Shape Memory Polymers in 4D Printing: Investigating Multi-Material Lattice Structures

Abstract

This paper evaluates the design and fabrication of a thermoplastic polyurethane (TPU) shape memory polymer (SMP) using fused deposition modeling (FDM). The commercially available SMP filament was used to create parts capable of changing their shape following the application of an external heat stimulus. The characterization of thermal and viscoelastic properties of the SMP TPU revealed a proportional change in shape fixity and recovery with respect to heating and cooling rates, as well as a decreasing softening temperature with increasing shape memory history due to changes in the polymer microstructure. Inspired by the advancements in 3D and 4D printing, we investigated the feasibility of creating multi-material lattice structures using SMP and another thermoplastic with poor adhesion to TPU. A variety of interlocking lattice structures were evaluated by combining SMP with another thermoplastic that have poor adhesion with TPU. The tensile strength and failure modes of the fabricated multi-material parts were compared against homogenous SMP TPU specimens. It was found that the lattice interface failed first at approximately 41% of the ultimate strength of the homogenous part on average. The maximum recorded ultimate strength of the multi-material specimens reached 62% of SMP TPU’s ultimate strength. These characterizations can make 4D printing technology more accessible to common users and make it available for new markets.

1. Introduction

1.1. Four-Dimensional Printing

The research on fused deposition modeling (FDM) 4D printing technology was pioneered by Tibbits to create stimulus-responsive shape morphing objects using a multi-material FDM Stratasys 3D printer [1,2]. His work allowed for printed objects to change from a one-dimensional string to a two-dimensional surface, or from one three-dimensional shape to another. In 4D printing, this process of shape morphing is referred to as the shape change effect (SCE), also known as the shape memory effect (SME). It is achieved by using a specific type of smart material called a shape-responsive material (SRM) - a material that can perform a shape change upon receiving an external stimulus. SRM describes a wide range of materials that is categorized based on the stimulus to which each material is responsive to, which is governed by the physical-chemical mechanisms that occur on a molecular level [3].

Materials that can change their physical form in response to a stimulus in a non-instant and non-spontaneous manner are referred to as shape memory materials (SMM) and are frequently used in 4D printing applications. SMMs come in the forms of shape memory alloy (SMA), shape memory polymer (SMP), shape memory hybrid (SMH), shape memory ceramic (SMC), and shape memory gel (SMG), all of which can be combined with an additive to create a shape memory composite (SMC). Of the many available materials for 4D printing, the most commonly used ones are SMA and SMP [4], with a specific SMP being the focus of this study. SMPs offer transformative capabilities, as seen in a recent review where examples are provided, such as temperature-sensitive vascular stents that can alter their shape for minimally invasive surgeries and automotive components designed for adaptive aerodynamics [5]. The authors also describe potential uses in tags for ant legs, self-opening holes triggered by heat, and morphing wings. In a different article, Hu et al. lists the following advantages of SMPs that make them so versatile: diverse trigger stimuli, tunable properties, biocompatibility, and good mechanical properties [6].

1.2. Interlocking Lattice Structures

Given the distinctive shape changing nature of 4D parts, SMMs greatly benefit from being used in conjunction with other materials, as shown by the development of low-cost self-deployable structures in aerospace [7], self-assembling chairs, reversible water-responsive composite inks [8], stimulus-responsive hinges [9], as well as flavor and color changing food products [10]. The SMP to be used in this research is based on a thermoplastic polyurethane (TPU) polymer, which naturally exhibits shape memory through its viscoelastic properties, making it a prime candidate for 4D printing. This means that the usefulness of multi-material 4D parts is largely dependent on interlayer or intralayer bonding. Since some polymers used in FDM 3D printing filaments do not adhere well to one another in this manner, creating parts that combine these materials using conventional methods has proved challenging, especially with TPU-based flexible materials. When printing a part of two incompatible materials, a failure will likely occur along the vertical interface between the two materials, as the adhesion would be weakest there. Ultimaker has provided a list of such incompatible materials, where, notably, all combinations of TPU are labeled as “experimental” and are not officially recommended [11]. For these reasons, it would be difficult to produce a dynamic 4D-printed multi-material part, as it would likely unmerge into separate components when stresses are applied.

