

110 vS 0.92 m s-1

100 99.1
100 vS 1.10 m s-1
94.73
90 vS 1.39 m s-1
mass output Al foil and

separator m Al, Sep [%]
81.31
80 75.31
70 68.23
60
50
40
31.77
30
23.69
20 18.69
10 5.27
0 0.9
0
aluminum foil separator
Fig. 12.8 Mass output of separator and aluminum in light and heavy product of the 2nd
air-classification with 2nd crushing step (Diekmann et al. 2016)
Fig. 12.9 Product fractions of the separation processes and their compositions
12.6 Products of the Separation Processes
The separation processes generate intermediates and products. Together with the
products of the disassembly process, these are listed below for the generic battery
system (Fig. 12.9).
The products of the manual disassembly mainly consist of aluminum from the
housing components. Further components are steel, cables, and the battery man-
agement unit (BMU). The so called heavy parts, the products of the first
air-classification, also mainly consist of aluminum, but also steel, copper, plastics,
and the inclusions. Probably the most valuable fraction is generated by sieving and
contains the coating materials and impurities. The second air-classification gener-
ates two products: the separator fraction and the current collector fraction. The latter
contains the fragments of aluminum and copper foil, but also the remaining coating
and separator. As this fraction needs further treatment, tentative attempts to separate
aluminum and copper foil were carried out (Sect. 12.7).
Overall, the mechanical separation treatment reaches a material recycling rate of
75-80% of an EV battery system, not including graphite, electrolyte, and separator.
12.7 Separation of Cu and Al Foil
The following experiments were carried out with model aluminum and copper foils
from lithium-ion batteries with specified fragment sizes. As a first approach to the
separation, air-classification via zig-zag-sifting was conducted. The results at dif-
ferent air velocities are shown in Fig. 12.10.
The results show that only concentration of aluminum or copper foil in the light
and the heavy product seems to be possible. The share of the overall copper foil in
the light product are too high at every tested air velocity. Also, balling of the current
collector foil fragments did not lead to an improvement of the results (Mämpel
2013).
To maximize the purity of the aluminum and copper foil fractions, further
experiments via optical sorting were carried out. Therefore, the different colors of
the isolated copper and aluminum fragments are detected by a CCD
(Charge-Coupled-Device) camera. A following blowout system affected the tra-
jectory of the fragments by directed air blast. These fragments were the rejected
fraction.
The results of a first run with model aluminum and copper foils are shown in
Fig. 12.11. It can be seen that even one run leads to an increased purity of the two
fractions. The reject fraction already reaches a copper content of 97.5 wt% with
only small impurities. Also, the accept fraction shows a relatively high content of
aluminum (93.1 wt%).
Fig. 12.10 Mass output of 110 vS 3.13 m s-1

mass output Al and Cu foil mAl, Cu [%]
aluminum and copper foil in 100 vS 5.22 m s-1

light and heavy product after vS 6.96 m s-1
90 85.7 87.1
separation via 83.1
air-classification; data 80 77.7

75.3
obtained from (Mämpel 2013) 70

60 59.3
50
40.7
40
30 24.7
22.3
20 16.9
14.3 12.9
10
0
aluminum foil copper foil

Fig. 12.11 Mass output and 110 mass output fraction
mass output and content Al and

content of Al and Cu foil in 100
content Al foil
97.5
accept and reject fraction of 93.1 content Cu foil
90
an optical separation (1st run);
data obtained from (Mämpel 80
69.6
Cu foil [%]
2013) 70
60
50
40
30.4
30
20
10 6.9
2.5
0
accept reject
Fig. 12.12 Mass output and 110 mass output fraction

content of Al and Cu foil in
mass output and content Al and
99.7 content Al foil

100 97.2
content Cu foil
accept and reject fraction of
90
an optical separation (2nd
run); data obtained from 80 78.2
Cu foil [%]
(Mämpel 2013) 70
60
50
40
30
21.8
20
10
2.8
0.3
0
accept reject
As the copper fraction is more valuable, higher copper purity is recommended.

Therefore, a second run with the reject fraction was executed. The results of this run
can be seen in Fig. 12.12.
A copper fraction with high purity, only 0.3 wt% aluminum, was generated.
These tentative attempts show that separation of aluminum and copper foil of
lithium-ion batteries is possible, but have to be carried out with real recycling
fractions before this technique can be used on a greater scale.
12.8 Conclusion
Subsequently to the process steps of crushing and drying, mechanical processes

enable a valuable separation of battery materials. Therefore, a combination of
several different processes is needed to achieve valuable fractions, because of many
similar materials properties.
A first air-classification recovers the so called heavy parts, which are mainly the
module and the cell housings. A following sieving step regains the valuable
materials of the electrode coatings. These contain the transition metal oxides
(lithium, nickel, cobalt, manganese). A preceding mixing step increases the shear
stress on the electrode fragments, which results in high yield of the coating
materials during sieving. An additional second crushing step not only increases the
yield of the coating materials, but also homogenizes the fragment size distribution.
This leads to a higher selectivity during the removal of the valueless separator by
the second air-classification.
One resulting product of these mechanical treatments contains copper and alu-
minum foil fragments of the anode and the cathode. First separation test with this
material lead to the conclusion, that separation via optical sorting results in very
high purity of the copper fraction, while air-classification is only able to concentrate
the metals.
The multiple separation processes described here, in combination with the dis-
assembly process, lead to material recycling rates of a battery system of 75–80%,
depending on the composition of the battery systems and cells.
References

Mämpel C (2013) Recycling von Lithium-Ionen-Batterien aus Elektromobilen. Bachelor thesis,
Technische Universität Bergakademie Freiberg
Chapter 13
Hydrometallurgical Processing
and Thermal Treatment of Active
Materials
Sergej Rothermel, Steffen Krüger, Martin Winter and Sascha Nowak
Abstract In this chapter, electrodes containing the cathode material Li

[Ni0.33Co0.33Mn0.33]O2 (NCM) were recycled in order to test a newly developed
recycling concept which is aiming towards commercial application. The possibility
of graphite recovery from spent LIBs by means of three different treatment methods
is demonstrated.
13.1 Recycling of the Cathode Material
In response to the relatively high value of the cathode material, to date industrial
and scientific research has been focused on development of efficient ways for its
recycling. Different leaching techniques using oxalate (Sun and Qiu 2012), ascorbic
acid (Li et al. 2012) or citric acid (Li et al. 2010) as soft acidic leachants or
concentrated acids have been used to dissolve the active materials (Ferreira et al.
2009; Shin et al. 2005; Dorella and Mansur 2007). Besides chemical routes, bio-
logical decomposition has been investigated (Zeng et al. 2012; Krüger et al. 2014).
There are several investigations focusing on physical and mechanical separation
methods (Shin et al. 2005). Castillo et al. developed a thermal method to remove
electrolyte and electrolyte decomposition products, binder and carbon additives
(Castillo et al. 2002). But also mechanochemical routes (Saeki et al. 2004) or
dissolution processes with the utilization of N-Methyl-2-pyrrolidone (NMP) in
order to remove binder additives, soluble in NMP (Contestabile et al. 1999) were
proposed. Another concept is based on a pyro-metallurgical route in which the cells
are burned and the valuable materials are recycled from their melts
(Georgi-Maschler et al. 2012; Krüger et al. 2014).
S. Rothermel S. Krüger M. Winter S. Nowak (&)

M. Winter
Helmholtz-Institute Münster (HI MS), Forschungszentrum Jülich, IEK-12,
Münster, Germany

https://doi.org/10.1007/978-3-319-70572-9_13
However, these studies are usually presenting parts of a recycling concept in a

laboratory scale, and trying to optimize the single parameters of each step for
example the leaching step or the hydrometallurgical separation. In this chapter, a
complete recycling concept was used, based on industrially feasible separation and
leaching techniques. To investigate the quality of the proposed recycling concept,
the electrochemical data, as a key-indicator for the quality of the recovered active
material, is rated. The detection and quantification of impurities regarding their
impact on the electrochemical performance of recycled active material are key
issues of this chapter (Krüger et al. 2014).
Since Li[Ni0.33Co0.33Mn0.33]O2 (NCM) is a widely commercially used cathode
material in lithium-ion batteries, it was used as probe material for the new recycling
concept. Commercial spent lithium-ion pouch-bag cells, containing a NCM cath-
ode, a graphite anode and an LiPF6/organic carbonate solvent based electrolyte, as
well as production rejects of the NCM electrode fabrication were taken as source for
the recycling process. The cells had an initial capacity of 2 Ah. The used cells were
formatted and the calendar life was at least over 1 year (Krüger et al. 2014).
The cells were discharged over an ohmic resistance to 0 V, were opened and the
electrodes were detached from the tabs using a scalpel. The electrode sheets were
sorted to anode and cathode. The cathodes were processed to recycling and briefly
washed with dimethyl carbonate (DMC). As presented in Fig. 13.1, the cathodes
were mechanically stressed using a cutting mill, resulting in a separation of the
electrode compounds (NCM, binder and carbon conductive additive) and current
collector foil (Hanisch et al. 2011). The stressed material was sieved through a
200 µm mesh and the resulting powder consisting of the electrode compounds was
dissolved in 10% sulphuric acid with addition of H2O2 with a liquid to solid ratio of
50 g L−1 at 60 °C. The separation of the transition metals was done by precipitation
Fig. 13.1 Schematic illustration of the material process chain of the proposed recycling concept
starting from electrode processing (Krüger et al. 2014)
13 Hydrometallurgical Processing and Thermal Treatment … 221
as hardly soluble carbonate salts under alkaline conditions. To release the precip-
itate from attached lithium, the filtrate was washed several times with distilled
water. The lithium containing solutions were merged and the lithium was extracted
with the use of an ion-exchange device. The same procedure was followed with the
material of the cathode production rejects (Krüger et al. 2014).
The gained transition metal solutions were characterized in terms of their ele-
mental and ionic composition using inductively coupled plasma—optical emission
spectrometry (ICP-OES) and ion chromatography (IC) techniques. To obtain the
needed amount of active material, defined by the lab scale re-synthesis reactor, the
new solutions were matched in their ionic and elemental composition. Therefore,
transition metal carbonates, sulfates and hydroxides were added as pure transition
metal salts to distilled water until the previously detected concentration of ions in
the solution, as well as its pH value was reached (Krüger et al. 2014).
Re-synthesis of the active material was carried out by a hydrometallurgical-
precursor synthesis. Stoichiometric amounts of LiOH were added to the precursors
and calcination was performed, subsequently. The obtained material was de-
agglomerated by using a jet mill (Krüger et al. 2014).
The sources for the active materials were the synthetically merged transition
metal solutions. Finally, two new active material samples were prepared: One
material from re-synthesized electrode production reject’s material (Rejects) and the
other from re-synthesized spent cells (Cycled). They were compared to the initial
(uncycled and unprocessed) active material synthesized from pure metal solutions
(Reference) (Krüger et al. 2014).
The synthesized materials were characterized by ICP-OES, scanning electron
microscopy (SEM), x-ray powder diffraction (XRD) and electrochemical methods
such as cyclic voltammetry, constant current cycling and rate performance exper-
iments in Li-ion battery half and full cell. The complete process is schematically
shown in Fig. 13.1 (Krüger et al. 2014).
The characterization of the raw materials is an essential step in the re-synthesis
of new active materials with spent lithium-ion batteries as resources. ICP-OES
measurements were performed to analyze the elemental composition of the
mother-solutions for NCM synthesis (Krüger et al. 2014).
Table 13.1 presents the composition of transition metals and the impurities in the
mother solutions for material re-synthesis. Nickel, cobalt and aluminum remained
as the main contents in the solutions. The lithium content was lower than 5% in the
strongly contaminated sample. The main metal impurity was found to be aluminum.
The concentration of aluminum is higher for the cycled samples than for the
electrode production reject material samples. Furthermore, the concentration of iron
as well as magnesium and silicon are increasing with a higher aging degree. Since
the processing of the electrode production reject active material is similar to the
processing of the cycled active material in the recycling circuit, the origin of the
impurities has to be searched in the battery cell chemistry. The electrode production
reject material did not have any contact to electrolyte, and therefore was not washed
with DMC. For this reason, all influences and impurities caused by side reactions of
Table 13.1 Measured /g L−1 Reference Rejects Cycled

concentrations of metals and
impurities in the Li – 0.56 6.46
mother-solutions used as Ni 34.00 39.00 39.00
source for the re-synthesized Co 34.00 38.00 37.70
materials. The concentrations Mn 32.00 32.00 33.60
were measured by ICP-OES
Al <0.02 0.24 1.48
and listed in g L−1 (Krüger
et al. 2014) Cu <0.01 <0.01 0.10
Fe <0.01 <0.01 0.03
Mg – 0.01 0.03
Si <0.01 0.01 0.11
the electrolyte or its decomposition products with other cell components could not
take place. Especially, the increased value of Si is a result of electrolyte decom-
position, since it is well known that HF, as a decomposition product of the elec-
trolyte (Lux et al. 2012), excellently dissolves silicates (Dippel et al. 2013). The
silicate source can either be inactive cell components or parts of the machinery used
in the recycling circuit, e.g., the steel from the milling equipment contains Si. The
copper and aluminum contents arise from the anode and cathode current collector,
whereas iron is a result of the material processing and originates most probably in
the machinery (Krüger et al. 2014).
To match the optimum 4:4:2 stoichiometry, the missing metal sulfates were
added and the pH value was adjusted. From the gained solutions, a re-synthesis was
performed (Krüger et al. 2014).
Figure 13.2 presents the morphology of the particles after re-synthesis from the
various sources. The secondary particle shape, caused by the de-agglomeration and
jet milling step, of the reference material is clearly outlined (Fig. 13.2a). In com-
parison to the recycled materials, originating from degraded material (Fig. 13.2b)
and cycled material (Fig. 13.2c), in average the secondary particles of the reference
material are bigger. The difference of the material synthesized from the pure salts is
much bigger to the cycled material sample than to the production rejects material
sample. In fact, the secondary particle shape is barely formed when taking cycled
material as source for the active-material re-synthesis (Krüger et al. 2014).
The SEM findings are underlined by the BET surface area results. Table 13.2
presents the BET surface areas of the three measured samples. The measured BET
surface area of the cycled sample is more than twice of the sample from rejects.
Like for the particle morphology, the reference material and the rejects sample are
closer in their BET surface area values. This indicates that the differences in the
measured BET surface area values are explainable by the secondary particle shape
and are a result of the particle conditioning step (Krüger et al. 2014).
To detect the origin of the different behavior of the re-synthesized particles,
ICP-OES and IC investigations on the recycled active material have been carried
out (Krüger et al. 2014).
Fig. 13.2 SEM-images of re-synthesized NCM materials from a pure metal salts b rejects
material and c cycled material as source. The images were taken at an acceleration voltage of 3 kV
(Krüger et al. 2014)
Table 13.2 BET surface area of the reference, rejects and cycled materials. Measurements were
performed under nitrogen and the used P/P0 pressure range was 0–0.3 (Krüger et al. 2014)
BET Surface Area/m2g−1
Reference 0.23
Rejects 0.28
Cycled 0.60
Table 13.3 lists the measured elemental distribution in the re-synthesized active
materials in comparison to the reference material. Table 13.3 shows, which ele-
ments persists in the active material after the synthesis and their percentage of the
Table 13.3 Anion and cation Reference Rejects Cycled

impurities of the
resynthesized active materials Li 7.60 7.56 7.57
from cycled cells and Ni 20 20 18
electrode rejects compared to Co 20 20 18
the reference material. All Mn 18 19 16
values are listed in percent
Al 0.01 0.01 0.57
Fe <10−3 <10−3 1.12 10−2
CO2
3
<0.10 <0.10 0.51
NO3− <2 10−2 <2 10−2 <2 10−2
sample. Table 13.3 shows that the elemental composition of the reference sample
and the sample originating from reject’s material is almost identical. Only the
cycled material has an increased alumina, carbonate and iron content. The nickel
and manganese contents are lower but in similar ratios as for the other samples
(Krüger et al. 2014).
The impurities were also determined via XRD measurements, showing that they
are also present in crystalline matter (Krüger et al. 2014).
Table 13.3 shows the X-ray powder diffraction data of the re-synthesized
materials. The lattice reflection of the resynthesized samples matches to the lattice
reflections of the reference sample. All lattice reflections were compared to litera-
ture, which matches perfectly. The active material, which source is cycled NCM,
exhibits small crystalline impurities. The lattice diffractions are not present in the
reference and rejects samples (Krüger et al. 2014) (Fig. 13.3).
Taking all measurements into account, it can be concluded, that low amounts of,
at least partly crystalline impurities, are present, which inhibit the formation of
secondary particles for the cycled samples. The impurities contained a high amount
of aluminum, which is known to precipitate as hydroxide and was found to disturb
the secondary particle formation. It is proposed that aluminum hydroxide, which is
amphoteric and co-precipitating with the other metal hydroxides, is incorporated in
the precursor. These concentrations of aluminum impurities are disturbing the
calcination process since the adhesion of the different crystallites breaks apart and
smaller secondary particles are obtained (Krüger et al. 2014).
The gained NCM materials were processed to electrodes using the following
procedure: NCM, carbon black Super C65 and graphite SFG-6L as conductive
additives and PVdF binder (Kynar 761) in a ratio of 87/4/4/5 were elutriated in
NMP. The slurries were coated on aluminum foil and dried at 0.1 mbar at 120 °C.
The electrodes were pressed at 1500 kg m−2 (Krüger et al. 2014).
Electrochemical tests were carried out in Swagelok three-electrode set-up uti-
lizing a lithium counter and reference electrode. The separator was the
Freudenberg FS 2226 E and the electrolyte was 1 M LiPF6 in EC/DMC in a 1:1
(wt/wt) mixture (Krüger et al. 2014).
Fig. 13.3 X-ray powder diffraction patterns of the reference, rejects and cycled NCM materials in
different stage of aging (Krüger et al. 2014)
Cycling experiments were performed with a Maccor 4000 Series battery cycler.
The cells were cycled at a current rate of 1C in a voltage window of 3–4.2 V in
three-electrode Swagelok cells. Lithium was used as reference and counter elec-
trode respectively. The nominal capacity of NCM was 160 mAhg−1 (Krüger et al.
2014).
Cyclic voltammetry was carried out using a BaSyTec MDS battery cycler. The
electrodes were analyzed in 3-electrode Swagelok cells in a potential region from 3
to 4.6 V versus Li/Li+ with a sweep rate of 0.025 mV s−1. The reference and
counter electrode was elemental lithium (Krüger et al. 2014) (Fig. 13.4).
Figure 13.5 presents the cycling behavior of the synthesized materials for 500
cycles at 1C. The specific capacity of the reference material was found to be
130 mAhg−1 in the beginning cycles. In comparison, the material with rejects NCM
as resource for the recycling shows a slightly lower discharge capacity, especially
after 500 cycles. The sample with cycled NCM as active material exhibits a
capacity of 120 mAhg−1 in the beginning cycles. The fading rate was found to be
similar to the fading rate of the reference sample. The slightly decreased capacity
might be a result of decreased electrode kinetics as a result of the secondary particle
shape. This was already detected in Fig. 13.5, where lower and broader peaks are
observed for the rejects and cycled samples (Krüger et al. 2014).
The C-rate performance of the active materials was tested in 2-electrode
lithium-ion battery full cells using a pouch-cell housing. The electrolyte was LiPF6
in EC/DMC 1:1 (wt/wt) and the counter electrode a graphite anode. The cells were
balanced with 10% anode excess capacity in order to avoid lithium plating. The
corresponding electrochemical data is demonstrated in Fig. 13.5. The absolute
capacity of the rejects sample is close to the reference sample, whereas the cycled
Fig. 13.4 Schematic cutaway illustration of a Swagelok® type T-cell with a three-electrode setup
in a fully assembled state. The magnification shows the electrode stack placed between the current
collectors in the sequence from left to right: metallic lithium/separator/graphite coating/NCM
coating/aluminum foil representing a half-cell setup. The reference electrode equipped with
metallic lithium is placed perpendicular to the electrode stack, separated by a separator membrane
(Rothermel et al. 2016)
sample provides slightly decreased performance. This trend is remaining at

