AMB2018-03: Benchmark Physical Property Measurements for Material Extrusion Additive Manufacturing of Polycarbonate


Material extrusion (MatEx) is finding increasing applications in additive manufacturing of thermoplastics due to the ease of use and the ability to process disparate polymers. Since part strength is anisotropic and frequently deviates negatively with respect to parts produced by injection molding, an urgent challenge is to predict final properties of parts made through this method. A nascent effort is underway to develop theoretical and computational models of MatEx part properties, but these efforts require comprehensive experimental data for guidance and validation. As part of the AM-Bench framework, we provide here a thorough set of measurements on a model system: polycarbonate printed in a simple rectangular shape. For the precursor material (as-received filament), we perform rheology, gel permeation chromatography, and dynamical mechanical analysis, to ascertain critical material parameters such as molar mass distribution, glass transition, and shear thinning. Following processing, we conduct X-ray computed tomography, scanning electron microscopy, depth sensing indentation, and atomic force microscopy modulus mapping. These measurements provide information related to pores, method of failure, and local modulus variations. Finally, we conduct tensile testing to assess strength and degree of anisotropy of mechanical properties. We find several effects that lead to degradation of tensile properties including the presence of pore networks, poor interfacial bonding, variations in interfacial mechanical behavior between rasters, and variable interaction of the neighboring builds within the melt state. The results provide insight into the processing–structure–property relationships and should serve as benchmarks for the development of mechanical models.


A profound shift has occurred in additive manufacturing (AM) technologies over the past decade. Whereas the technologies were originally applied to prototyping, now they are increasingly used in low-to-medium volume production of useful parts and products. The reasons underlying the growth of AM for these purposes are well documented. They include the ability to customize and quickly iterate designs, to manufacture in non-traditional locations, and to manufacture certain geometries not easily achieved with traditional techniques [1,2,3]. The shift in AM to manufacturing and production means that the focus on final properties, including assurance and prediction of such properties, becomes paramount.

A popular AM technique for polymer structures is thermoplastic material extrusion (MatEx), also referred to as fused deposition modeling (FDM)™ Stratasys, Inc, or fused filament fabrication (FFF). MatEx is an extrusion-based AM technique whereby a polymer feedstock is heated and then forced through a print nozzle. The nozzle is manipulated in 3-D space as the melted feedstock is extruded onto a substrate. The popularity of the method is due to the ease of implementation and the ability to print a large number of thermoplastics. The manner by which the extruded polymer beads solidify into the final part is a complicated process that depends on the material properties, proximity between neighboring extrudate, and thermal history of the material, among other factors [4].

Relative to conventional processing techniques such as injection molding (IM), poor mechanical performance and high anisotropy are well known in MatEx structures and may be caused by molecular alignment, volumetric shrinkage in preferred directions, poor bonding between builds, and high porosity [5, 6]. Several studies have focused on the mechanical behavior of MatEx structures with respect to raster and build orientation [7,8,9,10]. Limited studies have investigated mechanical performance as a function of thermal and flow history [11, 12], or due to microscale behavior of the printed interfaces [13, 14]. Several techniques have been proposed for improving mechanical performance of MatEx structures, including through the addition of low molecular weight additives [15], post-deposition annealing [16], laser-assisted preheating [17], and microwave irradiation of printed welds [18]. However, comparison of mechanical performance of structures processed through various MatEx techniques is greatly limited by lack of testing standards. Further shortfalls of MatEx relative to IM are low speed of manufacturing and higher surface roughness.

A general challenge for MatEx parts is the lack of understanding of the processing–structure–property relationships [1]. There is a need for theoretical and computational models that can provide robust predictions of part properties based on fundamental knowledge of processes and material input parameters. There are now nascent efforts to develop such modeling capabilities in areas such as thermal modeling [19,20,21], flow modeling [22], weld strength [23,24,25,26], and crystallization processes [27]. These have proven powerful, but they have not yet bridged the length and time scales required to make fundamental predictions that start with molecular feedstock and process parameters as input and result in final material property predictions.

The AM-Bench framework is designed to help bridge the gap between experimental and theoretical works by posing a set of challenge problems, each targeted to a particular methodology and material. In a given problem, build parameters and starting material characteristics are specified and the modeling community can use these to predict final properties. The data from this challenge—from material characterization to final properties—are published in order for refined model building [28, 29].

This work presents on the results of the challenge AMB2018-03 that relates to the prediction of properties for a simple rectangular print of a polycarbonate by the MatEx process. We provide, to our knowledge, the most comprehensive single set of measurements conducted on a single print. For the precursor material (the input filament), we provide rheology, gel permeation chromatography, and dynamical mechanical analysis. For the produced part, at the nanoscale level, we provide modulus information via AFM and local mechanics via depth-sensing indentation (DSI). At the mesoscale level, we provide information on the pore structure and shape of roads (i.e., an individual strand of the polymer after printing) through X-ray computed tomography and scanning electron microscopy (SEM). And at the macroscopic level, we provide information on mechanical properties via tensile testing. Three different build orientations are printed and measured in this data set.


The overall geometry of the part is dictated by the requirements for the tensile test. One well-used shape is the dogbone, as prescribed in ASTM-D638. However, there are documented issues whereby stress concentrations develop and cause fractures in the gauge region [30]. Figure 1 describes a few common processing challenges that complicate the mechanical response of MatEx dogbone samples. This complexity can prevent fundamental mechanisms from being characterized.

Fig. 1

Schematic representing various challenges encountered in designing dogbone test specimens processed through extrusion-based manufacturing techniques. a Inner-tab, b outer-tab, c external raster edge, d internal raster edge

Another geometry is suggested by the protocols for fiber-reinforced polymers (FRPs), which are also characterized by strength anisotropy. A typical unidirectional carbon FRP can expect to have two orders of magnitude higher strength and stiffness in the reinforcement direction compared to transverse and shear directions [31]. In simulation and experimental design of FRP materials, there are practical issues that arise from complicated structural geometry due to termination of continuous reinforcement at points such as ply drops/chamfers, bolt holes, and other joints [32,33,34,35,36,37,38]. Typically, this is due to shear stresses associated with load transfer between plies, which carry load along an entire structure, and those that are not continuous. Introduction of shear stress becomes a problem due to the anisotropy in shear versus axial strength. Strength anisotropy can be manipulated [39] by changing the reinforcement angle; a shear failure can be induced from axial loading by exceeding the ratio of shear stress/strength to axial stress/strength as axial load is increased.

Here, we utilize a simple rectangular shape, which acknowledges these discontinuity and strength anisotropy issues. This testing protocol is described in ASTM D3039, which suggests the use of straight sections reinforced in the grip section with tabs and allows for more straightforward simulation verification and validation. The straight section is used to avoid any stress concentrations associated with discontinuous reinforcement and the tabs distribute the grip force evenly to avoid grip failures. The lower the anisotropy in strength and stiffness, the easier it is to avoid unwanted failures from test specimen geometries like a traditional “dogbone” in ASTM-D638 [30].

For the material, we utilize polycarbonate because it is an amorphous polymer and thus avoids the complexities of crystallization transitions. Furthermore, as a single component material, it has a simpler rheology than some of the other commonly used polymers, such as acrylonitrile butadiene styrene (ABS). Finally, due to its toughness and relatively high glass transition, it is considered an engineering thermoplastic.