A novel study published by Kuipers et al., which later became a feature in the Ultimaker Cura 5 slicing software package, presents an algorithm to increase the strength of vertical material interfaces [12]. The paper discusses an interesting approach to multi-material printing by “weaving” together two dissimilar materials and generating a 3D internal structure. A study performed by Aharoni et al. investigated a similar optimization algorithm for joining parts in assemblies, but the algorithm is not available as open-source [13]. This interlocking topologically interlaced lattice (ITIL) is built up layer-by-layer taking inspiration from textile weaving, brick masonry, and a children’s board game involving stacking overlapping wooden blocks. It interlocks all three degrees of freedom, preventing any translation caused by a lack of interlayer bonding. Figure 1A shows a visualization of the top/bottom pattern, the first alternating pattern, and the second alternating pattern, respectively. The part is then built by stacking layer patterns in an ABCBCBCA format. As will be discussed later, it has attempted to manually add similar interlocking structures into polylactic acid (PLA) and TPU multi-material parts in a comparison with the ITIL, but was met with limited success. The materials would separate with repeated handling and shape change actuation. By comparison, fabricated parts with the interlocking material interface exhibited significantly better structural stability by not disassembling, even after repeated use and shape change.

1.3. Novelty and Significance

The novelty of this work is the combination of ITIL and 4D printing concepts. To the authors’ knowledge, this is the first work exploring this unique application of additive manufacturing. This coupling of techniques will improve the practicality and usability of 4D printing, while also showing the versatility of the ITIL in its compatibility with exotic materials like SMPs. Further research in this field will only expand the possibilities of material combinations and end-uses of 3D- and 4D-printed parts.

This study will focus on investigating the following research questions.

• What are the chemical, thermal, and mechanical properties of SMP TPU? It is important to characterize these properties before attempting to create ITIL due to the following two reasons: (a) to ensure that a regular 3D printer can be used to extrude an SMP TPU filament, and (b) to ensure that the properties of this material is compatible with that of PLA so that the properties of the ITIL are not compromised.
• What material properties of SMP TPU influence the shape memory characteristics of the printed samples? It is important to understand this as this characterization will dictate how the ITIL structure behaves.
• What are the tensile and ultimate strengths of the parts printed using the ITIL technique with SMP TPU and PLA? How are these strengths affected by various slicer parameters?

We aim to create a method that can be used to create strong and reliable 4D-printed parts with ITIL structures. The findings from this study provide guidelines for FDM 4D printing of SMP TPU parts and determine which slicing parameters are the most effective at increasing the tensile strength of the ITIL interface.

2. Materials and Methods

2.1. Material Selection

Due to the greater availability and documentation of various polymers in 3D printing, a one-way shape memory polymer was used in this study. SMPs provide advantages over the other classes of SMMs, including lighter weight, low cost, high shape deformability and recovery, and customizable key mechanical properties such as glass transition temperature and melting temperature [14]. Shape memory polymers are not commonly available in a 1.75 mm diameter filament form for FDM printing. In fact, there is only one shape memory polymer filament available off the shelf from the vendor filament2print [15]. This material is advertised as a flexible polyurethane-based 4D filament manufactured by SMP Technologies Inc., Tokyo, Japan and sold by Convena Polymers in Denmark for shape memory applications. A 300 g spool of this unique 4D SMP TPU filament was purchased for use in this project.

Since the material is relatively new in research and its material properties are not provided by the manufacturer or prior studies, it was important to characterize the chemical, thermal, and mechanical properties of it using appropriate characterization techniques. This characterization would help to determine the printability of this material using a regular 3D printer and its compatibility with PLA. Section 2.2, Section 2.3 and Section 2.4 will focus on the methodology used for the characterization of SMP TPU.

2.2. Chemical Characterization

Fourier transform Infrared (FTIR) spectroscopy was performed to determine the composition of this novel material and compare it with library references. The sample was examined in attenuated total reflection (ATR) mode using a Thermo-Nicolet iS50 Fourier Transform Infrared (FTIR) spectrometer. A diamond crystal was used with a typical depth of penetration on the order of 2 microns and a spot size of 2 mm. OMNIC 9.12 software was used to perform data analysis.

Elemental composition to evaluate the presence of fillers or other additives was performed using an X-ray Fluorescence (XRF) on a Rigaku Primus II wavelength dispersive spectrometer (WDXRF), which detects elements with a range from atomic number (Z) 4 (beryllium) through atomic number 92 (uranium) at concentrations from the low parts per million (ppm) range up to 100% by weight. The primary X-ray source comes from a rhodium X-ray tube. The sample analysis was tested using a 20 mm diameter sample holder under a vacuum. Fundamental Parameters (FP) standardless quantification software was used for element quantification. ZSX software was used for data processing.

