Electrochemistry Communications 50 (2015) 11–14

Contents lists available at ScienceDirect

Electrochemistry Communications
journal homepage: www.elsevier.com/locate/elecom

Short communication

Comparison of the growth of lithium filaments and dendrites under

different conditions
Jens Steiger a, Gunther Richter b, Moritz Wenk a, Dominik Kramer c,⁎, Reiner Mönig a,c
a
Karlsruhe Institute of Technology, Institute for Applied Materials, Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany
b
Max Planck Institute for Intelligent Systems, Heisenbergstr. 3, 70569 Stuttgart, Germany
c
Helmholtz Institute Ulm, Helmholtzstraße 11, 89081 Ulm, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Although lithium dendrites have important implications on the safety and reliability of lithium-based batteries,
Received 7 October 2014 an understanding of their growth mechanism is still lacking. Electron microscopy and in situ light microscopy
Received in revised form 24 October 2014 were used to investigate the growth of lithium filaments and dendrites. Lithium was deposited by thermal evap-
Accepted 3 November 2014
oration in vacuum as well as electrochemically using two different electrolytes. Filaments grow in all three cases
Available online 9 November 2014
by an insertion mechanism, suggesting that neither a solid-electrolyte interphase (SEI) nor electrolytes are
Keywords:
required to form lithium filaments. The role of the electrolyte becomes apparent in the detailed morphology of
Lithium metal the deposits. These findings indicate that instead of ionic transport and electrochemistry, lithium diffusion and
Battery crystallization are key processes which need to be modified in order to control the growth of lithium dendrites.
Electrodeposition © 2014 Elsevier B.V. All rights reserved.
Dendritic growth
Growth mechanisms
Light microscopy

1. Introduction solid/gel electrolytes. More fundamental approaches include tuning

Metallic battery materials have gained considerable attention
recently as they enable extreme volumetric and gravimetric energy
densities [1]. Especially the lightest metal – lithium – will be the
anode of choice for many next generation batteries such as lithium–air
and lithium–sulfur if safe and reliable recharging can be achieved [2].
Unfortunately, metals – especially lithium [2] – suffer from the phenom-
enon of dendritic growth during charging. Dendrites are structures with
the shape of needles or bushes that can cause severe problems in batte-
ries such as poor cycling performance and short circuits. Lithium den-
drites prevent a widespread commercialization of lithium metal
anodes in rechargeable batteries and afflict today's lithium-ion systems
under adverse charging conditions [3], i.e. low temperature and/or high
charge rates [3,4]. This was shown in the investigations of the Boeing
Dreamliner groundings in 2013, where dendrite formation has come
under suspicion [5] of having caused short-circuits and ignition of
lithium-ion batteries.
Strong efforts have been dedicated to mitigate this issue [2] since
its first observation in the 1960s [6]. Attempts that were somewhat
successful used mechanical suppression by optimized separators and
⁎ Corresponding author. Tel.: +49 721 608 248 94.
E-mail address: dominik.kramer@kit.edu (D. Kramer).
the solid electrolyte interphase (SEI), stirring the electrolyte or pulsed
charging. These countermeasures have beneficial effects on the charg-
ing behavior, but so far no attempt has succeeded in tackling both the
problem of efficiency and the danger of short circuits. Consequently, it
was tried to understand the origins of this phenomenon. However, the
varieties of published models are often fundamentally contradicting,
e.g. it is still under debate where lithium atoms are incorporated into
the structures. Models based on concentration gradients in the electro-
lyte, electrical field enhancement at protrusions or SEI inhomogeneity
predict that lithium atoms aggregate preferably at the tip. In the
whisker-based model of Yamaki et al. [7] atoms are added at the base
of needle-like structures in the first growth stage, followed by nucle-
ation of spheres at tips and kinks. Some difficulties in understanding
lithium dendrite growth are inherent to lithium which shows a very
weak interaction with electrons and X-rays. It is challenging to obtain
high resolution in situ images as the electronically nonconducting
electrolyte is opaque for electrons. In our recent in situ study by light mi-
croscopy we showed that lithium atoms can be added to a dendrite at
the base, the tip and into kinks [8,9]. This unexpected behavior
was not predicted by previous models of dendritic growth. Hence, we
suggested [8,9] that lithium atoms are inserted into the metal lattice
of dendrites or filaments at crystalline defects, in particular at interfaces
such as high angle grain boundaries (e.g. at kinks, cf. Fig. 1) or the inter-
face at the substrate.