We characterize the input material (in the form of filament) with three tests that are important for modeling considerations: rheology, gel permeation chromatography, and dynamical mechanical analysis. Rheology provides information on the printability of the material and the shape of the final printed road; the material viscosity must be low enough to allow flow, yet high enough once printed to maintain an acceptable shape [4]. Rheology also provides information for modeling the flow fields in the nozzle [4, 40]. Viscosity has an inverse relationship to diffusivity, which is believed to be a controlling factor in the formation of weld strength [4, 26, 41, 42].

Next, we consider gel permeation chromatography, a commonly used method to assess molar mass distributions (MMD) and the presence of impurities. MMD is important due to its well-documented relationship to viscosity, in particular to the zero shear viscosity, as well as shear thinning and shear-induced crystallinity, which was recently documented in MatEx [43]. It is also an important factor in the strength of the material, and in particular for MatEx, it is important for molecular diffusivity across the interface. Strategies to strengthen welds via manipulation of MMD are now being explored [15, 44].

Finally, we consider dynamic mechanical analysis, commonly used to measure solid-state polymer transitions such as the glass transition point, Tg. We are not aware of published reports that use this method to characterize input polymers. Nevertheless, it is useful as a benchmark because it serves as a check that the measured Tg is in accord with published values; if not, there may be additives. Due to the great dependence of viscosity near Tg, any model that considers both thermal transport and viscous effects will need this value to be correct.

There are numerous possible endpoints for modeling, and so we employ a variety of characterization tools to measure a broad set of properties that can potentially be predicted. For pore structure, we utilize X-ray computed tomography, which provides a three-dimensional perspective [45,46,47], along with scanning electron microscopy (SEM), which provides primarily a higher-resolution 2D perspective [48, 49]. For modulus measurements, we use tensile tests at the macroscopic level [50, 51] and AFM and DSI to provide local values [50]. For fracture and strength measurements, we again use tensile testing and use SEM to visualize the nature of the fracture [10].

Experimental Procedures


Polycarbonate (PC) filament (PC White filament feedstock) was purchased from Stratasys.Footnote 1 The support material used was SR-100 (soluble release, Stratasys), which is a proprietary PTFE-based material.

Thermoplastic Material Extrusion and Specimen Preparation

Sheets of PC were deposited onto soluble release substrates using a Fortus 400 mc MatEx system. During the deposition, the filament was heated to approximately 365 °C, while the bed temperature was kept at 145 °C. A T-10 tip with an inner diameter of approximately 0.13 mm was used for the MatEx process. Square sheets (17.8 cm by 17.8 cm by 1.9 mm) consisting of 15 layers of the PC were deposited. Samples were processed in 0° and in 45° raster orientations with respect to the PC sheet edge. While the support material was designed to be dissolved for removal, for the purposes of this study the SR-100 material was peeled from the PC sheets after processing in order to prevent further variability.

In order to avoid heterogeneities that commonly arise from processing dogbone specimens via MatEx (e.g., raster edge effects, contour singularities, infill-contour bonds), rectangular specimens were cut from the PC sheets using a water-cooled diamond saw. The outer 1.3-cm perimeter of the sheet was discarded. Figure 2 shows a schematic of the material used for the study. In order to prevent premature or unwanted specimen failures, 5.2 cm by 1.3 cm by 0.15 cm fiberglass/epoxy tabs with an 8.5° taper were attached to both sides of each specimen with methacrylate adhesive. Approximately 5 cm of each specimen was un-tabbed and represented the specimen gauge section.

Fig. 2

Schematic showing tensile specimens used in the current study; specimens were cut from sheets printed in a 0° and b 45° orientations. Note that the 90° specimens were similarly prepared by cutting samples transverse to 0° orientation sheets


Rheological samples were prepared by cutting the filament into pellet-sized sections using clean snips. The pellets were dried at 120 °C for 4 h under dynamic vacuum. The dried pellets were then pressed into 25 mm by 1 mm disks at 250 °C under 9 metric tons of pressure for 5 min and relaxed under nominal pressure for 5 min. The disks were stored in an electronic desiccator and re-dried at 120 °C for 4 h under dynamic vacuum prior to use. Rotational rheology measurements were taken on a strain-controlled ARES-G2 (TA Instruments, New Castle, DE) fitted with 25-mm parallel plates and a forced convection oven. The gap was zeroed after the plates were equilibrated at 160 °C for 30 min. The oven gas was switched over to N2, and a 25-mm disk was loaded on the hot plates. Frequency sweeps were run at the following temperatures 160 °C, 170 °C, and 180 °C to 280 °C in 20 °C increments. At each temperature, the frequency range was 0.1 rad/s to 100 rad/s with 3 points per decade. A constant gap was maintained by automatically compensating for the predetermined thermal coefficient of expansion of the 25-mm parallel pates. The strain started at 0.5% for 160 °C, 170 °C, 180 °C, and 200 °C and then doubled for each temperature increment reaching 8.0% for 280 °C. No bubbles were observed in the sample at the end of the experiment. A master curve was produced by horizontally shifting tan(δ) then making small vertical shifts for the moduli and complex viscosity. The plateau modulus was determined using the min method, identifying the modulus value that corresponds to the minimum in tan(δ) [52]. The horizontal shift factors were fit using the Williams–Landel–Ferry (WLF) equation [53]. The results are compared to literature values, other 3D printer filaments, and neat bisphenol-A-polycarbonate (BPA-PC) pellets.

Gel Permeation Chromatography (GPC)

GPC was performed on a Tosoh EcoSEC GPC system with RI detection coupled to a Wyatt Viscostar III differential viscometer and a Wyatt Dawn Heleos II Multiangle light scattering instrument with 18 angles and a laser at 661 nm. The mobile phase is THF at 35 °C. The polycarbonate was separated using 2 Tosoh TSK Gel GMH(HR)-H 5-µm bead linear mixed bed columns. Data analysis was conducted in ASTRA software to determine molar mass and MMD. The dn/dc was 0.1857 mL/g as determined by 100% recovery from the RI signal and agreed with literature values (dn/dc ≈ 0.1855 mL/g).

Dynamic Thermal Mechanical Analysis (DTMA)

DTMA was performed on a strain–controlled ARES-G2. Filament sections were mounted directly on a cylindrical torsion geometry. Filaments were prepared in two different ways. One section of filament was dried at 120 °C for 4 h under dynamic vacuum and then loaded under ambient conditions. Another section of filament was loaded under ambient conditions and then annealed at 150 °C for 1 h under N2. Injection-molded and printed samples were used as received. DTMA was run from -150 °C to 160 °C at 3 °C/min, 0.02% strain, and 10 rad/s with an axial tension of 2.0 N ± 1.75 N. All DTMA experiments were run under an N2 atmosphere.