2.3. Thermal Characterization

Differential scanning calorimetry (DSC) was performed on SMP TPU filament using a TA Instruments Q20. Specimens were weighed on an analytical balance and then sealed in aluminum pans. The pan was punctured to release any moisture that may evaporate during the run. Samples were exposed to a heat–cool–heat cycle from room temperature to 200 °C using a ramp rate of 10.00 °C/min.

Thermogravimetric analysis (TGA) was performed on a PerkinElmer TGA 8000 equipped with an autosampler. Samples were heated from room temperature until 600 °C at a ramp rate of 10 °C/min under a helium atmosphere. The atmosphere was changed to air, and the sample was heated until 800 °C. The experiment aimed to compare the thermal stability of SMP TPU in its filament and printed forms. While 3D and 4D printing rarely require exposing the material to temperatures above 260 °C, except for engineering grade and exotic filaments, 4D-printed parts may be expected to perform in extreme environments exceeding that value.

Dynamic mechanical analysis (DMA) was performed on the shape memory polymer to quantify the loss and storage moduli. In addition to measuring these quantities, DMA also provides a secondary measurement of the glass transition temperature. Twelve 45 mm × 10 mm × 1.6 mm test specimens were printed on the Creality Ender 3 V2 using the SMP TPU filament. The samples were printed using a 0.2 mm layer height with a 0.6 mm nozzle at 30 mm/s. The temperature of the hardened steel nozzle was set to 215 °C, and the bed was heated to 30 °C. These samples were inserted into a DMA 8000 with the geometry plate configured for the dual cantilever bending test at the maximum distance of 40 mm. The length, width, and thickness of each sample were measured and entered into the Pyris software version 7.0 prior to each test run, and the method was configured to heat the sample from the ambient temperature to a maximum of 60 °C at varying rates with a strain of 0.005. The final two groups of parts, 7–9 and 10–12, were analyzed to observe the effect of shape memory programming and actuation cycles had on the loss and storage moduli. Specimens 7–9 underwent five one-way shape memory cycles, whereas specimens 10–12 experienced 10 cycles. The programming phase consisted of heating a specimen in a hot water bath at 60 °C for 20 s to equalize the temperature of the material, subjecting it to a constant strain, then removing it from the water bath and allowing it to cool in ambient air for 20 s. The time of 20 s is a minimum, and increasing these times should improve the magnitude, quality, and longevity of the shape change by ensuring the temperature falls below or rises above the T_g. Following the 20 s of cooling, the strain on the specimen was released. If the specimen held its temporary shape, it was placed on the table. If the specimens did not hold their shape, the strain was reapplied for another 20 s while the material cooled. When all of the specimens had undergone the first programming of the temporary shape, they were placed back in the hot water bath to apply the heat stimulus to return them to the permanent shape. When the actuation was complete and no more motion was visible, all specimens were removed from the water and left to cool in the air once more, marking the conclusion of a single shape memory cycle.

2.4. Tensile Testing

After the characterization of SMP TPU was completed, the next step was to print samples with ITIL structures and then characterize the mechanical performance of those samples using tensile tests. Tensile testing was performed on multi-material specimens using an Instron 4411 Tensile Strength Tester at 5 mm per minute [16]. The specimens followed the ASTM D638-14 Type IV standard, but were modified to include two vertical material interfaces at both ends of the center gage section, as shown in Figure 2 [17].

These tests aim to rank the effect of ITIL generation parameters on the ultimate tensile strength of the specimen. The engineering stress was calculated using the formula in Equation (1), where $A_0$ is the measured cross-sectional area of each specimen and F is the force applied by the Instron. This value was then compared to yield strength, as follows:

$$\sigma ={\displaystyle \frac{F}{A_0}}$$

(1)

The slicer settings used in Cura to generate the gcode used the default 0.2 mm profile, but with the added changes specified in

Table 1. Most critical slicer parameters for printing SMP TPU and their optimal values.

