Professional Documents
Culture Documents
NETZSCH-Gerätebau GmbH
Version 1 (2023)
1 INTRODUCTION 1
5 VAT PHOTOPOLYMERIZATION 79
Thermal analysis and rheology are powerful techniques that provide information on the
behavior of materials under various thermal and mechanical conditions. The
combination of these techniques has been widely used to characterize the properties
of materials used in AM. Different thermal analysis methods, such as Differential
Scanning Calorimetry (DSC), Thermogravimetric Analysis (TGA), Thermomechanical
Analysis (TMA), Laser Flash Analysis (LFA), and Dynamic Mechanical Analysis
(DMA), can provide valuable information about the thermal behavior of materials.
Rheology techniques such as Rotational and Oscillation Rheometry and Capillary
Rheometry provide information on the flow behavior and viscoelastic properties of
materials.
This overview will be relevant to researchers and professionals working in the field of
AM, materials science, and engineering. It will provide a comprehensive understanding
of the state-of-the-art technology, challenges, and opportunities in AM, with a focus on
the importance of material characterization and in particular thermal analysis and
rheology. We hope that this book will be a valuable resource for those looking to
improve the quality and performance of AM-produced parts and to advance research
in this exciting field.
It is intended to grow over time as more studies are performed. Check back regularly,
if a new version is available or let us know if you’d like to contribute.
Page 1
We would like to thank all our contributors for their time and effort to contribute to this
compendium. They are all experts in their fields and enrich this book with their insights.
Thanks to
Dr.-Ing. Felipe Wolff-Fabris, Tobias Sommer, European Center for Dispersion
Technologies (EZD)
Prof. Dr.-Ing. Katrin Wudy, Robert Setter, M.Sc. and Joseph Hofmann, M.Sc.,
Professorship of Laser-based Additive Manufacturing (LBAM), Technical University of
Munich
Prof. Dr.-Ing. Andreas Blaeser and Anna Fritschen, M.Sc., Institute of Printing Science
and Technology, Technical University Darmstadt
We also like to thank all NETZSCH employees, who offered their time in making this
book a reality – from extensive measurements all the way to its layout.
Thanks to Dr. Stefan Schmölzer, Fabia Beckstein, Doreen Rapp, Rüdiger Sehling,
Senol Gezgin, Claire Strasser, Dr. Jürgen Blumm, Christina Strunz, Doris Forst,
Daniela Gäbler, Victoria Murr, Kathrin Frenzl and Philipp Köppe
Page 2
1.2 Additional Resources for Beginners
For readers, who are new to the field of Additive Manufacturing, some basic information
can be found in these entry level videos.
This video provides and overview and definition of the 7 main AM categories: Episode
2: Explaining the 7 Categories of Additive Manufacturing Technologies ‒YouTube
This next video explains why there are so many different names for the same
processes: Episode 3: Why are Tradenames in Additive Manufacturing Confusing ‒
YouTube
Here we give an overview of the typical workflow for any AM process: Episode 4:
Typical Process Flow in Additive Manufacturing ‒YouTube
Page 3
2 Thermal Analysis and Rheology Methods
Until the start of melting (t1), the difference between the sample and reference
temperatures is constant, Figure 2-2. During the melting process (t1 - tp), the sample
temperature does not change. The supplied heat is used exclusively for the phase
transition. The reference temperature, however, continues to rise.
The temperature difference increases until complete melting of the sample (tp). The
sample temperature returns to the reference temperature (t2). The temperature
difference decreases. From (t2), the sample and reference temperatures are identical.
The peak area, A, is proportional to the heat of fusion.
Page 4
Figure 2-2 Signal generation in a heat-flux DSC
The actual properties measured are the temperature of the sample and the
temperature difference between the sample and reference. From the raw data signals,
the heat flow difference between the sample and reference can be determined which
represents the caloric changes in the sample. Therefore, a DSC is well-suited to
measure endothermic and exothermic effects that are the result of an increase or
decrease in enthalpy of the sample. Typical effects measured in polymers are:
Glass transition temperature
Melting and crystallization behavior
Curing or cross-linking behavior
Specific heat capacity cp
Crystallinity
Oxidative Induction Time/Temperature (OIT)
Relaxation effects
All of these effects are relevant in the Additive Manufacturing processes of Powder
Bed Fusion, Material Extrusion, Sheet Lamination, and depending on the curing
system, in Vat Photopolymerization, Material Jetting and Binder Jetting.
2.1.2 Photo-DSC
In the case of these curing polymer systems, the cure can not only be initiated by
temperature or after passing a certain holding time, but also with the exposure to light;
typically UV light. These light-curing systems react very quickly (within a few seconds
Page 5
at low isothermal temperatures) and are solvent-free. This makes them very attractive
for many industrial applications; originally as adhesives in the electronics industry.
Three different types exist: radical, cationic and dual-cure systems. Dual-cure systems
exhibit a combination of thermal and light-curing reactions and can be found, for
example, in 3D printing resins for Vat Photopolymerization. Radical and cationic
systems are used in Vat Photopolymerization as well as Material Jetting and Binder
Jetting.
The capability of measuring the curing kinetics and degree of cure is essential for the
development of UV-curable resins and identification of optimal curing times and
conditions. Photo-Differential Scanning Calorimetry (Photo-DSC) is a powerful
analytical tool for accomplishing these measurements.
In photo-DSC measurements, both the sample and reference are located in one
furnace and are irradiated simultaneously. The fiber optics are firmly installed in the lid
to ensure reproducible distances between the fiber optics and the sample and
reference, Figure 2-3.
Figure 2-3 Setup of the photo-DSC including a schematic of the DSC cell and the UV light
source
The DSC measurement software communicates with the UV lamp, triggering its pulses
and – for certain models – automatically controlling the pulse length and intensity.
During the course of a measurement, the signals detected are the sample temperature
and heat flow difference. By integrating the heat flow peak, the heat of curing can be
determined, providing meaningful data for development or process optimization.
Recommended are mercury lamps such as OmniCure® S 2000 which allow for the
selection of wide and narrow ranges; these should have filters in the UV-A range
between 280 nm and 315 nm and in the UV-B range between 315 nm and 500 nm.
However, other commercially available lamps can also be adapted. For optimized
curing, a defined wavelength (e.g., 447 nm) can be used with diode laser systems or
high-power LED (e.g., 365 nm, 400 nm or 460 nm) – depending on the resin to be
cured. The OmniCure® lamp is fully software-controlled; selection of temperature, light
intensity, wave length and exposure time can be carried out by the DSC Proteus®
software.
Page 6
To mimic the conditions in the respective AM process, it is important to correlate the
two setups; in production and in the lab measurement.
All of these effects are relevant in the different Additive Manufacturing processes,
especially in regards to material development; but also in qualification of environmental
or health risks. This can be further refined by analyzing the evolved gases using Fourier
Transform InfraRed (FTIR) or Gas Chromatography Mass Spectrometer (GCMS)
analysis.
Page 7
method for measuring the coefficient of linear thermal expansion in plastics, a
fundamental material property describing the capability of a polymer to expand or
shrink when exposed to varying temperatures.
Irrespective of the type of deformation selected (expansion, compression, penetration,
tension or bending), every length change in the sample is communicated to a highly
sensitive inductive displacement transducer (LVDT) via a pushrod and transformed
into a digital signal. The pushrod and corresponding fused silica sample holders can
be quickly and easily changed out, in order to optimize the system for the respective
application, Figure 2-5.
TMA offers a higher degree of sensitivity than the DSC method when it comes to
detecting weak physical transitions that are associated with changes in modulus, post-
curing and delamination, which are hard or impossible to measure with other methods.
In Additive Manufacturing, it can help designing parts understanding their shrinkage or
during processing to comprehend warpage. It is useful for any of the processes, but in
particular for the ones using thermoplastic materials.
Page 8
2.4 Light/Laser Flash Analysis (LFA)
The measurement of the thermal diffusivity of a material is usually carried out
by rapidly heating one side of a sample and measuring the temperature rise curve on
the opposite side. The time it takes for the heat to travel through the sample and cause
the temperature to rise on the rear face can be used to measure the through-plane
diffusivity and calculate the through-plane thermal conductivity if specific heat
and density are known.
All three quantities as well as the thermal diffusivity can be functions of temperature:
λ(T) = a(T) · cρ(T) · ρ(T)
where
λ = thermal conductivity [W/(m·K)]
a = thermal diffusivity [mm²/s]
cp = specific heat [J/(g·K)]
ρ = bulk density [g/cm3].
In Light Flash Analysis (LFA), the front surface of a plane-parallel sample is heated by
a short energy light pulse. The resulting temperature increase on the rear surface is
measured versus time with an infrared (IR) detector, Figure 2-6.
Once the sample has been stabilized at the desired temperature, the laser fires several
times over a span of a few minutes and the necessary data is recorded for each laser
„shot“. The laser beam energy strikes and is absorbed by the front surface of the
sample, causing a heat pulse to travel through the sample thickness. The resulting
sample temperature rise is fairly small, ranging from 0.5 K to 2 K. The temperature rise
is amplified and recorded.
Page 9
Figure 2-7 Evaluation of the temperature rise as a function of half-time used for calculation of
the thermal diffusivity a
The Light Flash (LFA) technique is a fast, non-destructive, non-contact, and absolute
method for determining these thermophysical properties, including specific heat. This
data can then be used for:
In Additive Manufacturing of polymers, all these applications are useful for part design,
process optimization as well as modeling and simulation of 3D printing processes.
Page 10
into an “in-phase” and an “out-of-phase” part by means of Fourier Transformation. A
schematic of a NETZSCH DMA is shown in Figure 2-9.
The results of this mathematical operation are the storage modulus, E’, (related to the
reversible, “in-phase” response) and the loss modulus, E’,’ (related to the irreversible,
“out-of-phase” response). The loss factor, tan δ, is the ratio between the loss modulus
and the storage modulus (tan δ = E’’/E’). Generally, the storage modulus (E’) refers to
the material’s stiffness whereas the loss modulus (E’’) is a measure for the oscillation
energy transformed into heat. Tan δ characterizes the mechanical damping or internal
friction of a viscoelastic system.
DMA measurements provide data and information about the following effects:
Page 11
Glass transition temperature of highly cross-linked, amorphous or semi-
crystalline polymers and composites
Curing/post-curing
Aging
Creep and relaxation
Prediction of the material behavior using Time-Temperature-Superposition
2.6 Rheology
Rheological analysis of samples is a fundamental part of developing many types of
products. Unlike a viscometer, a rheometer can actually measure sample properties at
extremely low shear rates, as in sedimentation, or the high shear rates seen in
pumping, mixing and application. Making measurements in the correct shear range,
we can adequately simulate a flow process and thus differentiate between good
products and poor ones. The rheometer can also determine the effect of adding
different quantities of an additive, or process changes and can thus be used to optimize
the formulation and production of a product.
A rheometer measures not only the viscosity of the product at room temperature, but
can also be used to evaluate the viscosity during a programmed temperature profile.
This can also be used with polymers to evaluate processability and glass transition
temperatures.
Figure 2-10 Illustration showing a sample loaded between parallel plates and shear profile
generated by rotation across the gap
Viscometry can be used to investigate the yield stress, i.e., the stress required to
initiate sample flow, simulate a shearing process, measure shear stability or analyze
how viscosity changes with temperature. Oscillation tests usually investigate the
Page 12
viscoelastic structure of a sample without breaking it down. Initially, an amplitude
sweep is run to determine how large an oscillation the sample can withstand before
the structure breaks down; this is known as the linear viscoelastic region. Once the
linear viscoelastic region has been determined, a frequency sweep, time sweep or
temperature sweep may be performed to investigate how the viscoelastic structure and
viscosity change under dynamic conditions.
Polymers are often characterized using rotational rheometers when information about
the molecular structure, e.g., molecular weight changes or normal stress differences,
is needed or the processing is done with low shear rates, e.g.. thermoforming, Powder
Bed Fusion 3D printing or Material extrusion 3D printing. For processes with higher
shear rates like extrusion, also filament extrusion, injection molding or material jetting
3D printing, capillary rheometers are used.
2.6.1.1 UV-Rheology
UV rheology is a characterization method that combines rheology with UV light
exposure. This technique is used to study the flow and curing behavior of light-curing
resins. It has its origin in the coatings, adhesives and dental industry, but is now also
particularly useful for materials used in Additive Manufacturing processes such as Vat
Photopolymerization and Material Jetting, where the resin system itself is the enabler
for the 3D printing technology.
The working principle of UV rheology involves exposure of the resin to UV light while
simultaneously measuring its rheological properties, Figure 2-11. During exposure
from below, the liquid resin undergoes photopolymerization, where the reactive
monomers in the resin are converted into a solid and rigid polymer network. The degree
of polymerization is directly related to the intensity and duration of the UV light
exposure.
Figure 2-11 Schematic of the UV cell setup in a Kinexus Rotational Rheometer showing the
UV light beam coming through a glass plate in the lower geometry
Page 13
Figure 2-12 Kinexus Prime ultra+, equipped with the UV curing System
The rheometer applies a controlled deformation through the upper plate geometry to
the resin while measuring its response. The entire Kinexus system with the UV cell is
shown in Figure 2-12. By simultaneously monitoring the changes in viscosity, elasticity,
and other rheological properties over time during UV exposure, the curing kinetics of
the resin can be analyzed. This information can be used to optimize the curing process
and improve the properties of the final product.
Page 14
Figure 2-13 Paddle with interchangeable paddles and cylinder (cup) for powder rheology in the
Kinexus rheometer
During the measurements, the 2-blade paddle descends at a defined speed into the
cup containing the sample until it is completely immersed in the powder, and then the
paddle ascends, returning to its starting position. A controlled rotational speed is
applied additionally to the axial speed. The torque and normal force required for the 2-
blade paddle to rotate and to travel into the powder at controlled rotational and axial
speeds are recorded. Both torque and normal force signals are related to the
resistance of the powder to flow. The deeper the paddle is immersed into the sample,
the more resistance the powder offers to the axial and rotational movements of the
geometry.
Page 15
The pressure drop across the die is continuously monitored and measured with a
pressure transducer placed just above the die. A cut away is shown in Figure 2-14.
Capillary rheometers are useful for understanding the material and print behavior of
thermoplastic polymers in Material Extrusion AM processes.
Page 16
The Friedman method is a common way to calculate the activation energy (E) as a
function of the extent of conversion (α).
For the example of the curing behavior of polymer resins, the measurements of which
are typically obtained by DSC, plots at different values of T, for isothermal reactions,
or β, for dynamic heating rates, can be created to estimate the isoconversional
activation energy.
It is not always possible to accurately describe more complex systems with the model-
free approach, such as those with multiple steps, simultaneous reactions, or diffusion
influence. Diffusion control is critical in providing accuracy in models of many thermoset
systems. If the glass transition temperature of the polymer increases faster than the
sample temperature, the reaction will no longer be controlled by the kinetics of the
chemical reaction, but by diffusion processes. This can lead to
vitrification, slowing the expected reaction rate, and resulting in an insufficiently
crosslinked part.