X-ray Computed Tomography

Samples for structural analysis were cut from the gauge section of two of the tensile specimens, and four regions of interest (ROIs) were scanned in each sample using X-ray computed tomography (XCT). The scanner used was a Zeiss Versa XRM500, using a tube voltage of 60 kV in a cone-beam format. This low voltage was used since the specimens were made from a polymer and greater contrast is achieved at smaller tube voltages. Polymers have a low enough average atomic number and density so that enough X-ray photons were transmitted through the polymeric material at this tube voltage to form clear images. A power of 5 W was used, and the voxel size of the reconstructed images was 13.1 µm. The gauge length samples were placed upright in the path of the X-ray beam, and the four ROIs were evenly spaced along the sample for a total of eight collections of images. One sample was from the 0° build, and the other sample was from the 45° build. Each collection of images had ~ 1000 images, and each image comprised ~ 1000 pixels by ~ 1000 pixels. These images cut across the width and thickness of the specimen, perpendicular to the gauge length, which was vertically oriented. Because of cone-beam artifacts, the first and last 140 images were discarded, such that segmentation could be achieved on the remaining images. Cone-beam artifacts caused too much systematic gray-scale variation to allow for successful segmentation of the affected images. The remaining images were segmented with a single threshold, defining the difference between pore and solid materials.


Scanning electron microscopy (SEM) imaging was performed using a Hitachi 4700 FE-SEM. The fracture surface of each specimen type was sputter-coated with Au/Pd for improved contrast.

Global Mechanical Characterization

Static Tensile Testing

Specimens were tested in tension displacement control using a hydraulic test frame at a rate of 1 mm/min, roughly according to ASTM Standard D3039. Tests were terminated at specimen failure, defined by complete fracture, with measurements of displacement, force, and strain collected at 2 Hz. Values of elastic modulus (linear fit to a strain range of 0.002-0.005), strength, Poisson’s ratio, and strain to failure were recorded.

Strain data were collected using digital image correlation (DIC) with a single 5 megapixel (MP) camera mounted 48 cm from the specimen surface. The surface of each specimen was prepared with black spray paint speckling a target pattern where the feature size of interest was ideally much larger than the average dot pattern. Strains down to approximately 0.00005 were resolved with spatial resolution determined by the smallest computational steps, which resulted in convergence of the strain calculation. In this work, a subset, step, and filter size of 11, 3, and 9, respectively, were chosen.

Local Mechanical Characterization

Instrumented Indentation

Specimens for local mechanical testing were prepared using a Leica Ultracut UCT Ultramicrotome. The 0°, 45°, and 90° samples were processed such that surfaces perpendicular to the tensile loading direction were exposed for indentation and atomic force microscopy (AFM). A glass knife was used to remove slices of material until uniform glassy polycarbonate surfaces were achieved. The orientation, uniformity, and smoothness of the specimens were examined with an optical microscope prior to local mechanical characterization.

Instrumented indentation tests were performed using a Hysitron 950 Triboindenter fitted with a diamond Berkovich tip with a 150-nm radius of curvature. The indenter was operated in load control mode with a peak force of 4000 µN, which resulted in indent depths of approximately 1 μm. Indents were applied using a trapezoidal load profile with a loading rate of 400 µN/s followed by a 5-s hold period prior to unloading. Elastic properties were calculated according to the Oliver–Pharr method [54], assuming Ei= 1140 GPa, νi = 0.07 and νs = 0.40, which correspond to the indenter elastic modulus, indenter Poisson’s ratio, and sample Poisson’s ratio, respectively. νs was calculated on an injection-molded (IM) PC sample. Anelastic effects during indentation were accounted for according to Feng et al. [55] by measuring the probe velocity over the final 2 s of the hold period.

Indentation data were collected from within the builds and at build–build interfaces in the 0°, 45°, and 90° print samples. Tests were performed at multiple locations within the builds and within interfaces, with each indentation performed a minimum of 4 µm away from previous indents in order to prevent prior test interference. For interfacial indentations, data were collected both within a build layer (intralayer) and between adjacent build layers (interlayer).

Atomic Force Microscopy (AFM)

AFM studies were performed on the microtomed 0°, 45°, and 90° samples using an Asylum Research Cypher system. AFM tests were performed in alternating contact (AC) and amplitude modulation–frequency modulation (AM–FM) modes using silicon probes with typical natural frequency, tip radii, and stiffness of 75 kHz, 10 nm, and 4 N/m, respectively. For the mechanical tests, tip radii were calibrated on spin-coated polystyrene films.


Viscoelastic Properties

The rheological behavior of the PC feedstock can be seen in the master curve shown in Fig. 3. In general, the master curve for the PC had features common to a disperse and linear homopolymer [56]. The complex viscosity plateaued at low frequency, with a terminal viscosity (\( \eta_{0} \)) of 774 Pa-s, and transitioned to shear–thinning behavior at higher frequencies with a power law index (\( n \)) of 0.25. At low frequency, the moduli, storage (\( G^{\prime} \)), and loss (\( G\hbox{''} \)) followed the predicted terminal behavior with slopes of 1.93 and 0.99, respectively. The moduli crossed over twice, with a reptation or disengagement time (\( \tau_{d} \)) of 0.368 ms, calculated from the first moduli crossover, and an equilibration time (\( \tau_{e} \)) of 0.318 ns calculated from the second crossover. Finding the minimum of tan(δ) [52, 57], we identified the plateau modulus, \( G_{N}^{0} \)  = 2.10 MPa. Alternatively, the reptation or disengagement time was calculated from \( \eta_{0} = 0.822G_{N}^{0} \tau_{d} \), resulting in a value of 0.448 ms. From the plateau modulus, we calculated the entanglement molecular weight via Eq. 1, which is based on ideal rubber elasticity:

$$ M_{e} = \frac{\rho RT}{{G_{N}^{0} }} $$

where the density (ρ) is described by \( \rho \left( T \right) = 10^{3} /e^{{\left( { - 0.307 + 1.86 \times 10^{ - 5} T^{3/2} } \right)}} \)  = 1.1 g/cm3 at 260 °C, where \( R \) is the gas constant, and \( T \) is the temperature [58]. A \( M_{e} \) of 2300 g/mol was calculated from these values, which agrees well with the literature values of 1810 g/mol [59] to 2500 g/mol [60]. From the horizontal shift factors in Fig. 4, the constants for the WLF equation can be found through Eq. 2:

Fig. 3

Viscoelastic master curve of Stratasys PC (white) at a reference temperature of 260 °C. The terminal viscosity \( \eta_{0} \) = 774 Pa s, disengagement time \( \tau_{d} \) = 0.368 ms, reptation time \( \tau_{e} \) = 0.318 ns, plateau modulus (min tan(δ)) \( G_{N}^{0} \) = 2.10 MPa, power law index \( n \) = 0.25, and terminal slopes 1 and 2 are depicted in the figure

Fig. 4

Horizontal shift factors from multiple BPA-PCs plotted against temperature. Dashed line indicates WLF fit through all the data

$$ \log \left( {a_{T} } \right) = \frac{{C_{1} \left( {T - T_{\text{ref}} } \right)}}{{C_{2} + \left( {T - T_{\text{ref}} } \right)}} $$

Here, \( a_{T} \) is the horizontal shift factor which relates arbitrary relaxation processes (\( \tau_{i} \)) from a temperature (\( T \)) to a reference temperature \( (T_{{{\text{re}}f}} ) \), \( \tau_{i} \left( T \right) = a_{T} \left( T \right)\tau_{i} \left( {T_{\text{ref}} } \right) \), and \( C_{1} \) and \( C_{2} \) are empirical constants: \( C_{1} \)  = 2.52 ± 0.06 and \( C_{2} \)  = 150.3 K ± 1.5 K. The shift factors for the PC were fit along with other BPA-PC measured using the same techniques and agree well with previously reported data [61].