increased rates. Only at very high rates, reference, production reject’s and cycled
material provide the same capacity. Going back to lower rates, the capacity of the
cells is similar to the initial values (Krüger et al. 2014) (Fig. 13.6).
It has been proved that an experimental recycling circuit, consisting of several
industry-oriented steps, produces NCM active material with an electrochemical
performance close to material synthesized from pure solutions. Furthermore,
resulting impurities, caused by the material processing, were detected. Their
influence on the cycling performance was monitored. It is proposed, that especially
high aluminum contents disturb the material conditioning and therefore influences
the secondary particle shape. The cycling performance of recycled material from
spent Li-ion batteries at 1C is slightly decreased, caused by lower electrode kinetics
(Krüger et al. 2014).
Fig. 13.5 Cycling Performance of reference NCM material compared to material with electrode
rejects and cycled NCM as resource. Cycling was performed at 1C in a 1 M LiPF6 EC:DMC 1:1
(wt/wt) electrolyte with lithium as reference and counter electrode. The potential range was 3.0–
4.2 V versus Li/Li+ (Krüger et al. 2014)
Fig. 13.6 C-rate test of the recycled materials originating from rejects NCM and cycled NCM
electrodes compared to the reference material. The NCM/graphite full cells were investigated in an
LiPF6 EC:DMC 1:1 (wt/wt) electrolyte in a voltage region from 3.0 to 4.2 V (Krüger et al. 2014)
13.2 Recycling of the Anode Material
This chapter presents three different graphite recycling methods, which differ in
their purification procedure for graphite. The purification procedures include:
• Thermal evaporation of volatile electrolyte components (non-extracted sample)
• Electrolyte extraction with subcritical CO2 in combination with acetonitrile
(ACN)
• Electrolyte extraction with supercritical helium head pressure CO2.
In addition, the above mentioned purification procedures were supplemented by
consecutive washing steps consisting of washing with water, NMP and ethanol in
order to remove residues of the SEI, the potentially present PVdF binder and
non-volatile but soluble products deriving from the electrolyte decomposition. The
last step in graphite purification was a thermal treatment under an inert atmosphere
at 1000 °C. The objective of this step is the elimination of predominantly
oxygen-containing functional groups. An overview of the active surface nature of
graphite and the correlated thermal decomposition points of the specific chemical
groups is given in Fig. 13.7.
The need for surface treatment is necessary since the surface functionality of
graphite is directly associated with SEI formation and irreversible capacity loss
during electrochemical cycling. This effect is attributed to the increased formation
Fig. 13.7 Decomposition temperatures of specific surface oxygen groups. Adapted from Ref.
(Collins et al. 2015; Fuente et al. 2003; Figueiredo et al. 1999)
Fig. 13.8 Potential faradaic and electrostatic reactions of Li+ cation and PF6+ anion with
carboxilic, quinone and pyrone surface oxygen groups. Adapted from Ref. (Collins et al. 2015)
Fig. 13.9 Electrochemical cycling of the Panasonic CGR18650CH Li-ion MH12210 cell until
70% SOH. The charge/discharge cycling was performed with a constant charge current of 0.44 A
(1C) and a constant discharge current of 2.2 A (1C). The charging and discharging cut-off voltages
were set to 4.2 and 3.0 V respectively in accordance with the manufacturers specifications
of lithium rich compounds such as carboxylates or alkoxides on the electrode

surface during the initial charge as shown in Fig. 13.8. These lithium-containing
compounds act as anchoring points for the development of the SEI, thus increasing
the irreversible capacity (Collins et al. 2015). In fact, it was demonstrated that
reduced surface functionality of graphite corresponds to a decrease in irreversible
capacity (Xing and Dahn 1997; Collins et al. 2015; Lu and Chung 2001).
Since LIBs of various levels of age are expected to be recycled in practice, the
influence of the state-of-health (SOH) on the electrochemical performance of the
recycled graphite was investigated. The commercially available Panasonic
CGR18650CH Li-ion MH12210 cell was chosen to form the basis for practical
experiments. To acquire conclusive results, the effect of ageing was simulated by
continued electrochemical charge/discharge cycling until 70% SOH as exemplary
shown in Fig. 13.9.
As a reference, new, non-cycled cells (100% SOH) of the same type were used.
The non-cycled cells as well as the electrochemically aged cells were opened under
inert atmosphere. The electrodes were unwound and separated. Being coated on the
current collector foil, the anode active material of each cell was mechanically
detached and collected for the following investigations.
13.2.1 Analytical Characterization of Recycled Graphite
As discussed above, the thermal treatment is inevitable for various reasons. On the
one hand, the binder has to be decomposed in order to degrade the agglomerates.
On the other hand, the residual SEI components, mostly of organic nature, can be
removed by thermal decomposition. To determine the extent of these processes
thermogravimetric analysis (TGA) and differential scanning calorimetry
(DSC) were performed with two untreated graphite samples originating from cycled
(70% SOH) and non-cycled (100% SOH) Panasonic CGR18650CH Li-ion
MH12210 cells. TGA is a thermal method based on the principle to measure the
sample weight as a function of temperature or time. DSC determines the deviation
in heat flow between the sample and a reference (Stuart 2007). In Fig. 13.10, the
corresponding weight loss (in %) and heat flow curves (in W g−1 min−1) are
illustrated (Rothermel et al. 2016).
For both samples, substantial losses in weight of 90.0 ± 0.0% for the non-aged
cell and 87.3 ± 0.0% for the aged cell related to the initial weight are registered at
1000 °C. Almost identical changes in weight occur up to 160 °C, each associated
with consecutive events of heat consumption, heat release and heat consumption
again, whereby the corresponding amounts for the aged sample (70% SOH) are
more pronounced. The initial heat release can be attributed to the evaporation of
residual carbonates. The release of heat below 120 °C can be associated with
exothermal decomposition reactions within the SEI, such as polymerization,
Fig. 13.10 TGA and DSC analysis of the untreated negative electrode coating from a
CGR18650CH Li-ion MH12210 cell after electrochemical ageing (70% SOH) and as delivered
(100% SOH). The amount of samples: 27.143 ± 0.271 mg (70% SOH), 27.640 ± 0.276 mg
(100% SOH); inert atmosphere (flow): argon (100.0 mL min−1); sample pan: alumina (90 µL);
temperature ramp: 10.0 ± 0.1 °C min−1 (Rothermel et al. 2016)
oxidation or other exothermal processes. Beyond 160 °C, the loss in weight is
almost constant up to 450 °C, whereby this trend is more pronounced for the aged
sample. The possibly associated reactions can be correlated with those demon-
strated in Fig. 13.7. Accordingly, the carboxyl groups decomposing at temperatures
between 200 and 400 °C are assumingly decisive for the progression of the weight
curve. Since the SEI continuously grows along with the cycle life, the higher weight
loss of the aged sample appears reasonable. Beyond 450 °C other functional groups
such as anhydrides, phenols and lactones are decomposed, whereby an exothermal
reaction in both measurements occurs at 800 °C which cannot be attributed to an
individual event (Rothermel et al. 2016).
In order to find the best graphite recycling method, different purification
approaches were applied to the graphite particles. The impact of these methods is
investigated by scanning electron microscopy (SEM). As a reference, an electrode
taken from a non-cycled 18650 cell was visualized after drying under vacuum
conditions at an elevated temperature of 50 °C in order to avoid significant mor-
phology changes caused by decomposition reactions of the SEI and the conductive
salt. The corresponding SEM image is shown in Fig. 13.11 (Rothermel et al. 2016).
The electrode shows a uniform distribution of graphite particles, which vary in
size from 10 to 30 µm.
Applying a thermal evaporation of volatile electrolyte components and subse-
quent thermal treatment at 1000 °C it can be seen that the surface condition of the
graphite particles undergoes morphological changes. The particles form agglom-
erates and the surface covered with crystal-shaped structures as can be seen in
Fig. 13.12 for the 100% SOH specimen. The observed compounds appear in bright
tones, meaning that they have less electronic conductivity. This suggests, that the
origin of these compounds may be either the SEI or the conductive salt LiPF6, since
Fig. 13.11 SEM image of the negative electrode originating from the CGR18650CH Li-ion
MH12210 cell (100% SOH) (Rothermel et al. 2016)
Fig. 13.12 SEM image of the thermally treated graphite originating from the CGR18650CH Li-
ion MH12210 cell (100% SOH). White colored (less electronically conductive) particles
distributed over the particle surface can be observed (Rothermel et al. 2016)
the electrolyte solvents are volatile and mostly evaporated in vacuum (Rothermel
et al. 2016).
To provide an indication of the elemental composition of the observed struc-
tures, energy dispersive X-ray analysis (EDX) was applied. The results of this
measurement are demonstrated in Fig. 13.13. The image section (a) depicts the
analyzed area. In part (b) and (c) the corresponding distribution and relative
intensity of the elements oxygen and phosphorus over the scanned area are shown.
These elements are essential constituents of the related structures (Rothermel et al.
2016).
With regard to LiPF6 degradation mechanisms reported in literature the forma-
tion of organophosphates and phosphates appears reasonable (Kraft et al. 2015).
Being soluble in water, these compounds should have been leached out during the
washing procedure before treatment, but the amount of used water (100 mL/
sample) was probably chosen too low.
In Fig. 13.14, the corresponding SEM image of the graphite sample (100%
SOH) processed with a subcritical carbon dioxide assisted electrolyte extraction
with co-addition of acetonitrile and subsequent thermal treatment is demonstrated.
The surface of a representative particle appears to be free from “non-carbon”
compounds. This implies, that the previous electrolyte extraction prevented the
formation of phosphorous and oxygen containing compounds (Rothermel et al.
2016).
Related to the recovery rate, the mentioned extraction method was reported to be
the most efficient in literature, being able to recover all electrolyte components up to
90wt% (Grutzke et al. 2015).
Fig. 13.13 Energy dispersive X-ray analysis (EDX) image of the thermally treated graphite
originating from the CGR18650CH Li-ion MH12210 cell (100% SOH) shown in section a for the
elements oxygen b and phosphorus c of the scanned are shown as well (Rothermel et al. 2016)
Fig. 13.14 SEM image of graphite originating from the CGR18650CH Li-ion MH12210 cell
(100% SOH) after subcritical carbon dioxide assisted electrolyte extraction with acetonitrile
addition as a co-solvent and subsequent thermal treatment. The particle surface is free from
crystallites observed in Fig. 13.12 (Rothermel et al. 2016)
In the last recycling method, supercritical carbon dioxide (scCO2) was used as
extractant followed by a leaching procedure and subsequent thermal treatment
(Fig. 13.15).
Here again, the surface of the representative graphite particle is found to be free
from extraneous matter which is not in agreement with the expectations, since
scCO2 is rather selective to organic carbonates and does not extract the conductive
salt (Grützke et al. 2014). Nevertheless, the washing step performed after the
electrolyte extraction seems to be sufficient to wash out the residual LiPF6
(Rothermel et al. 2016).
As the SEM/EDX analyses reveal, the particle surface of the non-extracted
graphite is covered with oxygen- and phosphorus-containing impurities. For the
Fig. 13.15 SEM image of the thermally treated graphite originating from the CGR18650CH
Li-ion MH12210 cell (100% SOH) after supercritical carbon dioxide assisted electrolyte extraction
and subsequent thermal treatment. The particle surface is free from crystallites observed in
Fig. 13.12 (Rothermel et al. 2016)
Table 13.4 Summary of the elemental ICP-MS analyses. All data are given in ppb with respect to
the previously digested amounts of the graphite samples. Limit of detection (LOD) and limit of
quantification (LOQ) are shown. The LODs were calculated according to the three-sigma criterion.
For the calculation of LOQ the nine-sigma criterion was applied (Rothermel et al. 2016)
Concentration/(ppb)
7 31 55 58 59
Sample (SOH) Li P Mn Ni Co
non-ex. (100%) 676.68 ± 6.7 1 8.41 ± 0.3 <LOQ 1.81 ± 0.1
128.24 ± 25.5
non-ex. (70%) 450.83 ± 5.4 786.37 ± 4.9 39.08 ± 0.8 29.93 ± 0.3 35.10 ± 0.5
subcCO2 <LOQ 5.59 ± 0.4 0.47 ± 0.5 2.66 ± 0.0 <LOQ
(100%)
subcCO2 (70%) <LOQ 3.99 ± 0.4 0.04 ± 0.0 <LOQ <LOQ
scCO2 (100%) 0.48 ± 0.0 5.64 ± 0.5 <LOQ 1.70 ± 0.3 <LOQ
scCO2 (70%) 0.55 ± 0.0 3.47 ± 0.1 1.19 ± 0.1 4.33 ± 0.1 <LOQ
7 31 55 58 59
Li P Mn Ni Co
LOD [µg] 0.068 0.326 0.042 0.057 0.002
LOQ [µg] 0.226 1.086 0.141 0.190 0.005
quantification of lithium and phosphorous as the most likely constituent elements,

inductively coupled plasma mass spectrometry (ICP-MS) analysis was performed.
Being a powerful method for elemental analysis, this technique was simultaneously
used to verify whether the transition metals Mn, Ni and Co deriving from the
cathode are present in the samples. The results are summarized in Table 13.4.
In general, the assumptions made based on the SEM/EDX measurements are