The NETZSCH Kinetics Neo software provides the capability of combining different
chemical reactions and testing the accuracy of the corresponding model. Using
Kinetics Neo, it is possible to calculate different reaction orders, including autocatalysis
and power indices. It also provides the capability of considering diffusion-controlled
processes with the inclusion of glass transition data by using the Rabinowitsch
equation for the total rate constant. It has been shown that chemical reaction models
can be created with high accuracy and a good prediction of the curing behavior. These
predictions are often used to describe the material at isothermal conditions (e.g.,
application temperature), to describe the material behavior during the curing process,
or
to optimize the curing process with the aim of increasing efficiency and lower cycle
times.
Reference
[1] Analysis of the dynamics of the FT4 powder rheometer, C. Hare, U. Zafar, M.
Ghadiri, T. Freeman, J. Clayton, M.J. Murtagh, Powder Technology 285 (2015) 123-
127
Page 17
3 Powder Bed Fusion
In the PBF process, a thin layer of powder is applied on the build platform and heated
to just below the melting temperature of the material, which is often referred to as the
build temperature (heaters not shown in the schematic), Figure 3-1. Next, a laser traces
the cross-section of the part geometry of the first layer, providing enough energy to
locally melt the material. Without any shear forces, the melt needs to have a low
viscosity and surface tension to coalesce and form a uniform melt pool. The
surrounding powder stays solid and keeps the shape of the molten geometry.
Therefore, no support structures are needed. This can be seen by the three N-shaped
built parts in the powder bed. Now the build platform is lowered by one layer height
making room for the next layer. A sweeper or recoater roller moves across the surface,
picks up excess material from the reservoir and deposits new and colder powder on
top of the build platform to create the next layer. Again, the powder is heated to keep
it at the build temperature. This is important to hinder crystallization. The whole build
envelope is kept in a nitrogen atmosphere to reduce effects of aging.
These process steps of powder coating and laser melting are repeated over and over
until the whole part is built. Only then is the build envelope cooled down, which initiates
Page 18
crystallization and thus solidification process of the part. After the part and surrounding
powder is cooled completely, the part is unpacked.
Episode 10: Process Prinicple of Polymer Powder Bed Fusion (SLS) - YouTube
3.2 Materials
The first material used in this process was PA12, because of its good mechanical
performance and the capability of generating powders by precipitation. This yields
powder with close to perfect spherical shape, which is necessary to create a uniform
layer during coating. Then PA11 followed and together they still make up 90-95% of all
materials used in PBF today. However, in recent years, more and more materials have
been qualified for the process including high-performance materials such as PEEK,
elastomeric materials such as TPUs and even commodity materials such as PP. Most
of them are produced by cryogenic grinding and show more or less pronounced
deviations from the circular shape [1].
Episode 12: Materials Used in Polymer Powder Bed Fusion (SLS) - YouTube
3.3.2 How to Determine the Process Window for PBF Powders Using DSC
Author: Dr. Stefan Schmölzer
The upper limit of the process window in Powder Bed Fusion is determined by the
temperature the powder needs to reach in order to melt. The lower limit is defined by
the temperature needed in the build chamber to avoid premature crystallization and
warpage.
In order to characterize a polymer powder for its suitability for PBF and to determine
the possible process window, Differential Scanning Calorimetry (DSC) measurements
are needed, which can be performed with the NETZSCH DSC 303 Caliris® instruments.
Episode 11: Temperature Field in Polymer Powder Bed Fusion (SLS) - YouTube
Page 19
A dynamic measurement on PA12 powder is performed to determine the melting and
crystallization behavior during the 1st heating and cooling, respectively. The sample
was cooled from room temperature to start the measurement at 0°C. It was then
heated to 200°C at a heating rate of 10 K/min and cooled down at the same rate of 10
K/min back to 0°C. This cycle was repeated several times. All measurement conditions
are summarized in the following table:
Table 3-1 Measurement conditions
In Figure 3-2, the results of the 1st heating (blue) and cooling cycle (green) are
depicted. The onset of melting occurs at 181°C and the onset of crystallization at
153.4°C (here labelled as “End” due to the analysis from low to high temperatures).
Figure 3-21st heating and cooling cycle (blue and green, respectively) of PA12 powder at a
heating and cooling rate of 10 K/min
The measured PA12 powder (Figure 3-3) offers a process window of 27.6°C, where
the process temperature must be set between the onset of melting and crystallization.
The recommended build temperature for this material is 168°C, which lies in the middle
of the process window. If the build temperature is too close to the onset of
crystallization, the parts can warp due to larger temperature gradients. On the other
Page 20
hand, if the build temperature is too close to the onset of melting, hot spots can occur,
resulting in sintering of the surrounding solid powder on the surfaces and lateral growth
of the parts.
Figure 3-3 Schematic of the PBF process window and build envelope temperature based on
the DSC results depicted in Figure 3-1
While for the determination of the process window, the first heating of the powder is
most important, it is advisable to compare the second heating as well. In the case of
this measurement, the heating and cooling cycles were repeated multiple times and
Figure 3-4 shows the results of three runs.
First, it can be seen that the melting peak of the powder (1st heating) is shifted to higher
temperatures due to the slightly reduced contact surface of the powder with the pan as
well as the higher surface energy of the powder. Second, it can be seen that
the 2nd and all subsequent heating cycles show a double peak with lower onset
temperature. This peak indicates a different crystalline structure than that of the
powder, which is rather unique for this particular PA12 powder and not observed for
other PA12s. It shows that during cooling, in addition to the common α- and γ-
spherulites, an intermediate crystal structure is formed that can be observed as a small
peak at the beginning of melting. However, while this is of scientific interest, it is not
relevant for the PBF printing process.
Page 21
Figure 3-4 1st, 2nd and 3rd heating and cooling cycles at 10 K/min (dark blue: 1st heating; light
blue and green: 2nd and 3rd heatings); 1st heating curve shifted horizontally
for visualization purposes only. The exothermic crystallization peaks stay the same during all
cooling cycles
Page 22
Figure 3-5 Crystallization peaks at different cooling rates of 10, 5, 2 and 1 K/min show the shift
to higher temperatures at lower cooling rates
While the actual temperature on the surface of the powder bed can be measured with
IR thermometers, the temperature in lower layers is unknown on a commercial PBF
printer. Over the whole duration of the build, which can be up to 12 hours without
subsequent cooling, isothermal crystallization can occur after some time, especially if
the build temperature fluctuates too much due to coating with cold powder, non-uniform
distribution of parts inside the build envelope, or unbalanced heaters, just to name a
few reasons. Therefore, isothermal crystallization studies are needed to evaluate this
behavior for the selected polymer powders and thus qualify them for PBF.
This study is summarized here: Determining the process window of Polymer Powder
Bed Fusion - YouTube
Step 1: The sample was heated from room temperature to above melting at
200°C at 20 K/min. It was kept there for 1 minute to erase the sample history.
Step 2: It was then rapidly cooled to the isothermal temperature step (168, 167,
166, 165, 164, 163, 162°C in Figure 1) using a high cooling rate of 125 K/min to
prevent reorganization processes that occur with PA12 at slow cooling
rates. Both the capability of achieving a fast cooling rate with regular sample
sizes and the capability of precisely controlling the temperature are features of
the NETZSCH DSC 300 Caliris® with P-module that are extremely valuable for
this analysis.
Page 23
Step 3: Next, the sample was kept at the isothermal temperature for 30 minutes
to study the crystallization process.
Step 4: The sample can then be cooled down OR the sample can be heated
back up to 200°C at 10 K/min (as was done here) to get the full picture and
observe the melting behavior after the isothermal crystallization step.
Figure 3-6 shows the isothermal crystallization behavior at different temperatures from
165°C to 162°C right below the build envelope temperature. The crystallization peak
temperature, tmax, is analyzed as the peak of the curve from the start of the
measurement. Therefore, the values depicted here were normalized in the Proteus®
software for the actual start of the isothermal step.
Page 24
DSC /(mW/mg)
↑ exo
0.20
0.00
-0.05
0 5 10 15 20 25 30 35 40
Time /min
Figure 3-6 Isothermal crystallization behavior of PA12 powder at 162, 163, 164, 165, 166, 167
and 168°C
Figure 3-7 shows the corresponding normalized temperature profile. The isothermal
temperatures were reached about 10 minutes after the measurement start. Even at
these high cooling rates of 125 K/min, the temperature only overshoots by ± 0.1
K and hits the set temperature in less than 30 s.
Page 25
Temp. /°C
170
168°C
168
166
164
162
162°C
160
0.0 0.2 0.4 0.6 0.8 1.0
Time /min
Figure 3-7 Normalized temperature curves for the transition to the isothermal step at
temperatures from 168 to 162°C
These results highlight that even at the typical build envelope temperature of 168°C
for PA12, crystallization starts after about 10 minutes (Figure 3-6) and reaches its peak
after 23.7 minutes. While the top layers will be reheated closer to the melting
temperature with each additional layer, it becomes obvious that lower layers will
eventually stay at 168°C or could even cool further. Thus, given the long build durations
of typically several hours, crystallization will occur and has to be taken into account.
This study is summarized here: Crystallization Behavior in Polymer PBF - YouTube
Page 26
In isothermal analysis of crystallization, the first dependence is typically represented
by the Avrami equation, which represents crystallization nucleation rate.
𝑛−1
𝑓(𝛼) = 𝑛 ∙ (1 − 𝛼) ∙ [−𝑙𝑛(1 − 𝛼)] 𝑛 (Eq 3-2)
The extended version of the Avrami equation is the Sestak-Berggren equation. This
extended equation is used in the current analysis because it provides a better fit for
the experimental data.
𝑓(𝛼) = (1 − 𝛼)𝑛 ∙ 𝛼 𝑚 ∙ [−𝑙𝑛(1 − 𝛼)]𝑞 (Eq 3-3)
The dependence K(T) in 𝑑𝛼𝑑𝑡=𝑓(𝛼) ∙ 𝐾(𝑇) (Eq 3-1) is a formal Arrhenius equation
as the decreasing function of temperature with pre-exponent A and apparent activation
energy E:
−𝐸
𝐾(𝑇) = 𝐴 ∙ 𝑒𝑥𝑝 ( 𝑅𝑇 ) (Eq 3-4)
This kinetic model (Eq 3-1) presents the dependence of the current crystallization rate
on the temperature and current degree of crystallization.
The equations contain unknown parameters, which are found by the software in order
to determine the best fit for the experimental curves.
If this simulation is performed for temperature conditions of the isothermal experiments
with optimal parameters, there will be very good agreement between the experiment
and simulation with R2=0.998. In Figure 3-8, the points represent the experimental
data and the solid lines, the simulation, according to Eqs. (3-1, 3-3, 3-4).
Figure 3-8 Conversion rate: Experimental data and simulation according to the kinetic model
for isothermal crystallization of PA12. The points represent the experimental data and the solid
lines the simulation according to the Eqs. (3-1, 3-3, 3-4)
Page 27
This single model now works for different temperatures. Therefore, it can be used for
simulation of crystallization in the PBF process. The temperature profile of the powder
surface can be measured over the duration of multiple cycles. We can then run a
simulation of the crystallization process for this powder layer. It can be assumed that
each lower layer has a similar temperature profile, but with slightly reduced
temperatures due to the powder application for each layer. Thus, one can simulate the
crystallization process of a single layer during several laser cycles. Figure 3-9 presents
the simulation of the crystallization degree over 5 cycles where for each new cycle or
layer, the temperature was reduced by 2 K.
We see that one layer cannot completely crystallize during the time constraints of one
cycle, when this layer is on top of the powder bed. However, crystallization continues
throughout the PBF process, as each cycle produces further layers. Crystallization
during several cycles is one of the advantage of PBF, when the resulting 3D object has
very strong layer adhesion and isotropic mechanical properties in all directions like
hardness, tensile strength and elongation [3].
Figure 3-9 Simulation of the crystallization degree over 5 laser cycles for standard 3D-printing
where each new cycle has a temperature 2 K below the previous cycle
However, if the thickness of the powder layer is increased, then the temperature
difference between layers will be higher. This can happen during high-speed sintering.
The simulation over 5 cycles with a temperature difference of 5 K (Figure 3-10) shows
that the main crystallization is already finished during the second cycle, while the third
layer is already solid. This asynchronous crystallization could be the reason for the
mechanical stresses, warpage or curling in the sample because of its shrinkage during
the PBF process. Additionally, using thick powder layers could decrease the isotropy
of the final material.
Page 28
Figure 3-10 Simulation of the crystallization degree over 5 laser cycles for high-speed sintering
(HSS) where each new cycle has the temperature 5 K below the previous cycle
The combination of NETZSCH Kinetics Neo with DSC assists in studying the
crystallization rate of polymers and simulate their behavior for such complex industrial
processes such as PBF. This is extremely valuable in the development of new
materials as well as improve our understanding of the process.
Two quantities that are essential for the simulation of the PBF process are the specific
heat capacity (cp) and thermal conductivity (k) as a function of temperature. When
fillers are added, they do not only change the crystallization rate, but also require a
higher build temperature due to the change in cp and k.
Where
𝑄̇ is the heat flux or measured DSC signal,
m is the mass of material and
𝑇̇ is the heating rate.
Page 29
In accordance with DIN EN ISO 11357-4 (and ASTM E1269), the sample is measured
relative to a second (reference) sample of known heat capacity. A typical reference
sample would be sapphire. Therefore, one experiment consists of three different runs
in the temperature range of interest. The first is a scan with two empty pans (baseline),
the second a scan with one pan containing the sapphire sample (reference), and finally
the third run with the actual sample (sample) in the same type of pan.
𝑚 𝑄̇ −𝑄̇
𝑐𝑝 (𝑇) = 𝑐𝑝 𝑟 (𝑇) ∙ 𝑚𝑟 𝑄̇3 −𝑄̇1 (Eq 3-6)
𝑠 2 1
Where
𝑄̇1, 𝑄̇2 , 𝑄̇3 are the three DSC signal differences of the baseline, sapphire reference
and the sample. The masses mr and ms are of the reference and sample, respectively,
and cpr is the theoretical heat capacity of the reference.
The analysis in the NETZSCH Proteus® software is shown in Figure 3-11. It shows the
„apparent“ specific heat capacity, superimposed by the melting peak and glass
transition.
Page 30
Cp /(J/(g*K))
10
Figure 3-11 Specific heat capacity, cp, obtained from two repetitive measurements (green and
blue lines) as well as the calculated average (black line)
The cp data can be easily deduced from this curve. However, in the temperature range
between 90-190°C, the effect of the increasing cp and the endothermic effect of melting
are opposing each other. Therefore, the values in the melting range are typically
interpolated. This can be done by hiding the peak in the Proteus® software. The
resulting cp curve is shown as a purple line in Figure 3-12.
Figure 3-12 Specific heat capacity, cp, calculated by hiding the peak (purple line)
Page 31
The values can then be exported for subsequent use in material and process
simulations, e.g., for the temperature field and solidification in the PBF process.
Another application of the data is the calculation of the thermal conductivity from
thermal diffusivity and density data. More information about this example can be found
in chapter 3.3.12.
This study is summarized here: Specific Heat Capacity of PA12 Powder - YouTube
In the PBF process, the laser tracing the cross-section of the 2D geometry needs to
provide sufficient energy to locally melt the material. The melt needs to have a low
viscosity and surface tension to coalesce and form a uniform melt pool. For example,
if the viscosity is too high, the particles melt but do not spread, which leads to micro
voids and porosity in the final parts.
