1 Introduction

Shape memory materials (SMMs) are widely used and popular subsets of smart materials. These materials have the ability to save a secondary shape and shift to the original shape when faced with a suitable stimulus [1, 2]. Many efforts have been done over the years to expand and improve them in various ways, and many materials and structures have been developed and studied to create the shape memory effect (SME) [3, 4]. Among these vast volumes of work, shape memory polymers (SMPs) can show more special capabilities such as response to stimuli other than direct heat, high deformability, biocompatibility, and simple and cheap processing properties [5, 6]. Poly lactic acid (PLA) is one of the most widely used biocompatible SMPs with unique properties, but due to being semi-crystalline, it has very poor SME [7,8,9]. It is likely to have good fixity properties but very low recovery due to the absence of a strong soft phase for recovery, and the presence of another polymer with desirable properties is necessary to solve this problem and boost the SME of PLA. The synthesis and polymer blending is one of the primary solutions to improve the weak points and achieve better mechanical properties for different applications [10,11,12]. Thermoplastic elastomers (TPE) such as thermoplastic polyurethane (TPU) seem to be a good option for generating sufficient potential energy to restore the original shape of PLA, which is due to the high elasticity of TPU segments [13,14,15]. As a result, it is interesting to study the SME of the PLA/TPU blend. This compound can have high performance as an SMP; because on the other hand, the base polymer has biodegradability and other components have very unique capabilities such as good strength, high ductility, and excellent abrasion resistance [7, 16]. Furthermore, one of the characteristics of SMPs is good formability, which PLA lacks this advantage due to its high brittleness, and its combination with TPU can solve this defect well [16].

3D printing is an additive fabrication method for manufacturing a vast span of complex geometries and structures from a CAD 3D designed model [17, 18]. This technology involves layered printing of materials to create the final structure and in 1986 was introduced by Chuck Hull in a method known as stereolithography that was followed by further advances [19]. 3D printing, which includes a variety of methods, materials, and instruments, has developed over the years and can transform assembly processes and production. Additive manufacturing (AM) methods have been evolved to meet the production appeal of complex geometries with good resolution. The ability to print smart materials, medical equipment materials, rapid prototyping, printing large parts, reduce printing time and defects, and increase mechanical strength are some of the main goals in the evolving of AM processes [20]. The fused deposition modeling (FDM) known as most common 3D printing process mainly uses thermoplastic filaments [21]. In the FDM method, a continuous thermoplastic filament is used for the 3D printing of layered structures [22, 23]. The filament is heated in an extruder (liquid fire) to a semi-liquid state and then deposited onto a previously printed layers or printer bed [24, 25]. Low cost and process simplicity are the main advantages of FDM. The rheology of the polymer filament is an essential feature of this method, allowing the filaments to bond together during printing and solidify at room temperature after printing. Layer thickness, shell, printing speed, nozzle temperature, raster width, direction, infill, and pattern are the essential printing parameters that affect the mechanical properties of final structures. A large number of parameters and the dependence on physical and mechanical properties and even the cost of these parameters are the main challenges that have occupied a large part of the research in this field in recent years [26,27,28]. Ma et al. assessed the influence of the infill pattern on the crushing and energy absorption behavior [29]. They used two materials, PLA and PLA, reinforced with carbon fiber, and they concluded that the printing parameters had a greater effect on the investigated properties than the material. In comprehensive research, it was shown that by changing the rater layup in both unidirectional and 90° alternating layup, it is possible to control the mechanical properties included strength, toughness, and even failure mechanism [30]. By doing this, they achieved an ultimate tensile strength (UTS) range of 10 to about 45 MPa for 3D-printed PLA parts. Domingo et al. investigated the effect of velocity, layer thickness, nozzle diameter, and infill density in three levels for two different patterns of honeycomb and rectilinear on the fatigue properties using the design of experimental (DOE) [31]. Their results revealed that the infill density has the greatest effect on the fatigue behavior of 3D-printed ABS parts, and the honeycomb pattern behaves a better performance. Ghorbani et al. proved that the printing parameters strongly affect the cavity and microhole defects created during FDM printing, and an increase in the speed, nozzle diameter, and temperature increases the feeding rate and reduces the inter and intra-layer defects [32]. Deshwal et al. used hybrid statistical models to optimize the printing parameters of FDM [33]. They considered the three parameters of temperature, speed, and infill density as input parameters and UTS as the output parameters. They claimed that the predicted data for GA-ANN models were in perfect agreement with the experimental results [33]. All the research done so far shows the role of printing parameters on mechanical and physical properties and failure mechanisms. In another research, the DOE and ANOVA were used to investigate the influence of the input parameters of nozzle temperature, infill density, and layer thickness on the output parameters of strength, elongation, final 3D printed part weight, and printing time (cost) [34]. Their results also confirmed that the mutual effect of layer thickness and infill density on mechanical behavior is remarkable. By the combination of RSM and Genetic Algorithm Π, Yang et al. optimized the printing parameters with the goal of maximum final strength and minimum time and surface roughness [35]. The considered parameters were temperature, speed, layer thickness, nozzle diameter, and infill density. The results of their investigation confirmed that the nozzle diameter has the greatest effect on surface roughness. In spite of countless researches that have been done to investigate the effect of printing parameters on mechanical behavior and to optimize them, this issue has not been given much attention in the field of printing smart polymers or 4D printing. Since in addition to the influence of mechanical behavior, thermomechanical behavior and cooling and heating rates are also strongly affected by printing parameters, and the importance of investigating the effect of printing parameters in the shape memory cycle is twofold.