confirmed by the ICP-MS analysis. The amount of phosphorus in the sample is
quantified with 1128.24 ± 25.50 ppb for the non-aged sample (non-ex. 100%
SOH) and 786.37 ± 4.94 ppb for the aged counterpart (non-ex. 70% SOH).
Lithium is also found in significant quantities of 676.68 ± 6.72 ppb (non-ex. 100%
SOH) and 450.83 ± 5.37 ppb (non-ex. 70% SOH). Substantially smaller quantities
of phosphorous varying from 3.47 ± 0.09 ppb to 5.64 ± 0.49 ppb were found for
the other samples which have undergone electrolyte extraction with subcritical and
supercritical CO2. Li is not quantified in samples extracted by subcritical CO2. Only
small amounts of lithium are found in samples extracted by scCO2:
0.48 ± 0.04 ppb (100% SOH) and 0.55 ± 0.01 ppb (100% SOH). With respect to
the contamination by the cathode active material, significant amounts of manganese
(39.08 ± 0.76 ppb), nickel (29.93 ± 0.28 ppb) and cobalt (35.10 ± 0.48 ppb)
which roughly correspond to the stoichiometric ratio of the cathode material LiNi1/
3Mn1/3Co1/3O2 were found in the non-extracted aged sample (non-ex. 70% SOH).
The only explanation for this finding is that during mechanical separation of the
corresponding electrodes and the following detachment of the anode material, this
material specimen has come in contact with the cathode material and has been
contaminated. The subsequent electrochemical investigations will display in which
way the contamination affects the electrochemical behavior of the recycled anode
material (Rothermel et al. 2016).
As it determines its electrochemical performance, the crystallinity of graphite is
considered to be of crucial importance when used as the anode material (Kohs et al.
2003). This parameter can potentially be influenced by high pressurized carbon
dioxide, since besides its use as an extracting agent, application of supercritical CO2
is also a reported technique for exfoliation of layered graphite (Pu et al. 2009). In
order to analyze the effect of the three different recycling strategies on the crys-
tallinity degree of graphite, Raman investigations of the six graphite samples were
performed (Rothermel et al. 2016).
Raman spectroscopy is a powerful technique suitable to observe vibrational
modes in graphite (Jorio et al. 2011). In general, two main features can be identified
in the Raman spectrum: the G peak at around 1580 cm−1 and the D peak at
1350 cm−1 as can be seen in Fig. 13.16 for the non-aged, thermally treated graphite
sample without electrolyte extraction.
Since the D peak is observed in the presence of defects in an otherwise perfect
and infinite graphite structure, the ratio of D and G band is reported to be inversely
proportional to the crystallite size La. A general equation for the determination of
the in-plane crystallite size La by Raman spectroscopy is suggested by Cançado
et al., derived from systematic XRD and scanning tunneling microscopy
(STM) investigations. The equation is given as:
1
10
ID
La ðnmÞ ¼ 2:4 10 k4l
IG
Fig. 13.16 Raman spectrum of the thermally treated non-extracted graphite originating from the
CGR18650CH Li-ion MH12210 cell (100% SOH). Integrated peaks D and G are shaded in grey.
Laser wavelength: 532 nm; laser power: 15 mW; diffraction grating: 330 gr mm−1; total
acquisition time: 25 s (5 image frames, 5 s per frame) (Rothermel et al. 2016)
where kl is the laser wavelength and ID/IG the integrated intensity ratio of D and G
bands (Cançado et al. 2006). In case of the thermally treated graphite sample (100%
SOH), the ID/IG ratio is determined with 0.0177 (Fig. 13.16). Considering the used
laser wavelength of 532 nm, the La is calculated to be 1086.3 nm. For reasons of
comparability, equal calculations were made for the remaining samples according to
the Raman spectra demonstrated in Fig. 13.17 (Rothermel et al. 2016).
According to the equation introduced before, the integrated peak areas were
determined from Raman spectra demonstrated in Fig. 13.17. Their ratios and the
in-plane crystallinity sizes La were calculated using kl = 532 nm. The corre-
sponding values are summarized in Table 13.5 (Rothermel et al. 2016).
By comparing the resulting La values, a trend to smaller crystallite sizes with
increasing environmental pressure conditions is shown. For instance, the crystallite
size for the thermally treated graphite (100% SOH) decreases from 1086 to 635 nm
when subcritical carbon dioxide (60 bar) is used as extractant and halved again to
279 nm when supercritical carbon dioxide (120 bar) is applied. The aged graphite
samples follow the same trend, but with a generally reduced crystallite size varying
from 800 to 175 nm compared to their non-aged counterparts. Taking these results
into consideration, the unfavorable effect of high pressurized carbon dioxide on the
crystallinity of graphite is confirmed. It can further be concluded that the crys-
tallinity degree is also affected by the SOH condition of the original cell (Rothermel
et al. 2016).
For the evaluation of the electrochemical performance of recycled graphite, half
cell investigations of the six treated graphite samples were performed and, addi-
tionally, commercially available TIMREX® SLP50 graphite was used as the
benchmark material. For all measurements the commercial electrolyte SelectiLyte™
LP47 (1 M Lithium hexafluorophosphate in EC:DEC 3:7) was used. The charging
and discharging cut-off potentials were set to 0.02 V and 1.5 V versus Li/Li+,
respectively (Rothermel et al. 2016).
Fig. 13.17 Raman spectra of thermally treated graphite without electrolyte extraction (non-ex.),
graphite after subcritical carbon dioxide assisted electrolyte extraction with acetonitrile addition as
a co-solvent and subsequent thermal treatment (subcCO2) and graphite after supercritical carbon
dioxide assisted electrolyte extraction and subsequent thermal treatment originating from the
CGR18650CH Li-ion MH12210 cell (70% SOH/100% SOH). Laser wavelength: 532 nm; Laser
power: 15 mW; Diffraction grating: 330 gr mm−1; Total acquisition time: 25 s (5 image frames,
5 s per frame) (Rothermel et al. 2016)
Table 13.5 Integrated areas of D and G peaks from Raman spectra shown in Fig. 13.17, their
ratio and the calculated in-plane crystallite size La of each investigated graphite sample (Rothermel
et al. 2016)
D G ID La
IG ratio
Sample Peak FWHM Peak FWHM (nm)
area area
non-ex. 100% SOH 40.36 42.10 2280.57 17.49 0.0177 1086
non-ex. 70% SOH 57.95 48.83 2409.91 18.42 0.0240 800
subcCO2 100% 70.43 41.77 2324.59 17.63 0.0303 635
SOH
subcCO2 70% SOH 79.58 42.12 2345.00 17.84 0.0339 567
scCO2 100% SOH 189.87 47.81 2754.28 21.15 0.0689 279
scCO2 70% SOH 293.28 44.38 2669.57 19.92 0.1099 175
Fig. 13.18 Constant current discharge capacities and Coulombic efficiencies curves of thermally
treated non-extracted graphite originating from the CGR18650CH Li-ion MH12210 cell (70%
SOH/100% SOH) used as WE in a half cell setup. Measurements were performed using a
three-electrode setup with metallic lithium as RE and CE. Electrolyte: SelectiLyte™ LP47 (1 M
Lithium hexafluorophosphate in EC:DEC 3:7); WE: graphite/Super C65/Na-CMC (90/5/5);
discharging cut-off potential: 1.5 V versus Li/Li+; charging cut-off potential: 0.02 V versus Li/Li+;
specific current: 186 mA g−1, 0.5C (three initial formation cycles 37 mA g−1, 0.1C); temperature:
20 °C (Rothermel et al. 2016)
In Fig. 13.18, the constant current cycling behavior of the thermally treated
graphite samples without previous electrolyte extraction is demonstrated. The
discharge capacity and Coulombic efficiencies of the aged (70% SOH) and
non-aged (70% SOH) graphite sample are shown (Rothermel et al. 2016).
At this point it is useful to specify the term specific current or C–rate more
closely, which is often used hereafter. This measure is commonly used to represent
the charge and discharge rate equal to the nominal capacity of one cell or battery in
one hour. According to this, for a battery having 3 Ah discharge capacity, the
C-rate of 1 is equal to 3 A. Correspondingly, 0.5C is equivalent to 1.5 A and 2C to
6 A (Young et al. 2013; Rothermel et al. 2016).
It can be seen that stable cycling of both samples is possible over at least 100
cycles. When comparing the discharge cycling curves with each other, the cycling
performance of the aged graphite sample significantly outperforms not only the
non-aged counterpart but also the benchmark (blue marked data in Fig. 13.18). In
fact, the aged sample reaches an average specific discharge capacity of
374 mAh g−1, which slightly exceeds the theoretical value of graphite given with
372 mAh g−1. It is important to note that the specific discharge capacity is related
to the ratio of graphite used during the electrode preparation, which is 90%. The
conductive carbon C-NERGY™ SUPER C65 co-added in the amount of 5% is not
taken into consideration, but also provides a certain discharge capacity. The average
specific discharge capacity of the non-aged graphite sample is determined with
349 mAh g−1, which is below the benchmark level of 358 mAh g−1 (Rothermel
et al. 2016).
With respect to the first cycle Coulombic efficiency, the trend is reversed: the
aged sample reaches 78.7%, the non-aged sample 82.0% and for the benchmark a
value of 84.8% is found. Whereas the Coulombic efficiencies of the non-aged
sample and the benchmark become aligned in the following two cycles (both
around 97.2% in the 2nd cycle), the aged sample remains at 96.2%. After 50 cycles
all the samples operate at a comparatively equal level of 99.9% (Rothermel et al.
2016).
Correlating the electrochemical performance to the analytical characterization,
the lower Coulombic efficiency of the aged graphite sample during the initial cycles
can be attributed to the contamination of graphite with the cathode material in the
beginning of the recycling procedure. Being admixed with graphite, the cathode
material enters overdischarge conditions while cycling as negative electrode.
A comparative study investigating overdischarge behaviors of LiNiO2 and LiMnO2
cathode materials was reported by Shu et al. (2011) They concluded that these
transition metal compounds undergo an irreversible and electrochemically driven
structure change showing large hysteresis and low Coulombic efficiencies in the
first two cycles, which coincide with the tendency observed (Shu et al. 2011).
Presumably, the reported transformations are completed after few cycles with the
result that the impurities do not longer participate in electrochemical reactions and
become inactive (Rothermel et al. 2016).
Despite contamination with the cathode material, considerable amounts of
lithium and phosphorous were found in the graphite particles, whose influences are
difficult to assess. As discussed before, the SEI is associated with irreversible
consumption of lithium to form its organic or inorganic components. It is possible
that the present Li and P impurities are involved in the SEI formation processes,
thus additionally affecting the cycling efficiency (Rothermel et al. 2016).
In general, despite the contamination with transition metals, the discharge
capacity of the aged graphite electrode is significantly surpassing the benchmark.
There is no obvious explanation for this phenomenon (Rothermel et al. 2016).
Subcritical carbon dioxide extracted graphite
Comparable to the electrochemical characterization of the non-extracted thermally
treated graphite, the cycling behavior of two further graphite samples, which have
undergone a subcritical carbon dioxide assisted electrolyte extraction with ace-
tonitrile as co-solvent before thermal treatment was investigated. The corresponding
data is demonstrated in Fig. 13.19 (Rothermel et al. 2016).
The cycling behavior of both graphite samples is stable. Here again, the aged
graphite sample (subcCO2 70% SOH) outperforms the non-aged sample (subcCO2
70% SOH) in view of the discharge capacity reaching 380 mAh g−1 versus
372 mAh g−1 on average. Both samples significantly surpass the benchmark of
349 mAh g−1. Also here, the contribution of the conductive carbon is neglected
thus explaining practical discharge capacities exceeding the theoretical value
(Rothermel et al. 2016).
In view of the cycling efficiency in the first cycle, values of 82.9% and 81.6% are
found for subcCO2 extracted graphite with 70% SOH and 100% SOH, respectively.
Fig. 13.19 Constant current discharge capacities and Coulombic efficiencies of the graphite
originating from the CGR18650CH Li-ion MH12210 cell (70% SOH/100% SOH) after subcritical
carbon dioxide assisted electrolyte extraction with acetonitrile addition as a co-solvent and
subsequent thermal treatment used as WE in a half cell setup. Measurements were performed using
a three-electrode setup with metallic lithium as RE and CE. Electrolyte: SelectiLyte™ LP47 (1 M
Lithium hexafluorophosphate in EC:DEC 3:7); WE: graphite/Super C65/Na-CMC (90/5/5);
discharging cut-off potential: 1.5 V versus Li/Li+; charging cut-off potential: 0.02 V versus Li/Li+;
specific current: 186 mA g−1, 0.5C (three initial formation cycles 37 mA g−1, 0.1C); temperature:
20 °C (Rothermel et al. 2016)
In contrast to the thermal treated samples discussed before, the aged sample
demonstrates a better first cycle efficiency. This is in agreement with the argu-
mentation before where the influence of contaminants shows an adverse impact on
the Coulombic efficiency. No significant contaminations were found in the present
samples (Rothermel et al. 2016).
In general, the aged graphite sample provides the best electrochemical perfor-
mance following the trend that has been observed before. The overall better per-
formance compared to the non-aged sample is an unexpected result too. Compared
to the thermally treated non-extracted samples, the determined crystallinity size La
is remarkably reduced, which should lead to a decreased specific discharge capacity
due to reduced extension of the graphite galleries. This effect does not occur. Thus,
despite the earlier considerations, the electrolyte extraction with high pressurized
carbon dioxide offers an attractive recycling method (Rothermel et al. 2016).
The last investigated recycling method utilizes supercritical carbon dioxide
electrolyte extraction prior to thermal treatment. For this reason, again, the treated
graphite samples were subjected to electrochemical investigations. The corre-
sponding data is demonstrated in Fig. 13.20 (Rothermel et al. 2016).
The overall cycling performance remains significantly below the samples
examined so far. But the trend towards a better electrochemical performance of the
aged sample compared to the non-aged counterpart is maintained. The aged sample
(scCO2 70% SOH) reaches an average specific discharge capacity of 345 mAh g−1.
The cycling performance of the non-aged sample is less stable, showing a discharge
Fig. 13.20 Constant current discharge capacities and Coulombic efficiencies of graphite
originating from the CGR18650CH Li-ion MH 12210 cell (70% SOH/100% SOH) after
supercritical carbon dioxide assisted electrolyte extraction and subsequent thermal treatment used
as WE in a half cell setup. Measurements were performed using a three-electrode setup with
metallic lithium as RE and CE. Electrolyte: SelectiLyte™ LP47 (1 M Lithium hexafluorophosphate
in EC:DEC 3:7); WE: graphite/Super C65/Na-CMC (90/5/5); discharging cut-off potential: 1.5 V
versus Li/Li+; charging cut-off potential: 0.02 V versus Li/Li+; specific current: 186 mA g−1, 0.5C
(three initial formation cycles 37 mA g−1, 0.1C); temperature: 20 °C (Rothermel et al. 2016)
capacity of about 333 mAh g−1 (40th–70th cycle). Both samples remain signifi-
cantly below the benchmark level of 358 mAh g−1 (Rothermel et al. 2016).
The first cycle Coulombic efficiency observed for the non-aged sample is at a
noticeable low level of 56.1%, whereby the aged sample surpasses the benchmark
with 85.4% (vs. 84.8%) and remains at the highest level during the following
cycles. However, the non-aged sample remains significantly below the benchmark
level up to the 30th cycle. Thereafter, all the Coulombic efficiencies remain close to
99.8% (Rothermel et al. 2016).
Considering the results, it can be concluded that supercritical carbon dioxide has
such a destructive impact on graphite crystallinity that the discharge capacity is
adversely affected. Compared with samples subjected to the subcritical carbon
dioxide extraction, the present samples demonstrate a considerably reduced cycling
performance. The discussed values are summarized in Table 13.6 (Rothermel et al.
2016).
The final goal of the LithoRec II project is to complete the raw material cycle. In
an ideal case, the recycled materials enter the remanufacturing process of LIBs.
Whilst there is no doubt that inactive materials such as copper, aluminum or iron
containing compounds can be reused, the electrochemical performance of the active
materials such as electrolyte, cathode and anode materials and their interplay has to
be ensured (Rothermel et al. 2016).
As explained in chapter before, the recycling strategy of the cathode material has
already been successfully investigated and is therefore not a focus of this work.
However, the recycled cathode material synthesized by the company H.C. Starck
GmbH is used for the full cell investigations using recycled graphite as anode and
Table 13.6 Overview of the discharge capacities and associated Coulombic efficiencies obtained
from constant current charge/discharge cycling experiments according to data demonstrated in
Figs. 13.18, 13.19 and 13.20 (Rothermel et al. 2016)
Sample Disch. Cap. Coulombic efficiency (cycle)/%
50th cycle/ 1st 2nd 3rd 50th
mAh g−1
non-ex. 100% 348.8 ± 1.9 78.7 ± 1.2 96.2 ± 0.4 97.4 ± 0.4 99.9 ± 0.1
SOH
non-ex. 70% 375.0 ± 1.0 82.0 ± 1.4 97.2 ± 0.3 98.2 ± 0.2 99.9 ± 0.1
SOH
subcCO2 372.7 ± 2.5 81.6 ± 3.1 96.3 ± 1.2 97.6 ± 0.8 99.9 ± 0.1
100% SOH
subcCO2 70% 379.9 ± 4.4 82.9 ± 0.9 97.6 ± 0.1 98.5 ± 0.1 99.9 ± 0.1
SOH
scCO2 100% 332.7 ± 0.3 56.1 ± 1.8 91.8 ± 0.4 94.7 ± 0.3 99.8 ± 0.1
SOH
scCO2 70% 346.8 ± 7.8 85.4 ± 0.5 97.8 ± 0.3 98.5 ± 0.3 99.9 ± 0.1
SOH
Benchmark 357.6 ± 1.4 84.8 ± 0.8 97.3 ± 0.2 98.2 ± 0.2 99.9 ± 0.1
Fig. 13.21 Constant current and constant potential discharge capacities and Coulombic
efficiencies of the recycled cathode material used as WE. Measurements were performed using
a three-electrode setup with metallic lithium as the CE and RE. Recycled LiNi1/3Co1/3Mn1/3O2
(NCM) is used as the WE. Electrolyte: SelectiLyte™ LP47 (1 M Lithium hexafluorophosphate in
EC:DEC 3:7); WE: NCM/PVdF/Super C65/SFG-6L (87/5/4/4); discharging cut-off potential:
3.0 V versus Li/Li+; charging cut-off potential: 4.3 V versus Li/Li+; specific current: 150 mA g−1,
1 C (three initial formation cycles 30 mA g−1, 0.2C); constant potential charging step: 1 h;
temperature: 20 °C (Krüger et al. 2014; Rothermel et al. 2016)
recycled LiNi1/3Co1/3Mn1/3O2 (NCM) as cathode. For a better assessment of the full

cell performance, half cell investigations of the recycled cathode material were per-
formed. The corresponding data is demonstrated in Fig. 13.21. The related recycling
concept is also reported in literature (Krüger et al. 2014; Rothermel et al. 2016).
In a half cell setup, NCM undergoes lithiation during discharge while a