To study the formation of the melt pool and evaluate the suitability of a material to be
processed with PBF, the viscosity of the melt needs to be known. In contrast to many
other polymer processing technologies, there is no flow or even deformation acting on
the melt. Therefore, the zero shear viscosity is of interest, which is determined in cone
plate or parallel plate rheometer measurements for unfilled and filled materials,
respectively. These measurements can again be performed using a Kinexus Prime
ultra+ rheometer. The measurement temperature is chosen based on the temperature
that the material is heated with the laser. In this example of PA12 powder, it is 200°C,
just above the melting temperature.
The sample is heated to 200°C and a strain sweep is performed to determine the Linear
Viscoelastic region (LVER). It can be seen in Figure 3-13 that a strain of 1% is suitable
for all measurements to stay within the LVER.
Figure 3-13 Strain sweep at 200°C from 0.1 to 50% strain at a frequency of 1 Hz on neat PA12
powder
Page 32
Then a frequency sweep was performed to determine the zero shear viscosity. All
measurement conditions are summarized in the following table:
Table 3-4 Measurement conditions
Figure 3-14 Complex viscosity as a function of frequency at 200°C for neat PA12 powder
The same PA12 material with glass spheres as fillers was measured and the viscosity
is compared to the viscosity without fillers in Figure 3-15. As expected, the viscosity of
the glass-filled PA12 is higher. Processing trials have shown that both materials
process well, so in both cases, the viscosity is sufficiently low, but the flow or
coalescence of the material with fillers will be slower.
Page 33
Figure 3-15 Comparison of the complex viscosity as a function of frequency at 200°C for neat
PA12 powder (black) and PA12 powder filled with glass spheres
The position of the G´/G´´ crossover obtained during a frequency sweep is indicative
of the molecular weight and the mass distribution of a polymer.
The lower the frequency of the crossover, the higher the molecular weight Mw.
The lower the storage (or the loss) modulus at the crossover, the higher the
molecular weight distribution MWD.
For PA12 powders, it is known that post-condensation leads to increased chain length
of the aged powders. In this example, PA12 with glass spheres was investigated to
study the effect such a filler for shrinkage reduction can have on the ageing process.
The purpose of glass spheres as fillers is discussed in more detail in chapter 3.3.10.
In addition to the virgin powder, aged material was produced by storing it in an oven
for 24 and 96 h in an inert atmosphere. On these samples as well as the virgin material,
the same oscillatory measurements as explained in chapter 3.3.6 were performed. In
Figure 3-16 the change in complex shear modulus with increasing ageing time can be
observed.
Page 34
Figure 3-16 Complex shear modulus of PA12 w/ fillers as a function of aging time at 200°C,
1% strain
The increase in shear modulus correlates well with the molecular weight Mw values of
the material. To better understand this effect when the actual Mw is not known, the
storage (G’) and loss modulus (G”) are plotted and the crossover point is determined
in the rSpace software of the Kinexus rheometer.
Page 35
Figure 3-17 top: Storage (G’) and loss modulus (G”) of PA12 powder with glass fillers as a
function of aging time at 200°C, 1% strain; bottom: evaluation table in rSpace software listing
the values of frequency, elastic and viscous component as well as phase angle at the
crossover
While the crossover point for the virgin material lies outside the measurement range,
the shift to higher frequencies can be observed. Zooming in on the two crossover points
of the 24- and 96-h aged material, two shifts can be observed, Figure 3-18. The shift
on the X-axis indicates an increase in molecular weight with increased ageing time. An
increase in molecular weight by 20% leads to a decrease in the crossover frequency
by one decade. The shift on the Y-axis indicates a slight increase in molecular weight
distribution MWD, which can be confirmed by a small increase in the polydispersity
𝑀
index (𝑃𝐼 = 𝑀𝑤 ).
𝑛
Page 36
Figure 3-18 Detail view of Figure 3-17 on results of powder aged for 24 h (square) and 96 h
(circle)
It becomes clear that the chain growth typically observed during aging of pure PA12
powders is even more pronounced in the material with glass spheres investigated here
(results on the same pure PA12 powder not shown here).
Direct comparison of different materials can such give valuable insights into aging itself
as well as the significance on processing. At the same time, these same
measurements offer additional information about the flow behavior as shown in Figure
3-15 in chapter 3.3.6.
Other possible studies using rheology are summarized here: Melt Rheology in PBF
using Rotational Rheometry - YouTube
Page 37
material (above 186°C for PA12) and the preheated surrounding solid powder (around
168°C). This results in heat exchange between different regions, which introduces local
transient temperature states, even though the IR heaters on top of the build envelope
work to heat the new layer as quickly as possible.
During the first few layers of printing, when the first layer is melted, it is supported by
a cooler bottom layer, and colder powder is deposited on top during coating, which
rapidly cools the layer and likely initiates crystallization. When the newly deposited
material is melted, a steep temperature gradient between the first and second layers
creates compressive stresses at the top and tensile stresses on the bottom, leading to
upward curling of the layer. This curling causes the build to stick out of the powder bed
at the corners, leading to displacement or failure of the print.
The degree to which the layer curls out of the plane, δ, depends on the temperature
gradient, ΔT (between top and bottom), the length of the part, L, its thickness, d, as
well as the thermal expansion coefficient, α, following this relationship shown in Figure
3-19:
𝛼∙𝐿2 ∙∆𝑇
𝛿= Eq 3-7)
(
4𝑑
Figure 3-19 Schematic of the curling effect in PBF with relevant parameters
This shows that for the same layer (or part) geometry, a high thermal gradient, ΔT,
leads to more pronounced curling (high δ) if the coefficient of thermal expansion (α) is
high as well. It is also obvious that the longer the part (large L) or to a lesser degree
the thinner the part (small d), the higher is δ. Once more layers are stacked on top of
each other, the thermal gradient is smaller and curling will only be a problem in thin
sections of the part.
To determine the CTE of a material a dilatometer or in the case of polymers, more
typically a Thermomechanical Analyzer (TMA) such as a NETZSCH TMA 402 F1
Hyperion® equipped with a steel furnace is used. This measurement is commonly
performed on solid polymer samples that are being heated to temperatures close to
melting. Although it should represent the reversal of the thermal and crystallization
shrinkage happening during the process, it is not a direct measure of it and is not
capable of capturing the full melting process.
Page 38
Samples of 10 mm x 5 mm x 4.5 mm were prepared from the dogbone specimen in
one build orientation (X-direction) as the material is considered isotropic:
Figure 3-20 Location of the TMA sample and its orientation in the global build coordinate
system
The measurements started with an initial cooling step, after which the temperature was
increased at 5 K/min from -15…170°C, which corresponds to the temperature of the
build envelope. The thermal expansion coefficient was calculated using the mean CTE
analysis (m. CTE), which computes the slope between two data points using a
reference temperature Tref of 20°C. Following DIN 51054-1, the mean is given as
𝑑𝐿(𝑇1 ) 𝑑𝐿(𝑇𝑟𝑒𝑓 )
− 1 𝑑𝐿(𝑇1 )−𝑑𝐿(𝑇𝑟𝑒𝑓 ) ∆𝐿
𝐿0 𝐿0
𝛼(𝑇1 ; 𝑇𝑟𝑒𝑓 ) = =𝐿 ∙ =𝐿 (Eq 3-8)
𝑇1 −𝑇𝑟𝑒𝑓 0 𝑇1 −𝑇𝑟𝑒𝑓 0 ∙∆𝑇
where Tref is the reference temperature (typically around 20 to 25°C for measurements
that start below 0°) and L0 is the length at room temperature.
The measurement results are depicted in Figure 3-21. The glass transition, Tg, is
located at 36°C (extrapolated onset), which is in good agreement with literature values
for PA12. The mean CTE was calculated below and above the T g. However, it was
observed that above 130°C, the slope changes again and therefore, an additional CTE
value in the temperature range of the process was analyzed between 150 and 170°C.
Page 39
Figure 3-21 Measured length change of the sample and detection of the Tg as well as of the
CTEs below and above the Tg
323∙10−6 𝐾−1 ∙50 𝑚𝑚2 ∙1𝐾 323∙10−6 𝐾−1 ∙50 𝑚𝑚2 ∙5𝐾
𝛿1𝐾 = =0.02 mm; 𝛿5𝐾 = = 0.1 mm
4∙10𝑚𝑚 4∙10𝑚𝑚
Using these assumptions, it can be seen that the degree to which the layer curls out
of the plane, δ, can be significant even for small temperature gradients. Especially,
curling up by 0.1 mm can be enough for the roller to bump into and drag the previous
build layers with it.
The study is summarized here: Warpage prediction of PBF parts using CTE - YouTube
Page 40
the Young’s modulus is calculated as the slope of the curve between 0.05…0.25%
strain. It can be used for quality assurance, material development and optimization as
well as some dimensioning tasks. However, it cannot be used for component design
and simulation.
For this, it is important to obtain time- and temperature-dependent data that predicts
the material behavior over the service life at realistic loading conditions. The method
of choice is Dynamic Mechanical Analysis (DMA) such as a NETZSCH DMA 303
Eplexor, which allows for the sample to be subjected to a sinusoidal load and the
viscoelastic response of the material is detected. By varying the temperature and
frequency of the measurement, the temperature and time dependence can be
analyzed as well.
For understanding the PBF part property development during printing and in particular
shrinkage and warpage, a temperature-dependent DMA measurement is required.
During the constant cycle of powder coating and laser melting, the temperature is
constantly changing within the part and a temperature gradient is forming from the
bottom of the part to the top. This can cause warpage, which was already explained in
chapter 3.3.8 on thermal expansion.
However, another effect of warpage is the build-up of residual stresses, σ, within the
part, which are affected by the E modulus and the temperature gradient. The following
simplified relationship exists:
𝑧
𝜎 = 𝐸 ∙ ∆𝑇(𝑧) ∙ (Eq 3-9)
𝑑
where ΔT is the temperature gradient between top and bottom, d is the thickness of
the part and z denotes a certain location across the thickness of the part. From this
relationship, it can be seen that for a given geometry, the larger the temperature
gradient and/or the higher the modulus, the higher are the residual stresses.
Figure 3-22 Schematic of the residual stress build-up with relevant parameters
The middle pieces were cut out of the PBF dogbones to obtain beams of 50 mm x 10
mm x 4.5 mm dimensions. Even though the surface shows the typical roughness of
PBF parts, no additional surface treatment was selected, because the surfaces were
plane-parallel.
The samples were then loaded in the 40-mm wide flexural fixture of the DMA. Following
an initial cooling and equilibration step, the samples were heated from -50°C to 180°C
Page 41
at 2 K/min, which is just below the melting temperature of the material. All
measurement conditions are summarized in the following table:
Figure 3-23 depicts the measurement results of storage modulus, E’, loss modulus, E”,
as well as the damping factor, tan . They show the typical behavior of a semi-
crystalline thermoplastic material. The storage modulus exhibits a drop at the transition
temperatures, glass transition and melting, and the loss modulus and tan shows a
maximum. The factor used for the analysis is chosen based on the effect that is of most
interest. To understand shrinkage and residual stress build-up, the elastic response
(E’) is the most important and will be analyzed here.
Figure 3-23 Storage modulus, E’, loss modulu,s E” and tan delta as a function of temperature
for a PA12 dogbone produced by PBF
With increasing temperature, the storage modulus continuously decreases. The value
of E’ at room temperature is 1438 MPa. The data sheet of the measured sample
typically shows different values (here: 1650 MPa) because the Young’s modulus is
measured in tension. During a DMA measurement in flexural mode, both the
compressive and tensile load act on the sample, especially when measuring thicker
samples. The onset of the glass transition was determined to be 27°C. After the drop
in modulus, the values further decrease from 500 MPa to 114 MPa at the onset of
melting (167°C).
Page 42
While the value of the storage modulus, E’, just below melting is very important for a
successful printing process, the whole progression is important during the cooling
stage. Due to the big change in modulus at the glass transition, the cooling process
has to be very slow (> 12 h for the whole build) to reduce or eliminate warpage and
residual stress build-up during this stage. Understanding this behavior can help
optimize the process and potentially accelerate this time-consuming process step.
The study is summarized here: Estimating Residual Stresses in PBF Parts using DMA
- YouTube
Due to the still limited number of available materials for the Powder Bed Fusion (PBF)
process, there is a constant demand for materials with different properties. One
demand could be higher ductility (e.g., PA11), others could be better dimensional
stability (e.g., addition of glass beads), higher electrical or thermal conductivity (e.g.,
fillers such as aluminum or copper) or increased stiffness and strength (e.g., addition
of glass or carbon fibers). In the next chapters, effects of such fillers on the properties
and processing behavior are being investigated with various analysis methods.
The addition of any filler to PBF powder typically has an effect on the crystallization
behavior as the surfaces of the fillers act as heterogeneous nucleation
sites. Conductive fillers can also affect the material temperature and therefore, show
additionally changed behavior. Dynamic Differential Scanning Calorimetry (DSC) such
as the NETZSCH 300 Caliris can provide a first indication of a changed behavior,
which can then be further analyzed through isothermal measurements at or close to
the build temperature.
To understand the effect of conductive fillers, researchers at the Institute of Polymer
Technology (LKT) at the University of Erlangen-Nuremburg experimented with copper
fillers [2]. They used different mixtures of copper spheres and flakes (Figure 3-24) in
varying contents and analyzed the processing behavior as well as the resulting
conductive properties with the aim to generate complex components for thermal
management. A variety of samples were prepared by adding copper spheres in 5 and
10 vol% and copper flakes at 5 vol% to the PA12 powder. The energy density of 0.043
J/mm2 was kept constant for all materials to detect any changes in the process
behavior due to the fillers.
Page 43
Figure 3-24 Scanning Electron Microscope images of copper flakes (top) and copper spheres
(bottom) [1]
At NETZSCH Analyzing & Testing, the process window and the crystallization behavior
of these different mixtures of PA12 powder with copper particles were studied in
comparison to the neat PA12 material.
The sample was cooled from room temperature to start the measurement at -20°C. It
was then heated to 200°C at a heating rate of 10 K/min and cooled down at the same
cooling rate of 10 K/min to -20°C.
The following table summarizes the measurement conditions. The results for the neat
PA12 were discussed in detail in chapter 3.3.2.
In Figure 3-25, the results of the 1st heating and cooling are depicted for all 4 samples.
As can be seen, the peak melting temperature is not affected by the addition of the
fillers (The difference in area is related to the varying filler contents and was not
normalized in this graph.).
Page 44
Figure 3-25 1st heating and cooling of PA12 PBF parts at a heating and cooling rate of 10 K/min
However, it can be seen that the peak crystallization temperature as well as the onset
of crystallization are shifted to higher temperatures with increasing filler content.
The peak is shifted from 145.8°C for the neat material to 149.1°C for 5 vol% copper
fillers and to 151.3°C for 10 vol%, respectively.
It can be concluded that the fillers act as nucleation sites and accelerate crystallization.
This leads to a slightly reduced process window and needs to be taken into account
during the selection of the processing conditions.