http://dx.doi.org/10.1016/j.elecom.2014.11.002
1388-2481/© 2014 Elsevier B.V. All rights reserved.
12 J. Steiger et al. / Electrochemistry Communications 50 (2015) 11–14
LP30
a)
LiTFSI
c) d)
Fig. 1. Image sequences of the growth of lithium filaments in the shape of a loop (marked with parallel white lines) in LP30 (above) and LiTFSI (below). Both filaments elongate by addition
of new segments (marked in blue) without elongation of the surrounding segments (white). Hence, the segments must grow by atom insertion into the lattice of the filament, probably at
kinks. The arrows mark spheres that were plated at the side of a filament in LiTFSI. Image b) was taken 60 s after a). Image d) was taken 30 s and e) 45 s after c). Similar observations were
made on filaments. Changes in segment length can be detected more easily on loops.
2. Experimental information
We used an in situ optical cell based on a polymer frame that
connects two glass plates to allow for the observations. The details
of the setup can be found in [8]. LP30 (1 M lithium hexaflu-
orophosphate LiPF6 in EC/DMC 1:1) and 1 M LiTFSI (lithium
bis(trifluoromethanesulfonyl)imide) in DOL/DME (1,3-dioxolane/1,2-
dimethoxyethane) were used as electrolytes. PVD (physical vapor
deposition) was performed by thermal evaporation of lithium from a
tungsten boat inside a commercial thin film deposition system (NANO
36, Kurt J. Lesker). Different commercial foil substrates (battery grade
copper, tungsten and tantalum) were used as-received. Polycrystalline
copper substrates (Alfa Aesar Puratronic 99,999%) were prepared by
several polishing steps and subsequent vibratory polishing to achieve
smooth surfaces. Traces of oxides were removed by a heat treatment
in a vacuum at small partial pressures of forming gas (5% H2 und 95%
Ar) at 250 °C for 15 min. The nominal thickness of the lithium and the
deposition rate were monitored by a quartz crystal microbalance. Depo-
sition was performed at room temperature; the nominal deposition rate
was kept around 1 nm s−1 (which is equivalent to a current density of
0.74 mA cm−2) for 200 s. The samples were tilted to 0°, 45° and 65°.
Needles were found on all substrates with a tendency of increasing
density with increasing tilt angle. The residual gas pressure before
deposition was 3.3 · 10−6 mbar.
3. Results and discussion
To investigate the mechanisms of filament growth and to what ex-
tent they are affected by the chemical composition of the electrolyte,
we compared dendrite formation in two chemically very different elec-
trolytes, the widespread LP30 (1 M LiPF6 in EC/DMC) and a 1 M LiTFSI
solution in DOL/DME using the electrochemical cell described in [8]. In
both electrolytes lithium exhibits the tendency to grow needles and
b) 60 s
30 s e) 45 s
dendrites. As observed before [8,9], lithium filament growth can occur
by insertion at kinks (shown in Fig. 1), at the substrate interface and
by growth at the tip. The electrolytes have comparable ionic conductiv-
ities (11 mS cm−1 for LP30 [10] and 15 mS cm−1 for LiTFSI [11]), which
implies that morphological differences are related to the different SEI
compositions (containing e.g. LixPFy in LP30 [12] and Li2SO3 and Li2S
in LiTFSI [13]). Characteristic differences in the growth morphology
were identified: In contrast to LP30, spheres (marked by arrows in
Fig. 1) are distributed along the filaments in LiTFSI. In LP30, needles
can be elongated by insertion at the tip whereas in LiTFSI, the only
growth observed at the tip was the deposition of lithium spheres in
the tip area. Besides filament growth, also bushes or mossy lithium
can form (Fig. 2). Here, strong differences were found: While LP30
shows strongly branched bushes where the branches resemble the
filaments [9], electrodeposition in the LiTFSI exclusively leads to more
compact agglomerates containing spheres. Fig. 2 B is an extreme exam-
ple of such a structure not showing any elongated segments (as the ones
in the lower part of Fig. 1).