de-lithiation process takes place during charging. The electrochemical character-
istics were performed in a three-electrode setup with metallic lithium as the CE and
RE. The charging cut-off potential is set to 4.3 V versus Li/Li+ followed by a
constant potential step for 1 h in order to minimize kinetic effects reported to
influence the de-lithiation process (Kasnatscheew et al. 2016). In this way, a better
comparability of the recycled NCM with a commercially available benchmark
material NCM of the same stoichiometry (111) is given (Rothermel et al. 2016).
From Fig. 13.21 it can be seen that recycled NCM strongly outperforms the
benchmark NCM with regard to the average discharge capacity (e.g.,
160.3 ± 0.2 mAh g−1 vs. 147.6 ± 0.5 mAh g−1 in the 20th cycle) while the
progression of the benchmark data is more stable. Comparing the Coulombic
efficiencies, it becomes apparent that the recycled NCM is significantly less efficient
reaching 99.1 ± 0.0% as the highest value in the 50th cycle. This drawback is
attributed to the contamination of the cathode material with 0.57wt% aluminum
derived from the insufficient aluminum separation during precipitation of the
transition metals, which is rather a technical issue (Krüger et al. 2014).
Nevertheless, the recycled NCM is considered to be suitable for the following full
cell investigations (Rothermel et al. 2016).
Based on previous results, the electrochemical performance of a recycled Li-ion
full cell is investigated in this chapter. Two graphite samples (aged, non-aged) with
the best performance were selected as the constituent part of the anode. Those,
subjected to a subcritical carbon dioxide assisted electrolyte extraction with ace-
tonitrile addition and subsequent thermal treatment were identified as suitable
candidates as can be concluded from Table 13.6. The recycled LiNi1/3Mn1/3Co1/
3O2, which was discussed in the preceding chapter, formed the cathode. The cor-
responding cycling data is presented in Fig. 13.22 (Rothermel et al. 2016).
In view of the trend to better cycling performance of aged recycled graphite than
non-aged samples, a similar situation is reflected in a full cell. Although discharge
capacities of ca. 140 mAh g−1 can be reached in both cases during the initial cycles,
the performance at a higher rate of 150 mA g−1 reveal a strong capacity retention
when non-aged graphite (100% SOH) is used as anode. Thus, the related discharge
capacity in this case reaches 64.9 ± 0.8 mAh g−1 after 50 cycles. In contrast, a
value of 108.4 ± 0.6 mAh g−1 (50th cycle) is reached when aged graphite (70%
SOH) is used as anode (Rothermel et al. 2016).
In general, it can be concluded that the combined operation of recycled active
materials can be successfully performed. As already stated before, recycled graphite
originating from a previously electrochemically aged cell outperforms the non-aged
counterpart (Rothermel et al. 2016).
In summary, it can be stated that graphite originating from a previously elec-
trochemically aged commercial cell subjected to a subcritical carbon dioxide
assisted electrolyte extraction in combination with a thermal treatment demonstrates
the best electrochemical characteristics. It even outperforms the commercial syn-
thetic graphite TIMREX® SLP50 used as a benchmark. Of particular importance is
the fact that in all cases the previously aged graphite demonstrated a better
Fig. 13.22 Constant current discharge capacities and Coulombic efficiencies of a full cell.
Measurements are performed using a three-electrode setup with metallic lithium as the RE.
Recycled graphite originating from the CGR18650CH Li-ion MH12210 cell (70% SOH/100%
SOH) after subcritical carbon dioxide assisted electrolyte extraction with acetonitrile addition as
co-solvent and subsequent thermal treatment is used as the CE. Recycled LiNi1/3Co1/3Mn1/3O2 is
used as the WE. Electrolyte: SelectiLyte™ LP47 (1 M LiPF6 in EC:DEC 3:7); counter electrode:
graphite/Super C65/Na-CMC (90/5/5); WE: NCM/PVdF/Super C65/SFG-6L (87/5/4/4); discharg-
ing cut-off voltage: 4.25 V; charging cut-off voltage: 1.7 V; specific current: 150 mA g−1, 1C
(three initial formation cycles 30 mA g−1, 0.2C); temperature: 20 °C (Rothermel et al. 2016)
performance than its non-aged counterpart. Considering that the primary returning
LIBs are in an electrochemically aged condition, the benefit of graphite recycling is
significant (Rothermel et al. 2016).
Furthermore, SEM/EDX measurements reveal inorganic residues derived from
the conductive salt decomposition during the thermal treatment as small
crystal-shaped particles distributed on the surface of graphite particles when not
performing any electrolyte extraction prior to thermal treatment. The presence of
the elements lithium and phosphorous on the particle surface is slightly reflected by
the first cycle Coulombic efficiency when compared to samples subjected to sub-
critical carbon dioxide assisted electrolyte extraction, where virtually no elemental
residues are identified neither by SEM/EDX nor ICP-MS measurements (Rothermel
et al. 2016).
It addition, it was found that the application of high pressurized carbon dioxide
is unfavorable for the crystallinity of the graphite particles and therefore has an
adverse effect on the electrochemical performance. The tendency to lower crys-
tallinity with increasing carbon dioxide pressure was confirmed by Raman spec-
troscopy. Despite resulting in smaller crystallite sizes than without electrolyte
extraction, the application of subcritical carbon dioxide is considered to be the best
recycling method, as the recycled graphite shows the best electrochemical perfor-
mance and the electrolyte is recovered by 90% including the conductive salt. Thus,
this method offers the possibility to reutilize nearly the entire components from a
spent LIB, which is of great benefit with regard to the recycling efficiency and a
suitable method to meet the political and legislative targets (Rothermel et al. 2016).
References
Cançado LG, Takai K, Enoki T, Endo M, Kim YA, Mizusaki H, Jorio A, Coelho LN,
Magalhães-Paniago R, Pimenta MA (2006) General equation for the determination of the
crystallite size La of Nanographite by Raman Spectroscopy. Applied Physics Letters 88
(16):163106. doi:http://dx.doi.org/10.1063/1.2196057
Castillo S, Ansart F, Laberty-Robert C, Portal J (2002) Advances in the recovering of spent lithium
battery compounds. J Power Sources 112(1):247–254
Collins J, Gourdin G, Foster M, Qu D (2015) Carbon surface functionalities and SEI formation
during Li Intercalation. Carbon 92(1):193–244. doi:http://dx.doi.org/10.1016/j.carbon.2015.
04.007
Dippel C, Krueger S, Kraft V, Nowak S, Winter M, Li J (2013) Aging stability of Li2FeSiO4
polymorphs in LiPF6 containing organic electrolyte for lithium-ion batteries. Electrochimica
Acta 105:542–546. doi:http://dx.doi.org/10.1016/j.electacta.2013.05.013
Dorella G, Mansur MB (2007) A study of the separation of cobalt from spent Li-ion battery
residues. J Power Sources 170(1):210–215
Ferreira DA, Prados LMZ, Majuste D, Mansur MB (2009) Hydrometallurgical separation of
aluminium, cobalt, copper and lithium from spent Li-ion batteries. J Power Sources 187(1):
238–246. doi:http://dx.doi.org/10.1016/j.jpowsour.2008.10.077
Figueiredo JL, Pereira MFR, Freitas MMA, Órfão JJM (1999) Modification of the surface
chemistry of activated carbons. Carbon 37(9):1379–1389. doi:http://dx.doi.org/10.1016/
S0008-6223(98)00333-9
Fuente E, Menéndez JA, Suárez D, Montes-Morán MA (2003) Basic surface oxides on carbon
materials: a global view. Langmuir 19(8):3505–3511. https://doi.org/10.1021/la026778a
Georgi-Maschler T, Friedrich B, Weyhe R, Heegn H, Rutz M (2012) Development of a recycling
process for Li-ion batteries. J Power Sources 207:173–182. doi:http://dx.doi.org/10.1016/j.
jpowsour.2012.01.152
Grützke M, Kraft V, Weber W, Wendt C, Friesen A, Klamor S, Winter M, Nowak S (2014)
Supercritical carbon dioxide extraction of Lithium-Ion battery electrolytes. J Supercrit Fluids
94(1):216–222. doi:http://dx.doi.org/10.1016/j.supflu.2014.07.014
Grutzke M, Monnighoff X, Horsthemke F, Kraft V, Winter M, Nowak S (2015) Extraction of
Lithium-Ion battery electrolytes with liquid and supercritical carbon dioxide and additional
solvents. RSC Adv 5(54):43209–43217. https://doi.org/10.1039/c5ra04451k
Hanisch C, Haselrieder W, Kwade A (2011) Recovery of active materials from spent Lithium-Ion
electrodes and electrode production rejects. In: Hesselbach J, Herrmann C (eds) Globalized
solutions for sustainability in manufacturing. Springer Berlin Heidelberg, pp 85–89. doi:10.
1007/978-3-642-19692-8_15
Jorio A, Saito R, Dresselhaus G, Dresselhaus MS (2011) Raman Spectroscopy: From Graphite to
sp2 Nanocarbons. In: Jorio A, Saito R, Dresselhaus G, Dresselhaus MS (eds) Raman
Spectroscopy in Graphene Related Systems. Wiley-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, pp 73–101. doi:10.1002/9783527632695.ch4
Kasnatscheew J, Evertz M, Streipert B, Wagner R, Klopsch R, Vortmann B, Hahn H, Nowak S,
Amereller M, Gentschev AC, Lamp P, Winter M (2016) The truth about the 1st cycle
coulombic efficiency of LiNi1/3Co1/3Mn1/3O2 (NCM) Cathodes. Phys Chem Chem Phys 18(5):
3956–3965. https://doi.org/10.1039/c5cp07718d
Kohs W, Hofer F, Schrötner H, Doninger J, Barsukov I, Albering JH, Besenhard JO, Winter M
(2003) Graphite’s crystallinity influences on anodes electrochemical properties in Lithium-Ion
Cells. In: Julien CM, Prakash J (eds) New trends in intercalation compounds for energy storage
and conversion: proceedings of the international symposium, vol 2003–20. Electrochemical
Society, Pennington
Kraft V, Grützke M, Weber W, Menzel J, Wiemers-Meyer S, Winter M, Nowak S (2015)

Two-dimensional ion chromatography for the separation of ionic organophosphates generated in
thermally decomposed Lithium Hexafluorophosphate-based Lithium-Ion battery electrolytes.
J Chromatogr A 1409(1):201–209. doi:http://dx.doi.org/10.1016/j.chroma.2015.07.054
Krüger S, Hanisch C, Kwade A, Winter M, Nowak S (2014) Effect of impurities caused by a
recycling process on the electrochemical performance of Li[Ni0.33Co0.33Mn0.33]O2.
J Electroanal Chem 726(1):91–96. doi:http://dx.doi.org/10.1016/j.jelechem.2014.05.017
Li L, Ge J, Wu F, Chen R, Chen S, Wu B (2010) Recovery of cobalt and lithium from spent
Lithium ion batteries using organic citric acid as leachant. J Hazard Mater 176(1–3):288–293.
doi:http://dx.doi.org/10.1016/j.jhazmat.2009.11.026
Li L, Lu J, Ren Y, Zhang XX, Chen RJ, Wu F, Amine K (2012) Ascorbic-acid-assisted recovery
of cobalt and lithium from spent Li-ion batteries. J Power Sources 218:21–27
Lu W, Chung DDL (2001) Anodic performance of vapor-derived carbon filaments in Lithium-Ion
secondary battery. Carbon 39(4):493–496. doi:http://dx.doi.org/10.1016/S0008-6223(00)
00157-3
Lux SF, Lucas IT, Pollak E, Passerini S, Winter M, Kostecki R (2012) The mechanism of HF
formation in LiPF6 based organic carbonate electrolytes. Electrochem Commun 14(1):47–50.
https://doi.org/10.1016/j.elecom.2011.10.026
Contestabile M, Panero S, Scrosati B (1999) A laboratory-scale lithium batterie recycling process.
J Power Sources 83:4
Pu N-W, Wang C-A, Sung Y, Liu Y-M, Ger M-D (2009) Production of few-layer graphene by
supercritical CO2 exfoliation of Graphite. Mater Lett 63(23):1987–1989. doi:http://dx.doi.org/
10.1016/j.matlet.2009.06.031
Rothermel S, Evertz M, Kasnatscheew J, Qi X, Grützke M, Winter M, Nowak S (2016) Graphite
recycling from spent Lithium ion batteries. ChemSusChem [accepted]. doi:10.1002/cssc.
201601062
Saeki S, Lee J, Zhang Q, Saito F (2004) Co-grinding LiCoO2 with PVC and water leaching of
metal chlorides formed in ground product. Int J Mineral Process 74, Supplement (0):
S373–S378. doi:http://dx.doi.org/10.1016/j.minpro.2004.08.002
Shin SM, Kim NH, Sohn JS, Yang DH, Kim YH (2005) Development of a metal recovery process
from Li-ion battery wastes. Hydrometallurgy 79(3–4):172–181. doi:http://dx.doi.org/10.1016/j.
hydromet.2005.06.004
Shu J, Shui M, Xu D, Wang D, Ren Y, Gao S (2011) A comparative study of overdischarge
behaviors of cathode materials for Lithium-ion batteries. J Solid State Electrochem 16(2):
819–824. https://doi.org/10.1007/s10008-011-1484-7
Stuart BH (2007) Degradation. In: Stuart BH (ed) Analytical techniques in the sciences. Wiley,
Ltd., Chichester, pp 191–208. doi:10.1002/9780470511343.ch7
Sun L, Qiu K (2012) Organic oxalate as leachant and precipitant for the recovery of valuable
metals from spent lithium-ion batteries. Waste Manag 32(8):1575–1582
Xing W, Dahn JR (1997) Study of irreversible capacities for Li insertion in hard and graphitic
carbons. J Electrochem Soc 144(4):1195–1201. https://doi.org/10.1149/1.1837572
Young K, Wang C, Wang LY, Strunz K (2013) Electric vehicle battery technologies. In:
Garcia-Valle R, Peças Lopes AJ (eds) Electric vehicle integration into modern power networks.
Springer New York, New York, NY, pp 15–56. doi:10.1007/978-1-4614-0134-6_2
Zeng G, Deng X, Luo S, Luo X, Zou J (2012) A copper-catalyzed bioleaching process for
enhancement of cobalt dissolution from spent lithium-ion batteries. J Hazard Mater
199–200:164–169
Chapter 14
Realization in a Demonstration Plant
Jan Diekmann, Thomas Loellhoeffel and Arno Kwade
Abstract This chapter describes the areas and aggregates, as well as the operating
cycle of the realized temporary demonstration plant in Braunschweig. This plant
was run by employees of TU Braunschweig and Lion Engineering GmbH. As
especially safe processing of entire battery systems had to be assured, the developed
safety concept is presented. Overall, the demonstration plant processed 1.4 tons of
the battery systems of electric vehicles, reaching a material recycling rate of up to
80%.
14.1 Design of the Demonstration Plant
The demonstration plant was realized in a refitted and vacant hall at Technische
Universität Braunschweig. Therefore, the requirements of the process like funda-
ments for aggregates, and the technical conditions of the hall had to be arranged.
These arrangements led to the installation of different aggregates in areas, as seen in
Fig. 14.1.
The division in areas mainly follows the process steps and their auxiliary pro-
cesses, or combines several process step alternatives: the disassembly area con-
tained manual, as well as automated disassembly equipment. As during crushing
and drying gaseous components are also released, these areas also included the
specific gas treatment. The separation area combined all process steps for the
material separation except sieving.
J. Diekmann (&) T. Loellhoeffel A. Kwade

Technische Universität Braunschweig, Brunswick, Germany

https://doi.org/10.1007/978-3-319-70572-9_14
Fig. 14.1 Schematic drawing of the LithoRec demonstration plant
14.1.1 Discharge
The first process step is the discharge of the entire battery system. Therefore, partial
disassembly and evaluation of design, contacting, and cells have to be carried out.
Discharging in the demonstration plant was executed using an electrical load which
is capable of putting the power back into the grid. During this process, discharge
current, voltage, and temperatures of the battery systems and, if possible, the
modules were monitored. After reaching a specific voltage level, the battery system
was short circuited.
14.1.2 Disassembly
Following disassembly, recovering the battery system periphery was conducted

manually on a specifically designed disassembly table. This table enables working
simultaneously on every side of the system. The disassembled parts were separately
sorted in stillages and transported to further recycling processes. Dismantling of
battery modules was carried out by two skilled workers with insulated tools. Every
disassembled battery module was short circuited directly.
The hybrid-disassembly station was built to demonstrate the automated disas-
sembly via a robotic arm. This was able to disassemble a battery module with
human support.
14 Realization in a Demonstration Plant 249
14.1.3 Crushing
The resulting battery modules were crushed in an inert rotary shear with four shafts
and a 20 mm discharge screen. The crusher was equipped with an air lock to enable
inert conditions inside the crushing space (oxygen concentration <4%). The
resulting fragments were collected in a barrel. Due to safety reasons, the barrel was
equipped with an emergency flooding system. Cleaning of the loaded atmosphere
was carried out via active carbon. After each crushing process, the inner space of
the crusher was purged to avoid concentration of the carbonates. These concen-
trations were monitored via Fourier transform infrared spectroscopy (FTIR).
14.1.4 Separation of the Electrolyte
Different techniques for the removal of the electrolyte were tested in the demon-
stration plant: one process of the Lion Engineering GmbH and another with a
thermal treatment and lowered pressure. For both methods, a gastight ploughshare
mixer was used. The generated gas components, like hydrogen fluoride were
cleaned via an alkaline jet scrubber. Further, solvent extraction of the electrolyte
components was carried out in laboratory scale.
14.1.5 Material Separation
As mentioned before, the separation area combined all mechanical process steps for
the material separation except sieving. The first, recovery of the heavy parts, and
second air classification, removal of the separator, was executed via a zig-zag-sifter
with nine separating steps. Because of the carcinogenic materials of the cathode
coating, the sifter operated with circulating air. To increase the mechanical stress on
the electrode fragments and to homogenize the fragment size distribution, an
enclosed cutting mill was used.
14.1.6 Sieving
The sieving area combined different sieving technologies. One process was an air
jet sieving, which increases the mechanical stress on the fragments via impact
stress. Sieving with this aggregate was carried out with a mesh size of 200 µm.
Sieving with mesh size of 500 µm and larger was executed via a vibration sieve.
14.2 Operating Cycle
The demonstration plant operated according to a specific cycle. Because no storage

capacity for loaded battery system existed, their delivery had to be just in time. The
battery systems were discharged, as well as short circuited in flame retardant
containers. The short circuited battery system was stored in the container for at least
24 h to avoid voltage relaxation. The short circuited battery modules, resulting from
the disassembly, were stored in safety cabinets before crushing. As the capacity of
the drying step was limited to 50 L, only 10 kgs of battery modules were crushed
and dried afterwards. Because of the harmful decomposition products, no storage of
non-dried fragments was allowed. The drying process takes around 60 min,
depending on the drying technology used. After complete removal of the elec-
trolyte, the resulting fragments could be stored in marked barrels before the sepa-
ration processes were carried out. The resulting fractions were analyzed and then
transported to further processes.
Overall, 1.4 tons of battery systems were processed in the demonstration plant.
14.3 Safety Concept
For the realization of the process in pilot scale, Technische Universität

Braunschweig developed a safety concept. Therefore, the specific hazards and
requirements were evaluated for every area and aggregate, especially concerning
the instruction of the workers. Every process part had to suit the requirements of the
so called 3P-Model (Fig. 14.2).
This model covers three safety levels: prevent, perform, and protect. The first
level is the most important as it involves all precautions to prevent an incident. If it
comes to a safety-critical state, the second level provides actions that have to be
performed. The third level is the same for all process parts and becomes effective if
the actions of the second safety level are without risk minimizing effect.
Fig. 14.2 3P-model for

occupational safety in the
recycling of lithium-ion
batteries
14 Realization in a Demonstration Plant 251
This third level involves inducing the evacuation alert, informing the rescue
forces, and starting the 20-fold air change in case of gas release without fire.
The measures of the first two levels (prevent and perform) are associated with
the different process parts and the specific hazards, and are determined by hazards
analysis.
The first process part is discharging and short circuiting of the battery system.
This is the most important step to prevent incidents, because the stored energy is
removed and therefore the electrical hazard, as well as the interaction hazard is
minimized. The following short circuit and storage avoids the relaxation behavior
of lithium-ion batteries. Because of the high voltage, these tasks have to be carried
out by an electrically qualified person with specific equipment like insulated tools
and electrician gloves. To detect safety critical states, the temperatures of the
modules should be observed via an infrared camera during discharging.
If the temperature exceeds a specific limit, the discharge process has to be
interrupted as the first action of the second safety level (perform). Further rising of
the temperature is an indicator of exothermal reactions inside a battery cell. These
can only be stopped by rapid cooling. For this purpose, a Dewar flask with liquid
nitrogen has to be in place for emergency use.
Nearly the same hazards are present for the following disassembly, although the
electrical hazard should be minimized. These short circuits stay in place until the
module is crushed. Although the electrical hazard potential is minimized, special
qualified persons are working in this process part and electrically insulated tools are
used. In case of a heating cell, the emergency cooling with liquid nitrogen has to be
performed.
During the next process part, crushing, the deliberate destruction of the battery
modules takes place. Before the crushing is carried out, the voltage of every module
has to be checked. Further safety aspects regarding detecting and performing are
considered by the design of the crusher. The crusher itself is gastight and operates
with inert atmosphere. In case of temperature rise and detection of CO, the gas
inflow as well as outflow to the gas scrubber are increased until the reaction stops.
This enables a safe transport and neutralization of the gases that are set free. For
maintenance, specific personal protective equipment is needed due to the decom-
position products of the electrolyte components.
The following process part is determined by the chemical hazard. During the
drying process, the electrolyte components are removed and the conducting salt is
decomposed. Here, a combination of gas treatment is needed. For the neutralization
of HF, a gas scrubber is recommended. To prevent gas leakage, the dryer and its
pipework should be checked regularly. Detection of such gas leaks can be done by
CO sensors.
After drying, the mechanical process steps are carried out. The only remaining
hazard in this process part is the carcinogenic coating materials. To prevent leakage
of these materials, housing of every process step and pipework has to be carried out.
Detection of leakage can be assured by white painted floors and machines.
Further machine specific precautions can be found in the chapters of the process
steps.
Chapter 15
Economic Assessment of the LithoRec
Process
Christian Thies, Karsten Kieckhäfer, Claas Hoyer and