Additional studies of the isothermal crystallization behavior can be useful to understand
this behavior even better. A detailed description of the measurement procedure is
given for an unfilled PA12 powder in chapter 3.3.3.
The researchers at LKT performed these additional isothermal studies at 165°C on the
copper filled samples and found that the crystallization half time is reduced due to the
addition of fillers, further confirming the nucleation effect. The whole study can be found
here [Open Access]!
3.3.11 How Specific Heat Capacity of Filled Powders Affects PBF Processing
Parameters
Author: Dr. Stefan Schmölzer
Page 45
For this study, the same copper-filled materials as described in chapter 3.3.10 from the
Institute of Polymer Technology (LKT) at the University of Erlangen-Nuremberg were
used. The process temperature for the mixture with copper spheres was determined
to be 167°C and with the copper flakes 173°C, respectively.
A NETZSCH DSC 300 Caliris was used to measure the specific heat capacity, cp, as
a function of temperature of these different mixtures of PA12 powder with copper
particles in comparison to the neat PA12 material. The measurements were performed
in accordance with ASTM E1269 and ISO 11357-4.
After an initial cooling step to -25°C, the temperature was increased to 215°C at 10
K/min. Two different samples were measured and the average was calculated.
The following table summarizes the measurement conditions.
The analysis in the NETZSCH Proteus® software is shown in Figure 3-26. It shows the
„apparent“ specific heat capacity of a PA12 sample with 5 vol% copper spheres,
superimposed by the effects for melting and glass transition.
Cp /(J/(g*K))
Peak: 183.1 °C
8
4
75.0 °C; 1.804 J/(g*K)
Figure 3-26 Specific heat capacity, cp, of PA12 w/ 5 vol% copper spheres obtained from two
repetitive measurements (green and blue lines) as well as the calculated average (black line)
Page 46
The cp data can be easily deduced from this curve. Figure 3-27 shows the cp values
after interpolation for all four samples, which is explained in chapter 3.3.5.
3,0
specific heat capacity cp [J/g*K]
2,5
PA12 neat
2,0
PA12 + 5% Cu
spheres
1,5
PA12 + 10% Cu
spheres
1,0 PA12 + 5% Cu
flakes
0,5
10 60 110 160 210
Temperature [°C]
Figure 3-27 Measured cp values for all four samples as a function of temperature including the
interpolated values between 90-190°C
As expected, it can be seen that the cp is increasing with increasing temperature. The
additional copper content reduces the cp and no effect of the filler geometry can be
detected. The researchers at LKT even confirmed that the decrease in c p with
increasing copper content follows the rule of mixture. However, they only measured
the cp at 25°C. The temperature-dependent measurements shown in Figure 3-27
indicate further that the slope of the cp increase with temperature is slightly reduced
the more copper particles are in the mixture.
The measurements confirm that the change in cp can contribute to the higher energy
input required during 3D printing. However, additional information about the thermal
conductivity is needed to evaluate the impact of both effects on the thermal conditions.
The relevant study can be found in chapter 3.3.12.
It should be noted that this behavior is universal for all plastics materials modified with
thermally conductive fillers. Therefore, it is an important quantity to be measured for
the design as well as injection molding simulation of heat sinks or other components
needed in thermal management.
1Energy Density = How much energy a system contains in comparison to its volume
3.3.12 How Thermal Diffusivity Affects the Build Temperature in the PBF
Process
Author: Fabia Beckstein
Page 47
3.3.12.1 Why Thermophysical Properties Need to Be Assessed in Different
Directions
Many researchers have shown that the build orientation of the samples in the build
volume has an effect on the mechanical properties of the Powder Bed Fusion (PBF)
parts. The build orientation refers to the position of a part in regards to the plane of
powder application (xy) and the movement of the build platform (z), Figure 3-28.
Figure 3-28 Global coordinate system of mechanical test specimen; the LFA sample is drawn
in green
In very general terms, it can be said that parts where the test direction of the specimen
aligns parallel to the xy-plane of the build (as seen in Figure 3-28) are stronger than
parts where the test direction is perpendicular to the xy-plane of the build (or parallel
to the zy-plane).
Because of this directionality of mechanical properties, various other properties are
also best assessed in different directions; especially when additional fillers are used to
further enhance the properties as explained in chapters 3.3.10 and 3.3.11. In PBF, the
filler size has to be very small – on the same scale as the powders – to enable a
uniform layer during coating. In the case of PA12 powders, that would be around 60 µm
length.
For fibers as fillers, there are currently two options available. One is a powder that has
the fibers already incorporated into the particle and the other is a dry mixing of fibers
and powder. The first requires or yields even shorter fibers than 60 µm, but the
orientation is random inside the powder bed. The dry mixing, however, results in a
preferential orientation of the fibers in the direction of powder application (commonly
denoted as the x-direction). This can be also observed for other fillers: e.g., flakes in
comparison to spheres (see Figure 3-29 below).
Page 48
Figure 3-29 Single layer of PA12 and 5 vol% copper flakes (top) and spheres (bottom),
respectively [2]
Where the flakes are mostly oriented in the xy-plane, the spheres as isotropic fillers
are evenly distributed within the layer. Therefore, special care has to be taken during
the preparation of the samples to be able to detect any effects that are a result of the
orientation during powder coating.
At NETZSCH Analyzing & Testing, samples were cut on a milling machine using a
side-milling cutter from dog bone specimens in three different directions to allow
analysis of the coating operation on the filler orientation, Figure 3-29. If a cooling
medium is used during cutting, the samples shall be cleaned afterwards.
The samples in z-direction can be measured in the standard sample holder, while the
samples in x- and y-direction need to be cut into smaller strips, turned and measured
in the new arrangement in the laminate sample holder as explained visually in Figure
3-30.
Figure 3-30 Schematic of the sample preparation for all 3 measurement directions: z-direction
just cut out of the dogbone (2.4-mm thickness and milled on both sides); x- and y-direction cut
into strips, turned in 90° around the appropriate axis and rearranged (2-mm thickness based
on the cut width)
Page 49
To ensure that all strips are planar and there are no gaps, burrs are removed around
the sample edges with sand paper. Since the samples in x- and y-direction consist of
individual strips, the laminate sample holder is designed to fix the strips in position,
Figure 4.
Figure 4: Sample Holder for LFA 467 HyperFlash®; a) Standard sample holder 12.7
mm for samples specimens in z-direction; b) laminate sample holder for samples
specimens in x- and y-direction
This particular sample holder allows for measurement of one particular sample in at
least two different directions. For example, one could measure the sample in z-
direction, then cut it into strips and measure it in another direction. However, the more
common reason is that the samples can only be prepared in specific dimensions, which
is often the case with dogbone samples or other thinner plate-like parts. In these cases,
the laminate sample holder is the only option.
Thus, the x-direction indicates the measurement in the direction of the coating
operation from left to right or vice versa; y is in the same xy-plane, but in perpendicular
direction to the coating from front to back and vice versa; z is the thickness direction of
the layer.
The dimensions of the samples in x- and y-direction are 12.6x12.6x2 mm3 and in z-
direction 12.6x12.6x4 mm3 (only the neat PA12 sample has such a low thermal
diffusivity and thus too long measurement duration with 4-mm thickness that the
sample was grinded to 2.4-mm thickness).
The next preparation steps are necessary for all samples where the surface texture or
degree of transparency could have an effect on the reflection of the laser beam.
The surface of the PBF samples have a rough texture compared to injection molding
parts due to the powder raw material and the process. Since the Laser Flash Analysis
is an optical method that needs to eliminate reflection, the samples are spray-coated
with a thin layer of graphite, which has high absorptivity for the energy pulse of the
LFA.
Page 50
At NETZSCH Analyzing & Testing, an LFA 467 HyperFlash® was used to measure the
thermal diffusivity of these different mixtures of PA12 powder with copper particles in
comparison to the neat PA12 material.
As explained earlier, a short light pulse heats the bottom surface of the sample and the
temperature rise on the rear surface is measured as a function of time using an IR
detector.
This is repeated for each temperature step after the sample temperature is stabilized
and the flash lamp is also fired several times over a span of minutes.
After loading the samples, the measurement is started using the conditions
summarized in the following table:
The NETZSCH Proteus® software automatically fits a suitable model to the measured
data to allow calculation of the half-times, Figure 3-31.
T inf=1,577
1,5
IR-Detector/V
0,5
0
-10000 0 10000 20000 30000
Time /ms
Figure 3-31 Example of detector signal as a function of time at 25°C with the fitted curve (red)
for the sample with 5 vol% Cu spheres
Figure 3-32 shows the analyzed thermal diffusivity as a function of temperature and
sample orientation for the neat PA12 in comparison to the PA12/copper sphere
mixtures.
Page 51
Figure 3-32 Temperature dependence of the thermal diffusivity in three measurement
directions: Comparison of the neat PA 12 sample and the PA 12/Cu spheres mixture
As expected, the neat PA12 samples show no directionality and the thermal diffusivity
values are lowest. They show the typical decrease with increasing temperature up to
the melting temperature.
The samples with 5 vol% Cu spheres show slightly higher values for the thermal
diffusivity than the neat PA12 and the samples with 10 vol% Cu spheres show the
highest values of the three materials. This is a result of the higher thermal diffusivity of
the copper compared to the insulating matrix. For most samples, no directionality is
observed due to the isotropic properties of the spheres. However, for the sample with
10 vol% Cu spheres in the thickness direction z, the thermal diffusivity is slightly lower
than for the other two directions. This is likely related to the higher porosity of these
samples, which was measured by Lanzl et al. [2]. The LFA results indicate higher
porosity between the layers in the z-direction than within a layer in the xy-plane.
A different behavior is observed with the copper flakes as shown in Figure 3-33, where
the thermal diffusivity measurements of all samples in the x-direction and of the flakes
in all three directions are compared.
Page 52
Figure 3-33 Temperature dependence of thermal diffusivity in three measurement directions:
Comparison of the PA 12/Cu flakes and isotropic materials (blue – only x-direction)
The flakes show much higher values for the thermal diffusivity than the other mixtures
with spheres and the neat PA12. The high degree of anisotropy is expected based on
the 2D character of the filler. The highest thermal diffusivity is measured in the y-
direction followed by the x-direction. The lowest values are achieved through the
thickness of a layer in the z-direction. This indicates a higher preferential orientation in
the xy-plane, which is likely due to the powder application process.
Figure 3-29 at the top shows a microscopy image of the cross-section of a single layer
of the PA12/Cu flake mixture as reported by Lanzl et al. [1]. The image shows that the
particles touch each other and therefore, the overall thermal resistance of the material
(or here cross-section) should be minimized. The majority of the fillers is oriented
horizontally, which corresponds to the xy-plane. However, it can be seen that some
flakes are tilted at an angle, which results in the higher thermal diffusivity in the z-
direction compared to all other samples.
Thermal diffusivity measurements provide significant insight into both the orientation
of the fillers and their vicinity to each other without the need for additional optical
imaging.
Page 53
λ(T)=a(T)∙cp(T)∙ρ(T) (Eq 3-10)
Both the thermal diffusivity and specific heat capacity are measured as a function of
temperature. The measurement and results of the cp measurements are explained in
chapter 3.3.11. However, the density requires reusing the density at room temperature
as well as the thermal expansion coefficient for the investigated temperature range:
The density at room temperature was measured by the buoyancy-flotation method with
water, the thermal expansion coefficient, α, is measured with a Thermomechanical
Analyzer (TMA), which is explained in chapter 3.3.14. The coefficient of expansion is
direction-dependent und is calculated as follows
αv = (αx + αy + αz)/3 (Eq 3-12)
The resulting computed thermal conductivity values are plotted in Figure 3-34 as a
function of temperature for the different materials and mixtures.
Figure 3-34 Temperature dependence of the Thermal conductivity in three directions for neat
PA 12 and PA12/Cu mixtures
The same trends as seen for the thermal diffusivity are observed:
With increasing copper content, the thermal conductivity is increasing.
The Cu spheres show mainly isotropic behavior. Differences in the values are
related to the porosity of the samples.
Page 54
The copper flakes show the highest increase in thermal conductivity as the
fillers partly touch and reduce the conductivity resistance of the composite
material.
The copper flakes show anisotropic behavior due to their 2D geometry and the
powder application process.
However, the reduced temperature dependence as well as the slight curvature at low
temperatures is related to the temperature dependence of the cp values.
For the application of such conductive fillers in thermal management, it is important to
adjust the orientation of the 3D-printed parts to account for any anisotropy due to the
coating process and filler geometry.
Regarding the process settings and in particular the build temperature, it was observed
that the mixture of the flakes needed to be processed at a build temperature of 173°C,
which was 6°C higher than the mixtures with spheres. The higher thermal conductivity
and the lower specific heat both lead to a reduced capability of heat storage in the
compound and better discharging of the heat. Especially in the xy-plane, where the
highest conductivities with Cu flakes were obtained, it is to be expected that the energy
input from the laser is more rapidly distributed, leading to a lower temperature. Thus,
increasing the build temperature is counteracting this effect.
For better understanding the influence of the different filler shapes on the energy input,
Lanzl et al. [1] analyzed the thickness of a single layer. It was found that the layer
thickness of the mixture with Cu flakes is significantly thinner. The researchers
attributed this to the increased thermal conductivity in the xy-plane compared to the
thickness direction and also to the increased diffuse reflection of the laser, which
results in lower energy input. This additional analysis highlights the importance of
understanding the changes in thermal diffusivity and conductivity for all aspects of the
PBF process and the most suitable process settings.
A summary of the study can be found here: Measuring Thermal Diffusivity using LFA -
YouTube
Polymers shrink. Most of the shrinkage happens during the cooling step of polymer
processing. Depending on the processing conditions, the finished plastic part may
continue to shrink very slightly until temperature and moisture content stabilize, or
during use, if re-crystallization or relaxation effects occur. A way to reduce shrinkage
and increase dimensional stability of polymers is the addition of inorganic fillers. While
this has been common practice in conventional polymer processing, it has also been
picked up by the Additive Manufacturing community; for example, in the PBF process.
In order to have the maximum reduction of shrinkage, but the lowest possible
interaction with processing, isotropic, bead-shaped fillers can be used that have a
similar size or size distribution as the polymer powder itself.
Such filler materials used in PBF are glass beads (as discussed in various previous
chapters) and specifically hollow glass beads that do not counteract the light weight
potential of polymers.
The thermal expansion of a filled system (c) follows roughly the rule of mixture:
Page 55
αc= αm∙Vm+αf∙Vf (Eq 3-13)
where
m denotes the matrix,
f denotes the filler and
V is the volume fraction in %.
The samples were prepared as part of a study [3] by researchers at the Polymer
Engineering Center (PEC) at the University of Wisconsin-Madison. They used different
mixtures of glass beads (true density = 0.456 g/cm3) with PA12 powder (density = 0.95
g/cm3) in a Sinterit Lisa desktop PBF printer with a diode laser instead of the otherwise
typical CO2 laser. The print bed temperature was set to 177.5°C.