In thermoplastic polymers, DTMA is sensitive to changes in segmental mobility and can be used to identify thermal transitions such as the glass transition temperature, \( T_{g} \), or crystalline melting temperature, \( T_{m} \). Characteristic temperatures are typically defined by peaks in the loss modulus, \( G^{\prime\prime} \), or loss tangent, \( \tan \left( \delta \right) \), and the onset of softening is seen in the behavior of the storage modulus, \( G^{\prime} \). The DTMA curve for the PC, Fig. 5, has features typical of many amorphous thermoplastics. At low temperatures, a broad sub-\( T_{g} \) γ relaxation with a peak in \( G^{\prime} \) and \( \tan \left( \delta \right) \) occurs near − 100 °C [59, 62,63,64,65]. At 140 °C, features corresponding to \( T_{g} \) (alternatively denoted \( T_{\alpha } \) in DTMA literature) begin to appear, starting with the onset of softening in \( G^{\prime} \) at 142.2 °C, a peak in \( G^{\prime\prime} \) at 145.4 °C, and a peak in \( \tan \left( \delta \right) \) at 152.9 °C. The resulting DTMA curves from samples prepared using other methods have the same features, and the resulting characteristic temperatures are reported in Table 1. The temperatures measured fall into the range reported in the literature with values for the \( T_{g} \) from differential scanning calorimetry (DSC) of 142 °C [60] to 145 °C [61, 65] and DTMA values of 149 °C [60] to 150 °C [64].

Fig. 5

Storage and loss modulus and phase angle from dynamic thermal mechanical analysis (DTMA) on Stratasys PC (white). Onset of G′, peak G″ and peak tan(δ) noted on figure. Results are representative of all samples measured

Table 1 Stratasys PC DTMA results prepared under different conditions (IM = injection-molded)


The molar mass distribution, Fig. 6, is typical of addition condensation polymers, broad with a slight shoulder at higher molecular masses. The two moments of the molar mass distribution are Mn = 17,940 g/mol and Mw = 24,710 g/mol, resulting in a dispersity of 1.3.

Fig. 6

Differential mass fraction from GPC of PC


The pore structure of the MatEx samples was characterized through XCT. In MatEx, the pores tend to form in lines parallel to the motion of the print head, since parallel ribbons of polymeric material are extruded in this direction. For 0o samples, these lines ran parallel to the length of the specimen, while in the 45o sample, these lines ran at 45o to the length of the specimen. Clear variations in the pore structure were evident when comparing the 0 and 45 degree specimens. For example, examining a cross-sectional image of the 0o specimen, 58 pore channels were observed in the cross section; in the 45o specimen, 40 channels were observed in the cross section. This discrepancy was expected since the cross section for the 45o specimen was cut at 45o with respect to the pore channels [58/40 = 1.45, and (1.45)2 = 2.10. Note that the square root of 2 = 1.414, which is the ratio of a square diagonal to a square edge].

Using the segmentation method outlined above, the 0o specimen had a porosity of 0.0690 ± 0.0046, and for the 45o specimen the porosity was calculated to be 0.0695 ± 0.0042. The uncertainties were taken from an average over the porosity per cross-sectional image (slice) for each sample. There was an additional uncertainty based on the precise value of the threshold used, which was the same for every image, since variation in the threshold caused variations in the calculated porosity. This uncertainty was estimated to be approximately ± 0.006 or 10% for each image. One more source of uncertainty arose from the total volume of the specimen, and any variations from slice to slice, since the porosity was determined as a fraction of the total volume. The uncertainty in the total sample volume in the image stack was dominated by the measurement uncertainties of the width and thickness dimensions, as measured and averaged over several locations, since the length of each image stack was given more precisely by the number of images multiplied by the voxel size.

Another way to consider the per-slice information was to obtain an approximate measure of the uniformity of the build process by graphing how the porosity varied along cross sections of the gauge portion of the tensile specimens. Note that the qualitative slice-to-slice variation in porosity was only indirectly dependent on the uncertainties of the threshold and sample dimensions. If the pore channels were exactly of the same size and number and arranged in a perfect square array, a graph of the porosity variation versus slice would be a horizontal line, with exactly the same porosity per slice. Even if the pore channels were randomly arrayed and of random size in each image, but still had their total cross-sectional area the same for each slice and were still parallel, this graph would still be a horizontal line. The slice-by-slice porosity variation was therefore sensitive to the size, number, total cross-sectional area, spatial arrangement of the pore channels, and alignment of pore channels. Figure 7 demonstrates this computation for one ROI each for 0o and 45o specimens.

Fig. 7

Porosity vs. slice for one region of interest (ROI) for 0o and 45o specimens

Note that the 0o specimen had a fairly regularly spaced peak structure, with a peak width approximately 60–100 slices (i.e., 0.8 mm to 1.3 mm). The nozzle diameter of the print head was approximately 0.127 mm, so this peak spacing was about 7 to 10 nozzle diameters. The tail of this curve for higher slice number deviated to higher porosity. The 45o specimen gave similar results, but with fewer and wider peaks, likely due to the cross-sectional images cut across the pore channels at 45o. The flat region in the middle of this second curve indicates a region of relatively constant porosity. There was no deviation to higher porosity in the 45o curve as was seen with the 0o curve.


The XCT images were stacked for each ROI in order to characterize the 3D pore structure. This was accomplished by using a burning algorithm, which essentially “lights a fire” at one end of the 3D image stack in the pore region and then “burns” to the opposite end in order to determine continuity of the pore channels throughout the specimen—the pores are deemed to be “combustible” [66]. This procedure is limited by the voxel size, since for a smaller voxel size, a channel that was closed as imaged with the original voxel size may now appear open, if its size was smaller than the original voxel size [67]. Therefore, at this voxel size, the number of open channels is a lower bound. Note that this procedure did not conclude that any one pore channel was continuous, but only that the pore channels taken as a whole had continuity. In Fig. 8, the upper segmented image shows all the visible pore channels (white) in cross-sectional slice #1. After the burning process was carried out, it was found that many pore openings, at least 50%, did not lead to continuous channels, so they are blacked out and only pore channels that are part of a continuous network are shown in white. Figure 9 shows similar images for the 45o sample, where the upper image is time-unsegmented. In the lower segmented image in Fig. 9, it is clear that only a few pore channels are non-continuous, with the non-continuous pore channels changed to black. It may be that building at 45o and then slicing at a 45o angle across pore channels introduced a larger degree of interconnectivity between pore channels, by having larger diameter pore channels that were better resolved by the fixed XCT voxel size used.

Fig. 8

0o sample. Upper image (segmented) shows all the pore openings (white) visible on the segmented slice #1. Lower image (segmented) shows all the pore openings on slice #1 that are part of a connected pore system

Fig. 9

45o sample. Upper image (unsegmented) shows all the pore openings visible on the segmented slice #1 (black). Lower image (segmented) shows all the pore openings (white) on slice #1 that are part of a connected pore system

Quasi-Static Tensile Testing

Typical quasi-static tensile behavior of the MatEx samples can be seen in Fig. 10. Irrespective of raster orientation, the samples responded fairly linearly for much of the stress–strain response, with a brittle final failure for each raster orientation.