In the design of experiment (DOE), it is possible to evaluate the combined effect of two or more experimental variables that are used simultaneously. Today, DOE is widely used in various fields of engineering to improve and increase process performance. Adequate knowledge of the parameters investigated in the experiment is one of the basic requirements to start the DOE. This knowledge helps to determine the parameters and levels correctly. DOE and statistical analysis are tools for the correct diagnosis of effective factors and the percentage of influence of each parameter. For the first time, the role of FDM printing parameters on SME of PLA/TPU blend is evaluated in this paper by the DOE model. The analysis is carried for three basic FDM parameters: shell, nozzle temperature, and infill density in three levels. Thermomechanical experiments are conducted to evaluate shape memory behavior including loading stress, recovery stress, shape fixity, and shape recovery.

2 Research methodology

2.1 Blending, filament fabrication, and 3D printing

In this paper, commercial poly-ester based 90A TPU and PLA granules were provided by Xiamen Keyuan Plastic Co. Ltd. (Fujian, China) and Khatam Polymer Co. (Iran), respectively. PLA-TPU compounds with 30% by weight of TPU were blended by melt mixing. According to the literature, this compound has optimal mechanical properties and shape memory effect [36]. Melt mixing was done at a speed of 100 rpm and a temperature of 200 °C for 15 min with the internal mixer. After that, to prepare the filament, three steps of turning lumps into sheets using a hot press, sheets into granules with a crusher, and granules into filaments with an extruder were performed. The diameter and surface roughness of the filament strongly affects the quality of the printed samples. Filament preparation was done using a hand-made single extruder (a length over diameter ratio of 15) with a nozzle diameter of 1.75 mm and speed (25 rpm) and temperature (200 °C) control. The diameter of the produced PLA/TPU filament was checked in the range of 1.75 ± 0.25 mm. The PLA-TPU samples were printed using a hand-made laboratory FDM printer. The constant and variable parameters considered are presented in Tables 1 and 2. In Fig. 1, raw materials including PLA and TPU granules, extruder and filament production process, printer, and printed parts with different parameters are presented.

Table 1 Constant printing parameters
Table 2 Designed input variables and levels
Fig. 1
figure 1

Prepared PLA-TPU filament and printed samples, (a) granules, (b) lab-made extruder, (c) lab-made FDM 3D printer, and (d) 3D printed sample

2.2 DMTA and SEM

To check the thermal behavior and determine the glass transition temperature, the raw materials and the PLA-TPU compound were subjected to the dynamic mechanical thermal analysis (DMTA) test. According to ASTM D4065-01 standard, DMTA was done in a constant temperature range (-100 °C to 100 °C). Temperature changes and constant frequency were applied at the rate of 5°/min and 1 Hz, respectively, on the beam-shaped sample with dimensions of 40 × 10 × 1 mm under bending mode. Also, SEM was used to evaluate the morphology of the PLA-TPU compound. Before that, the samples were broken down into liquid nitrogen and then coated with gold.

2.3 Design of experiment

In this study, the Box-Behnken design (BBD), as an experimental design of the response surface method (RSM), is used to fit the model between variables (inputs) and responses (outputs). Table 2 presents the variables of the current research and their intervals. Printing of PLA/TPU samples based on DOE via BBD model for three basic FDM printing parameters: shell, infill density, and nozzle temperature in three levels was done according to Table 2. One of the advantages of using DOE is reducing the number of tests and costs and the possibility of optimizing the parameters and estimating the effectiveness of each factor. Therefore, 17 samples with the same dimensions of 20 × 10 × 10 and different parameters were printed to study the shape memory effect. In order to increase the accuracy of the results and eliminate the effect of printing quality, simple geometries were used and the tests were repeated at least three times for each group. Also, Fig. 2 illustrates the schematic of the fitted model between the input and output parameters. In the fitted model, the effect of the shell (variable A), infill density (variable B), and nozzle temperature (variable C) on the value of loading stress, recovery stress, shape fixity, and shape recovery ratio was determined. In the DOE, one shell unit is considered for every 200 microns to simplify the modeling. Table 3 represents the variable parameters specified according to the DOE method.