At NETZSCH Analyzing & Testing, unfilled samples as well as samples filled with
5 wt% hollow glass beads were analyzed using the NETZSCH TMA 402 F1 Hyperion
®. In order to determine the coefficient of thermal expansion (CTE), samples were cut
The resulting graph for the measurement of the sample with 5 wt% glass beads is
shown in Figure 2. It can be seen that the thermal expansion is almost isotropic as
would be expected with such 3D fillers with an aspect ratio close to 1. However, it can
also be seen that the expansion through the thickness of the parts in z-direction is
lower than for the other two directions. This could be related to porosity effects between
the layers or changes in bonding strength within a layer compared to between layers.
An indication of the validity of this assumption is given by additional mechanical testing
results from the study [3]. The researchers showed that with increasing filler content
up to 5 wt%, the brittleness of the samples increased due to porosity.
Page 56
Figure 3-35 Measured length change of the sample with 5 wt% glass beads
The authors also showed that the fillers act as nucleation sites and affect the
crystallization behavior of the PA12 powder [3]. The same effect was observed with
copper spheres as fillers and the required DSC analysis is explained in chapter 3.3.10.
The effect of isotropic fillers like beads (or any shape with an aspect ratio of 1) was just
explained in the previous chapter. Here, the effect of anisotropic fillers with higher
aspect ratios like flakes and fibers will be studied. The addition of such fillers not only
reduces the overall shrinkage, but rather reduces it differently in different
directions depending on the filler orientation in the parts.
For the analysis, again samples from the Institute of Plastics Technology (LKT) at the
University of Erlangen-Nuremberg were used, chapter 3.3.12.1. Researchers
produced different mixtures of PA12 powder with isotropic copper spheres and
anisotropic flakes in varying contents.
At NETZSCH Analyzing & Testing, all samples were analyzed using the NETZSCH
TMA 402 F1 Hyperion ®. For the determination of the coefficient of thermal expansion
(CTE), samples were cut from dog bone specimens in three different directions, Figure
3-20, x- and y-direction: 10x5x4.5 mm3, z-direction: 4.5x5x5 mm3.
The thermal expansion was measured in a range from -20 to 170 ºC using a heating
rate of 5 K/min. All measurement conditions are summarized in the following table:
Page 57
Table 3-11 Measurement conditions
Figure 3-36 shows the results for the unfilled PA12 and the mixture with isotropic fillers.
dL/Lo /%
1.5
1.0
0.5
0.0
-0.5
-20 0 20 40 60 80 100 120 140 160
Temperature /°C
Created with NETZSCH Proteus software
Figure 3-36 Measured length change as a function of temperature of the neat PA12 sample in
comparison to the sample with 5 vol% Cu spheres in 3 different directions
It can be seen that the thermal expansion is smaller for the filled system than the
unfilled system even though the volume content of 5 vol% is quite small.
Comparing the different directions, we find that the thermal expansion in the thickness
direction is lower for both materials. However, the difference is even bigger for the
copper filled sample. This can be explained with the different solidification and particle
adhesion within a layer (in the xy-plane) compared to the adhesion between layers.
This is typically observed by changes in the mechanical properties, but was also
observed by Lanzl et al. [2] as a change in porosity. Since the researchers found that
the porosity is higher with the copper filled composites, it explains also the bigger
difference between the z and xy-direction. The same effect was observed with glass
beads as isotropic fillers in the previous chapter 3.3.13.
Page 58
The comparison between the different volume contents of Cu spheres is shown in
Figure 3-37. There is no significant change observed between the samples.
dL/Lo /%
1.5
1.0
0.5
0.0
-0.5
-20 0 20 40 60 80 100 120 140 160
Temperature /°C
Created with NETZSCH Proteus software
Figure 3-37 Measured length change as a function of temperature of the two samples with 5
and 10 vol% Cu spheres in 3 different directions
The comparison of different copper shapes at the same volume content of 5 vol% filler
material is displayed in Figure 3-38.
Page 59
dL/Lo /%
1.5
1.0
-0.5
-20 0 20 40 60 80 100 120 140 160
Temperature /°C
Created with NETZSCH Proteus software
Figure 3-38 Measured length change as a function of temperature of the samples with 5 vol%
Cu spheres and flakes, respectively, in 3 different directions
At the same volume content, directionality becomes quite evident. The Cu spheres
show isotropic behavior. In comparison, the flakes lower the CTE in the x- and y
direction and increase it in the z-direction. The reason is the alignment of the fillers.
During the coating process, the flakes are aligned in the xy-plane, thus having the most
pronounced effect in these directions. However, they do not cross over into
neighboring layers or show a significantly enough alignment in the z-direction to make
a huge contribution to the thermal expansion. The value of CTE in the thickness
direction is almost that of the matrix material PA12. As explained earlier, this behavior
is a direct consequence of processing and the alignment of fillers due to that.
To compare the two materials, the coefficient of thermal volume expansion needs to
be taken into account. As both samples have the same copper content of 5 vol%, the
volume CTE should be approximately the same.
For isotropic materials, the volume CTE is calculated as
αv = 3 αl or αv = 3 αx (Eq 3-14)
Page 60
3.3.15 How Fillers Increase the Modulus of PBF Parts through Their
Alignment
Author: Rüdiger Sehling
Page 61
E' /MPa
1500
500
PA12
PA12 Cu spheres 5vol%
PA12 Cu spheres 10vol%
PA12 flakes 5vol% E'
0
Figure 3-39 Storage modulus E’ of neat PA12 and filled samples with 5 and 10 vol% copper
spheres and 5 vol% copper flakes as a function of temperature
Created with NETZSCH Proteus software
Looking at the modulus values of the different samples shows – as expected – the
lowest values for the neat PA12 (e.g., 1480 MPa at 27.5°C and 135 MPa at 167.7°C).
The modulus of the samples filled with 5 vol% Cu spheres shows slightly higher values.
A significant increase is observed with 10 vol% Cu spheres, which shows that even
fillers with aspect ratio = 1 can increase the stiffness of the material when a high
enough filler content is used. However, it can be seen that the highest modulus values
are obtained with 5 vol% copper flakes (e.g., 2278 MPa at 26.7°C). This modulus at
the onset of the glass transition is 54% higher with the flakes compared to the neat
PA12. This can be explained by the predominant orientation of the flakes in the xy-
plane, which is aligned with the axis of testing in the flexural fixture.
Figure 3-40 shows the results of the loss modulus E“ and tan δ of the same samples
as presented in Figure 3-39.
Page 62
E'' /MPa tan d
160 0.20
E'' (1.000 Hz) Peak: 40.7 °C
E'' (1.000 Hz) Peak: 40.1 °C
140 E'' (1.000 Hz) Peak: 41.7 °C
E'' (1.000 Hz) Peak: 41.7 °C
120 0.15
60
40
0.05
20 PA12
PA12 Cu spheres 5vol%
PA12 Cu spheres 10vol% E''
0 PA12 flakes 5vol% tan d
0.00
-50 0 50 100 150
Temperature /°C
Figure 3-40 Loss modulus E” (dashed curves )and tan δ (dotted curves) of neat PA12 and
filled samples with 5 and 10 vol% copper spheres and 5 vol% copper flakes as a function of
temperature Created with NETZSCH Proteus software
These results also show that the characteristic maxima are independent of the sample
variations investigated.
The measurements confirm that the stiffness of the PBF parts is increased by the
addition of fillers regardless of their aspect ratio. It is further shown that for fillers with
higher aspect ratios such as the Cu flakes, even small filler contents can have
significant influence such as a 54-% increase in modulus. This can be used to modify
the material performance without changing to a completely new polymer, which might
be difficult to process in the PBF process.
Page 63
3.4 Powder Bed Fusion with Thermoplastic Elastomers (TPEs)
Authors: Joseph Hofmann, M.Sc., Robert Setter, M.Sc., Prof. Dr.-Ing. Katrin Wudy,
Professorship of Laser-based Additive Manufacturing (LBAM), Technical University of
Munich
In previous sections, the general principles of Powder Bed Fusion were discussed, and
methods to analyze and enhance part attributes were explored. In this section, a new
class of materials, thermoplastic elastomers, will be introduced. The methodology for
their characterization and the intricacies of the processing being researched
Professorship of LBAM will be detailed.
Laser-based powder bed fusion of plastics (PBF-LB/P) enables the production of
architected cellular structures, which have potential applications as replacements for
foams in products that mitigate vibrations and shocks. However, products such as bike
saddles, shoe soles, and paddings tend to be costlier when PBF-LB/P is employed.
The additional costs are attributed to the raw material, the extended manufacturing
duration, and the required post-production steps. To justify the elevated prices of
3D printed products, an additional value must be added through features such as
customization, weight reduction, enhanced ventilation, or superior shock absorption.
To realize these features via PBF-LB/P, a powder that can be passively melted and
shaped like plastic but possesses the elastic characteristics of rubber is needed.
Consequently, architected cellular structures also known as lattice structures are often
fabricated using thermoplastic elastomers, notably thermoplastic polyurethane (TPU).
TPUs are characterized as having nanophase-separated systems with alternating hard
and soft segments. The phase separation between these hard and soft segments gives
TPU its unique elastic properties. The hard phases are responsible for TPU's strength
and toughness, while the soft phases provide elasticity. The reason for this behavior is
that TPUs are linear segmented block copolymers. The hard segments are made by
combining a short-chain diol with an isocyanate. The soft segments consist of polyether
or polyester chains. By varying the composition and combination of these segments,
TPUs with a diverse range of attributes can be synthesized. As a result, thermoplastic
elastomers show a comparatively wide range of thermal and rheological properties,
making determining optimal processing conditions for such powders in PBF-LB/P
particularly challenging.
Page 64
TPU, dist. H2O, 3000 rpm, 10 %
Differential volume dV/dD / %
6 60
4 40
2 20
0 0 100 µm
0.1 1 10 100 1000 10000
Diameter D / mm
Figure 3-41 Particle size distribution (left) and microscopy image (right) of an commercial TPU
powder
Experiments show that this usually results in reduced powder flowability, while
semicrystalline materials seem to have less interparticle adhesion and friction and thus
better flowability [5]. Although TPEs' powder flowability is lower than commercial PA12
powders, the packing densities reach a similar level since the broad particle size
distributions and the compressibility allow more efficient packing [5].
Page 65
transition into the hard phase's melting and crystallization temperature ranges [5].
Suppose a TPU contains larger fractions of hard segments. In that case, it will show a
more abrupt melting in the melting temperature range with a larger melting and
crystallization peak of the respective constituent. Additional low-temperature peaks are
often due to short-range order crystallites, caused for example, by processes such as
annealing [9]. In this case, the other high-temperature peaks correspond with
disrupting the long-range order crystallites in the hard phase [9]. Moreover, some TPUs
might exhibit varied crystalline structures within the hard phase, leading to diverse
melting patterns [10]. Occasionally, TPUs might show peaks stemming from
microphase mixing and demixing [6].
Figure 3-43 represents a commercial TPU powder's DSC diagram (Type: TPU1301,
Manufacturer: EOS, Germany). Such TPU powders typically display a tight process
window between the onset of melting and crystallization. Additionally, the melting
peaks appear notably wide. Furthermore, the processing temperature range
determined using DSC measurements only indicates the actual processing
temperature in a PBF-LB/P process. The reason for this is, that the cooling rate
influences the onset temperature of crystallization since the crystallization behavior is
time and temperature-dependent. However, the cooling rate in PBF-LB/P is highly
dynamic, as shown by the exemplary melt pool temperature during processing in
Figure 3-44. While the cooling rate is constant during a DSC measurement, the cooling
rate in PBF-LB/P is high shortly after laser exposure and small after decay to an
equilibrium state. Reducing the cooling rate generally shifts the onset of crystallization
to higher temperatures, thus reducing the process window further. This makes
pinpointing the precise onset of melting challenging. Consequently, determining the
ideal processing temperatures using DSC becomes more complex than with
PA powder. Hence, understanding the patterns of solidification and shrinkage during
crystallization becomes crucial for an efficient process.
Material: TPU1301; Sample mass: 10.4 mg;
Atmosphere: N2; Heating/cooling rate: 10 K/min
0.6
Specific heat flow / mW/mg
0.4
Onsetc:
0.2 cooling
119 °C
0.0
heating
-0.2 Onsetm:
121 °C
-0.4
-0.6
-50 0 50 100 150
exo
Temperature / °C
Figure 3-42 Representative DSC diagram for a commercially available TPU powder (Type:
TPU1301, Manufacturer: EOS, Germany)
TPUs exhibit minimal crystallization shrinkage [5]. Also, TPUs can display gradual
crystallization steps, but these are insignificantly small compared to the shrinkage
Page 66
observed in semi-crystalline thermoplastics such as PA12. The likely reason for this
limited crystallization shrinkage is the relatively low volume fraction of crystallites in
TPU materials, as only the hard phase crystallizes [5]. As a result, even when the
TPU's hard phase crystallizes, it doesn't lead to significant shrinkage. This
characteristic reduces issues like curling during processing and warping in the cooling
phase after processing.
During laser exposure in TPU processing, fume generation is commonplace. However,
the changes in molecular weight largely depend on the kind of TPU and the processing
settings [6], moisture or oxygen presence [7], and the inclusion of additives [8]. Thus,
TPU powders should include antioxidants and stabilizers to optimize processing [4].
However, the degradation mechanisms are generally difficult to predict. As such,
PBF-LB/P components made from TPU powders show a comparatively large residual
porosity. Consequently, the mechanical properties are usually significantly lower than
conventional processes (e.g., UTS: 30 to 70%) [5].
Page 67
Figure 3-43 Exemplary melt pool temperatures for TPU (Type: TPU1301, Manufacturer: EOS,
Germany) processed with various energy inputs evaluated using in-situ process monitoring
(infrared thermography)
The combined effects of processing settings, material specifics, and part design on
mechanical properties necessitate a more comprehensive experimental design in
PBF-LB/P. Yet, the impact of these interdependencies on the properties of unit cell
struts has not been thoroughly studied for TPU. TPU is frequently chosen to design
components with elastic features by customizing architected cellular structures.
However, the lack of a complete understanding of the influences on the mechanical
properties of TPU components in PBF-LB/P limits the use of topology optimization for
the automated design of architected cellular structures. As a result, the design of
applications involving lattice structures is often still based on a trial-and-error approach.
In recent years, there has been an increase in research efforts aimed at better
understanding the effect of unit cell design, precisely cell type, cell size, and strut
diameter, on the compression and damping properties, as well as user comfort, of TPU
architected cellular structures. Hofmann et al. [11] found that using smaller cell sizes
or larger strut diameters increases the compressive modulus, indentation hardness,
compression set, and hysteresis loss rate of the architected cellular structure.
Meanwhile, changes in unit cell type (face-centered cubic and Kelvin cell lattices) have
minimal effect on mechanical properties. To further improve the uniformity of local
pressure distribution, an additional shell layer can be combined with smaller cell sizes
and larger strut diameters, reducing skin irritation for users in applications such as
cycling. These findings are used to design a topology-optimized bicycle saddle tailored
to a specific load case, Figure 3-44.
Page 68
Figure 3-44 Prototype of 3D printed bike saddle showing a topology optimized architected
cellular structures based on a customized load model [11]
References
[1] Schmid, M. (2018): Laser Sintering with Plastics - Technology, Processes and
Materials, Carl Hanser Verlag, Munich.