Slicer Parameter	Value
Layer height	0.2 mm
Infill percentage	10%
Nozzle temperature	215 °C
Bed temperature	30 °C
Print speed	30 mm/s

. Higher infill values were not necessary, as the failure was expected to appear at the material interfaces. The bed temperature must be set below the T_g, otherwise the material will never solidify and the print will fail. A buffer of no less than 10 degrees below the T_g is specified by the manufacturer to avoid heat transfer from recently deposited material. The filament showed a tendency to backlash or creep back to a stress-free shape upon being extruded, which was controlled primarily by the lateral speed of the toolhead; therefore, the print speed should be no higher than 30 mm/s for best results on non-specialized equipment. In addition to these changes, the setting titled “Generate Interlocking Structure” was enabled.

Table 2. Slicing parameters for the fabrication of the 12 tensile test dogbone specimens.

Sample Number	ITIL Parameter	Value
1	All (IBW, ISO, IBLC)	Default (0.8 mm, 2, 22.5°)
2
3
4	Interlocking Beam Width	0.4 mm
5
6
7	Interlocking Beam Layer Count	1
8
9
10	Interlocking Structure Orientation	90°
11
12

shows the interlocking structure parameters’ default and the changed values of the parameters.

A visualization of how interlocking beam width (IBW), interlocking structure orientation (ISO), and interlocking beam layer count (IBLC) change the structure of the generated structure is shown in Figure 3. The values of the parameters were left at their default values and varied by 1 DOF to generate 12 specimens across 4 groups, as can be seen in Table 2.

3. Results and Discussion

3.1. Characterization of the SMP TPU

3.1.1. Chemical Characterization

The FTIR analysis revealed that the TPU was similar to a poly (ether urethane) based on 4,4′-diisocyanato diphenylmethane and polyoxytetramethylene, as indicated by a good match with a library reference shown in Supporting Information Figure S1.

The XRF results are also provided in Table S1 in the Supplementary Information, supporting the FTIR finding of a poly (ether urethane) primarily composed of C, O, and N with no significant fillers. In addition, no major changes were observed between the filament and the printed part, although the nitrogen content decreased slightly upon printing.

3.1.2. Differential Scanning Calorimetry (DSC)

DSC analysis of the SMP TPU filament was performed using dual heat–cool–heat cycles, and the results are shown in Figure 4. The first scan (as-received properties) showed a T_g of 46.8 °C and a small endotherm at 162.4 °C. When the sample was heated again, thereby erasing the thermal history, the T_g increased to 54 °C and the endotherm disappeared. These results suggest that the processing of the filament induces some plasticization, which causes the T_g to be artificially lower than it is when the thermal processing is more controlled. Note that the observed value of 54 °C is very close to the “softening temperature” of 55 °C specified by the vendor.

Polyurethane (PU) has been shown to exhibit shape memory in both amorphous and semicrystalline forms [18,19,20,21]. The existing literature also tells us that polymers with a single thermal transition range are known as one-way SMPs, while multiple-way SMPs employ either multiple transition temperatures or a single very broad (glass or melting) transition range [22].

3.1.3. Thermogravimetric Analysis (TGA)

SMP filament and SMP-printed samples were heated from ambient room temperature to 800 °C, and the change in the mass of the sample is shown in Figure 5. The results of the analysis show that both forms of the polymer remain stable until approximately 280 °C, with a minimal difference in the thermal decomposition rates. These findings are significant for several reasons. First, the results establish a maximum temperature for any applications of this material. This allows for the SMP TPU to be considered for some high-temperature applications such as automotive or aerospace engineering. Additionally, the similar thermal stability of both forms of the material suggests that the FDM printing process does not significantly alter the decomposition temperature of the polymer, meaning that the performance will be more reliable and consistent in different end-use environments. A study performed by Slavkovic et al. drew similar conclusions [23]. This study provides insights into the stability and mechanical resilience of SMPs under varied thermal conditions, emphasizing the material’s suitability for high-temperature applications.

3.2. Dynamic Mechanical Analysis (DMA)

The main parameters and average T_g of each group are provided in

Table 3. Test conditions and average glass transition temperatures for all 12 DMA test specimens.