These differences indicate that the SEI has a strong influence on the
growth morphology as reported before [14,15], but filamentary growth
was found for both electrolytes which raises the question whether it
depends on the composition of the SEI. To fully exclude electrochemical
effects, namely electric fields, concentration gradients and the SEI,
deposition was additionally performed by thermal evaporation (PVD).
Fig. 3 shows SEM images of lithium needles on Cu grown by PVD in
comparison with those obtained by electrodeposition. The deposits
look very similar: Both contain lithium needles with a length in the
micrometer range and a diameter of ca. 0.1–0.2 μm. PVD filaments –
which we observed in all PVD runs on different substrates – can have
kinks as the electrodeposited ones.
The strong resemblance of PVD and electrodeposited filaments as
exemplified in Fig. 3 indicates that the underlying mechanism could
be controlled by the same insertion mechanism. As a consequence,
 J. Steiger et al. / Electrochemistry Communications 50 (2015) 11–14
a)
Fig. 2. SEM images of bush-like structures electrochemically grown on tungsten substrates. a) grown in LP30, b) grown in LiTFSI.
electrochemical needle growth would not only be independent of SEI
composition as suggested by the comparison of the two electrolytes
above, but would not require a SEI at all. Accordingly, the tendency to
form needle-like deposits would be an inherent property of metallic
lithium. In this context lithium is no exception from other materials
forming whiskers: During vapor deposition, the growth of whiskers,
e.g. of copper or silicon, is a known phenomenon; it can be a useful
fabrication route for nanostructures [16]. It occurs for deposition at
substrate temperatures above 65% of the melting point of the deposit
[17]. Nanoscale whiskers have been observed for many metallic and a
few inorganic materials on different substrates [16,17]. This phenome-
non is not well understood but experimental evidence [16,17] supports
the following description: During PVD, atoms are “raining” down
homogenously onto the substrate which should result in an even
layer. Hence, any protrusion in the deposit must be a result of the mobil-
ity and the rearrangement of adatoms on the substrate. On the
substrate, the wetting properties vary, and a sufficiently high mobility
of adatoms leads to the formation of agglomerates in the regions of
the lowest interface energy. Due to the elevated temperatures and the
associated high mobilities, crystalline configurations with low energies
develop adopting the form of Wulff constructions [18] with the differ-
ence of an interface with the substrate at a location where wetting is
favored. The non-wetting regions in the vicinity of the crystallites
supply atoms for the growth. The interface of the crystallite with the
substrate is assumed to be the defect where adatoms are inserted and
the facetted crystal is pushed forward to form an elongated whisker.
Usually, surface treatments (e.g. carbon coating [17]) are needed to
adjust the wetting properties to obtain whiskers; however, in the case
of lithium, we observe them without pretreatment. The formation
of needles in the PVD process described here is based on i) highly
mobile adatoms so that low energy configurations can develop and
ii) persisting defects that allow for enhanced insertion. The similarity
a)
Fig. 3. SEM images of lithium needles grown on copper substrates at room temperature. Image a) shows needles grown electrochemically in LP30 electrolyte; image b) shows needles
grown by PVD in vacuum. In a), a SEI is present on the Cu surface.
13
b)
in the morphologies in Fig. 3 suggests that these arguments also apply
for the electrochemical deposition. In our previous publications [8,9],
we showed that mechanisms like concentration gradients, radial diffu-
sion, SEI cracking or electric fields are not the main causes of needle
growth. However, if these effects have only a minor impact, there is
no explanation for the inhomogeneous deposition as there is no mech-
anism directing the lithium ions in the electrolyte towards the insertion
sites. Hence, it seems plausible that in electrodeposition, adatoms are
first adsorbed at the entire electrode surface but remain mobile. Defect
sites then act as sinks which form the initial needles. A high mobility of
lithium adatoms is plausible due to the low melting point of lithium
(Tm = 453.7 K, i.e. RT ≈ 65% Tm). The suggested adatom mobility mech-
anism also challenges the existence of the usual critical current density
which needs to be exceeded to trigger dendritic growth [19,20] and
which is related to the ionic transport in solution [20].
Although we observed kinked, needle-like deposits during PVD of
lithium, we did not find any bush-like deposits. This strongly suggests
that bush growth is a phenomenon associated with electrodeposition.