Thomas S. Spengler
Abstract The successful implementation of an industrial-scale recycling process

depends strongly on its economic feasibility. Thereby, the economics are influenced
by many drivers such as the amount of returned batteries, the prices that can be
obtained from selling the recycled material fractions, and investments and operating
costs resulting from the process configuration. In this chapter, an analysis of the
economic feasibility of the LithoRec process is carried out using an optimization
model for technology and capacity planning. The model is applied to different
scenarios describing the development of the amount of spent batteries and factor
prices for the European market. From this, optimal investment plans and financial
performance indicators are determined. Overall, the results indicate that the process
can be operated economically in the long run. In most scenarios, the operation pays
off by selling the recycled materials. Only when the amount of returned batteries is
low, moderate gate fees or economic incentives by policy makers would be
required.
15.1 Introduction
The LithoRec process allows for a safe and efficient recycling of lithium-ion bat-
teries from electric vehicles. While it has been implemented successfully in a pilot
plant, the scale-up to an industrial-sized recycling operation requires a careful
economic assessment. For potential investors, it is important to know whether the
process can be operated profitably and by which drivers its profitability is
influenced.
From an economic perspective, recycling of lithium-ion batteries appears
promising due to the high value of the materials. For example, the market
prices (from 2016) for key materials that are used in lithium-ion batteries are:
C. Thies (&) K. Kieckhäfer C. Hoyer T. S. Spengler

Institute of Automotive Management and Industrial Production,
e-mail: ch.thies@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_15
254 C. Thies et al.
1.50 EUR/kg for aluminum, 5 EUR/kg for copper, 6.00 EUR/kg for lithium car-
bonate, 9.00 EUR/kg for nickel, and 33 EUR/kg for cobalt (LME 2017; USGS
2017). Based on these prices, the value of the materials contained in a 300 kg
battery pack exceeds 700 EUR. Even if not all materials can be recovered and the
price for secondary materials is typically lower than for primary raw materials, there
is significant potential for generating revenues from battery recycling.
On the other hand, the implementation of the LithoRec process requires
sophisticated machinery and equipment that involves high investments. Moreover,
its operation induces various expenses for staff, energy, and other input factors.
Therefore, the investments and expenses must be evaluated against the potential
revenues, which is a complicated task due to various interdependencies between the
decisions that are involved. For example, a certain technological variant of the
process may improve the quality of the recycled materials and thus lead to higher
revenues but also requires more sophisticated equipment, leading to higher
expenses. Or, the installed capacity determines how many batteries can be
processed as well as the rate at which the equipment is utilized. Furthermore,
surrounding conditions, like the amount of returned batteries that is available for
recycling or the price developments of input and output factors, need to be con-
sidered. Therefore, a structured methodology for economic assessment of the
LithoRec process is proposed in this Chapter.
The remainder of this Chapter is organized as follows. The influencing factors on
economic performance are discussed in Sect. 15.2. Then, an optimization model for
technology and capacity planning is proposed in Sect. 15.3. The underlying dataset
and the investigated scenarios are described in Sect. 15.4. Results from applying
the model are presented in Sect. 15.5. Last, managerial implications are discussed
and conclusions are drawn in Sect. 15.6.
15.2 Drivers of Economic Performance
Several drivers of the economic performance of the LithoRec process can be

identified (Hoyer et al. 2015). First, the types and the amount of the returned
batteries need to be considered in an economic assessment. The battery type refers
to the cell chemistry and the size of the battery. The cell chemistry is an important
driver because some cell chemistries, like lithium nickel manganese cobalt oxide
(NCM), contain a higher share of valuable metals than other cell chemistries, like
lithium iron phosphate (LFP). The size of the battery mainly depends on its
application. Battery packs that are used in full electric vehicles (EV) are much
larger than battery packs used in plug-in hybrid electric vehicles (PHEV) or hybrid
electric vehicles (HEV). The total amount of returned batteries is important because
it determines the recycling capacities to be installed. Larger recycling machinery
usually benefits from economies of scale, but there is also a higher risk of poor
utilization if the amount of returned batteries drops below the expectations.
15 Economic Assessment of the LithoRec Process 255
Second, the prices of input and output factors are also an important driver. The
prices for the recycling fraction determine the revenues, the prices for input factors
(e.g. electricity) and undesired output factors (e.g. waste disposal) determine the
costs of the process. Particular importance should be paid to the “black powder”
fraction that contains the electrode coating materials due to the high concentration
of high-value materials.
Third, the process configuration must be considered. For example, the disas-
sembly process can be carried out manually or semi-automated with collaborative
robots. While the first option requires little initial investments, it has much higher
operating cost due to the specially trained staff that is needed for the disassembly
operation. Another example relates to the question whether the remaining energy
should be fed into the electricity network (higher investments, positive revenues) or
be dissipated in heat (lower investments, no revenues).
The difficulty in dealing with these influencing factors is that they are dynamic
(e.g. battery volume expected to increase, changing mix of battery types over time)
and that they are interrelated (e.g. more sophisticated process leads to better quality
products and consequently higher prices). Therefore, these drivers and the relations
between them are included in the proposed model in Sect. 15.3.
15.3 Optimization Model for Technology

and Capacity Planning
An optimization model is used to assess the economic performance of the process in

different market and price scenarios. The model is adopted from Hoyer et al. (2015)
and here, only its main features are described. An overview of the basic concept and
the system boundaries of the model is given in Fig. 15.1. The recycling process is
divided into three steps, and for each step several process modules that are dis-
tinguished by their technological variant and their capacity can be installed. The
basic idea of the model is to find optimal investment plans for the process modules.
To this end, the model decides on the time of installation, the technological variant,
and the capacity for each process module. Despite the original model’s capability to
decide on the recycling program as well, it is assumed that all available packs go
through all three recycling steps due to the current lack of suitable recycling routes
and corresponding prices for intermediate products.
The decisions are taken so that net present value (NPV ) of future cash flows is
maximized. Formally, the NPV is the sum of cash flows CFt over all planning
periods (t ¼ 1; . . .; TÞ. The cash flows are discounted with interest rate i, reflecting
the investor’s expectation on financial return. An investment is economically
advantageous if the NPV is positive.
256 C. Thies et al.
Input factors
(e.g. electricity from grid, DMC, water, nitrogen)
Handling & Disassembly Mechanical

discharge processing
Hydro /pyro
metallurgical
L L L treatment
S S S
Available packs Discharged packs Discharged modules Electrode coaƟng powder

System boundary for
economic assessment
Output factors
(e.g. electricity to grid, steel scrap, aluminum scrap, solid waste)
Fig. 15.1 Conceptual model with system boundaries for economic assessment
XT
CFt
NPV ¼ ð14:1Þ
t¼1 ð1 þ iÞt
The cash flows CFt in each period t are composed of the investments into new
Liq
m;t ), liquidation revenues (CFm;t ), the fixed operating
process modules (CFInvest
expenses (CFFixOp VarOp
m;t ), variable operating expenses (CFm;t ) and the income and
expenses from input factors and output factors (CFFactors
m;t ) over all process modules m.
X Liq FixOp

CFt ¼ m;t þ CFm;t þ CFm;t
CFInvest þ CFVarOp
m;t þ CFFactors
m;t ð14:2Þ
m2M
Complementary to the NPV, the discounted payback period is analyzed as an

economic performance indicator. The discounted payback period T Payback is defined
as the first period s in which the cumulative discounted cash flow turns positive.
The discounted payback period is a useful indicator if the investor has expectations
regarding the time after which the investment shall amortize.
Xs
CFt
T Payback ¼ min s 2 f1; . . .; T gj 0 ð14:3Þ
t¼1 ð 1 þ i Þt
In the model, the maximization of the NPV is subject to several constraints.

These comprise the conservation of material flows, the provision of sufficient
capacities for the execution of recycling activities and minimum operating times.
For a full description of the model, we refer to Hoyer et al. (2015).
The model is implemented in AIMMS 4.28 x64. For the settings described in
Sect. 15.4, the resulting mixed-integer linear program contains about 16,300
decision variables (including 2,700 integer variables) and 11,800 constraints (for
each considered scenario). These problem instances can be solved with CPLEX
12.6 on a standard PC (Intel i5 3320M, 2.6 GHz, 8 GB RAM, Windows 10 x64) in
less than 5 seconds, using approximately 100 MB RAM.
15.4 Data and Scenarios for Model-Based Analysis
The data used to run the model is compiled from different sources. Material and
energy flows stem from the LithoRec pilot plant (cf. Chap. 14). Prices are derived
from market prices and literature values. Investments for recycling equipment are
derived from quotes of the manufacturers and estimates in literature.
The amount of spent batteries that are available for recycling is derived from past
and future sales of electric vehicles, the useful life of the batteries in electric
vehicles, and the potential re-use of the batteries in stationary applications (Hoyer
et al. 2011). The development of the electric vehicles stock in Europe is illustrated
in Fig. 15.2. The European market is considered because it is unlikely that each
member state will have its own recycling facilities. Instead, it is assumed that
especially in an initial market phase, a limited number of operators will process the
batteries from multiple countries. For the future development, three market sce-
narios are considered: a pessimistic market scenario in which the stock of electric
vehicles will reach 4.7 million by 2020, a realistic market scenario with a stock of
8.5 million electric vehicles in 2020, and an optimistic market scenario with a stock
of 11.9 electric vehicles in 2020.
The useful life of the batteries is highly uncertain and will be different for each
battery as it depends on how the battery is used (annual mileage, driving behavior,
12
11.9
Electric vehicles [in millions]
Market scenario EV type 10

optimistic BEV
PHEV 8.5 8
realistic
pessimistic HEV
6
ACTUAL PLAN 4.7 4
0
2010 2012 2014 2016 2018 2020
Fig. 15.2 Past and future development of the electric vehicle stock in Europe (EU28 + Norway)
for three market scenarios
258 C. Thies et al.
charging patterns, temperatures). The useful life is modeled using a binomial dis-
tribution with a maximum useful life of 20 years and an average useful life as
indicated in Table 15.1. It is also assumed that a certain share of batteries that have
reached the end of their useful life in vehicle applications will have a second life in
stationary applications and be available for recycling with some delay. The assumed
second life shares in the three scenarios are also indicated in Table 15.1. Finally, it
is assumed that there will be competition between multiple recycling processes and
LithoRec will have a limited market share.
The resulting volume of spent batteries is illustrated in Table 15.2 for the years
2020 and 2025. To account for the different battery sizes (EV: 300 kg; PHEV:
170 kg; HEV: 30 kg), the total volume is expressed in EV-equivalent units. Due to
the combination of different uncertainty sources, the range of values is quite large.
In this sense, the pessimistic and the optimistic market scenarios describe extreme
developments in which all negative or all positive factors accumulate. For all
scenarios, it is assumed that 80% of the batteries are based on NCM cell chemistry
and 20% are based on LFP cell chemistry.
The investments and fixed operating expenses of the process modules depend on
the technology variant and the capacity (Table 15.3). For the handling and dis-
charge step, two technology variants are considered. In the first variant without
energy recuperation, the batteries are discharged via a large resistor and the residual
energy dissipates as heat. In the second variant, the residual energy is refed into the
electricity grid. In the next process step, only manual disassembly is considered for
economic assessment because the hybrid human-robot disassembly still lacks valid
data. In the mechanical processing step, two technology variants are considered that
differ in the way how solvents from the electrolyte are removed from the shredded
battery modules, either via thermal drying (cf. Chap. 8) or via solvent extraction
(cf. Chap. 9).
Each process step can be installed in two capacity classes. The small capacity
class is comparable to the LithoRec pilot plant and allows for processing 1 EV-eq
battery pack or 200 kg of battery modules per hour, which translates into 6000
EV-eq battery packs or 1200 tons of battery modules per year. The large capacity
class represents an industrial scale operation and allows for the fivefold amount.
Annual capacities are derived under the assumption that the process is operated in a
3-shift-system on 250 days of the year, i.e. that the equipment is available for
6000 h per year. Downtimes for maintenance and breakdowns are not accounted
Table 15.1 Assumptions for estimating the amount of spent batteries in three market scenarios
Parameter Unit Pessimistic Realistic Optimistic
Electric vehicles stock 2020 [millions] 4.7 8.5 11.9
Avg. useful life of LIB (EV/PHEV/ [years] 10/15/15 8/10/15 5/8/10
HEV)
Second life share (EV/PHEV/HEV) [%] 30/20/5 10/5/2 5/2/0
Market share LithoRec [%] 10 20 30
Table 15.2 Annual volume of spent battery packs for recycling [EV-eq]
Year Pessimistic Realistic Optimistic
2020 1800 14,600 90,000
2025 8100 51,500 172,500
Table 15.3 Process modules with capacities, initial investments and fixed operating expenses
Process step Variant Capacity Investment Fixed operating
(EUR2020) expenses (EUR2020/a)
Building – – 3,648,000 66,000
Handling Without energy Small (6000 140,500 1280
and recuperation EV-eq/a)
discharge Large (30,000 386,500 2400
EV-eq/a)
With energy Small (6000 146,500 1400
recuperation EV-eq/a)
Large (30,000 416,500 3000
EV-eq/a)
Disassembly Manual Small (6000 10,000 120
EV-eq/a)
Large (30,000 44,000 480
EV-eq/a)
Mechanical Thermal drying Small (1200 t 1,693,000 310,000
processing modules/a)
Large (6000 t 3,232,000 537,000
modules/a)
Solvent Small (1200 t 2,750,000 363,000
extraction modules/a)
Large (6000 t 6,171,000 684,000
modules/a)
for. The model allows for installing multiple process modules of different capacities
for each process step.
The initial investments include not only the catalogue price of the equipment
itself, but also the expenses for auxiliary systems, installation, and engineering,
which are added via contingency factors that are multiplied with the catalogue
price. The fixed operating expense comprises a component that is proportional to
the initial investment (maintenance, insurance, etc.) and a component that covers
the salary of permanent staff (production engineer, supervisor). Here, all permanent
staff is allocated to the mechanical processing step. For the handling and discharge
step as well as the disassembly step, staff can be hired more flexibly. Thus the
wages are considered as variable expenses that are proportional to the throughput.
The prices of input and output factors are listed in Table 15.4. Similar to the
market scenarios, three price scenarios (pessimistic, realistic, and optimistic) are
considered to capture the range of possible price developments for each factor.
260 C. Thies et al.
Table 15.4 Initial prices of selected materials in EUR2020. Negative prices indicate that the
material must be disposed
Material Unit Pessimistic Realistic Optimistic
Input factors
Electricity, from grid kWh 0.25 0.20 0.15
Heating steam kWh 0.03 0.03 0.03
Nitrogen m3 0.23 0.23 0.23
Freshwater L 0.01 0.01 0.01
DMC (make-up) kg 0.47 0.53 0.58
Output factors
Electricity, to grid kWh 0.15 0.20 0.25
(Stainless) steel scrap kg 0.60 0.70 0.80
Aluminum scrap kg 0.50 0.60 0.70
Mixed electronics and cables scrap kg 0.20 0.30 0.40
Aluminum fragments (compacted) kg 0.60 0.70 0.80
Copper fragments (compacted) kg 2.80 3.80 4.80
DMC kg 0.47 0.53 0.58
EMC kg 0.67 0.75 0.83
EC kg 1.33 1.50 1.67
Waste, for incineration kg −0.30 −0.25 −0.20
Waste, for disposal kg −0.30 −0.25 −0.20
Waste, for special treatment kg −1.19 −1.19 −1.19
Table 15.5 Assumed prices for electrode coating powder in EUR2020/kg

Without fluoride impurities With fluoride impurities
From NCM cells 0.80 0.80
From LFP cells 0.00 0.00
The prices are related to the reference year 2020 and are adjusted by a yearly
inflation factor of 2% for future periods of the planning horizon.
The price of the electrode coating powder is difficult to determine because there
is no established market yet. The powder contains a high concentration of valuable
metals. In fact, for NCM batteries, the value density in the electrode coating power
(ca. 6 EUR per kg) is much higher than in complete battery cells (ca. 4 EUR per kg)
or entire battery packs (ca. 3 EUR per kg). However, further hydro- or pyromet-
allurgical treatment is needed to recover the metals and there is high uncertainty
regarding the cost of this step. Therefore, we assume a price of 0.8 EUR per kg of
NCM electrode coating powder and 0 EUR/kg for LFP electrode coating powder
for our baseline analysis. The potential influence of fluoride impurities on the
achievable price will be discussed in Sect. 15.5, but is neglected for the baseline
analysis. The assumed prices are summarized in Table 15.5.
15.5 Results from Model Application
Several results can be obtained from applying the model. An aggregate perspective
on economic performance is presented first before exploring the influence of some
drivers in more detail.
Figure 15.3 shows the NPV and payback period for different market and price
scenarios. The NPV is positive in the realistic and the optimistic market scenario,
indicating that investing would advantageous under these circumstances. Only in
the pessimistic market scenario, the NPV is negative. In this case, additional
measures like gate fees would be required to make the investment profitable. Next
to the market scenario, the price scenario has a major influence on the NPV.
Obviously, the optimistic price scenario leads to higher NPVs than the realistic or
pessimistic price scenario. The differences in NPV between the price scenarios are
related to the market scenario and increase as more batteries are processed. The
payback period ranges from 17 years to 1 year and can only be calculated if the
NPV turns positive at all. In the realistic market scenario and realistic price sce-
nario, the investment would pay back after 5 years.
For a better understanding, the composition of the NPV for the realistic market
scenario and realistic price scenario is presented in Fig. 15.4. The NPV in this
scenario is 25.9 million EUR. The contribution of the building in which the process
is operated is −3.9 million EUR. Handling and discharging contribute with −12.6
million EUR. The main driver for this step are the variable operating expenses,
more particularly, the wages of the staff who transports and opens the battery
systems and connects them to the discharge device. The variable operating
expenses of the disassembly step are even higher than in the handling and discharge
step, because this process is very labor intensive. Contrary, in disassembly sig-
nificant revenues can be made due to the sale of aluminum, steel, and electronic
scrap from the housing and the battery management system. Overall, the
200 184
NPV [EUR in millions]
150
113 Price scenario
Pessimistic
100
Realistic
45
50 Optimistic
26 19
1
0
8 4 1 17a 5a 3a 7a 1a 1a PaybackPeriod
50
Pessimistic Realistic Optimistic
Market scenario
Fig. 15.3 NPV and payback period for different market scenarios and price scenarios
262 C. Thies et al.
80 53.6
60