Fig. 10

Typical (top left) axial stress–strain and (top right) axial–transverse strain behavior (data shown for 0° specimen). (Bottom) typical axial stress–strain behavior for each MatEx sample; data are compared to typical behavior of the PC processed through IM

Figure 10 (bottom) also compares the typical behavior of the PC structures processed through MatEx to the PC processed through IM. The average modulus response across IM and 0, 45, and 90° specimens was 2.73 GPa, 2.04 GPa, 1.44 GPa, and 1.51 GPa, respectively. Some discrepancy in the MatEx values comes from the compaction and co-mingling of neighboring rasters as well as the void content. In general, if an accurate representation of the contact area throughout the volume was resolved, the area-normalized values like stress would likely be more similar across MatEx and IM specimens as the continuous area varies, as seen in SEM and XCT imaging.

The tensile strength was the most obvious distinction in mechanical performance of the MatEx samples processed through various raster orientations. The 90° specimen strength was 19% of the 0° average, whereas the ratio of the 90° / 0° tensile elastic modulus was 74%. In general, the MatEx process creates more anisotropy in strength than stiffness, which is the expected response given the high porosity and poor interfacial welds. Thus, the MatEx specimens showed less ability to deform plastically compared to the IM specimens, which were able to deform to approximately 50% plastic strain. The geometric detail, e.g., the co-mingling of neighboring rasters and void content, likely dominated the plastic response of the MatEx samples. The results of the global mechanical tests are summarized in Table 2.

Table 2 Summary of bulk quasi-static tensile tests showing elastic modulus (E), tensile strength (σ), Poisson’s ratio (ν), and strain to failure

Scanning electron microscopy studies provided insight into the discrepancy in mechanical behavior between the MatEx samples processed in various raster orientations. The fracture surface of a specimen with 0° raster orientation is shown in Fig. 11. Close observation of the fracture images shows there was large deformation throughout the fracture surface, evident in the increased diameter of deformed rasters (Fig. 11a) and interfacial plastic deformation between rasters (Fig. 11c). As expected, localized plastic deformation led to higher toughness in the 0° specimen as compared to the 45° and 90° raster orientations, which was evident in the increased strength and %-elongation in Fig. 10. As shown in Fig. 11c, at the interface between rasters, large deformation resulted in reorientation and stretching of polymer chains, which lead to crazing. Localized yielding at these interfaces provided the toughening mechanism necessary to sustain more plastic deformation and thus increased elongation. A limitation of a typical MatEx process is the resulting voids between rasters. As observed in Fig. 11b, failure can initiate at these sites, leading to premature failure of the specimen relative to the IM specimens (Fig. 10).

Fig. 11

Fracture surface of specimen with 0° raster orientation showing a large deformation of rasters, b initiation of cracks at voids, and c localized interfacial yielding

One unique feature in the 45° specimen not observed in the 0 and 90° cases was a sawtooth-type fracture, both at a global and local scale. At a global scale, fracture did not remain in the same plane of fracture but propagated perpendicular and parallel to the loading direction, in a staircase fashion (Fig. 12a). At a local scale, fracture within an individual raster seemed to occur by formation of large voids in the raster and elongation of the surrounding polymer until fracture occurred in a direction perpendicular to the orientation of the raster. This resulted in the spaced roughness observed in Fig. 12b. As with the 0° raster orientation samples, 45° raster orientation led to localized yielding at the interface between rasters, as shown in Fig. 12c.

Fig. 12

Fracture surface of specimen with 45° raster orientation showing a sawtooth-type fracture, b ductile fracture within raster, c yielding at raster interface

In the 90° orientation specimen, the strength of the part was dependent on the adhesion strength between rasters. Due to the contact area between rasters being small and at times discontinuous (Fig. 13a), failure occurred prematurely, unlike the 0° orientation, where the strength was dependent largely on the material properties within the raster. Localized yielding was not significant in the 90° rasters due to the poor interfacial bonding, yet there was evidence of ductile fracture in the region where rasters were adhered, as shown in Fig. 13b. Another notable feature observed in the fracture surface of 90° raster samples was the lack of crazing as observed in both the 0 and 45° rasters.

Fig. 13

Fracture surface of specimen with 90° raster orientation showing a small contact area and partial/discontinuous bonding of rasters, b ductile fracture within bonded region

Local Mechanical Testing

The local mechanical behavior of the MatEx samples was characterized through nanoindentation tests performed on the printed builds and interfacial welds. Tests were performed on microtomed cross sections of the specimens such that indents were made in the same direction as the tensile loading axis. Typical indent depths were approximately 1 µm. Anelastic effects were accounted for by observing the velocity of the probe during the indentation hold segment, although this behavior was not significant. Figure 14 summarizes the nanoindentation data from tests on the 0°, 45°, and 90° MatEx samples. Comparison of the indentation modulus data between interfaces and builds for a given raster orientation was statistically significant (P < 0.05); elastic behavior comparisons between raster orientations were also statistically significant with the exception of the 0° build and 45° interface case. The mechanical behavior of the 0° specimens was fairly consistent across the builds and interfaces. The 90° specimens typically displayed a relatively compliant response, with indentation modulus approximately 20% lower than the 0° specimens at locations within the builds and at the interfacial welds. The 45° sample exhibited the highest average modulus and hardness, but also resulted in relatively high variation in the mechanical response.

Fig. 14

Modulus (left) and hardness (right) data from nanoindentation experiments on AM PC samples. B represents indentations from within a build, where I represents indentations at an interface between builds

The discrepancy in mechanical behavior could be due to several factors. First, the MatEx process could cause partial polymer alignment, which may be particularly acute in an outer shell of the deposited filament [23]. Another complicating factor is the residual stress of the structure, which may cause apparent variations in the mechanical behavior due to pile-up or sink-in around the probe during indentation [68]. The high thermal gradients and shear rates experienced by the melt during the MatEx process are likely to cause variations in the residual stress across the printed builds and welds. The indentation response would also likely be sensitive to substrate [69, 70] and boundary [71] effects, which could further complicate the local mechanical behavior. The typical indent depths (~ 1 µm) imply that the elastic response is influenced by material of several micrometers from the indent due to the relatively long-range elastic stresses that form under the tip [69]. The substrate and boundary effect issues are particularly of concern for tests on the 45° interfaces, given that material at the testing surface and under the testing surface (i.e., a neighboring filament) may have various alignment and residual stress states.

The local mechanical response can be further understood by dividing the data from tests performed within (intra) and between (inter) build layers (Table 3). In addition to the complications described above, the interfaces within a build layer typically experience greater time in the melt state relative to the interlayer interfaces. Note that the 90° specimen data were strictly from interlayer tests because the tensile specimen cross section did not contain intralayer regions. Figure 15 demonstrates the variation in mechanical behavior across an individual intralayer interface in the 45° sample (the inset in Fig. 15 shows a schematic of the intralayer and interlayer regions of interest). While this intralayer interface consisted of material from filaments deposited in consecutive rasters and thus had a relatively long time to comingle in the melt state compared to interlayer interfaces, the 180° angle between the prints resulted in a sharp variation in the mechanical behavior that extended several micrometers from the center of the weld.