As we have shown in [9], the multiplication of defects is needed to
switch from needle to bush growth. It may be assumed that the SEI is ac-
tive in defect formation and multiplication. Different SEI compositions
lead to differences in thickness, electrical resistance, mechanical proper-
ties (e.g. compliance or fracture toughness), interface energies and
wetting characteristics with lithium. Therefore differing SEI composi-
tions can easily explain different morphologies observed for the two
electrolytes in Figs. 1 and 2. The listed parameters of the SEI will affect
lithium insertion into the growing structure and local variations of
these parameters on a growing structure will define sites of accelerated
and retarded lithium insertion. For example, cracking and incorporation
of parts of the SEI layer into the lithium crystal may create further crys-
talline defects. Such an incorporation of SEI into lithium at dendrite
locations has been observed [21].
b)
14 J. Steiger et al. / Electrochemistry Communications 50 (2015) 11–14
4. Conclusions
We investigated the mechanisms behind one-dimensional needle
growth and three-dimensional bush growth. The concepts presented
here are of general nature and not only specific to lithium; it is plausible
that they are also valid for the RT electrodeposition of materials with
similar melting points, including sodium and tin also relevant for elec-
trochemical energy storage. Our results show that needle growth is an
inherent tendency of the metal. It combines a high atomic mobility
caused by the low melting point with the insertion at defect sites, in
particular, at the substrate–metal interface. This mechanism is not
depending on the deposition technique or the presence of a surface/
interface layer. Bush growth and the resulting deposits were found to
be a phenomenon controlled by the SEI. The results show that the SEI
is the major factor determining the morphology of the deposits. Physical
and electrochemical growth modes are superimposed during electro-
chemical deposition, with the needle growth being the precursor of
bush growth [9]. Conventional approaches to avoid dendrites seem
insufficient, since needle growth is an inherent property of lithium.
Such approaches include variations of ionic transport conditions (e.g.
charge rate), and SEI modifications (e.g. optimized mechanical proper-
ties). In order to attack the dendrite problem at its root, we suggest
reducing the mobility of lithium atoms. Blocking lithium diffusion on
the substrate including its SEI should prevent the formation of filaments
which are precursors of dendrites.
Conflict of interest
There is no conflict of interest with any financial organization re-
garding the material discussed in the manuscript.
Acknowledgments
We thank Christoph Weidmann and Sven Leuthner from BELLA for
supplying electrolytes. This work was supported by the Helmholtz
Initiative for Mobile/Stationary Energy Storage Systems.
References
[1] H. Kim, G. Jeong, Y.-U. Kim, J.-H. Kim, C.-M. Park, H.-J. Sohn, Metallic anodes for next
generation secondary batteries, Chem. Soc. Rev. 42 (2013) 9011–9034, http://dx.
doi.org/10.1039/c3cs60177c.
[2] W. Xu, J. Wang, F. Ding, X. Chen, E. Nasybulin, Y. Zhang, et al., Lithium metal anodes
for rechargeable batteries, Energy Environ. Sci. 7 (2014) 513–537, http://dx.doi.org/
10.1039/c3ee40795k.
[3] Z. Li, J. Huang, B. Yann Liaw, V. Metzler, J. Zhang, A review of lithium deposition in
lithium-ion and lithium metal secondary batteries, J. Power Sources 254 (2014)
168–182, http://dx.doi.org/10.1016/j.jpowsour.2013.12.099.
[4] M. Petzl, M.A. Danzer, Nondestructive detection, characterization, and quantification
of lithium plating in commercial lithium-ion batteries, J. Power Sources 254 (2014)
80–87, http://dx.doi.org/10.1016/j.jpowsour.2013.12.060.
[5] J. Ostrower, A. Pasztor, Microscopic 'dendrites' a focus in Boeing
Dreamliner probe, Wall Str. J. (2013) (http://online.wsj.com/article/
SB10001424127887324880504578298673566960476.html (accessed June
5, 2013)).
[6] Z. Stachurski, Rechargeable Current-Generating Electrochemical System with
Wiper Means, 3 (440) (1969) 098.