[2] Lanzl, L., Wudy, K., Greiner, S., Drummer D., Powder Bed Fusion of Copper Filled
Polyamide 12: Characterization of Powder Properties and Process Behavior, Polymer
Composites, pp. 1801-1809, 2019
[3] Klett, J., Osswald, T.A., Cholewa, S., Investigation of glass bubbles iM16K
polyamide 12 composites for Powder Bed Fusion, ANTEC conference 2020, March
31, 2020
[4] Drobny J. G. (2007): Handbook of Thermplastic Elastomers, Elsevier Inc.,USA,
ISBN: 978-0-323-22136-8
[5] L. Verbelen et al., “Analysis of the material properties involved in laser sintering of
thermoplastic polyurethane,” Additive Manufacturing, vol. 15, pp. 12–19, 2017, doi:
10.1016/j.addma.2017.03.001.
[6] A. Saiani, W. A. Daunch, H. Verbeke, J.-W. Leenslag, and J. S. Higgins, “Origin of
Multiple Melting Endotherms in a High Hard Block Content Polyurethane. 1.
Thermodynamic Investigation,” Macromolecules, vol. 34, no. 26, pp. 9059–9068, 2001,
doi: 10.1021/ma0105993.
[7] G. Lu, D. M. Kalyon, I. Yilgör, and E. Yilgör, “Rheology and extrusion of medical-
grade thermoplastic polyurethane,” Polym Eng Sci, vol. 43, no. 12, pp. 1863–1877,
2003, doi: 10.1002/pen.10158.
[8] Vasquez, M. Analysis and development of new materials for polymer laser sintering,
Dissertation, 2019, Loughborough University
Page 69
[9] A. Frick and A. Rochman, “Characterization of TPU-elastomers by thermal analysis
(DSC),” Polymer Testing, vol. 23, no. 4, pp. 413–417, 2004, doi:
10.1016/j.polymertesting.2003.09.013.
[10] R. M. Briber and E. L. Thomas, “Investigation of two crystal forms in MDI/BDO-
based polyurethanes,” Journal of Macromolecular Science, Part B, vol. 22, no. 4, pp.
509–528, 1983, doi: 10.1080/00222348308224773.
[11] Hofmann, J.; Koller, N.; Bucht, L.; Wudy, K.: Influence of the Design of Architected
Cellular Structures on the Mechanical Properties in Laser-based Powder Bed Fusion
of Thermoplastic Polyurethane, Presentation, FEMS EUROMAT 2023 Conference,
2023, Frankfurt, Germany
[12] C. A. Chatham, M. J. Bortner, B. N. Johnson, T. E. Long, and C. B. Williams,
“Predicting mechanical property plateau in laser polymer powder bed fusion additive
manufacturing via the critical coalescence ratio,” Materials & Design, vol. 201, p.
109474, 2021, doi: 10.1016/j.matdes.2021.109474.
[13] J. Hofmann, K. Taphorn, J. Herzen, and K. Wudy, “Influence of a Varying Scan
Vector Length on the Spatially Resolved Melt Pool Temperature and the Resulting
Porosity of Components in Laser-Based Powder Bed Fusion of Polyamide 12,”
Proceedings of the 19th Rapid.Tech 3D Conference (9 - 11 May 2023, Erfurt,
Germany), pp. 92–101, 2023.
[14] S.-L. Sindinger, C. Kralovec, D. Tasch, and M. Schagerl, “Thickness dependent
anisotropy of mechanical properties and inhomogeneous porosity characteristics in
laser-sintered polyamide 12 specimens,” Additive Manufacturing, vol. 33, p. 101141,
2020, doi: 10.1016/j.addma.2020.101141.
Page 70
4 Material Extrusion
4.2 Materials
Overall, Material Extrusion offers the widest selection of materials compared to all other
AM technologies. The materials are all thermoplastics, which means they can be
melted and solidified reversibly. This means the filament can be produced in a single
screw extruder and later be reheated and melted in the 3D printer nozzle. The most
common used materials are PLA and ABS. Other materials are PET-G, PC, PEEK,
PEI (Ultem), TPU, PP, PA and composite materials like wood plastic compound (WPC)
or graphite filled filaments.
As water soluble material, PVA and its variants are used.
A material overview is also given here: Episode 17: Materials Used in Material
Extrusion (FDM, FFF) - YouTube
Page 71
during printing (e.g., interdiffusion, relaxation time). These are all quantities that can
be investigated with either capillary or rotational rheometry.
Furthermore, the filament should be flexible enough to be spooled and stiff enough to
be used as a piston to pump material through the nozzle (DMA). It needs to melt
sufficiently fast to be able to reach high print speeds (LFA and DSC). In order to
understand and predict the warpage behavior, the temperature-dependent modulus
(DMA) and coefficient of thermal expansion (TMA) need to be known.
Page 72
4.4.1.2 Material Extrusion 3D printing
During the printing process, the temperature of the nozzle tip as well as the diameter
are most important. To quickly assess the shear rates occurring during a 3D print, the
following calculation can be made.
The apparent shear rate is a function of the volume flow rate and the radius of the
nozzle:
4∙𝑄̇
𝛾̇ = 𝜋∙𝑟 3
The volume flow rate depends on the bead width, layer height and print speed: 𝑄̇ =
𝑤𝑏𝑒𝑎𝑑 ∙ ℎ𝑙𝑎𝑦𝑒𝑟 ∙ 𝑣𝑝𝑟𝑖𝑛𝑡
Typical values found in Material Extrusion:
• Bead width: 0.3…0.5 mm
• Layer height: 0.2…0.5 mm
• Print speed: 50…100 mm/s
• Tip radius: 0.25 mm
For the 3D printer setup used at UW Madison, the following shear rate in the tip can
be calculated for r = 0.25 mm:
𝑄̇ = 0.5 𝑚𝑚 ∙ 0.25 𝑚𝑚 ∙ 100 𝑚𝑚⁄𝑠
𝑚𝑚3
4 ∙ 12.5 𝑠
𝛾̇ = = 1,020𝑠 −1
𝜋 ∙ 0.25𝑚𝑚3
In this case, the shear rate during the 3D printing process is twice as high as during
filament production.
left right
Die diameter 1 mm 1 mm
Die length 16 mm 0.3 mm
Pressure transducers 10 k psi 3000 psi (=20.64 MPa)
Page 73
Figure 4-2 Measured pressure as a function of time for the left and right bore
Figure 4-6 shows the measured data and Figure 4-3 the calculated and Bagley-
corrected shear viscosity data as a function of the shear rate at 205°C.
Figure 4-4 shows the shift of viscosity with increasing temperature to lower values.
Page 74
Figure 4-4 Comparison of shear viscosity as a function of the shear rate at 205, 220 and 230°C
Figure 4-5 shows the viscosity as a function of the shear rate at 250°C – the
temperature used for 3D printing. While the viscosity during filament extrusion ranges
between 300-400 Pa s, the viscosity during 3D printing is 88 Pa s as both the
temperature and the shear rate is higher.
A summary can be found in these two videos:
Episode 18: Relevant parameters for production of ME filament - YouTube
Episode 19: Melt rheology in Material Extrusion using Capillary Rheometry - YouTube
Page 75
4.4.2 Impact of Layer Infill Orientation and Layup on the (flexural) Elastic
Modulus
Author: Rüdiger Sehling
In material extrusion, the area of each layer is filled by the deposition of beads next to
each other until the whole layer is filled. Then the print bed is lowered and the process
is repeated for the subsequent layers. The orientation of the infill is typically
predetermined by the Slicer software. To get a smoother finish, each layer is normally
surrounded by an outline.
The ABS filament was printed on a LulzBot TAZ 5 with a hot end temperature of 205°C
and a build platform temperature of 110°C. All test specimens were printed with an XY
travel speed of 2000 mm/min. The bead orientation of the layers is varied between 0°
and 90°. The samples LxWxT was approximately 50 mm x 10.5 mm x 3 mm.
Figure 4-6 shows the results of one sample with 0° infill orientation. Two transitions
can be observed. The first is the glass transition of butadiene and can be detected by
the peak of the loss modulus at -88.4°C. The second is the glass transition of the blend
and can be determined by the peak of the loss modulus at 102.1°C.
Page 76
Figure 4-6 Storage E‘ and loss modulus E” of ABS sample with 0° infill orientation
The overall modulus values of the 0° samples should be highest as here, the material
behavior of the ABS is tested. In contrast, the modulus of the 90° samples should be
lowest as it tests the interfacial strength between the beads. Figure 4-7 shows the
comparison of the storage modulus, E’, of all different samples. As expected, the 0°
orientation is highest and the 90° orientation is lowest. In addition, the layup with 0°
orientation on the outside has higher modulus values than the samples with 90°
oriented layers on the surface, because bending is a mixed mode between tension and
compression and thus sensitive to the stacking sequence of anisotropic layers.
Page 77
Figure 4-7 Storage modulus E‘ of all ABS sample: 0°, 90°, 0/90° and 90/0° infill orientation
Page 78
5 Vat Photopolymerization
Page 79
Another process variation called Digital Light Synthesis (DLS) is explained in chapter
5.5.1.
Page 80
5.2 Critical Parameters to Study with Thermal analysis and Rheology
Typical photopolymer formulations have a low viscosity compared to thermoplastics.
However, they need to be sufficiently low to flow on top of the part in the top-down
configuration to generate the new layer or to easily fill the gap between the part and
the bottom window in the bottom-up approach. Shrinkage should be minimal to avoid
delamination. Both can be studied and optimized using rotational rheometry.
The curing behavior is of the utmost importance as it needs to cure quickly for fast
printing speeds and no excessive heat should be generated. At the same time, the
curing speed and the degree of cure need to be precisely controlled for good adhesion
between layers, etc. Therefore, different instruments like a UV-DSC, UV-DEA and a
rotational rheometer with UV lamp are useful.
The curing reaction and progression can be studied using a rotational rheometer with
a UV light source. This provides a controlled environment to simulate the VPP process.
It is a previously underutilized but important approach in this topic, because it makes
it possible to investigate the various influences of formulation components and
exposure conditions.
Depending on the use case, light sources with specifically controllable wavelengths or
broadband UV lamps can be used. As explained in chapter 2.6.1.1, the material is
irradiated from the bottom through a glass plate while its rheological properties are
measured through movement of the upper plate. The curing process is monitored by
measuring the loss and storage modulus over time. The time at which the two moduli
intersect is called gel point. The gel time is the time between irradiation start and gel
point.
The following graph shows an example measurement of a newly developed epoxy
resin, Figure 5-1. In this case, curing was measured via oscillatory measurements at a
Page 81
deformation of 0.5% and a frequency of 10 Hz, with a gap distance of 50 μm. The loss
and storage modulus were recorded as a function of time to identify the gel point
(intersection of the two moduli) and correlate it with the formulation and curing
parameters used. UV irradiation was started at the 30-second mark after the start of
the experiment and continued for 5-15 seconds. This is the approximate time per layer.
This was followed by an additional 10 minutes of measurement to validate the
attainment of the plateau of the two moduli.
Figure 5-1 Example measurement to determine the gel time using UV rheology
In liquid samples, the viscous component and, thus, the loss modulus exceeds the
storage modulus. As crosslinking commences, the storage modulus rapidly increases
and intersects with the loss modulus. This change means the formation of a gel-like
structure in the sample and is referred to as gel point. The faster the curing reaction,
the earlier the gel point will be reached.
Page 82
The investigated power range in this study was on the upper intensity ranges up to a
maximum of 7 W/cm² and was limited depending on the wavelength as part of the DoE.
The irradiation time was set to 5 to 15 s since in VPP printing, the layer times are in
this range.
In order to cover all relevant areas, an experimental plan with at least 82 trials was
prepared. To minimize possible sources of error, additional experimental points were
planned, depending on the expected uncertainty. A randomized experimental design
with 130 trials was created for this study. To analyze the results, a prediction plot was
used to represent different factors influencing the target variable, namely the gel time.
Each factor has its own graph showing the influence on gel time when all other factors
remain constant. The solid line shows the predicted influence of the factor on gel time,
while the dashed lines show the confidence intervals that account for measurement
uncertainty and other unpredictable factors.
Figure 5-2 Prediction plot with respect to calculated the gel time (incl. confidence interval) for
an exemplary combination of factor settings (vertical lines and upper labels)
Page 83
without sensitizers. Since the sensitizers, unlike the initiators, had sufficient absorption
at all wavelengths used here, samples with the addition of a sensitizer showed a low
influence on the wavelength.
Using the created polynomial model, the gel time can also be predicted depending on
the factors investigated for the entire operating range and optimized for a minimum gel
time.
This methodology and the lessons learned can be used for fast resin development –
now and in the future.
[3] J. P. Fouassier, J. Lalevée: Photoinitiators for Polymer Synthesis: Scope, Reactivity and
Efficiency, Wiley, Weinheim 2012.
Page 84
TG /%
100 -0.49 %
90
-10.56 %
85
80 -1.74 % 78.82 %
Figure 5-3 Mass change of a 3D-printed, solid rectangular alumina green body (approx. 5 mm
edge length) between room temperature and 800°C
At first, a small amount of water is released from the surface, up to around 100°C.
Then, more evaporation occurs, resulting in humidity and low-molecular hydrocarbons
being released up to 180°C. Thus, a large part of the water present during printing has
already evaporated due to the storage at room temperature. Afterwards, the organic
binder decomposes in multiple steps between 180°C and 550°C. There are three
distinct steps where the mass loss is 8.33%, 10.56%, and 1.74%. However, the
structure of the steps indicate that the binder decomposition and oxidation might be
even more complex. Once the temperature reaches above 550°C, the binder is
completely evaporated and decomposed, and the sample mass remains constant. It is
worth noting that the overall binder content in the sample is very high, at approximately
21.2%, when compared to traditionally manufactured ceramic green bodies.
Page 85
Figure 5-4 Length change of a 3D-printed alumina green body (approx. 11-mm length) between
room temperature and 1650°C, heating and cooling
Figure 5-4 shows the result with an overall shrinkage of ~19.2%. However, the gap
between the heating curve (red) and the subsequent cooling curve (blue) shows that
there is further shrinkage taking place throughout the isothermal segment following the
heating segment.
In Figure 5-5, we can see the density changes that occur during debinding and
sintering (green curve), as well as during the cooling phase (blue curve). To calculate
this, we used thermal expansion values from DIL in both the printing direction (shown
in Figure 5-4) and normal to the printing direction (the results are not shown, as they
show similar temperature dependence but ended up at lower total shrinkage) and
corrected for the mass change using results from TGA (Figure 5-3).
During debinding, density decreases as the organic binder is removed, but then it
increases again during sintering. Looking at the cooling curve, we see that the room-
temperature values for the sintered material amount to 4 g/cm³. This is slightly higher
than literature values for dense alumina (3.98 g/cm³), which could be due to slight
deformation of the green body during the melting decomposition of the binder inside
the DIL.
Page 86
4,5
4
Density /g/cm³
3,5
2,5
Heating
2
Cooling
1,5
0 200 400 600 800 1000 1200 1400 1600 1800
Temperature /C°
Figure 5-5 Calculated density of a 3D-printed alumina green body between room temperature
and 1600°C, heating (debinding and sintering) and cooling (sintered sample)
Figure 5-6 illustrates the thermal diffusivity development during debinding and
sintering, measured via LFA and corrected for the thermal shrinkage investigated via
DIL. Similar to the density development, we observe a decrease in thermal diffusivity
during the combustion of the organic binder (up to 500°C), followed by an increase
during sintering. At 1600°C, the thermal diffusivity reaches around 0.924 mm²/s.