Table 3 Summary of local mechanical properties of AM PC samples, as determined through nanoindentation
Fig. 15

Modulus and hardness response across an intralayer interface in the 45° specimen. Inset shows schematic representation of interfacial types in a 45° specimen

Figure 16 shows an AFM modulus map series performed across an intralayer interface in the 45° sample. The elastic modulus at each of the 512 pixels by 512 pixels is overlaid onto the corresponding topography. The discrepancy in the mechanical response observed through instrumented indentation is likely due to differences in the depths of material examined (10−9 m for AFM versus 10−6 m for indentation) as well as inherent differences in the mechanical test (purely elastic for AFM versus elastic–plastic for indentation). While the topography is fairly consistent across the scans due to the microtome, a distinct interface is apparent due to the variation in mechanical response. Variations in local mechanical properties could also be due to the presence of additives in the feedstock material that may not be apparent from the global mechanical tests but may contribute to interfacial behavior. The typical test in the instrumented indentation study would examine a significant portion of the testing surface shown in Fig. 16c, which helps explain the relatively high variation in mechanical response observed for the 45° specimen. While these local testing methods outlined in this study require further investigation to explore the various combinations of process parameters, this framework could be an effective means for characterizing mechanical behavior across MatEx interfaces and enabling reliable experimental data for ICME studies incorporating nano-/microscale behavior.

Fig. 16

Atomic force microscopy elastic modulus maps of an intralayer interface in the 45° specimen. Modulus maps overlaid onto corresponding height images (z scale height features less than 100 nm). a 15-μm image of the interface; highlighted region shown in b 5-μm image of the interface; highlighted region shown in c 2-μm image of interface


This work describes a methodology for examining the multiscale mechanical behavior of model thermoplastic structures processed through thermoplastic material extrusion (MatEx). Rheological studies performed on the polycarbonate (PC) feedstock reveal a homopolymer with viscoelastic behavior similar to other reported amorphous thermoplastics. PC test specimens were printed in 0o, 45o, and 90o orientations with respect to the tensile loading axis. Guided by similar challenges faced in the fiber-reinforced polymer testing community, global mechanical tests were performed on straight tensile specimens according to ASTM D3039 and compared to specimens processed through injection molding; the results revealed a wide variation in the measured tensile strength and strain to failure due to porosity and poor interfacial bonding between printed builds. X-ray computed tomography was used to characterize the regular pore networks that formed due to the MatEx process and resulted in significant anisotropic mechanical behavior. Instrumented indentation and atomic force microscopy elastic modulus mapping revealed variations in local mechanical behavior in and around the interfacial welds, which was attributed to partial chain alignment, boundary effects, and residual stress disparities resulting from the process parameters.

A repository of the test data used in this study has been made electronically available and may serve as a guide for researchers seeking to validate MatEx simulations [72]. The methodology outlined in this work provides a framework for characterizing multiscale mechanical behavior of specimens processed through MatEx and may serve as a guide for more direct comparison to structures fabricated via advanced AM materials and techniques. In particular, this study could inform constitutive models that incorporate MatEx process parameters and resulting printed structures as determined through XCT and SEM, in order to obtain the global response of the printed parts [73]. In addition, the thermomechanical properties of the precursor materials and the resulting local mechanical behavior of the printed structures (e.g., interfacial behaviors as determined through indentation and AFM) could facilitate advanced simulations of diffusion processes of the welds [26]. Taken together, this work could help enable multiscale testing standards for AM structures and aid in producing reliable experimental data to validate integrated computational materials engineering (ICME) efforts.


  1. 1.

    Certain commercial equipment, software and/or materials are identified in this paper in order to adequately specify the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the equipment and/or materials used are necessarily the best available for the purpose.


  1. 1.

    Goh GD, Yap YL, Tan HKJ, Sing SL, Goh GL, Yeong WY (2020) Process-structure-properties in polymer additive manufacturing via material extrusion: a review. Crit Rev Solid State Mater Sci 45(2):113–133

    CAS  Google Scholar 

  2. 2.

    Huang Y, Leu MC, Mazumder J, Donmez A (2015) Additive manufacturing: current state, future potential, gaps and needs, and recommendations. J Manuf Sci E T ASME 137(1):10

    Google Scholar 

  3. 3.

    Tofail SAM, Koumoulos EP, Bandyopadhyay A, Bose S, O’Donoghue L, Charitidis C (2018) Additive manufacturing: scientific and technological challenges, market uptake and opportunities. Mater Today 21(1):22–37

    Google Scholar 

  4. 4.

    Mackay ME (2018) The importance of rheological behavior in the additive manufacturing technique material extrusion. J Rheol 62(6):1549–1561

    CAS  Google Scholar 

  5. 5.

    Es-Said O, Foyos J, Noorani R, Mendelson M, Marloth R, Pregger B (2000) Effect of layer orientation on mechanical properties of rapid prototyped samples. Mater Manuf Process 15(1):107–122

    CAS  Google Scholar 

  6. 6.

    Sood AK, Ohdar RK, Mahapatra SS (2010) Parametric appraisal of mechanical property of fused deposition modelling processed parts. Mater Des 31(1):287–295

    CAS  Google Scholar 

  7. 7.

    Rankouhi B, Javadpour S, Delfanian F, Letcher T (2016) Failure analysis and mechanical characterization of 3D printed ABS with respect to layer thickness and orientation. J Fail Anal Prev 16(3):467–481

    Google Scholar 

  8. 8.

    Letcher T, Rankouhi B, Javadpour S (2015) Presented at the ASME 2015 international mechanical engineering congress and exposition, 2015 (unpublished)

  9. 9.

    Ziemian C, Sharma M, Ziemian S (2012) Mechanical engineering. InTechOpen

  10. 10.

    Riddick JC, Haile MA, Von Wahlde R, Cole DP, Bamiduro O, Johnson TE (2016) Fractographic analysis of tensile failure of acrylonitrile-butadiene-styrene fabricated by fused deposition modeling. Addit Manuf 11:49–59

    CAS  Google Scholar 

  11. 11.

    Seppala JE, Han SH, Hillgartner KE, Davis CS, Migler KB (2017) Weld formation during material extrusion additive manufacturing. Soft Matter 13(38):6761–6769

    CAS  Google Scholar 

  12. 12.

    Coogan TJ, Kazmer DO (2017) Bond and part strength in fused deposition modeling. Rapid Prototyp J 23(2):414–422

    Google Scholar 

  13. 13.

    Gardea F, Cole DP, Glaz B, Riddick JC (2019) Energy dissipation characteristics of additively manufactured CNT/ABS nanocomposites. Rapid Prototyp J 26(3):509–517

  14. 14.

    Mueller J, Courty D, Spielhofer M, Spolenak R, Shea K (2017) Mechanical properties of interfaces in inkjet 3D printed single-and multi-material parts. 3D Print Addit Manuf 4(4):193–199

    Google Scholar 

  15. 15.

    Levenhagen NP, Dadmun MD (2017) Bimodal molecular weight samples improve the isotropy of 3D printed polymeric samples. Polymer 122:232–241

    CAS  Google Scholar 

  16. 16.

    Hart KR, Dunn RM, Sietins JM, Mock CMH, Mackay ME, Wetzel ED (2018) Increased fracture toughness of additively manufactured amorphous thermoplastics via thermal annealing. Polymer 144:192–204

    CAS  Google Scholar 

  17. 17.