Sample #	Heating Rate (°C/min)	# Shape Memory Cycles	Average Tg (°C)
1–3	2	0	51.43
4–6	5	0	50.59
7–9	5	5	48.84
10–12	2	10	50.80

. These values of T_g are still slightly higher than the T_g of 46.76 °C measured using DSC, which is to be expected. A difference of 5–10 °C is the norm for a T_g measured with DSC and DMA. Regardless, future experiments should be performed with a heating rate of 10 °C/min or less to avoid misrepresentations of the true T_g. More importantly, the same heating rate should be used across tests to eliminate additional factors for differences in measured T_g. DMA results for samples 1–3 yielded an average T_g of 51.43 °C, while samples 4–6 had a T_g of 50.59 °C, less than a degree of difference. Results of sample 3 can be found in Figure 6, and charts for the remaining samples can be found in Figures S4–S14 in the Supplementary Materials. A summary table of all 12 experiments is provided in

Table 4. Data summary of all 12 DMA test specimens.

Sample	Storage Modulus at 25 °C (GPa)	Loss Modulus Peak (MPa)	Glass Transition Temp Tg	Storage Modulus at Tg + 5 °C (GPa)
1	33.7	5774	51.24	1.26
2	32.1	5575	50.99	1.40
3	34.6	5826	50.33	1.46
4	35.5	6523	52.15	1.71
5	35.3	6514	50.78	2.03
6	33.7	6668	50.80	1.94
7	30.3	5400	48.93	1.05
8	31.3	5550	48.24	1.12
9	38.2	7160	49.36	1.29
10	55.2	8570	51.36	0.787
11	45.0	8370	52.35	0.407
12	33.1	6170	51.63	0.361

.

The storage modulus is at its maximum, and the loss modulus is at a very low value at ambient temperature, indicating that the material is very rigid at room temperature and exhibits the greatest spring-like behavior. This fact also suggests that the lower the cooling temperature during the programming stage, the more rigid the material will become and store more energy, potentially leading to a higher shape recovery ratio.

As the number of shape memory cycles increased during the programming of samples 7–12, the specimens began to gradually lose their permanent shape by shrinking along their 45 mm length and volumetrically expanding in the other two axes. The results of specimens 10–12 show a few differences when compared to specimens 7–9, namely broader tan δ and loss modulus peaks, lower storage modulus at 25 °C, and a left-shifted loss modulus peak. Figure 7 shows the DMA results of sample 10 overlaid onto the results of sample 7. First, the earlier onset of loss modulus in Figure S11 indicates that increased shape memory history leads to earlier softening of the material while maintaining a similar T_g. Next, another quality of the increased shape memory was a higher initial value of storage modulus. In specimens 7–9, the average maximum is roughly 39 GPa, whereas specimens 10–12 exhibited a higher average maximum value of approximately 47 Gpa. However, the increased storage modulus also comes with the drawback of a much earlier onset of softening, leading to a decrease in the shape recovery ratio. In parts with extensive shape memory history, this would have to be mitigated by the cooling temperature or increasing the magnitude of the heat stimulus for shape recovery. This would give the part the best shape recovery ratio due to the higher storage of elastic energy at lower temperatures and less recovery energy absorbed by the part by quickly passing the most viscous regions. The starting temperature of the DMA experiments fluctuated between 18 and 25 °C, but this does not affect any results. To mitigate this effect, the storage modulus was compared between specimens at a common temperature of 25 °C.

3.3. Interlocking Lattice Tensile Testing

Test specimens 1–3 were printed with a ISO value of 22.5°, and specimens 10–12 were printed using a value of 90°. The results of the default value specimens (1–3) can be found in Figure 8. Specimens 4–6 were sliced with the interlocking beam width reduced from 0.8 mm to 0.4 mm. Those results are shown in Figure 9. The IBW refers to the width of the “prongs” that protrude from each of the materials into the other. Finally, samples 7–9 changed the IBLC from 2 to 1, meaning that the lattice pattern was repeated twice as frequently, alternating each layer instead of every other layer. The strength measurements for all test specimens with default and varied ITIL parameters can be seen in

Table 5. Tensile results of multi-material test specimens with generated interlocking structures of varying parameters.

Samples	Parameter	Average Ultimate Strength (MPa)	Maximum Ultimate Strength (MPa)	% of SMP TPU Ultimate Strength (Avg/Max)
1–3	Default	6.66	7.87	42/49
4–6	IBW	5.12	6.92	32/43
7–9	IBLC	8.80	9.99	55/62
10–12	ISO	5.36	6.60	34/41

.

The stress–strain curves showed a nonlinear curve that is not typical for most materials, but is expected for TPU. This behavior is present for the same reason that TPU has shape memory properties: its phase-segregated structure. It elicits a time-dependent response of properties as the microstructure changes. Still, the areas of interest on these plots are the ultimate strengths, which remain unchanged.