Page 87
2
1,8
Temperature /C°
Figure 5-6 Thermal diffusivity of a 3D-printed alumina green body between room temperature
and 1600°C, heating
Figure 5-7 shows the cooling curve for the sintered material, which exhibits a typical
curve shape for a sintered ceramic. At room temperature, the material has high thermal
diffusivity values of around 8.6 mm²/s, but these values sharply drop towards higher
temperatures.
10
9
Thermal diffusivity /mm2/s
8
7 Heating
6 Cooling
5
4
3
2
1
0
0 200 400 600 800 1000 1200 1400 1600 1800
Temperature /C°
Figure 5-7 Thermal diffusivity of a 3D-printed alumina green body between room temperature
and 1600°C, heating (debinding and sintering) and cooling (sintered sample)
Based on the previous measurements, all relevant data is available to calculate the
thermal conductivity throughout the course of thermal treatment of the ceramic green
bodies. To compute the thermal conductivity, the specific heat capacity, cp, (literature
values for pure alumina from the NETZSCH LFA software, the binder contribution was
not considered for the calculations) and density, , from DIL and TGA measurements
are required:
λ(T)=a(T)∙cp(T)∙ρ(T) (Eq 3-10)
Page 88
For more details, refer to the previous section 3.3.12.4.
Figure 5-8 provides a detailed illustration of the thermal conductivity values at different
temperatures during both the heating and cooling segments. As expected, the thermal
conductivity decreases during binder burnout due to the reduction in density and
thermal diffusivity up to 500°C. However, above 800°C, the thermal conductivity starts
increasing due to the rising thermal diffusivity.
30
Thermal conductivity / W/(m*K)
25
Heating
20
Cooling
15
10
0
0 200 400 600 800 1000 1200 1400 1600 1800
Tiemperature / °C
Figure 5-8 Calculated thermal conductivity of a 3D-printed alumina green body between room
temperature and 1600°C, heating (debinding and sintering) and cooling (sintered sample)
The whole study was published here: J. Blumm, Ch. Strunz. 2022. “Analysis and
Optimization of the Binder Burnout of 3D-Printed Ceramic Components”, Ceramics
Forum International, 3: 35
Due to this effect, the resin stays liquid at the interface to the window and can be easily
released during the upward movement of the platform. The rest of the resin in the gap
that is exposed to UV light, however, is being cured. The process is explained here:
Page 89
Episode 9: CLIP: A Unique Variation of the Vat Photopolymerization Process -
YouTube
To understand the UV curing behavior of a given resin, Differential Photocalorimetry
(given by a Photo-DSC) equipped with a UV light source can be used.
Page 90
In Figure 5-9, the result of UV cure and subsequent thermal cure is shown. At the
beginning of the isothermal segment, the sample is exposed for 6.8 s and the resulting
exothermal enthalpy is measured to be 78.4 J/g. During the following heating step,
thermal curing of the sample happens and is already finished when the end
temperature of 120°C is reached.
Figure 5-9 Photo-DSC measurement of a dual cure polyurethane resin that starts with UV
curing and followed by a thermal curing step
This can be better observed in Figure 5-10, where the deviation from the baseline is
highlighted. It can be seen that the exothermal enthalpy due to the thermal cure is
20.89 J/g.
During UV exposure, the measured energy generated by pure light needs to be
corrected for. For that reason, the UV exposure step is repeated on the fully cured
resin sample and the enthalpy increase is measured. The result is shown in Figure
5-11. The blue curve shows the initial measurement (see Figure 5-9) and the black
curve shows the enthalpy of the UV exposure of the fully cured resin. Using the
subtraction function in the Proteus® software, the corrected enthalpy is calculated and
visualized as the green curve. The exothermal enthalpy after correction is 70.29 J/g.
Page 91
Figure 5-10 Detail view of the thermal curing step of a dual cure polyurethane resin during
heating at 3 K/min
Figure 5-11 Results of UV exposure of the uncured resin (blue), of the fully cured resin (black)
and of the corrected Photo-DSC curve after subtraction (green)
Page 92
5.5.4 The Right Balance Matters
This example shows that with an exposure time of 6.8 s, the majority of curing happens
during the UV exposure compared to the thermal cure (21 J/g). It can be seen that the
combination of a Photo-DSC and conventional DSC functionality is capable of
analyzing such complex material systems. The complete study shows that lower
exposure times shift this ratio in the opposite direction: at low exposure duration, the
majority of the cross-links is formed during the thermal curing step.
The authors combined these results with mechanical tests on the samples and were
able to conclude: the more cure happened due to UV light exposure, the stronger the
resulting parts (see Figure 5-12).
This indicates that thermal cross-linking depends on the previously formed network
during UV exposure. However, the authors also found that if the amount of cross-links
from thermal cure gets too low, embrittlement can occur and thus, in turn, the
mechanical performance can be reduced as well. The whole study can be read here!
Page 93
AM techniques, driving innovation and pushing the boundaries of what can be
achieved in modern manufacturing processes. Their significance lies in enabling the
creation of complex, high-quality prototypes and end-use products with unmatched
precision and speed. However, as AM techniques advance, so do the challenges
toward the individual materials. One such challenge arises in advanced VPP
processes like Multiphoton Lithography as well as hybrid processes like the Fusion
Jetting process (compare chapter Fehler! Verweisquelle konnte nicht gefunden
werden.), where the development of the liquid monomers and oligomers involves
exposure to high temperatures. The inherent limitation in heat resistance of these
materials becomes a significant obstacle in the path of progress. Understanding the
curing kinetics and decomposition behavior of acrylate-based photopolymers under
these conditions is imperative to surmount this hurdle. In the realm of Multiphoton
Lithography, where ultra-short-pulsed lasers interact with materials, the demand for
heat-resistant plastics is paramount.
Page 94
Figure 5-13: Simultaneous Thermogravimetric Analysis (TGA) and Fourier Transformation
Infrared Spectroscopy (FTIR) of acrylate-based photopolymers (exemplary FTIR results in
image B only depicted for Acrylate A) [2]
Page 95
Figure 5-14: Impact of elevated temperatures and different UV-intensities (in mW/cm2) on the
exothermal curing behavior of acrylate-based photopolymers based on Photo-DSC results [2]
As depicted in Figure 5-14, the reaction enthalpy (area below curves) and the reaction
time is strongly influenced by the chosen isothermal temperature as well as the UV-
intensity. In general, increased UV-intensities lead to higher curing speed independent
from the temperature. Closer evaluation of the time until the highest reaction rate is
reached tmax as well as the curing half time t1/2 and the total reaction time ttotal confirm
this behavior (compare Figure 5-15). The reaction rate slightly increases with the
temperature. This effect correlates to the collision theory after Trautz [3] and Lewis [4].
The theory is based on the increased mobility of molecular groups at higher
temperatures and therefore a higher probability for reaction initiating collisions. Above
90 °C, the enthalpy reduces while t1/2 and ttotal increase. This behavior likely correlates
with the previously described mass loss with evolving gases after reaching 150 °C
(compare Figure 5-13). [2]
Page 96
Figure 5-15: Time at highest reaction rate is reached tmax as well as the curing half time t1/2 and
the total reaction time ttotal [2]
Page 97
applications, where consistency and reliability are paramount. Furthermore, the
models serve as predictive tools, offering insights into the behavior of acrylate-based
photopolymers under diverse conditions. Industries reliant on AM can benefit
significantly from these predictions, ensuring the reproducibility of high-quality products
and minimizing material wastage. In summary, developing kinetic models based on
Photo-DSC results provides a systematic and in-depth understanding of acrylate-
based photopolymer curing reactions. These models empower researchers and
industries alike, guiding the precise customization of curing processes, ultimately
advancing the efficiency, reliability, and applicability of additive manufacturing
technologies. For the above-described acrylate-based photopolymers, NETZSCH
Kinetics Neo software (see chapter 2.7) was used to and test the accuracy of the
corresponding model. In case of the analyzed acrylate-based photopolymer
(tradename: UV DLP Hard, from Photocentric 3D, United Kingdom), the highest
correlation was showcased for the Kamal-Sourour model [5]. Exemplified results are
presented in Figure 5-16. [2, 6] A more detailed analysis of different models can be
found in [6].
Figure 5-16: Applicability of different kinetic modeling approaches based on the NETZSCH
Kinetics Neo software (exemplified for a single measurement of Acrylate A) [5]
Page 98
Reference
[1] Obst, P., Riedelbauch, J., Oehlmann, P., Rietzel, D., Launhardt, M., Schmölzer, S.,
Osswald, T.A. and Witt, G. (2020): Investigation of the influence of exposure time on
the dual-curing reaction of RPU 70 during the DLS process and the resulting
mechanical part properties. Additive Manufacturing Volume 32.
https://doi.org/10.1016/j.addma.2019.101002
[2] Setter, R., Schmölzer, S., Rudolph, N., Moukhina, E. & Wudy, K. Thermal stability
and curing behavior of acrylate photopolymers for additive manufacturing. Polymer
Engineering & Science 63, 2180-2192, doi:https://doi.org/10.1002/pen.26355 (2023).
[3] Trautz M. Das Gesetz der Reaktionsgeschwindigkeit und der Gleichgewichte in
Gasen. Bestätigung der Additivität von Cv-3/2R. Neue Bestimmung der
Integrationskonstanten und der Moleküldurchmesser. Zeitschrift für anorganische und
allgemeine Chemie. 1916;96(1):1-28.
[4] Lewis WCM. XLI.—Studies in catalysis. Part IX. The calculation in absolute
measure of velocity constants and equilibrium constants in gaseous systems. Journal
of the Chemical Society, Transactions. 1918;113(0):471-92.
[5] Vyazovkin S, Burnham AK, Favergeon L, Koga N, Moukhina E, Pérez-Maqueda
LA, et al. ICTAC Kinetics Committee recommendations for analysis of multi-step
kinetics. Thermochimica Acta. 2020;689:178597.
[6] Setter, R., Schmölzer, S., Rudolph, N., Moukhina, E. & Wudy, K. Modeling of the
curing kinetics of acrylate photopolymers for additive manufacturing. Polymer
Engineering & Science 63, 2149-2168, doi:https://doi.org/10.1002/pen.26353 (2023).
Page 99
6 Material Jetting
In Material Jetting printheads or print, nozzles are used to supply the photoreactive
resin material. Once a layer of material is jetted onto the build platform, a UV lamp is
moved across the surface to cure the resin by light exposure. Afterwards, the build
platform is lowered by one layer height and the process is repeated. Due to the nature
of the process, support structures are needed to build complex parts. The material for
supports is printed simultaneously with the part and is either used as break away
support or dissolved afterwards in an ultrasonic bath without any marks.
Due to the supply of material through arrays of individual nozzles, both the color and
composition of each droplet can be changed – one drop at a time. In addition to full-
color prints, this also allows for the combination of different resin mixtures to create
multi-material or digital materials. One variation could be the degree of hardness, which
has been shown by Stratasys in their PolyJet technology.
Because of this feature set, material jetting is often used for full color, visual prototypes
and has also been used for injection molds and casting patterns.
Since the typical materials are quite prone to embrittlement, there is significant effort
to improve the material properties with new formulations. Another effort is the addition
of fillers similar to other 3D printing technologies, where the resin is only needed for
the shape generation and is later burned off.
The process principle is explained here: Episode 19: Process Princple of Material
Jetting (MJ, DOD) - YouTube
6.2 Materials
The first materials used in Material Jetting were waxes used for casting. Today, the
majority of materials are photopolymers based on acrylic chemistry similar to those
used in Vat Photopolymerization, which are explained in chapter 5.1.3.
One advantage of the deposition of droplets is the possibility to change the resin
formulation in each drop. This technology was developed by Stratasys and is referred
to as digital material or gradient materials. Thus it is possible to generate material
properties ranging from rigid to flexible with Shore hardness as low as Shore A20.
The materials for break away support are the same materials as the build materials.
However, soluble support material can be used as well, which is also acrylic based and
often dissolves.
The materials used are given in the first half of this video: Episode 20: Benefits and
Applications of Material Jetting (MJ, DOD) - YouTube
Page 100
6.3 Critical Parameters
The resin needs to have a low viscosity so it can be properly pumped and jetted out of
the nozzles. The drop needs to have precise dimensions, it should not splatter and
needs to keep its shape after deposition. Therefore, both capillary rheometry for shear
rate dependent viscosity and rotational rheometry in rotation and oscillation for
viscoelastic behavior is useful.
Similar to the resins used in Vat Photopolymerization, curing needs to happen quickly
and without excessive heat generation. Therefore, UV-DSC and rotational rheometers
with UV lamp can be used.
Again similar to VPP, the shrinkage needs to be minimal to reduce void formation and
thus porous parts. Thus, rotational rheometry and TMA can be useful tools as well.
In bioprinting in particular, the viability of the cells after printing is of importance and
thus are the stresses – mechanical and thermal – that are applied during deposition.
These can be best assessed using rotational rheometry.
In bioprinting, the gel time is of high importance. Some hydrogels gel thermally or
physically over a certain period of time. Therefore, fast gelation at elevated
temperatures as well as delayed gelation can be of interest. Both can be studied using
a heated rheometer.
7 Drop-on Demand (DOD) – Focus on Bioprinting
In cooperation with Prof. Dr.-Ing. Andreas Blaeser and Anna Fritschen, M.Sc., Institute
of Printing Science and Technology, Technical University Darmstadt
Page 101
because printing is done inside a support fluid that makes room for the new printed
structure exactly at the location where it is dispensed. This is called submerged
bioprinting.
7.2 Materials
The material used for bioprinting depends on the area of application; ranging from
cartilage and bone tissue to muscle and neural tissue. Therefore, scaffolds out of a
thermoplastic polymer, such as polylactic acid (PLA) or polyethylene glycol (PEG), and
a bioink, e.g., cell-laden alginate, collagen or modified gelatine, are often used in
combination. In either case, the rheological properties are key to understand and tailor
the 3D printing behavior of such materials.
Another application not typically associated with bioprinting, but closely related, is the
production of cultured meats or in-vitro meats, which are plant-based.
Page 102
Figure 7-1 Schematic of the effect of velocity u and shear stress distribution inside the nozzle
on the cells [6]
In order to study these effects, researchers at the Institute of Printing Science and
Technology at the Technical University Darmstadt investigate the rheological behavior
of hydrogels with and without cells. They use a Kinexus lab+ rotational rheometer with
cone and plate configuration (1°/60 mm cone) due to the very low viscosity of the
hydrogels.
To study the viscosity as a function of shear rate during the dispensing process, a
shear rate sweep at constant temperature was performed for 3 different materials and
the results can be seen in Figure 7-2. Due to the Cox-Merz rule, the shear viscosity
can be plotted as a function of shear rate.
Here, material A is a low concentrated polysaccharide with good printability, material
B is a highly fibrous protein that is especially liked by cells, and material C is a mixture
of B and another material to improve gelation time and printability.