    Ravi AK, Deshpande A, Hsu KH (2016) An in-process laser localized pre-deposition heating approach to inter-layer bond strengthening in extrusion based polymer additive manufacturing. J Manuf Process 24:179–185

    Google Scholar 

  18. 18.

    Sweeney CB, Lackey BA, Pospisil MJ, Achee TC, Hicks VK, Moran AG, Teipel BR, Saed MA, Green MJ (2017) Welding of 3D-printed carbon nanotube–polymer composites by locally induced microwave heating. Sci Adv 3(6):e1700262

    Google Scholar 

  19. 19.

    Yardimci MA, Hattori T, Guceri SI, Danforth SC (1997). In: Bourell DL, Beaman JJ, Crawford RH, Marcus HL, Barlow JW (eds) Sol freeform fabric. Univ Texas Austin, Austin, pp 689–698

  20. 20.

    D’Amico A, Peterson AM (2018) An adaptable FEA simulation of material extrusion additive manufacturing heat transfer in 3D. Addit Manuf 21:422–430

    Google Scholar 

  21. 21.

    Luo C, Wang X, Migler KB, Seppala JE (2020) Upper bound of feed rates in thermoplastic material extrusion additive manufacturing. Addit Manuf 32:8

    Google Scholar 

  22. 22.

    Xia HX, Lu JC, Tryggvason G (2019) Simulations of fused filament fabrication using a front tracking method. Int J Heat Mass Transf 138:1310–1319

    CAS  Google Scholar 

  23. 23.

    McIlroy C, Olmsted PD (2017) Deformation of an amorphous polymer during the fused-filament-fabrication method for additive manufacturing. J Rheol 61(2):379–397

    CAS  Google Scholar 

  24. 24.

    Bartolai J, Simpson TW, Xie RX (2018) Predicting strength of additively manufactured thermoplastic polymer parts produced using material extrusion. Rapid Prototyp J 24(2):321–332

    Google Scholar 

  25. 25.

    Coogan TJ, Kazmer DO (2019) Modeling of interlayer contact and contact pressure during fused filament fabrication. J Rheol 63(4):655–672

    CAS  Google Scholar 

  26. 26.

    McIlroy C, Olmsted PD (2017) Disentanglement effects on welding behaviour of polymer melts during the fused-filament-fabrication method for additive manufacturing. Polymer 123:376–391

    CAS  Google Scholar 

  27. 27.

    McIlroy C, Graham RS (2018) Modelling flow-enhanced crystallisation during fused filament fabrication of semi-crystalline polymer melts. Addit Manuf 24:323–340

    CAS  Google Scholar 

  28. 28.

    Bain ED, Garboczi EJ, Seppala JE, Parker TC, Migler KB (2019) AMB2018-04: benchmark physical property measurements for powder bed fusion additive manufacturing of polyamide 12. Integr Mater Manuf I 8(3):335–361

    Google Scholar 

  29. 29.

    Levine L, Lane B, Heigel J, Migler K, Stoudt M, Phan T, Ricker R, Strantza M, Hill M, Zhang F, Seppala J, Garboczi E, Bain E, Cole D, Allen A, Fox J, Campbell C (2020) Outcomes and conclusions from the 2018 AM-bench measurements, challenge problems, modeling submissions, and conference. Integr Mater Manuf Innov 9(1):1–15

    Google Scholar 

  30. 30.

    Miller A, Brown C, Warner G (2019) Guidance on the use of existing ASTM polymer testing standards for ABS parts fabricated using FFF. Smart Sustain Manuf Syst 3(1):122–138

    Google Scholar 

  31. 31.

    Daniel IM, Ishai O, Daniel IM, Daniel I (1994) Engineering mechanics of composite materials. Oxford University Press, New York

    Google Scholar 

  32. 32.

    Yang J, Song B, Zhong X, Jin P (2016) Optimal design of blended composite laminate structures using ply drop sequence. Compos Struct 135:30–37

    Google Scholar 

  33. 33.

    Sudhagar PE, Babu AA, Rajamohan V, Jeyaraj P (2017) Structural optimization of rotating tapered laminated thick composite plates with ply drop-offs. Int J Mech Mater Des 13(1):85–124

    Google Scholar 

  34. 34.

    Henry TC, Riddick JC, Mills BT, Habtour EM (2017) Composite driveshaft prototype design and survivability testing. J Compos Mater 51(16):2377–2386

    CAS  Google Scholar 

  35. 35.

    Henry T, Mills B (2019) Optimized design for projectile impact survivability of a carbon fiber composite drive shaft. Compos Struct 207:438–445

    Google Scholar 

  36. 36.

    Campilho RD, Banea MD, Pinto AM, da Silva LF, De Jesus A (2011) Strength prediction of single-and double-lap joints by standard and extended finite element modelling. Int J Adhes Adhes 31(5):363–372

    CAS  Google Scholar 

  37. 37.

    Ribeiro T, Campilho R, Da Silva L, Goglio L (2016) Damage analysis of composite–aluminium adhesively-bonded single-lap joints. Compos Struct 136:25–33

    Google Scholar 

  38. 38.

    Liu H, Ojha A, Li Z, Engler-Pinto CC Jr, Su X, Sun Q, Kang H, Wen W, Cui H (2019) Fatigue modeling for carbon/epoxy unidirectional composites under various stress ratios considering size effects. Int J Fatigue 120:184–200

    CAS  Google Scholar 

  39. 39.

    Chamis CC, Sinclair JH (1977) Ten-deg off-axis test for shear properties in fiber composites. Exp Mech 17(9):339–346

    Google Scholar 

  40. 40.

    Coogan TJ, Kazmer DO (2019) In-line rheological monitoring of fused deposition modeling. J Rheol 63(1):141–155

    CAS  Google Scholar 

  41. 41.

    Cicala G, Giordano D, Tosto C, Filippone G, Recca A, Blanco I (2018) Polylactide (PLA) filaments a biobased solution for additive manufacturing: correlating rheology and thermomechanical properties with printing quality. Materials 11(7):13

    Google Scholar 

  42. 42.

    Li H, Zhang S, Yi ZR, Li J, Sun AH, Guo JJ, Xu GJ (2017) Bonding quality and fracture analysis of polyamide 12 parts fabricated by fused deposition modeling. Rapid Prototyp J 23(6):973–982

    Google Scholar 

  43. 43.

    Northcutt LA, Orski SV, Migler KB, Kotula AP (2018) Effect of processing conditions on crystallization kinetics during materials extrusion additive manufacturing. Polymer 154:182–187

    CAS  Google Scholar 

  44. 44.

    Srinivas V, van Hooy-Corstjens CSJ, Harings JAW (2018) Correlating molecular and crystallization dynamics to macroscopic fusion and thermodynamic stability in fused deposition modeling; a model study on polylactides. Polymer 142:348–355

    CAS  Google Scholar 

  45. 45.

    Chen RK, Lo TT, Chen L, Shih AJ (2015) Nano-CT characterization of structural voids and air bubbles in fused deposition modeling for additive manufacturing. In: ASME international manufacturing science and engineering conference, paper no: MSEC2015-9462, V001T02A071.

  46. 46.

    Nouri H, Guessasma S, Belhabib S (2016) Structural imperfections in additive manufacturing perceived from the X-ray micro-tomography perspective. J Mater Process Tech 234:113–124

    CAS  Google Scholar 

  47. 47.