In all experiments, specimens failed along the interface of the two materials. This is expected behavior, as Kuipers et al. [12] showed that the failure mode of the ITIL always occurred at the interface before any other location on the test part. In the case of specimens 1–3, the failure occurred at an average value of 6.66 MPa, with a maximum of 7.87 MPa. Taking the ultimate tensile strength of the SMP TPU to be 16 MPa, as stated by the manufacturer, the maximum value reached 49%, and achieved 42% on average.

In the case of test specimens 4–6, the ITIL was able to support a maximum tensile strength of 6.92 MPa, or 43% of the maximum 16 MPa, and achieved an average ultimate strength of 5.12 MPa, or only 32%. Clearly, the surface area of the interlocking beams rather than the amount of beams contributes more to increasing the ultimate strength, as evidenced by the better performance of 0.8 mm IBW parts compared to the 0.4 mm IBW parts. In the future, it would be pertinent to measure the strengths of parts fabricated with higher values of IBW. Similarly, to decrease the IBW, the next parameter to be adjusted was the interlocking boundary layer count, or IBLC, which can be observed in Figure 10.

Specimens 7–9 were printed with 1 IBLC, and achieved a maximum strength of 9.99 MPa, or 62% of the expected. The parts also demonstrated an average ultimate strength of 8.8 MPa, or 55% of 16 MPa, which was higher than the averages of all nine other samples. The final specimens to be tested were those with a 90° interlocking structure orientation. As mentioned previously, this change caused the cell variant of the ITIL to change from triangular to straight, which demonstrated a lower maximum strength, but also a smaller variation in the results. Unfortunately, an issue occurred at some point between the data collection and processing that resulted in the loss of data of specimen 12, leaving only data from samples 10 and 11, which is shown in Figure 11. It was found that the specimens with the 90° angle had a maximum strength of 6.6 MPa at 41% and an average strength of 5.36 at 34%. Other values of ISO should also be investigated in the future, especially those that match the angle of the vertical material interface. In the case of the example in Figure 1, the material interface runs at a 45° angle to the front plane, so setting the ISO to this value results in a generated ITIL which runs collinear and perpendicular to the material interface.

The complete results of the tensile testing can be found in Table 4, where the key points are outlined. Figure 12 shows the stress–strain curves for all specimens, and Figure 13 outlines the ultimate strengths of each sample.

One way to improve ultimate strength is to simply increase the cross-sectional area. According to the stress formula presented in Equation (1), the force a material can support before reaching its ultimate strength as determined by its cross-sectional area. However, most infill geometric patterns are three dimensional, meaning the internal forces are not entirely axial, but also tangential in all other directions and not solely affected by area. Therefore, with a square change in area, it is likely that the ultimate strength will increase at a rate which is more than linear but less than cubic.

4. Discussion

4.1. Classification of the Shape Change Effect

After removing thermal history due to plasticization, it can be seen from the second heating cycle that this material only has a single thermal transition. The sole glass transition region indicates that this shape memory polymer has an amorphous nature as indicated by a lack of a melting temperature. Polyurethane (PU) has been shown to exhibit shape memory in both amorphous [18,19] and semicrystalline forms [20,21], but the manufacturer of the TPU SMP does not provide a description of the crystallinity or clarification on how many shapes can be remembered. The existing literature states that polymers with a single thermal transition range are known as one-way SMPs, while multiple-way SMPs employ either multiple transition temperatures or a single very broad (glass or melting) transition range [22]. The combination of these facts and the macroscopic qualitative observations of the printed parts confirm that this material is a one-way SMP capable of returning from a temporary shape to a permanent shape.

4.2. Creation of Multi-Material Parts

Following the analysis of homogeneous SMP TPU specimens, several PLA/SMP TPU multi-material parts were fabricated to gain insights on their manufacturability, macroscopic shape change behavior, and change with increased shape memory history. The first multi-material parts were simple rectangular geometries, as shown in Figure 14A. As mentioned previously, layers of PLA and TPU are not known to effectively bond when vertically stacked, so these parts were created to establish a baseline and examine the quality of the interface bond with increasing shape memory history. The printed part is shown in Figure 14B. Interestingly, two copies of this part were successfully fabricated with no separation of the PLA and SMP TPU sections. It was only after repeated shape memory actuations that any evidence of delamination could be seen. Even after 5+ shape memory cycles, the different materials remained in adhesion with each other even after the permanent shape was altered due to internal stress buildup, as shown in Figure 14C.