[7] J. Yamaki, S. Tobishima, K. Hayashi, K. Saito, Y. Nemoto, M. Arakawa, A consideration
of the morphology of electrochemically deposited lithium in an organic electrolyte,
J. Power Sources 74 (1998) 219–227, http://dx.doi.org/10.1016/S0378-7753(98)
00067-6.
[8] J. Steiger, D. Kramer, R. Mönig, Mechanisms of dendritic growth investigated by
in situ light microscopy during electrodeposition and dissolution of lithium, J.
Power Sources 261 (2014) 112–119, http://dx.doi.org/10.1016/j.jpowsour.
2014.03.029.
[9] J. Steiger, D. Kramer, R. Mönig, Microscopic observations of the formation, growth
and shrinkage of lithium moss during electrodeposition and dissolution,
Electrochim. Acta 136 (2014) 529–536, http://dx.doi.org/10.1016/j.electacta.2014.
05.120.
[10] Merck KGaA, Merck SelectiLyte™ — Materials for Li-Ion Batteries and Double Layer
Capacitors, 2009.
[11] L. Suo, Y.-S. Hu, H. Li, M. Armand, L. Chen, A new class of solvent-in-salt electrolyte
for high-energy rechargeable metallic lithium batteries, Nat. Commun. 4 (2013)
1481, http://dx.doi.org/10.1038/ncomms2513.
[12] D. Aurbach, Review of selected electrode–solution interactions which determine the
performance of Li and Li ion batteries, J. Power Sources 89 (2000) 206–218, http://
dx.doi.org/10.1016/S0378-7753(00)00431-6.
[13] H. Ota, Y. Sakata, X. Wang, J. Sasahara, E. Yasukawa, Characterization of lithium
electrode in lithium imides/ethylene carbonate and cyclic ether electrolytes — II.
Surface chemistry, J. Electrochem. Soc. 151 (2004) A437–A446, http://dx.doi.org/
10.1149/1.1644137.
[14] D. Aurbach, Y. Gofer, J. Langzam, The correlation between surface chemistry, surface
morphology, and cycling efficiency of lithium electrodes in a few polar aprotic
systems, J. Electrochem. Soc. 136 (1989) 3198–3205, http://dx.doi.org/10.1149/1.
2096425.
[15] D. Aurbach, I. Weissman, A. Zaban, O. Chusid, Correlation between surface chemis-
try, morphology, cycling efficiency and interfacial properties of Li electrodes in
solutions containing different Li salts, Electrochim. Acta 39 (1994) 51–71, http://
dx.doi.org/10.1016/0013-4686(94)85010-0.
[16] M. Schamel, C. Schopf, D. Linsler, S.T. Haag, L. Hofacker, C. Kappel, et al., The filamen-
tary growth of metals, Int. J. Mater. Res. 102 (2011) 828–836, http://dx.doi.org/10.
3139/146.110541.
[17] G. Richter, K. Hillerich, D.S. Gianola, R. Mönig, O. Kraft, C.A. Volkert, Ultrahigh
strength single crystalline nanowhiskers grown by physical vapor deposition,
Nano Lett. 9 (2009) 3048–3052, http://dx.doi.org/10.1021/nl9015107.
[18] R.F. Sekerka, Equilibrium and growth shapes of crystals: how do they differ and why
should we care? Cryst. Res. Technol. 40 (2005) 291–306, http://dx.doi.org/10.1002/
crat.200410342.
[19] J.L. Barton, J.O.M. Bockris, The electrolytic growth of dendrites from ionic solutions,
Proc. R. Soc. Math. Phys. Eng. Sci. 268 (1962) 485–505, http://dx.doi.org/10.1098/
rspa.1962.0154.
[20] K.I. Popov, M.G. Pavlović, M.D. Maksimović, Comparison of the critical conditions for
initiation of dendritic growth and powder formation in potentiostatic and galvano-
static copper electrodeposition, J. Appl. Electrochem. 12 (1982) 525–531, http://dx.
doi.org/10.1007/BF00614978.
[21] K.J. Harry, D.T. Hallinan, D.Y. Parkinson, A.A. MacDowell, N.P. Balsara, Detection
of subsurface structures underneath dendrites formed on cycled lithium metal
electrodes, Nat. Mater. 13 (2013) 69–73, http://dx.doi.org/10.1038/nmat3793.