40
20 -11.3
-3.9 -12.6
0
-20
-40
Handling & Mechanical
Building Disassembly
discharge processing
Investments + fixed op. -3.9 -0.7 -0.1 -13.6
Variable operating -12.9 -27.1
Input factors -3.5
Output factors 1.1 15.9 70.7
Sum -3.9 -12.6 -11.3 53.6
Fig. 15.4 Composition of NPV for realistic market scenario and realistic price scenario
contribution of the disassembly step to the NPV amounts to −11.3 million EUR.
With 53.6 million EUR, the mechanical processing step has by far the highest
contribution to the NPV. The discounted investments and fixed operating expenses
amount to −13.6 million EUR. Thereby, the fixed operating expenses include
wages of the production engineer and control room staff whose presence is required
at all times independent of the throughput. High revenues can be generated from
selling the output factors.
Figure 15.5 shows the contributions to NPV of the input and output factors in
the realistic market scenario and realistic price scenario. As expected for a recycling
process, the positive contributions of 91.2 million EUR are much higher than the
negative contributions of −7.0 million EUR. The main positive contributions arise
from the sale of copper fragments (from the anode current collector), aluminum
scrap (from pack and cell housing), and the electrode coating powder. The negative
contributions comprise the disposal of waste fractions and the procurement of the
input factors (mainly electricity).
Regarding the technology choices, the two variants of the handling and dis-
charge steps are compared first: discharge with energy recuperation that enables
feeding the remaining energy in the batteries back to the grid and without energy
recuperation (remaining energy dissipates as heat in a large resistor). The equipment
that is necessary for energy recuperation requires a higher investment at the
beginning. Whether this higher investment pays off, depends on price for feeding
electricity to the grid. Figure 15.6 shows the difference in NPV between installing
the handling and discharge step with and without energy recuperation. Obviously,
a) Negative contributions b) Positive contributions

Electronic
Aluminum scrap Electricity,
fragments 1,3 to grid
4,5 1,1
Steelscrap
Nitrogen 6,8
0,3
Copper
Waste Electrode fragments
Electricity, 7,0 disposal coating 91,2 43,0
from grid 3,5 powder
3,2 16,4
Aluminum
scrap
[EUR in millions] 18,1
Fig. 15.5 a Negative and b positive contributions to NPV of input and output factors in the
realistic market scenario and realistic price scenario
2.0
Optimistic
1.5 Realistic
Pessimistic
1.0
Required: 0.01
0.5
0.0
0 0.05 0.1 0.15 0.2 0.25 0.3
0.5
Electricity price [€/kWh]
Fig. 15.6 Break-even analysis for discharge with energy recuperation (compared to discharge
without energy recuperation) with regard to electricity price
when the electricity price is zero, the difference in NPV is negative because the
handling and discharge with recuperation requires higher investments. However,
the two options break even (DNPV ¼ 0) at an electricity price of 0.01 EUR/kWh.
This is well below the electricity prices assumed in the three price scenarios. So
even with a significant drop in the electricity price (as long as it remains above
0.01 EUR/kWh), the recuperation option should be preferred from an economic
perspective.
The second technology choice that is analyzed relates to the mechanical pro-
cessing module. The electrolyte can be separated from the shredded battery mod-
ules by thermal drying (cf. Chap. 8) or solvent extraction (cf. Chap. 9). The solvent
extraction variant is more complicated than thermal drying and requires higher
investments and more input factors. However, it allows for a reduction of undesired
fluoride impurities in the electrode coating powder, which might cause problems
during its further treatment. Therefore, the achievable price of the electrode coating
powder may be affected. The tradeoff between potentially lower prices for electrode
coating powder with fluoride impurities and the higher expenses for the solvent
extraction process is depicted in Fig. 15.7. When prices for the electrode coating
264 C. Thies et al.
NPV [Million EURR]

2
0.28 0.35 0.72
1 Market scenario
0 Pessimistic
Realistic
10 0.2 0.4 0.6 0.8 1
Optimistic
2
3
Price difference for electrode coating powder
with and without fluoride impurities [€/kg]
Fig. 15.7 Break-even analysis for mechanical processing with extraction (compared to mechan-
ical processing with thermal drying) with regard to price difference for electrode coating powder
with and without fluoride impurities in different market scenarios
2
1 0.06 0.24 0.47 Price scenario

Pessimistic
0 Realistic
0 0.1 0.2 0.3 0.4 0.5 0.6 Optimistic
1
2
Gate fee [EUR/kg]
Fig. 15.8 Minimum required gate fees to make the process economically feasible (NPV 0) for
the case that value of cathode coating powder is zero
powder are the same (zero price difference), the thermal drying variant leads to a
higher NPV than the solvent extraction variant, i.e. the difference in NPV is neg-
ative. At a certain price difference, the solvent extraction breaks even with the
thermal drying variant. The break-even point depends on the market scenario and is
0.28 EUR/kg in the optimistic case, 0.35 EUR/kg in the realistic case, and
0.72 EUR/kg in the pessimistic case. This means that if for the electrode coating
powder with reduced fluoride impurities a higher price of at least 0.35 EUR/kg can
be achieved in the realistic market scenario, then the higher investment in solvent
extraction pays off.
As described above, the LithoRec process can be operated profitably when there
is a sufficient amount of returned batteries. In the pessimistic market scenario,
however, the NPV would be negative if no further measures are taken. Such
measures could be the introduction of gate fees. In this case, a gate fee for each
battery system would have to be collected. The minimum required gate fees for
obtaining a positive NPV are illustrated in Fig. 15.8 for different price scenarios. If
all other prices develop optimistically, a gate fee of 0.06 EUR/kg would be
required. For the realistic and the pessimistic price scenario, the required gate fees
would be 0.24 EUR/kg and 0.47 EUR/kg respectively. This would translate into a
gate fee between 18 EUR and 141 EUR for a 300 kg EV system, which would
ultimately be charged to the customers by the OEMs. While the car industry favors
a situation without gate fees, it must be stated that the required gate fees are
comparably low with regard to the gate fees that are charged by other recycling
operators, which are reported at roughly 4 EUR/kg (Kushnir 2015).
15.6 Managerial Implications and Conclusions
From the economic assessment of the LithoRec process, its installation and oper-
ation at an industrial scale appears economically feasible. In most scenarios, the
operation pays off by selling the recycled materials and only when the amount of
returned batteries is low (pessimistic market scenarios), moderate gate fees would
be required to make the process profitable. Main contributors to the economic
feasibility of the LithoRec process are copper fragments, aluminum scrap, and the
electrode coating powder, which can be sold as a secondary feedstock. With regard
to technology choices, the handling and discharge module should be implemented
with energy recuperation, because the break-even electricity is well below the actual
current and the expected future electricity prices. Regarding mechanical processing,
the thermal drying variant is economically advantageous as long as the resulting
fluoride impurities do not affect the price of the electrode coating powder.
Therefore, potential investors should thoroughly investigate further treatment routes
for the powder.
Beyond the analyses presented is this chapter, the optimization model also
allows to derive detailed investment plans for investors over the planning horizon,
i.e. when to invest in a certain process or facility of a certain capacity. While these
investment plans are subject to high uncertainties with regard to the market and
price scenarios, more accurate information may be available in the future.
Therefore, the model can support investors in fixing initial investment plans, which
need to be revisited and adjusted to future circumstances in a rolling planning
process. To this end, the model can be run repeatedly (e.g. every year) with updated
data taking into account already implemented decisions.
To promote a fast and comprehensive establishment of effective recycling net-
works, car manufacturers and policy makers can adopt several measures to reduce
the risks for potential investors and operators. For example, car manufacturers could
engage financially in a joint venture for battery recycling or make contractually
guaranteed long-term reservations of recycling capacities. Policy makers could
grant direct subsidies for the installation of new recycling facilities, which would
reduce the initial investments and shorten the payback period. Also, the definition
of material-specific minimum recycling rates or the introduction of premiums for
the recovery of certain strategic materials would provide investors the incentive to
install sophisticated technology that enables a high quality recycling of lithium-ion
batteries (Hoyer 2015).
266 C. Thies et al.
References
Hoyer C (2015) Strategische Planung des Recyclings von Lithium-Ionen-Batterien aus

Elektrofahrzeugen in Deutschland. Produktion und Logistik, Springer, Wiesbaden
Hoyer C, Kieckhäfer K, Spengler TS (2011) Strategische Planung des Recyclings von
Lithium-Ionen-Trakionsbatterien. In: Sucky E, Asdecker B, Dobhan A, Haas S, Wiese J
(eds) Logistikmanagement: Herausforderungen, Chancen und Lösungen. University of
Bamberg Press, Bamberg, pp 399–419
Hoyer C, Kieckhäfer K, Spengler TS (2015) Technology and capacity planning for the recycling of
lithium-ion electric vehicle batteries in Germany. J Bus Econ 85(5):505–544. https://doi.org/
10.1007/s11573-014-0744-2
Kushnir D (2015) Lithium ion battery recycling technology 2015: Current state and future
prospects. Environmental systems analysis report #2015:18. Chalmers University of
Technology. http://publications.lib.chalmers.se/records/fulltext/230991/local_230991.pdf
LME (2017) Monthly average settlement prices for January 2017. London Metal Exchange.
https://www.lme.com/metals/reports/averages/
USGS (2017) Annual publications: Lithium. United States Geological Survey. https://minerals.
usgs.gov/minerals/pubs/commodity/lithium/
Chapter 16
Environmental Aspects of the Recycling
of Lithium-Ion Traction Batteries
Felipe Cerdas, Stefan Andrew, Sebastian Thiede

Abstract Recycling lithium-ion traction batteries is expected to contribute

decreasing the environmental impact of electric vehicles. Recycling might not only
help reducing the amount of primary material required to be supplied to the battery
industry but also preventing landfill and incineration activities. Nevertheless,
recycling does not imply per se an environmental benefit as its impact is affected by
different issues such as the quality of the material recovered, the energy and
material consumption by the process itself and the efforts caused by the required
logistics. This chapter presents an analysis of the most relevant aspects of the
recycling process of lithium-ion batteries from an environmental perspective. It first
introduces a framework to understand the different ways in which a recycling
industry might affect the environment. This framework is further applied to describe
the potential environmental effects of recycling traction batteries. Using primary
data, we conducted an energy and materials flow analysis of the process developed
within the LithoRec project. Finally, we discuss the results of the life cycle
assessment (LCA) performed within the context of the LithoRec project and
identify key issues to be considered in order to develop recycling processes that
contribute to develop an environmentally consistent recycling strategy parallel to
the rising traction battery industry.
16.1 Introduction
As the automotive industry gradually moves towards the electrification of the power
train, recycling is increasingly seen as a requirement to mitigate the additional
environmental impact caused by the battery system. Besides the potential economic
benefits, different advantages are normally attributed to recycling: (i) it can con-
tribute to reduce the energy embodied to a battery system by avoiding many of the
F. Cerdas (&) S. Andrew S. Thiede C. Herrmann

Chair of Sustainable Manufacturing and Life Cycle Engineering, Institute of Machine Tools
and Production Technology, TU Braunschweig, Brunswick, Germany
e-mail: juan.cerdas-marin@tu-braunschweig.de

https://doi.org/10.1007/978-3-319-70572-9_16
268 F. Cerdas et al.
energy intensive steps in the production of primary materials (Gutowski et al.

2013), (ii) it can divert batteries from undesired disposal treatments leading to
further environmental impacts, (iii) it enables the local production of raw materials
limiting the geopolitical dependency on materials. However, recycling per defini-
tion does not imply an environmental benefit. Recycling processes often require
large amounts of energy and auxiliary materials. Furthermore, the collection of
spent products from waste streams represents additional environmental efforts.
Having the inherent chemical and physical complexity of traction batteries in
mind, it comes as no surprise to learn that a recycling process for a traction battery
is time and resources intensive. It is therefore essential to develop and optimize
collection and separation technologies that are both economically attractive and
environmentally consistent. Ideally, a recycling process maximizes the amount and
quality of material recovered while having minimum environmental impacts. The
work done within LithoRec had pursued this objective. The project aimed at
developing a recycling process while achieving a high recovery rate for the majority
of the materials found in a stream of traction batteries. This chapter is dedicated to
the analysis of the environmental aspects of the recycling process developed in
LithoRec (II). We have performed an environmental characterization of each of the
unit processes developed for the project. Energy and material flows were measured
from the processes developed for the recycling plant. Additionally, the Life Cycle
Assessment (LCA) methodology was applied to estimate the environmental impacts
of the process. This chapter gives insights regarding the main environmental aspects
of the recycling process, the product and its materials on a life cycle perspective.
Conclusions and policy implications, as well as further research needs in this field
have been summarized at the end.
16.2 Environmental Relevance of the Recycling

of Traction Batteries
Recycling is seen in general as a way to decrease the environmental impact caused

by the consumption of material (Geyer et al. 2015). Recycling itself cannot be seen
as an environmental objective but rather as a way to achieve a specific environ-
mental goal. Parallel to the rise of a traction battery industry, recycling needs to
emerge and be guided in an environmentally sound direction. Although recycling
is, in fact, the only end-of-life strategy able to return spent materials back to the
supply chain (Ashby 2013), it causes environmental impacts during the collection
of products, it’s separation and the reprocessing of materials.
To understand the net impact of the recycling process of lithium-ion batteries we
must consider all influencing factors. These elements are represented in the
framework in Fig. 16.1. As discussed by Geyer et al. (2015), recycling causes net
positive impacts when the environmental effects caused directly by the material
recovering processes (e.g. collection, separation, crushing, etc.) are lower than the
16 Environmental Aspects of the Recycling of Lithium-Ion Traction … 269
Fig. 16.1 Framework for the analysis of environmental effects of recycling. Based on the work by
Geyer et al. (2015). Dotted red arrows represent the avoided material flow
avoided impact from the production of primary material substituted and the
non-material-recovery activities [(a), e.g. landfill, incineration, etc.] together. In the
following sections we use the framework in Fig. 16.1 to address the dimensions in
which the recycling of traction batteries interacts with the environment.
16.2.1 Recycling Avoids the Impact Caused

by Non-material-Recovery End-of-Life Activities
Under the absence of attractive economic benefits or policies that promotes recy-
cling, a battery system (or parts of it) could end up in common streams of electric
and electronic waste being dumped in landfills, incinerated or informally recycled
in secondary markets in developing countries.
If a product has none or little economic value, it will probably be sent to a waste
stream and treated according to a sanitary regulation. Non-material recovering
waste treatment activities such as landfilling and incineration have in the past
contributed to get rid of waste while allowing the generation of energy (e.g. through
capturing methane or recovering heat from combustion processes). The generation
of electricity from these activities is, however, largely inefficient and can contribute
to important environmental dangers. In incineration processes present energy
recovery efficiencies of around 50% for the recovery of heat and 35% for the
generation of electricity (Ashby 2013) and produces large amounts of GHG-related
and ozone depleting emissions such CO2, N2O, NOx, NH3, and organic C.
The leaching of high concentration of organic contaminants and metals into soil
and water as well as the intensive use of land are some examples of environmental
impacts from landfill activities. Although traction battery cells are free of hazardous
materials such as mercury, lead and cadmium, other metals used in traction batteries
such as zinc, manganese, copper, lithium and nickel can potentially leak into soil
and water causing unwanted health effects (European Commission 2014). For
industrialized countries with strong environmental regulations, it is highly unreal-
istic that traction batteries end up being dumped. In Europe, the directive on bat-
teries and waste batteries recommends to prohibit the disposal of industrial and
automotive batteries in landfills or their incineration and established that recycling
processes for such batteries are required by law to meet a material recovery effi-
ciency of at least 50% of its weight (European Commission 2014).
If a product, in turn, contains materials with some remaining economic value, it
is foreseeable that the product will find its way legally or illegally to secondary
markets. While the amount of electric and electronic waste has dramatically
increased since 2012 to a record of 50 million tons per year globally, it is estimated
that around 75–80% of these products are sent to developing countries mainly in
Asia and Africa for disposal (Perkins et al. 2014). The products are then manually
disassembled and the materials recovered using rather rudimentary recycling
techniques. Typical informal recycling activities like the open burning of plastics or
the use of toxic solvents and reagents to recover valuable metals, contaminate soil,
air and water and leads to severe impacts on human health (Perkins et al. 2014;
Kahhat and Williams 2012).
16.2.2 Recycling Avoids the Impact of Producing