Fig. 2
figure 2

Conceptual structure of the used RSM model

Table 3 Specified variable parameters using DOE

2.4 Shape memory evaluation

The shape memory test includes two main stages programming and recovery. The programming stage is also heating up to the rubber area (20 °C higher than the transition temperature), loading, cooling, and unloading. Programming for all PLA-TPU samples was done at a temperature of 90 °C and a heating rate of 8 °C/min, and after reaching this temperature, the sample was kept for 240 s. 50% strain was applied with a displacement rate of 3 mm/min. After the end of loading, the sample was quickly cooled to 25 °C at a rate of 15 °C/min; and after holding for two minutes, the unloading was done. At the end of the programming, the shape fixity ratio was calculated using Eq. 1. The recovery was performed with two different approaches of free and constrained recovery to calculate and evaluate shape and stress recovery, respectively. In free recovery, heating was done up to 90 °C and the shape recovery ratio was calculated using Eq. 2. In these equations, parameters A, B, C, and D are the initial height, the amount of deformation, the height after unloading, and recovery, respectively [37].

$$\mathrm{shape\;fixity\;ratio}=\frac{B}{A-C}\times\;100$$
(1)
$$\mathrm{shape\;recovery\;ratio}=\frac{A-C}{D-C}\times\;100$$
(2)

In the constrained test, during the recovery process, the sample was not allowed to recover its shape and the recovery force was read through the load cell connected to the constraints. Heating in both recovery approaches was done at a rate of 8 °C/min.

3 Results and discussion

3.1 DMTA and SEM

In Fig. 3, the results of DMTA analysis for TPU, PLA, and their blended with a weight ratio of 70PLA:30TPU are presented. In polymer blends, in addition to the storage modulus changes, we are often interested to have the relationship between the loss modulus and the storage modulus. For this purpose, Tanδ changes are also presented in terms of temperature, which defines the ratio of the storage modulus to the loss modulus. The storage and loss modulus also represent the stiffness and damping part of the polymer. According to Fig. 3, in all three materials, the storage modulus value decreases with increasing temperature, and the main difference is in the starting temperature of the changing rate of the storage modulus. TPU has two mild and severe reduction rates in the storage modulus diagram, respectively, in the temperature range of -100 to -25 °C and -25 to 0 °C, respectively. Also, the peak of the Tanδ diagram is observed at 0 °C, which indicates the glass transition temperature of TPU. For PLA, the same trend has been repeated, but the reduction rates of the storage modulus are much lower and the Tanδ peak occurs at a temperature higher than 100 °C. According to Fig. 3(c), the composition of PLA-TPU has an interstitial trend of both raw materials with a higher contribution of PLA. The Tanδ diagram has two mild and intense peaks at temperatures of -20 and 70 °C, which are related to the role of TPU and PLA, respectively. The presence of these two peaks, which are related to the components, indicates the immiscibility of the PLA-TPU compound. Due to the higher contribution of PLA, the corresponding peak is more intense and the glass transition temperature for the PLA-TPU is 67 °C. In addition, the drop of the storage modulus is much more severe in the temperature ranges related to the transition temperature of the components, and here, it is stronger and more severe in the transition temperature related to PLA.

Fig. 3
figure 3

DMTA results for (a) TPU, (b) PLA, and (c) PLA-TPU

In Fig. 4, the cross-section images of the PLA-TPU samples after molding and 3D printing are presented to evaluate the morphology. According to Fig. 4(a), TPU droplets are dispersed in the PLA matrix and the immiscible matrix-droplet morphology for PLA70TPU composition is quite evident. In Fig. 4(b), the presence of microholes and cavities between layers and between the rasters of one layer due to printing can be seen. The density and geometry of these cavities determine the printability of the compound, its mechanical properties, and porosity. The formation of cavity depends on the FDM feeding mechanism and the shrinkage of the molten polymer after cooling, and many efforts have been made to eliminate them. Most of this research is in the field of optimizing printing parameters such as speed, temperature, nozzle diameter, and melt flow rate. Of course, as can be seen, the density of holes is limited due to the high printability of raw materials, especially PLA, and PLA-TPU has an acceptable print quality compared to handmade and research filaments. In Fig. 4(c), the morphology after printing is presented for comparison with the same sample after molding. According to these figures, the TPU droplets in the molded sample have a uniform distribution with a completely circular geometry, while after printing, most of these droplets were stretched in printing different. In fact, the uniformity of the drops in terms of distribution and size has been lost, and their geometry has changed from circular to elliptical. Applying two steps of molten extrusion and rapid cooling to prepare the filament and perform printing causes TPU droplets and even polymer chains to be stretched, which is the origin of the difference in morphology in the two modes of molding and printing of PLA-TPU.