The next multi-material parts were created to better showcase the advantages of incorporating both types of filament in a single part. The part shown in Figure 15A was designed to be printed in the flat open orientation, be able to be programmed into a cube, and then returned to the original open design. This could be reversed by printing the part in the closed configuration and programming the flat version as the temporary part, but this may introduce the need for a support material during printing. The design of this part also took inspiration from the design of the part in Figure 14C and Figure 15B, this time utilizing multi-stage folding operations. Even without any reliefs in the SMP TPU base, the part was able to hold the temporary shape relatively well, and was able to return to its permanent shape after several cycles. Figure 15B shows a comparison of two identical parts in their permanent shapes with different shape memory histories. It is clear to see that the part with several cycles of shape memory history had its permanent shape altered. This is due to the high stresses in the SMP TPU by the edges of the blue PLA triangles where the softened SMP TPU may have been stretched past its yield strength. The warping of the SMP TPU base caused by the deformations at the edges of the triangles ultimately led to a loss of adhesion between the blue PLA portions and the semi-transparent SMP TPU. Some additional parts were also printed with a custom retained 3D structure meant to improve inter-filament adhesion, with one example shown in Figure 15C. The performance of these new parts did not improve compared to the parts with no retained structure, and showed signs of delamination after just the first shape memory cycle.

During the creation of initial parts, one particular effect became clear while observing the shape change. The raster angle, or the angle with respect to the horizontal axis at which the beads of extruded filament are deposited, heavily influences the direction of the shape recovery process. The parts would tend towards a U-shaped curve about the axis of the crease, and appeared to achieve the shape fixity with minimal creep and greater shape recovery. The raster angle is easily controlled as a parameter in the slicer software, and should be taken into account while designing a 4D part intended to bend about a specific axis.

4.3. Linking DMA Results to Macroscopic Behavior

The findings of DMA were consistent with an experiment wherein two matching 4D-printed parts were programmed by being folded completely in half in a hot water bath above 55 °C, but one was cooled by being held in ambient temperature air, whereas the other was cooled in a cold water bath below ambient temperature. Figure 16 shows the shape fixity of the two samples upon release of the constant programming stress, where it is clearly visible that one of the parts held the programmed shape much better. The part with the better temporary shape retention was, in fact, the part cooled with a water bath. The parts were then exposed to a heat stimulus in the form of a hot water bath to return the materials to their permanent shapes. Like the programming, the rate of actuation and the shape recovery ratio increased proportionally with the magnitude of the heat stimulus. A difference could even be seen between water baths at 55 °C and 60 °C. This effect was studied by Mendenhall et al. by varying printing temperature, post-printing heating, and cooling of three commercial filaments [24]. Further investigations of this nature would contribute towards the definition of SMPs for numerical modeling.

The creation of the multi-material parts focused on the macroscopic qualitative aspects of the shape change, rather than its quantitative properties such as shape recovery ratio and shape fixity ratio as performed by Rahmatabadi et al. [25]. This would be accomplished by producing a large number of test parts with a single degree of freedom between them. It is crucial to understand the impact of each parameter such as printing temperature, printing speed, shape memory history, thickness, programming velocity, and more on the value of fixity and shape recovery. Such an analysis will be incredibly useful in the long-term goal of creating a numerical model for simulation of the shape change effect, and deserves its own separate study and manuscript. One example of a study on the quantification of shape fixity and recovery is presented by Zolfaghari et al. [26]. Similar works have been conducted by Santiago et al. and Yu et al. on the measurement and analytical modeling of shape fixity/recovery, respectively [27,28]. Additionally, volumetric flow rate of the filament deposition influences the porosity of the printed part, which further affects its shape fixity and recovery, as found by Pivar et al. The researchers were able to vary the flow multiplier parameter in the slicer software to show that an increased flow multiplier of 1.15 was ideal for printing TPU. The parts fabricated in this study used the default flow rate value of 1 as it was suggested by the manufacturer, and because higher values resulted in over-extrusion. However, the effects of flow rate and flow multiplier on the porosity and shape fixity/recovery of SMP TPU would be interesting to investigate as well. Thus, the quantitative analysis of the shape fixity and shape recovery ratio of single material SMP TPU or multi-material SMP parts are saved for future work.