Virgin Material
Besides the potential economic benefits of recycling, recovering materials from a

lithium-ion battery system implies, under the precondition that there are no material
quality losses, that less primary material is required to be extracted from primary
sources. This contributes to reduce the emission intensity of mining and primary
material processing activities. This is particularly significant for current commercial
lithium-ion batteries as these contain large shares of materials that potentially
represent an environmental risk as their production processes are highly
energy-intensive. The analysis of the environmental impacts of a battery pack is
usually divided into two perspectives: (i) the amount of material used and (ii) the
environmental relevance of these materials. Table 16.1 gives an overview of the
material composition of a generic battery system. A battery pack is composed by a
structural casing and connecting components, a battery management unit
(BMU) including several power electronics devices, and battery modules which
consist of the battery cells. The battery cells are composed by the cathode, the
anode, electrolyte, separator and the cell housing among other components. Notice
that the battery cells (i.e. the materials in the battery cells) represent around 45% of
the whole weight.
Research done by Ellingsen et al. (2013) present results of a life cycle assess-
ment (LCA) performed to estimate the environmental impacts of a traction battery.
Table 16.1 Components and material shares of a generic battery system (Diekmann et al. 2017)
Component Material Share (%)
Battery system periphery Aluminum 18.00
Steel 5.70
Plastics 5.70
Electronics 2.70
Cables 2.30
Module periphery Aluminum 5.20
Steel 3.30
Plastics 1.50
Cell housing Aluminum 5.80
Cathode Aluminum 5.50
Oxygen 4.80
Ni 3.10
Co 3.10
Mn 2.80
Li 1.00
Anode Copper 9.20
Graphite 8.20
Electrolyte, separator and others Volatile components 8.30
Plastics 3.80
As they reported, the manufacturing of the cells itself is responsible of more than
60% of the emissions contributing to global warming potential. Copper, used as a
current collector for the negative electrode was found to present strong environ-
mental effects to toxicity related impact categories such as human toxicity potential
HTP and other impact categories such as eutrophication potential ETP. Finally, the
cathode coating material was found to present important environmental impacts in
categories such as terrestrial acidification potential (TAP) and particle matter for-
mation potential (PMFP). Research by Dunn et al. (2015b) presents estimations on
the environmental impacts of traction batteries on an embodied energy basis. Cradle
to gate embodied energy for the production of a battery system is provided
Fig. 16.2. For all of the battery systems compared, the embodied energy is driven
by both the production of the cathode and the aluminum required for the housing of
the battery and the modules. For instance, the production of primary aluminum has
an embodied energy that ranges from 187 to 206 MJ/kg (Granta Design Limited
2013) and makes around 35% of the total mass of the battery system as shown in
Table 16.1.
Notice in Fig. 16.2 that all the battery systems containing cobalt and nickel
result in general in a higher embodied energy. For these systems, the embodied
energy is driven on more than a 30% by the production of aluminum and around
45% by the production of the cathode materials.
Fig. 16.2 Cradle to gate embodied energy for the production of one kilogram battery pack. LFP:
LiFePO4, NCM: LiNi0.4Co0.2Mn0.4O2, LCO: LiCoO2, LMR-NCM: 0.5Li2MnO30.5LiNi0.
44Co0.25Mn0.31O2, HT: hydrothermal, SS: solid state. Data taken from Dunn et al. (2015a)
The materials required to produce the battery cells have as well high embodied
energies. Cast copper for example has 56–62 MJ/kg, primary nickel 159–175 MJ/
kg, primary cobalt 270–300 MJ/kg. In addition, mining processes yielding nickel,
copper and cobalt are known to have a large environmental footprint as they present
heavy production of SOx emissions and involve some sulfidic tailing (Hawkins
et al. 2013) which are usually linked to toxicity issues having repercussions in
humans and ecosystems (Beylot and Villeneuve 2017). Deeper in this matter,
Fig. 16.3 shows the contribution of the compounds required (in columns) to pro-
duce the different cathode materials to the cradle to gate embodied energy of the
cathode excluding the current collector. In the figure can be also noticed that,
besides NaOH, the compounds containing cobalt and nickel drive the embodied
energy for each cathode system (in rows).
16.2.3 The Impact of Recycling Activities
The Exergy framework as in Gutowski (2011) allows to illustrate how the necessary
efforts to recover a specific material change as products go through different life
cycle stages. In the context of thermodynamics, the exergy of a system is defined as
the maximum work that can be done by a system interacting within a reference
environment (Dincer and Rosen 2007) and is often used as a measure of quality for
material resources (Connelly and Koshland 1997). Exergy gauges the energy
Fig. 16.3 Cradle to gate embodied energy for the production of different cathodes materials. Data
taken from Dunn et al. (2015a)
Fig. 16.4 Changes in exergy values for materials in different life cycle stages
content of a material that may be used to perform a task (Connelly and Koshland
1997) or, in other words, gives a measure of potential available work embodied in a
material (Ayres et al. 1996). Figure 16.4 aims at depicting the concepts of material
separation and recycling schematically from the perspective of the exergy frame-
work for materials evaluation. The y-axis represents exergy values and the x-axis
represents the different phases of the life cycle of a material from its extraction at a
mine through its use phase while being mixed in a product until its re-purification
and the start of its second life.
Similar as presented by Gutowski (2011) and by Connelly and Koshland (1997),

the difference in exergy values between the points in “x” represent the minimum
work necessary to bring a material into an improved state. Improving the exergy
value of materials lifts the availability of doing work with them. A pure material has
an optimum level of exergy Epm (amount of exergy to be transferred to a material to
reach the required commercial quality) and is thus ready to be used in manufac-
turing. At this point the materials start a degradation phase resulting from different
mixing activities: (i) with other materials, to manufacture metal alloys, for example,
(ii) within a product and (iii) within a waste stream as the product enters the end of
its first life. Materials might experience as well degradation of their exergy value
due to reactions with the environment in which they operate. Corrosion and oxi-
dation are some examples of these as discussed by Gutowski (2011). The concept of
exergy removal as addressed by Connelly and Koshland (1997) gives a measure of
the degradation of a material using an standard energy scale. This means, exergy
removal gives an estimate of what has to be “returned” to a material in order to
reestablish its potential embodied available work. In a general sense, if the con-
centration of a specific material in an alloy, a product and a waste stream is low, the
efforts to recover this material are high.
Notice in Fig. 16.4, that recycling theoretically allows to restore the exergy
value of a material so that it is able to provide work again. In Fig. 16.4 we defined
this value as Esm (the exergy needed to be returned back to the material to restore it
to its optimum exergy value or Epm). High values of Esm implies not only costly
collection and recycling processes but also high efforts in terms of energy as well as
auxiliary materials required. As argued by Allwood et al. (2011), recovering
materials from multi-material products disposed in mixed waste streams is highly
challenging due to the high collection and separation costs. Consequently, recycling
processes are often limited to recover materials with a high market value. Research
from Dahmus and Gutowski (2007) founds in this regard that given a low mixture
complexity, products containing high amounts of valuable materials are currently
recycled opposed to products that present a higher mixture complexity and a higher
dilution of the material.
One direct consequence of this issue is represented in Fig. 16.4. Recycling
process will not necessarily return back all the exergy removed from a material
during the mixing and degradation stage. If we take a look back to Fig. 16.1, this
issue becomes important to consider when addressing the environmental implica-
tions of a recycling system. Pursuing an economic objective, a recycling process for
Material B, for instance, might recover the material with a different quality grade
(i.e. a different exergy value). While the process is able to recover material, the
recycled material will not necessarily displace the production of virgin Material B
as the potential embodied work in the recovered material may happen to not meet
the quality requirements of Material B. This is commonly known as open loop and
close loop recycling. Arguments on close loop recycling often make the assumption
that, contrary to open loop recycling systems, the recycled materials are used to
produce the same product from where they were recycled from. Although there is
no evidence to ascertain close loop recycling system are necessarily better than an
open loop system, there is a general hypothesis1 that recovering a material through
recycling and use it to produce the same product is environmentally better as this
guarantees avoiding the risk of downcycling (Geyer et al. 2015).
Regarding the information described in Fig. 16.4, different aspects are important
to mention in order to place the focus back on traction batteries. The restoration of
the exergy removed from the materials of lithium-ion batteries depends on the
different mixing events occurring during and after the manufacturing phase:
Battery systems mixing in a waste stream
On the one side, due to the large amounts of valuable materials such as aluminum,
copper, cobalt and nickel in a battery system, it is expected that recycling
lithium-ion batteries becomes increasingly economically attractive. Geopolitics
plays also an important role. If recycling is to be performed domestically, gov-
ernments are able to secure a growing stock of valuable materials to remain within
their territory. This is particularly important for materials such as cobalt and nickel
as the production of these materials and its refining is largely concentrated in few
countries (Schmidt et al. 2017), implying a risk of shortage due to political insta-
bility. In this regard, there is enough motivation to imagine that, similar to the
success of the recycling industry for lead batteries (Gaines 2014), a lithium-ion
battery recycling industry will find funding, investors and supporting policies to
emerge. An organized reverse supply chain backed by the establishment of col-
lection centers and the engagement of automotive manufacturers might decrease the
exergy removal implied by the dispersion of batteries in the society.
On the other side, as the quest towards achieving higher energy densities in
traction batteries accelerates, new materials are developed as reviewed by Placke
et al. (2017). This can result in heterogeneous streams of traction batteries. As a
consequence, the exergy removal value increases due to a more difficult separation.
Finally, collection and logistics might represent a challenge as batteries are heavy
and voluminous devices. An analysis of the current commercial electric vehicles
done by Grunditz and Thiringer (2016) shows that the weight of the traction battery
for around 40 models analyzed ranges between 145 and 620 kg.
Material mixing with other materials and material dilution within the battery
pack
Battery systems consist of four main components, as shown in Fig. 16.5. The
battery housing contains materials such as aluminum, steel and plastics. These
materials are mechanically joined. Although the disassembly process can be labor
and or time intensive, the materials recovered from the housing structure do not go
through an intensive degradation process caused by mixing. The modules are
1
ISO14044 for instance states that close-loop recycling, contrary to open loop recycling, does not
change the inherent properties of the material and that no changes in the inherent properties of the
material implies that the recycling process displaces the use of virgin material, as addressed in
Geyer et al. (2015).
Fig. 16.5 Change in separation effort within the material separation processes for lithium-ion
batteries
composed of housing materials (aluminum, steel and plastics) mechanically joined

but completely closed. In this case, a manual disassembly to separate housing
components and to gain access to the battery cells is very difficult. Destructive
mechanical processes are required in order to separate the housing components
from the cell materials. Finally, the electrode coating material is recovered as a
black mass containing all the components listed in Table 16.1 and Fig. 16.3. As
reported by Diekmann et al. (2017), the transition metals (NCM) and graphite can
be separated from the black mass through a hydrometallurgical process. However,
as they argued, the black mass must be homogeneous (i.e. not mixed with other
cathode materials such as LFP) and should be free of contaminants such as car-
bonates, copper and aluminum as these impurities have a negative effect on the
purity of the extracted material which then implies that the separation process must
be repeated several times. Material dilution is in this regard an important aspect. For
example, while it is technically possible to recover high grade lithium hydroxide
and lithium carbonate from the black mass of electrode coating, its high dilution
(just around 2% of the whole black mass) makes it economically unattractive which
might result in recovering lower quality lithium products.
Material separation processes for lithium-ion batteries consist of a combination
of different unit operations such as deactivation, mechanical treatment, hydromet-
allurgical and/or pyro-metallurgical processes (Hanisch et al. 2015). A detailed
description of current commercial processes can be consulted in the research done
by Hanisch et al. (2015). All these processes have a certain interaction with the
environment as they recover material and require energy. As reviewed by Hanisch
et al. (2015) the amount and quality of material recovered as well as the energy
needed in all of the processes are different from each other. We focus in this chapter
on the process developed within the LithoRec project. LithoRec had as an objective
the development of a recycling process that is able to recover the most variety and
quantity of materials from a stream of spent traction batteries as possible and aiming
at achieving a material quality grade that allows a close loop recycling. In this
regard, the analysis present in the following section offers important insights to
understand the relationship between recycling efforts, unit processes and material
composition for the case of traction batteries.
16.3 Environmental Characterization

of the LithoRec Process
In this section we describe the main process chain developed within the project
LithoRec from an environmental perspective. First, we present an energy and
material flow analysis based on data measured at the different equipment used. In a
second section, we present the results of the LCA performed to evaluate the process
developed.
16.3.1 Energy and Material Flow Modeling
The process chain developed for the recycling of lithium-ion traction batteries is
depicted in Fig. 16.6. A detailed description of the main technical aspects of every
unit process can be consulted throughout this book. Material and energy flows for
Fig. 16.6 Process chain developed in LithoRec. Based on Diekmann et al. (2017)
Fig. 16.7 Modeled energy and material flows from the LithoRec process
the recycling process allocated to one 346 kg battery pack are shown in Fig. 16.7
and the specific energy consumption profile for the respective unit processes are
shown in Fig. 16.8 and an energy portfolio analysis for the processes of the
demonstration plant is shown in figure is shown in Fig. 16.9.
The process starts with the deactivation of the battery. This step is meant to
reduce the electrochemical energy remaining in the battery and therefore mini-
mizing the risk of thermal runaways. A full deactivation of a spent battery can be
accomplished through discharging as well as through the application of thermal
treatments such as freezing and preheating (Hanisch et al. 2015). In LithoRec, a full
discharge of the battery pack was executed and values around 13 kWh.
Disassembly follows with a sequence of processes aiming at separate the battery
pack into its components. As seen in Chap. 6, the large number of components in a
battery pack and how these are connected to each other leads to a complex dis-
assembly procedure entailing a large number of steps with high requirements on
manual labor as shown by Herrmann et al. (2014). Automation and modern
human-robot collaboration in production processes bring in this regard opportuni-
ties to decrease dismantling times while reducing costs. One example of this
methodology applied to the recycling process of lithium-ion batteries is presented
by Wegener et al. (2014). A detailed description can be found in Chaps. 5 and 6. As
seen in Fig. 16.7, the disassembly step separates a mass stream of around 120 kg
including valuable materials such as steel, copper, aluminum approximately.
Most of these materials come from components like the battery housing and
cooling system as well as from the electrical cables that connect the modules. Other
components such as the BMU are also in this step separated but the further sepa-
ration and recycling of materials from these electronic components were not con-
sidered in the project as the required process is not different to current recycling
processes for electric and electronic waste.
The further mechanical treatment steps are thoroughly described in the research
by Diekmann et al. (2007) and Hanisch et al. (2015) and in Chaps. 7–13. Once the
battery pack has been disassembled and the housing components separated,
Fig. 16.8 Power operation profiles for the unit processes considered in LithoRec
Fig. 16.9 Energy portfolio classification for the mechanical treatment processes in LithoRec
the battery modules are crushed in an inert atmosphere to minimize the risk implied
by the presence flammable materials like the electrolyte. The crushing of the
modules was executed using a four shaft rotary shear that allows to dispense
modules and cells. The power of the shredder after the main switch has been turned
on reaches around 0.9 kW and with the motors ramp-up the power increases up to
around 1.6 kW where the machine enters its ready-for-production status. At point
number three in the energy consumption pattern of the shredder starts the value
creation process. At this point a battery module in the machine is crushed with a
peak of over 3.3 kW. After approximately 2 min the module has been completely
crushed and the power demand returns to its ready-for-production status.
The crushed materials are then dried in a furnace at temperatures ranging from
100 to 140 °C for approximately five hours with the objective of separating the
electrolyte. To hold the heat within this range, the power of the furnace was
measured to vary around the 2.8 kW. The vaporized solvents are collected by
condensation and might be combusted to recover thermal energy (Diekmann et al.
2017). From this process a total of 1 kg of electrolyte was separated for every 6 kg
of shredded modules fragments. As seen in Fig. 16.7, this represents around 30 kg
from the battery used as reference. Additionally, the HF generated after the
decomposition of the LiPF6 is then cleaned through a gas scrubbing process which,
as seen in Fig. 16.8, has a power demand above the 5 kW. This process requires
additionally around 0.45 kg of inert gas per kg of cell. This flow is not shown in
Fig. 16.7.
After the drying process and once the electrolyte has been removed, it is possible
to separate the rest of the components of the modules. The following separation
processes of two air classification processes executed with a zick zack sifter and one
sieving process as shown in Fig. 16.6. The first air classification is performed with a
relatively high wind speed for about 3 h. As noticed in Fig. 16.8, the fan demands
around 1.5 kW while running stationary during the first classification. The sieving
process separates the coating material from the rest of components. This process is
executed using a sieving machine that combines vibration and air classification for
approximately 1.3 h. As presented in Fig. 16.8, this process has a relatively high
power demand. The vibration motor was measured to demand a power of around
1.5 kW and with the ramp up of the fan, the overall power demand of the machine
during the separation reaches the 5 kW. The following step is the second air
classification process. This process runs for about 4.2 h and the power demand was
measured to be around the 250 W.
As seen in Fig. 16.7, both the first and the second air classification processes
allow to separate the big majority of modules and electrode materials from the
coating materials. The first classification step recovers the heavy pieces of materials
that are found in the mass of crushed modules. This includes aluminum from the
modules housing and the cell packaging, steel from the modules connection and
some copper from the conductive elements of the cells. The second air classification
is executed with a significantly lower air speed, which allows to separate the lighter
components from the coating materials such as pieces of the aluminum foil and the
separator. After the completion of the previously described mechanical treatment
steps an approximate amount of 83 kg of the so called black mass is separated. As
shown in Fig. 16.7, the total energy consumed for the separation of a complete
battery pack was estimated to be around 58.3 kWh.
As a final step, the black mass is treated in a hydrometallurgical procedure in
which graphite is recovered after a leaching process and the transition metals are
precipitated through pH manipulation. The lithium solution resulting is cleaned and
after its crystallization it is possible to recover battery grade lithium carbonate and
lithium hydroxide.
For the LithoRec process, an energy portfolio analysis is useful as it helps to
classify the different processes in a production chain regarding their leverage effect
on the absolute energy consumption per amount of throughput. The drying process
was identified to be a process that not only has a power demand above the average
power demand among the rest of the process, but also drying requires longer
processing time than the other process, which results in a high energy consumption.
Crushing and sieving are two processes that have lower operation times, on a per
battery system basis, however both processes have a high power demand. These
two process are considered to have a medium to high leverage effect on the final
energy consumption. Finally, the air classification processes are considered to have
a medium to low leverage effect on the energy consumption due to their generally
lower power demand.
16.3.2 Environmental Assessment of the LithoRec Process
Background of the LCA