    Xu N, Ye XJ, Wei DX, Zhong J, Chen YY, Xu GH, He DN (2014) 3D artificial bones for bone repair prepared by computed tomography-guided fused deposition modeling for bone repair. ACS Appl Mater Inter 6(17):14952–14963

    CAS  Google Scholar 

  48. 48.

    Hutmacher DW, Schantz T, Zein I, Ng KW, Teoh SH, Tan KC (2001) Mechanical properties and cell cultural response of polycaprolactone scaffolds designed and fabricated via fused deposition modeling. J Biomed Mater Res 55(2):203–216

    CAS  Google Scholar 

  49. 49.

    Shor L, Guceri S, Wen XJ, Gandhi M, Sun W (2007) Fabrication of three-dimensional polycaprolactone/hydroxyapatite tissue scaffolds and osteoblast-scaffold interactions in vitro. Biomaterials 28(35):5291–5297

    CAS  Google Scholar 

  50. 50.

    Cole DP, Riddick JC, Jaim HMI, Strawhecker KE, Zander NE (2016) Interfacial mechanical behavior of 3D printed ABS. J Appl Polym Sci 133(30):43671

  51. 51.

    Ziemian S, Okwara M, Ziemian CW (2015) Tensile and fatigue behavior of layered acrylonitrile butadiene styrene. Rapid Prototyp J 21(3):270–278

    Google Scholar 

  52. 52.

    Liu C, He J, Van Ruymbeke E, Keunings R, Bailly C (2006) Evaluation of different methods for the determination of the plateau modulus and the entanglement molecular weight. Polymer 47(13):4461–4479

    CAS  Google Scholar 

  53. 53.

    Williams ML, Landel RF, Ferry JD (1955) The temperature dependence of relaxation mechanisms in amorphous polymers and other glass-forming liquids. J Am Chem Soc 77(14):3701–3707

    CAS  Google Scholar 

  54. 54.

    Oliver WC, Pharr GM (1992) An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 7(6):1564–1583

    CAS  Google Scholar 

  55. 55.

    Feng Y, Siegmund T, Habtour E, Riddick J (2015) Impact mechanics of topologically interlocked material assemblies. Int J Impact Eng 75:140–149

    Google Scholar 

  56. 56.

    Dealy JM, Read DJ, Larson RG (2018) Structure and rheology of molten polymers: from structure to flow behavior and back again. Carl Hanser Verlag GmbH Co KG, Munich

    Google Scholar 

  57. 57.

    Lomellini P (1992) Effect of chain length on the network modulus and entanglement. Polymer 33(6):1255–1260

    CAS  Google Scholar 

  58. 58.

    Zoller P (1982) A study of the pressure-volume-temperature relationships of four related amorphous polymers: polycarbonate, polyarylate, phenoxy, and polysulfone. J Polym Sci Polym Phys Edn 20(8):1453–1464

    CAS  Google Scholar 

  59. 59.

    Plummer C, Soles C, Xiao C, Wu J, Kausch H-H, Yee A (1995) Effect of limiting chain mobility on the yielding and crazing behavior of bisphenol—a polycarbonate derivatives. Macromolecules 28(21):7157–7164

    CAS  Google Scholar 

  60. 60.

    Wimberger-Friedl R, Hut M, Schoo H (1996) Chain stiffness of copolycarbonates containing a spiro linkage. Macromolecules 29(16):5453–5458

    CAS  Google Scholar 

  61. 61.

    Lomellini P (1992) Viscosity-temperature relationships of a polycarbonate melt: Williams–Landel–Ferry versus arrhenius behaviour. Die Makromolekulare Chemie 193(1):69–79

    CAS  Google Scholar 

  62. 62.

    Illers K, Breuer H (1961) Ein Torsionspendel zur genauen und schnellen Bestimmung der dynamisch-mechanischen Eigenschaften visko-elastischer Stoffe (Messungen an Polycarbonaten). Kolloid-Zeitschrift 176(2):110–119

    Google Scholar 

  63. 63.

    Hoffman JD (1969) Anelastic and dielectric effects in polymeric solids. McCrum NG, Read BE, Williams G (1967) Wiley, New York, pp 617. J Appl Polym Sci 13(2):397–397

  64. 64.

    Yee A, Smith S (1981) Molecular structure effects on the dynamic mechanical spectra of polycarbonates. Macromolecules 14(1):54–64

    CAS  Google Scholar 

  65. 65.

    Hutchinson J, Smith S, Horne B, Gourlay G (1999) Physical aging of polycarbonate: enthalpy relaxation, creep response, and yielding behavior. Macromolecules 32(15):5046–5061

    CAS  Google Scholar 

  66. 66.

    Stauffer D, Aharony A (1994) Introduction to percolation theory. Taylor & Francis, London

    Google Scholar 

  67. 67.

    Garboczi EJ, Bentz DP (2001) The effect of statistical fluctuation, finite size error, and digital resolution on the phase percolation and transport properties of the NIST cement hydration model. Cem Concr Res 31(10):1501–1514

    CAS  Google Scholar 

  68. 68.

    Chen X, Yan J, Karlsson AM (2006) On the determination of residual stress and mechanical properties by indentation. Mater Sci Eng A 416(1–2):139–149

    Google Scholar 

  69. 69.

    Saha R, Nix WD (2002) Effects of the substrate on the determination of thin film mechanical properties by nanoindentation. Acta Mater 50(1):23–38

    CAS  Google Scholar 

  70. 70.

    Forster AM, Michaels CA, Sung L, Lucas J (2009) Modulus and chemical mapping of multilayer coatings. ACS Appl Mater Interfaces 1(3):597–603

    CAS  Google Scholar 

  71. 71.

    Jakes JE, Frihart CR, Beecher JF, Moon RJ, Stone D (2008) Experimental method to account for structural compliance in nanoindentation measurements. J Mater Res 23(4):1113–1127

    CAS  Google Scholar 

  72. 72.

    Cole DP, Gardea F, Henry TC, Seppala JE, Garboczi EJ, Migler KD, Shumeyko CM, Westrich JR, Orski SV, Gair JL (2018) 2018 AM-bench test descriptions for AMB2018-03. 25 September 2020.

  73. 73.

    Garzon-Hernandez S, Arias A, Garcia-Gonzalez D (2020) A continuum consitutive model for FDM 3D printed thermoplastics. Comput Part B 201:108373

    Google Scholar 

Download references


The authors would like to acknowledge useful discussions with Rachel Andrulonis at the National Institute for Aviation Research (NIAR), Erich Bain at CCDC Army Research Laboratory, and Vicky Nguyen and Sung Hoon Kang at Johns Hopkins University.

Author information



Corresponding author

Correspondence to Daniel P. Cole.

Ethics declarations

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cole, D.P., Gardea, F., Henry, T.C. et al. AMB2018-03: Benchmark Physical Property Measurements for Material Extrusion Additive Manufacturing of Polycarbonate. Integr Mater Manuf Innov 9, 358–375 (2020).

Download citation


  • Additive manufacturing
  • Thermoplastic material extrusion
  • Fused deposition modeling
  • FDM
  • 3-D printing
  • X-ray computed tomography
  • Rheology
  • Polycarbonate
  • AM-bench