Figure 7-2 Viscosity as a function of shear rate for 3 different materials (same as
depicted in Figure 7-4)
Page 103
It can be seen that material A has the lowest viscosity and shows no shear thinning
behavior. The viscosity of material C is higher and of material B is the highest with both
of these materials exhibiting shear thinning behavior.
With this data, the resulting shear stress for the cells can be approximated [7]. In this
case, the materials cover a large range of viscosities. However, they are either low
viscous or shear thinning, so it can be expected that at the estimated shear rates of up
to 200 000 1/s inside the nozzle, they have low viscosity and therefore inflict low stress
on the cells. At the same time, a higher viscosity at low shear rates also indicates a
certain stability of drops after printing, which can indicate a high print resolution.
As mentioned above, the transition temperature and also time from liquid (sol-state) to
solid (gel-state) are important to ensure that the materials do not gel inside the printer’s
nozzle. To study the sol-gel transition, the viscoelastic properties of the hydrogels are
determined using oscillatory measurements with PU 40 mm. By measuring the storage
G’ and loss shear modulus G” as well as the phase angle as a function of temperature,
the gel point can be identified. It is either the cross-over point of G’ and G” or the point
where the phase angle is 45°, which mostly coincides with the cross-over point. For
the material investigated in Figure 7-3, it can be seen that the gel point occurs at about
11°C.
The measurements further show that the material is liquid at 37 °C (G”>G’), where cells
are in their natural state and printing is preferred. However, this means that the build
platform needs to be cooled or the print has to occur at temperatures below 37 °C in
order for it to gel quickly after deposition.
Another important factor in bioprinting is the final elastic modulus of the gel, as this
influences the final stability of the construct, but also the cell behavior. For example,
liver or brain tissue are very soft, while muscle fibers are much firmer. Hydrogels are
very soft materials with an elastic modulus below 5 kPa, which makes it nearly
impossible to measure their elastic modulus via conventional methods such as tensile
testing. For hydrogels, the rotational rheometer is therefore a well-suited equipment
that can not only measure the modulus, but can also map how the modulus is changed
over time.
Page 104
As an example, the complex shear modulus above the gel point was measured as a
function of shear strain to assess the mechanical properties of three different materials,
Figure 7-4.
Figure 7-4 Complex shear modulus of 3 different materials relevant for the assessment of
tissue stiffness of the 3D printed part
References
[1] Blaeser, A., Campos, D.F.D., Fischer, H., 3D bioprinting of cell-laden hydrogels for
advanced tissue engineering, Current Opinion in Biomedical Engineering, 2017, p 1-
9; https://doi.org/10.1016/j.cobme.2017.04.003
[2] Leberfinger, A.N., Dinda, S., Wu, Y., Koduru, S.V., Ozbolat, V., Ravnic, D.J.,
Ozbolat, I.T., Bioprinting functional tissues, Acta Biomaterialia, Volume 95, 2019,
Pages 32-49; https://doi.org/10.1016/j.actbio.2019.01.009
[3] Matai, I., Kaur, G., Seyedsalehi, A., McClinton, A., Laurencin, C.T.,Progress in 3D
bioprinting technology for tissue/organ regenerative engineering, Biomaterials, Volume
226, 2020, 119536; https://doi.org/10.1016/j.biomaterials.2019.119536
[4] Unagolla, J.M., Jayasuriya, A.C., Hydrogel-based 3D bioprinting: A comprehensive
review on cell-laden hydrogels, bioink formulations, and future perspectives, Applied
Materials Today, Volume 18, 2020, 100479;
https://doi.org/10.1016/j.apmt.2019.100479
Page 105
[5] Gudapati, H., Dey, M., Ozbolat, I., A comprehensive review on droplet-based
bioprinting: Past, present and future, Biomaterials, Volume 102, 2016, Pages 20-42;
https://doi.org/10.1016/j.biomaterials.2016.06.012
[6] Blaeser et al, Controlling Shear Stress in 3D Bioprinting is a Key Factor
to Balance Printing Resolution and Stem Cell Integrity, Advanced Healthcare
Materials, 2015; https://doi.org/10.1002/adhm.201500677
[7] Fritschen, Anna, et al., Influence of the physico-chemical bioink composition on the
printability and cell biological properties in 3D-bioprinting of a liver tumor cell line,
Frontiers in Bioengineering and Biotechnology 11 (2023): 1093101
Page 106
8 Binder Jetting
Page 107
The differences to traditional Binder Jetting are that the deposited ink is not fusing the
particles together, but rather allows the melting through the deposited fusing agent.
But the process also shares similarities in regards to speed, accuracy and cost-
effectiveness.
One of the trends in Binder Jetting is the development of new materials and material
combinations that offer improved properties and functionality. The so-called "Fusion
Jetting" (FJ) represents a pioneering attempt in the field of additive manufacturing (AM)
to simultaneously process both thermosets and thermoplastics into multi-material
parts. The technology finds applications in creating hard-soft structures, smart
structures, and selectively varying mechanical, thermal, and electromagnetic
properties (especially when using inks/binders with filler components). The approach
was initially suggested by Wudy and Drummer [1] and involves integrating two
advanced powder-based AM technologies: Powder Bed Fusion with plastics (PBF-
LB/P) and Binder Jetting. The main differences to the previously described multi jet
fusion process are represented by the combined processing of thermosets and
thermoplastics, the utilization of a CO2 laser, as well as the ability to selectively deposit
material and not just globally. The process, outlined in Figure 8-1, begins with the
application of thermoplastic powder layers on the build surface, followed by the
selective deposition of reactive thermoset-based liquids. These liquids are then cured
using UV-light. The CO2 laser is used to selectively melt the thermoplastic powder,
creating multi-material parts combining thermosets and thermoplastics. A
demonstrator part illustrating this process is depicted in Figure 8-2 [2].
Page 108
Figure 8-1: Schematic representation of the fusion jetting process steps
Figure 8-2: Thermoset/thermoplastic multi-material parts created with the FJ technology [2]
Originally designed for vat photopolymerization, the materials offer rapid curing rates
and adequate thermal stability for processing in high-temperature environments. Its
Page 109
temperature-dependent viscosity ensures successful infiltration of selected
thermoplastic powders without unintended curing at elevated temperatures. The
chemical composition includes proprietary diacrylates (constituting 10 to 15% by
volume) and dimethacrylates (making up 50 to 70% by volume) [14]. This material is
commonly used in Powder Bed Fusion with Laser-Based Powder (PBF-LB/P)
processing and boasts excellent flowability for recoating and melting during
manufacturing. Key mechanical properties of all materials can be found in Table 8-1.
Page 110
Figure 8-3: Investigation of the risk for thermally induced curing reactions based on thermal
DSC analysis of liquid acrylate-based photopolymers without UV-curing [3]
As can be seen in Figure 8-3, exothermal peaks are visible for both acrylate-based
photopolymers which potentially represent thermally induced curing mechanisms
superimposed with decomposition of the material. Subsequent visual inspection of the
samples after testing showcased a change from a liquid to a cured glassy/rigid state
during the testing procedure. Therefore, it can be concluded that the exothermal peaks
can be at least partially related to thermally activated curing. The exothermal reactions
take place above temperatures of 150 °C respectively which also correlate well with
decomposition temperatures determined through TGA (compare results presented in
chapter 5.6). In consideration of the previously defined process window (compare
chapter 3.4), the results indicate minimum risks for unwanted curing and thermal
decomposition of both materials with a build temperature between melting and
crystallization of TPU. [3]
Page 111
with the environmental temperature and accordingly the temperature-dependent
viscosity. Therefore, in case of the infiltration speed, increased temperatures can be
considered beneficial towards rapid processing. Another beneficial aspect is the
increase of the curing speed by increasing the process temperature (as previously
discussed in Fehler! Verweisquelle konnte nicht gefunden werden.). [4]
Figure 8-4: Impact of the processing temperature on the infiltration speed and droplet height
over time [4]
In consideration of the results above, the choice of the right fluid temperature
represents a challenging aspect in the field of FJ processing. An ideal tradeoff must be
found between process stability and process speed based on the selected
temperature. The threshold can be identified in a combined analysis of the rheological
behavior as well as curing behavior of the photopolymers. Accordingly, the
temperature-dependent viscosity (compare results regarding complex viscosity in
Figure 8-5 A) needs to be tracked before, during and after initiation of the curing
reaction with UV-light. Only a limited number of analyzing techniques are capable of
simultaneously characterizing this behavior. One possibility is the combination of a
conventional rotational rheometer with UV-light source. In this case, the lower plate of
the rheometer is replaced with a transparent glass plate, capable of transmitting UV-
light to the sample. Exemplified results based on such an experimental setup are
depicted in Figure 8-5.
Page 112
Figure 8-5: A) Comparison of complex viscosity (rotational rheometer) and ion viscosity
(DEA) of acrylate-based photopolymers (Acrylate B) for increasing temperatures (without
curing); B) Comparison of complex viscosity (rotational rheometer) and ion viscosity (DEA) of
acrylate-based photopolymers (Acrylate B) depend on time with UV-curing;
DEA: DEA 288 Ionic with IDEX 115/35 T sensors, NETZSCH, Germany
Page 113
Figure 8-6: Schematic visualization of the working principle of an IDEX sensor with improved
accessibility of a liquid sample with an external UV-light source (Images: DEA 288 Ionic with
IDEX 115/35 T sensors, NETZSCH, Germany)
The penetration depth of the sensors is approximately 100 µm, which is close to the
desired layer heights of the future FJ process. Instead of the complex viscosity, the so-
called ion viscosity is determined based on the applied voltage 𝑈 (sinusoidal excitation)
as well as the resulting current 𝐼 (answer), and phase shift 𝜑. The ion viscosity 𝜂𝑖𝑜𝑛 is
calculated as follows:
1 1 𝑈∙𝐴
𝜂𝑖𝑜𝑛 = = =
𝜎 𝜀′′ ∙ 𝜀0 ∙ 2 ∙ 𝜋 ∙ 𝑓 𝐼 ∙ 𝑐𝑜𝑠(𝜑) ∙ 𝑑
𝜀 ′′ Loss factor
𝜀0 Permittivity of vacuum
𝑓 Measuring frequency
𝑈 Voltage
𝐼 Current
𝜑 Phase shift
𝑑 Electrode plate distance
𝐴 Electrode plate area
𝜎 Ion conductivity
With DEA, the movability of ions and the rotation of stationary dipoles can be correlated
the ion viscosity 𝜂𝑖𝑜𝑛 (compare Figure 8-7). During curing, an increase of the degree
of cure causes a decrease of the movability of the charge carriers and therefore a
reduced current 𝐼. Accordingly, the ion viscosity 𝜂𝑖𝑜𝑛 is increased (compare equation
for 𝜂𝑖𝑜𝑛 above). As can be seen in Figure 8-5 A, the ion viscosity shows an outstanding
correlation to the conventionally determined complex viscosity. Further, the curing
reaction of the photopolymer can be depicted with comparable stability during the liquid
as well as the crosslinked state (Figure 8-5 B). Another advantage of DEA is
represented by increased measuring frequencies. Accordingly, short-term phase
changes can be detected more precisely. For example, a momentary reduction of the
Page 114
viscosity is detectable just after initiation of the curing reaction which is barely
detectable with low measuring frequencies of a regular rotational rheometer (Figure
8-5 B). This drop in viscosity is called the “initiation phase” of the curing reaction
(definition based on Steinhaus et al. [5]) and can be correlated to the heat emitted
during start of the exothermal curing reaction. Parallel to the Photo-DSC results of 5.6,
an analysis of the UV-light dependent curing behavior of Acrylate A based on DEA is
depicted in Figure 8-8 A and B. A variation of the UV-intensity and the environmental
temperature is performed. In comparison to the Photo-DSC results, the DEA results
showcase additional information by depicting a clear impact of UV-intensity and the
process temperature on the final ion viscosity.
Figure 8-7: Sinusoidal voltage 𝑈 (excitation) and resulting current 𝐼 (answer) and phase shift
𝜑 (left); Increased movability of ions and the rotation of stationary dipoles due to applied
voltage 𝑈
Figure 8-8: Impact of the UV-intensity (A) and the process temperature (B) on curing behavior
of acrylate-based photopolymers Acrylate B (results generated on DEA 288 Ionic with IDEX
115/35 T sensors, NETZSCH, Germany)
Therefore, the ion viscosity showcases higher states of cross-linking even after the
exothermal curing reactions detected with Photo-DSC. The so-called “post-curing”
(compare definitions by Steinhauser [5]) will be part of future investigations. Besides
Page 115
that, a decrease of the initiation phase for higher temperatures can be highlighted
(compare Figure 8-8) which potentially correlates with increased mobility of the
molecular groups according to the collision theory by Trautz [6] and Lewis [7].
All in all, the FJ technology represents a promising AM approach for the creation of
complex multi-material parts. However, the broad range of combinable materials
requires extensive and combinatory experimental approaches for well-matched
processing conditions. A promising experimental method in this case is represented
by DEA, which simultaneously combines information regarding the viscosity and the
curing behavior. Compared to conventional rheometers and DSC, the informational
output is increased due to the outstanding measuring frequency, improved sensor
accessibility, and the ability to create process-like conditions. Future investigations will
be centered around the characterization of the characteristic post-curing reactions as
well as the analysis of powder/liquid samples.
References
[1] Wudy, K. & Drummer, D. Infiltration Behavior of Thermosets for Use in a Combined
Selective Laser Sintering Process of Polymers. JOM 71, 920-927,
doi:10.1007/s11837-018-3226-0 (2019).
[2] Setter, R. & Wudy, K. Simultaneous processing of thermosets and thermoplastics
in additive manufac-turing of multi-material polymer parts Annual Technical
Conference – ANTEC, Conference Proceedings (2023).
[3] Setter, R., Schmölzer, S., Rudolph, N., Moukhina, E. & Wudy, K. Thermal stability
and curing behavior of acrylate photopolymers for additive manufacturing. Polymer
Engineering & Science 63, 2180-2192, doi:https://doi.org/10.1002/pen.26355 (2023).
[4] Setter, R., Riedel, F., Peukert, W., Schmidt, J. & Wudy, K. Infiltration behavior of
liquid thermosets in thermoplastic powders for additive manufacturing of polymer
composite parts in a combined powder bed fusion process. Polymer Composites 42,
5265-5279, doi:https://doi.org/10.1002/pc.26221 (2021
[5] Steinhaus J., Hausnerova B., Haenel T., Großgarten M., and Möginger B., “Curing
kinetics of visible light curing dental resin composites investigated by dielectric analysis
(DEA),” Dental materials : official publication of the Academy of Dental Materials, vol.
30, no. 3, pp. 372–380, 2014, doi: 10.1016/j.dental.2013.12.013
[6] Trautz M. Das Gesetz der Reaktionsgeschwindigkeit und der Gleichgewichte in
Gasen. Bestätigung der Additivität von Cv-3/2R. Neue Bestimmung der
Integrationskonstanten und der Moleküldurchmesser. Zeitschrift für anorganische und
allgemeine Chemie. 1916;96(1):1-28.
[7] Lewis WCM. XLI.—Studies in catalysis. Part IX. The calculation in absolute
measure of velocity constants and equilibrium constants in gaseous systems. Journal
of the Chemical Society, Transactions. 1918;113(0):471-92.
Page 116