As part of the LithoRec project, a LCA (Buchert and Slutter 2017) was executed in
order to evaluate the process from an environmental perspective.2 As mentioned
before, one of the main objectives of the project, was to design and develop a
recycling process for lithium-ion traction batteries that enable the recovery of the
majority of materials contained in a battery pack. This include all the transition
metals contained in the battery cells as well as the lithium compounds. Furthermore,
the process was planned to recover materials with a high purity grade so they might
be re-inserted in the material supply chain of traction batteries. The LCA performed
in the context of LithoRec, aimed at identifying all the potential environmental
improvement opportunities of the process developed.
Setup of the study
A first challenge was the definition of a generic battery system. As battery types,
materials and their weights vary significantly among the many commercially
available traction batteries, defining a generic composition was essential for the
analysis. This generic composition decided is shown in Table 16.1. The LCA only
included in the analysis NCM batteries as these where the only kind of batteries
processed in the project. The functional unit was decided to be the recycling of 1
ton of spent batteries having the composition of the generic battery presented in
Table 16.1. Following processes were included in the analysis: discharge of the
battery, dismantling, mechanical separation and the final hydrometallurgical pro-
cess. The foreground data used for the study were taken mainly from measurements
at laboratory and/or pilot plant scale. Information regarding the application of
background datasets and the modeling of the transportation is given in the LCA
report (Buchert and Slutter 2017). In line with Fig. 16.1, the LCA executed esti-
mated the avoided impact due to the displacement of primary material. Table 16.2
lists the main materials recovered from the processes together with the recycling
processed modelled as the proxy applied to estimate the avoided impact. Additional
recycling processes that take place outside the scope of LithoRec were considered
as well in study. In this regard, a production process of secondary aluminum and
copper and the production of recycled plastics was taken from generic datasets.
For the BMU, the study adapted a recycling process for printed wiring boards in
order to evaluate the impact of recycling the BMU of the battery and calculated the
avoided impacts based on the displacement of the main metals contained in the
BMU such as copper, nickel, gold and silver. More detailed information regarding
2
The development of this section is mostly based on the report published by the German institute
for applied ecology (Öko-institut) (Buchert and Slutter 2017). The Öko-institut was enrolled as an
associated partner for the execution of an LCA and in charge of the coordination of the umbrella
group LCA.
Table 16.2 Recycled material, recycling process and avoided primary material. Data taken from
Buchert and Sutter (2017)
Material Recycling process Avoided process
Aluminum Secondary aluminum (processing and Production of primary
smelting) aluminum in European mix
BMS Recycling of printed wiring boards composed Primary extraction of the
of copper, nickel, gold and silver metals
Copper Secondary copper Primary copper production
Plastics Processing of plastics, PA Primary PA production
Cobalt Primary data from LithoRec Primary cobalt production
sulfate
Nickel Primary data from LithoRec Primary nickel production
sulfate
Manganese Primary data from LithoRec Primary manganese
sulfate production
Lithium Primary data from LithoRec Primary lithium hydroxide
hydroxide production
the use of specific processes and proxies from the databases to estimate avoided
impacts can be consulted in (Buchert and Slutter 2017).
Results and analysis
The results of the study are presented in Fig. 16.10. The absolute results are
indicated for both the recycling and the potential avoided impact. The results are
aggregated in three steps: dismantling, mechanical treatment and hydrometallur-
gical treatment. Figure 16.10 shows directly the comparison of the recycling
impacts (in red) against the potential avoided impact (in blue) for six impact cat-
egories: global warming potential (GWP), cumulative energy demand (CED),
eutrophication potential (EP), acidification potential (AP), photochemical ozone
creation potential (POCP) and abiotic depletion potential (ADP). The results are
normalized to the absolute maximum impact among both the caused and the
avoided impact, so it is possible to understand the contribution of the processing
steps to each of the categories while keeping the comparison of avoided versus
caused impacts in perspective. For more information regarding the contribution of
the specific elementary flows to each of the absolute results please refer to the LCA
report (Buchert and Slutter 2017).
The avoided impact is significantly higher in each of the environmental impact
categories. The GWP and CED impacts caused by the recycling process reach, for
instance, around half the potential impact avoided by the process.
For these two impact categories, the avoided burden is mostly driven by the
dismantling phase. On the same time, the impact caused by the dismantling process
itself is relatively low. From Figs. 16.2 and 16.7 we see that, although the dis-
mantling phase implies a relatively low environmental effort, it recovers around
34% of the battery by weight. This includes over 60 kg of aluminum which as seen
Fig. 16.10 Results of the Life Cycle Impact Assessment performed within the LithoRec project
in Fig. 16.2 has high embodied energy. In the same way, the mechanical treatment
presents a positive balance.
Although the process demands high amounts of energy to separate the active
materials from the rest of the housing and conducting elements and the separator,
the process recovers as well large quantities of copper and aluminum. The largest
part of the GWP is produced during the hydrometallurgical process. This is due to
the high requirements on electricity and auxiliary energy carriers as reported in
(Buchert and Slutter 2017). Some impact is in this phase avoided as the process is
able to recover important metals like cobalt and nickel. However, the hydromet-
allurgical step ends with a negative balance as it creates more impact than what a
potential recovery of lithium can possibly avoid.
The acidification potential avoided by the process strongly driven by the
hydrometallurgical step. This is mainly due to the recovery of nickel as this dis-
places primary nickel which is produced from the pyrometallurgical processing of
sulfide ores. This process is known for generating high amounts of SO2 emissions,
which accordingly, represents around 90% of the emissions contributing to avoid
AP impact in the LithoRec process. Avoided POCP is largely driven by the
recovery of lithium which in for this impact category impulses a positive balance.
Finally, the avoided ADP is determined largely by the recovery of copper from the
mechanical treatment and by the extraction of components like gold and silver
separated from the BMU.
16.4 Discussion and Conclusions
Recycling has an outstanding role in tackling and diminishing the life cycle envi-
ronmental impact of traction batteries and, therefore, of the rising electric mobility
industry. Recycling might not only contribute to displace the production of energy
intensive primary materials, but is also a solution to a continuously growing waste
stream. Nevertheless, recycling processes for lithium-ion batteries shall be imple-
mented considering a broader environmental scope. The project LithoRec and the
LCA done for its evaluation has a prospective character. LithoRec intended to
develop processes, that are expected to be commercially available in the future and
therefore most of the insights gained are originated from laboratory and pilot plant
data. The results of the project contribute to broaden the understanding of the
leverage effect of the recycling of traction batteries to reduce the impact of trans-
portation, a key issue in the quest towards sustainability. In this regard, following
points are worth discussing:
Quality matters
As mentioned throughout the chapter, one of the main objectives of LithoRec was
to create a process that is able to recover the majority of the materials contained in a
traction battery system, with the condition that these materials meet the purity grade
of primary battery materials. With this in mind, background datasets corresponding
to the production of high quality metals from the housing (aluminum, steel and
copper) recovered in the dismantling step where used for the LCA resulting in high
values of avoided impact.
This is important to considered as the largest amount of GHG emissions avoided
through the displacement of primary material production processes was found to be
driven by the materials recovered during the dismantling phase. The assumption,
that aluminum, steel and copper will be recovered in their higher quality is highly
uncertain and unrealistic. The effect of producing lower quality secondary materials
out the recovered materials in LithoRec to the total environmental impact of the
process should be evaluated in further research. As an example of this, through the
application of an alternative hydrometallurgical process, a lower non-battery-grade
lithium compound was recovered. The process however was measured to consume
much less energy and auxiliary materials than the actual process intended to recover
battery grade lithium reducing on a 50% the GWP impact produced by the
hydrometallurgical step.
This issue raises additional challenging questions to feed the debate regarding
the criticality status of lithium. As presented by Christmann et al. (2015), while
more than half of the lithium produced worldwide goes to dispersive uses from
where it cannot be recovered, recycling lithium from batteries is expected to remain
economically unappealing. Parallel to an environmental assessment of the recycling

processes for traction batteries, its impact on criticality issues should be explored
thoroughly.
Geography matters
Different issues are important to address in this regard. Infrastructure and regulation
is the first factor. A recycling industry for traction batteries is necessarily regulated
by directives defining limits to material recovery efficiency to the recycling com-
panies. In addition, this recycling industries are accompanied by large infrastruc-
tural developments that ease the collection and transportation processes of spent
batteries. This is likely to be rigorously implemented in developed countries, but is
definitively an issue to considered when planning recycling systems worldwide. On
second factor to consider is the effect of regional electricity mixes as the amount of
energy consumed during the recycling process is high enough to lead to increments
on the amount of emissions produced if the process is powered with electricity with
a higher environmental footprint. A third factor to consider is the fate of the metals
separated from the battery system as depending on the available regional recycling
infrastructure these materials might end up downgraded to a lower quality.
Sorting and labelling matters
As experienced in LithoRec, developing a process that is able to recover such
amount of high quality material was in a sense only possible because it was
assumed that all the batteries are composed of the same active components. As seen
in Placke et al. (2017), the variety commercial batteries is relatively large and
prompt to keep evolving. This of course is an extremely challenging issue that has
to be considered because, as mentioned, mixed waste streams will increase the
effort required to separate the materials involved.
References
Allwood JM, Ashby MF, Gutowski TG, Worrell E (2011) Material efficiency: a white paper.
Resour Conserv Recycl 55(3):362–381. https://doi.org/10.1016/j.resconrec.2010.11.002
Ashby MF (2013) Materials and the environment eco-informed material choice. Mater Environ.
https://doi.org/10.1016/B978-0-12-385971-6.00016-6
Ayres RU, Ayres LW, Martinas K (1996) Eco-thermodynamics: exergy and life cycle analysis (96/
04/EPS). Fontainebleau, France
Beylot A, Villeneuve J (2017) Accounting for the environmental impacts of sulfidic tailings
storage in the life cycle assessment of copper production: a case study. J Cleaner Prod
153:139–45 (Elsevier Ltd). doi:https://doi.org/10.1016/j.jclepro.2017.03.129
Buchert M, Sutter J (2017) Aktualisierte Ökobilanzen Zum Recyclingverfahren LithoRec II Für
Lithium-Ionen-Batterien (Stand 09/2016). Darmstadt, Berlin
Christmann P, Gloaguen E, Labbé J-F, Melleton J, Piantone P (2015) Global lithium resources and
sustainability issues. Lithium Process Chem 1–40. doi:https://doi.org/10.1016/B978-0-12-
801417-2.00001-3
Connelly L, Koshland CP (1997) Two aspects of consumption: using an exergy-based measure of

degradation to advance the theory and implementation of industrial ecology. Resour Conserv
Recycl 19(3):199–217. https://doi.org/10.1016/S0921-3449(96)01180-9
Dahmus JB, Gutowski TG (2007) What gets recycled: an information theory based model for
product recycling. Environ Sci Technol 41(21):7543–7550. https://doi.org/10.1021/es062254b
Dincer I, Rosen MA (2007) Exergy: energy, environment and sustainable development. Newnes,
2012
Dunn JB, Gaines L, Kelly JC, James C, Gallagher KG (2015a) The significance of Li-Ion batteries
in electric vehicle life-cycle energy and emissions and recycling’s role in its reduction. Energy
Environ Sci 8(1):158–68 (Royal Society of Chemistry). doi:https://doi.org/10.1039/
C4EE03029J
Dunn JB, Gaines L, Kelly JC, James C, Gallagher KG (2015b) The significance of Li-ion batteries
in electric vehicle life-cycle energy and emissions and recycling’s role in its reduction. Energy
Environ Sci 8(1):158–68. doi:https://doi.org/10.1039/C4EE03029J
Ellingsen LA-W, Wick GM-B, Singh B, Srivastava AK, Valøen LO, Strømman AH (2013) Life
cycle assessment of a lithium-ion battery vehicle pack. J Ind Ecol 18(1):1–48. https://doi.org/
10.1111/jiec.12072
European Commission (2014) Directive 2006/66/EC of the European Parliament and of the
Council of 6 September 2006 on batteries and accumulators and waste batteries and
accumulators and repealing Directive 91/157/EEC (Text with EEA relevance)
Gaines L (2014) The future of automotive lithium-ion battery recycling: charting a sustainable
course. Sustain Mater Technol 1–2:2–7 (Elsevier B.V.). doi:https://doi.org/10.1016/j.susmat.
2014.10.001
Geyer R, Kuczenski B, Zink T, Henderson A (2015) Common misconceptions about recycling.
J Ind Ecol 20(5). doi:https://doi.org/10.1111/jiec.12355
Granta Design Limited (2013) CES selector software. Cambridge
Grunditz EA, Thiringer T (2016) Performance analysis of current BEVs based on a comprehensive
review of specifications. IEEE Trans Transport Elect 2(3):270–289
Gutowski TG (2011) Material separation and recycling. In: Bakshi BR, Gutowski TG, Sekulic DP
(eds) Thermodynamics and the destruction of resources. Cambridge University Press (doi:sdf)
Gutowski, TG, Sahni S, Allwood JM, Ashby MF, Worrell E (2013) The energy required to
produce materials: constraints on energy-intensity improvements, parameters of demand.
Philoso Trans. Series A, Math, Phys Eng Sci 371(1986):20120003. doi:https://doi.org/10.1098/
rsta.2012.0003
Hanisch C, Diekmann J, Stieger A, Haselrieder W, Kwade A (2015a) Recycling of lithium-ion
batteries. Handb Clean Energy Syst 2011:1–24. https://doi.org/10.1002/9781118991978.
hces221
Hanisch C, Loellhoeffel T, Diekmann J, Markley KJ, Wolfgang H, Kwade A (2015b) Recycling of
lithium-ion batteries: a novel method to separate coating and foil of electrodes. J Cleaner Prod
108:301–11 (Elsevier Ltd). doi:https://doi.org/10.1016/j.jclepro.2015.08.026
Hawkins TR, Singh B, Majeau-Bettez G, Strømman AH (2013) Comparative environmental life
cycle assessment of conventional and electric vehicles. J Ind Ecol 17(1):53–64. https://doi.org/
10.1111/j.1530-9290.2012.00532.x
Herrmann C, Raatz A, Andrew S, Schmitt J (2014) Scenario-based development of disassembly
systems for automotive lithium ion battery systems. Adv Mater Res 907:391–401. doi:http://
doi.org/10.4028/www.scientific.net/AMR.907.391
Kahhat R, Williams E (2012) Materials flow analysis of e-waste: domestic flows and exports of
used computers from the United States. Resour, Conserv Recycl 67(May 2015):67–74. doi:
https://doi.org/10.1016/j.resconrec.2012.07.008
Perkins DN, Drisse MNB, Nxele T, Sly PD (2014) E-waste: a global hazard. Ann Glob Health 80
(4):286–95 (Elsevier Inc). doi:https://doi.org/10.1016/j.aogh.2014.10.001
Placke T, Kloepsch R, Dühnen S, Winter M (2017) Lithium ion, lithium metal, and alternative
rechargeable battery technologies: the odyssey for high energy density. J Solid State
Electrochem 1–26. doi:https://doi.org/10.1007/S10008-017-3610-7
Schmidt T, Buchert M, Schebek L (2017) Investigation of the primary production routes of nickel and
cobalt products used for Li-ion batteries. Resour, Conserv Recycl 112:107–22 (Elsevier B.V.).
doi:https://doi.org/10.1016/j.resconrec.2017.04.017
batteries using the example of the Audi Q5 hybrid system. Procedia CIRP 23(C):155–60. doi:
https://doi.org/10.1016/j.procir.2014.10.098

Recycling of Lithium-Ion Batteries Libro

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Recycling of Lithium-Ion Batteries Libro

Uploaded by

Copyright:

Available Formats

Sustainable Production, Life Cycle Engineering and Management

Series Editors: Christoph Herrmann, Sami Kara

More information about this series at http://www.springer.com/series/10615

ISSN 2194-0541 ISSN 2194-055X (electronic)

© Springer International Publishing AG 2018

Printed on acid-free paper

This Springer imprint is published by Springer Nature

Private transportation is going through a dramatic technological transformation.

consumption compared to the current commercial pyrometallurgical processes. This

Lithium-ion batteries are increasingly applied in electrical vehicles, stationary

Braunschweig, Germany Arno Kwade

9 Separation of the Electrolyte—Solvent Extraction . . . . . . . . . . . . . 155

μCT Micro-computer tomography

EMC Ethyl methyl carbonate

PLC Programmable logic control

Fig. 1.9 Electrode fabrication process . . . . . . . . . . . . . . . . . . . . . . . . . .. 16

Fig. 3.11 Concentrations of hydrogen fluoride (HF) during nail

Fig. 4.21 Overdischarge current/voltage characteristic curves for VW

Fig. 6.13 Experimental accuracy analysis of direct positioning control

Fig. 10.2 Compositions of the recovered electrolyte from commercial

Fig. 11.7 (a) Location of temperature peaks along the adsorption

Fig. 13.16 Raman spectrum of the thermally treated non-extracted

Fig. 13.20 Constant current discharge capacities and Coulombic

Fig. 15.1 Conceptual model with system boundaries for economic

Fig. 16.9 Energy portfolio classiﬁcation for the mechanical treatment

Table 4.1 Battery cells, modules, systems investigated at the time

Table 13.6 Overview of the discharge capacities and associated

Sergej Rothermel, Martin Winter and Sascha Nowak

S. Rothermel M. Winter S. Nowak (&)

© Springer International Publishing AG 2018 1

1.2 Introduction into Lithium-Ion Batteries

1.2.1 Liquid Nonaqueous Electrolytes

In general, the formulation of the electrolytes used in LIBs involves a mixture of

well as open-chain carbonates such as dimethyl carbonate (DMC), ethyl methyl

1.2.2 Negative Electrodes in Rechargeable Lithium

1.2.3 Positive Electrodes

Following the approach of mixed oxides the mixture of LiNi1/3Co1/3Mn1/3O2 (NCM

1.3 Industrial Production of Lithium-Ion Cells

1.3.1 Cell Design

Fig. 1.6 Cutaway view of a cylindrical 18650 cell

Usually, the occurrence of increased cell pressure is the result of overheating or

Fig. 1.7 Cutaway view of a wound prismatic cell

1.3.2 Electrode Fabrication

Fig. 1.8 Cutaway view of a stacked pouch cell

1.3.3 Cylindrical Cell Fabrication

Fig. 1.9 Electrode fabrication process

Fig. 1.10 Schematic illustration of the cylindrical cell fabrication process

1.3.4 Prismatic and Pouch Cell Fabrication

1.3.5 Cell Formation

electrolyte distribution between the electrodes. The initial charge is performed at

1.3.6 Battery System Manufacturing

1.4 Recycling of Lithium-Ion Batteries

1.4.2 Overview of Selected Lithium-Ion Battery Recycling

The generic term “Lithium-Ion” to be found on labels of LIBs refers to a broad

single-furnace pyrometallurgical process with subsequent hydrometallurgical

Cheret D, Santen S (2005) Battery recycling. United States Patent

Kathryn CS (2008) Solvent extraction in the hydrometallurgical processing and puriﬁcation of

Jan Diekmann, Sergej Rothermel, Sascha Nowak and Arno Kwade

2.1 Objectives and Results of LithoRec

J. Diekmann (&) A. Kwade

© Springer International Publishing AG 2018 33

2.2 Objectives and Project Progression of LithoRec II

2.3 The LithoRec Process Chain

The LithoRec process combines electrical, mechanical, mild thermal, and