4.4. Challenges and Limitations

Among the other suggestions for future work presented in this study, there are many others that would be beneficial to explore in reference to the items in this section. First, there is a large opportunity in the market for additional commercially available 4D printing filaments, as only a single one was available at the time of authorship. Low availability combined with high cost significantly limits availability, which restricts further research being conducted. Four-dimensional printing SMP filament should not be difficult to produce, only necessitating a viscoelastic response specialized for shape fixity and recovery. Another area that is limited is the literature on two-way shape memory polymers. A majority of papers discuss one-way SMMs, with fewer experimental papers on two-way and multi-way materials. It is the belief of the authors that two-way and multi-way SMPs are more useful than one-way SMPs and will see greater use in the future. As mentioned previously, one aspect of 4D printing that prevents wider adoption is the absence of a numerical model for computer simulation. The pre-shape change prediction and guaranteed repeatability would be one of the pillars that elevate 4D printing from a novel research topic to a viable process in industry. To accomplish this, a library of materials must be compiled which predicts the material’s shape fixity and recovery under the application of heat, load, and shape memory history. Lastly, to be investigated is the shape memory fatigue discussed in the previous section. If SMPs are to be used in any application or process, then the SCE must remain stable and precise. Once these improvements are made, 4D printing will be a serious consideration for a manufacturing process in many industries due to its versatility, low waste, and speed.

5. Conclusions

The research performed in this study found that entry into 4D printing is accessible and can work with an off-the-shelf shape memory polymer on a commercially available hobbyist-level machine. The greatest factors limiting further use of 4D printing are a lack of shape memory polymer filaments, especially with two-way SMPs, and the unavailability of a numerical model to simulate the shape change effect.

Research Question 1: What are the chemical, thermal, and mechanical properties of SMP TPU?

Through investigation, we found that the composition of TPU was similar to a poly(ether urethane) based on 4,4′-diisocyanato diphenylmethane and polyoxytetramethylene. It primarily includes C, O, and N with no significant fillers. No major changes were observed in the chemical composition due to the printing. During the thermal characterization, we observed that the processing of the filament induces some plasticization, which causes the T_g to be artificially lower than it is when the thermal processing is more controlled. The polymer remains stable until 280 °C, which means that the material is suitable for high-temperature applications. The mechanical properties of the TPU material were observed to be very similar to that of PLA.

Research Question 2: What material properties of SMP TPU influence the shape memory characteristics of the printed samples?

The DMA showed that the storage modulus is at its maximum, and the loss modulus is at a very low value at room temperature, causing the material to be very rigid. It has a phase-segregated microstructure and exhibits a spring-like behavior. The lower the cooling temperature, the more rigid the material becomes and the more energy it stores, potentially leading to a higher shape recovery ratio. We can conclude that the cooling rate is a critical factor in the shape memory behavior. Future studies need to explore this characteristic further.

Research Question 3: What are the tensile and ultimate strengths of the parts printed using the ITIL technique with SMP TPU and PLA?

Table 5 summarizes the average ultimate strength of samples printed with the ITIL technique. These parts have a fraction of the ultimate strength of the TPU material. As expected, all the tensile test specimens failed at the part with the ITIL structure. Their stress–strain curves show a nonlinear behavior due to the phase-segregated microstructure, which elicits a time-dependent response of properties as the microstructure changes.

Characterizing the one-way shape memory polymer thermoplastic polyurethane provided information about the chemical structure of the SMP TPU and its thermal properties. Also, the introduction of shape memory history in part led to a higher maximum storage modulus, but at the drawback of an earlier softening onset and larger average loss modulus. Combined with the observation that heating and cooling rates affect shape fixity and recovery, it becomes clear there is an opportunity for further exploration of the changing mechanical properties due to temperature gradients and shape memory history.

The research discussed in this paper also examined the generation of interlocking structures with SMP TPU and PLA, as well as the effect of three of its parameters on ultimate strength. It was found that the ultimate strength of the interlocking topologically interlacing lattice can be improved by decreasing the interlocking boundary layer count from the default value of 2 to 1. The data also suggest that increasing the interlocking beam width and changing the interlocking structure orientation may also increase the ultimate strength. Further research and testing were proposed throughout this paper further to understand the working mechanisms behind the shape change effect and make 4D printing more accessible to hobbyists and engineers alike. The future prospects of 4D printing are bright, and are guaranteed to surpass any of today’s expectations.