Fig. 4
figure 4

Cross-section SEM images of the PLA-TPU: (a) molded and (b, c) 3D printed samples

3.2 Shape memory cycle

In Fig. 5, schematic of the shape memory cycle (programming, free, and constrained recoveries) is presented. According to this figure, all samples are subjected to the same programming, and the recovery stage is applied with two different protocols, free, and constrained, to obtain shape and stress recovery parameters, respectively. In Fig. 5, shape memory cycle (programming and constrained recovery) for the different 3D printed PLA-TPU according to DOE. Also, the quantitative changes of temperature, strain, and stress in terms of time under programming and constrained recovery for four samples number 1, 6, 13, and 14 are presented. According to these figures, in the first stage (programming steps), the sample is heated to 90 °C and then it is kept for a 240 s. In the next step, loading is applied and it is observed that the amount of stress increases with the increase of strain, and due to the same elastic modulus, the linear slope is the same in all four samples. In sample 13, due to low filling density (40%) and buckling during force application, a wave is created in the stress diagram. After the loading is completed, the cooling starts without removing the force. After keeping the polymer at glass temperature (30 °C) for a sufficient time (120 s), this opportunity is given to maintain the new structure. The chains in the equilibrium state have a spiral and random structure, and after changing the shape, they lose this structure and the chains are stretched in the force direction. This change causes the polymer to move away from the equilibrium state and entropy decreases. The cooling causes the polymer chains to freeze in the deformed state, and as can be seen, with the passage of time, the amount of force decreases. In fact, ideally, this force should reach zero so that the total applied force is stored and most of the polymer chains are frozen. Chain deformation in the applying force direction, rapid cooling, and maintaining, this structure causes a severe decrease in entropy in the thermodynamic equilibrium system, and the entropy force plays a fundamental role in restoring the original shape as the driving force of recovery stage. Therefore, the stabilization and freezing can be considered as the basic stage in the shape memory cycle. After unloading, a part of the force is released, which is the caused by the polymer chains that return to their equilibrium state after unloading and cause a decrease in the shape fixity values.

Fig. 5
figure 5

Schematic of the shape memory cycle

In the second stage, the sample is heated in constrained and free form, and as soon as it enters the glass transition zone, the shape and stress recovery begin. Because upon entering this area, the constraints and hard points of the network begin to loosen and the polymer chains return to their original state depending on the degree of instability. In fact, with the increase in temperature, the free volume increases and the system moves toward thermodynamic equilibrium (entropy increase). According to Fig. 6, the highest amount of stress recovery is obtained at 90 °C and the main reason for that is the application of high strain, cooling rate, and keeping the deformed sample glassy region, which has caused more instability of the polymer chains from the equilibrium state, which requires higher energy to recover. In Table 4, the results of the free and constrained shape memory effects are presented quantitatively.

Fig. 6
figure 6

Shape memory cycle including programming, cooling, unloading, and recovery under constraint for different samples according to DOE via BBD: (a) sample 1, (b) 6, (c) 13, and (d) 14

Table 4 Quantitative experimental results of the SME tests performed based on the BBD

3.3 RSM and ANOVA

Output parameters are defined as a function of the shell (variable A), infill density (variable B), and nozzle temperature (variable C) in Eqs. 36.

$$\mathrm{Applied\;Stress\;}\left(\mathrm{A}.\mathrm{S}.\right):{\mathrm{f}}_{1}\left(\mathrm{A},\mathrm{B},\mathrm{C}\right)$$
(3)
$$\mathrm{Recovey\;Stress}\left(\mathrm{R}.\mathrm{S}.\right): \left(\mathrm{A},\mathrm{B},\mathrm{C}\right)$$
(4)
$$\mathrm{Shape\;Fixity }\left(\mathrm{S}.\mathrm{F}.\right): \left(\mathrm{A},\mathrm{ B},\mathrm{C}\right)$$
(5)
$$\mathrm{Shape\;Recovery\;}(\mathrm{S}.\mathrm{R}.): {\mathrm{F}}_{4} (\mathrm{A},\mathrm{ B},\mathrm{ C})$$
(6)

The equations fitted by the RSM are considered as the third-order polynomial regression equation in Eq. 7.

$$Y= {b}_{0}+\sum {b}_{i}{x}_{i}+\sum {b}_{ii}{x}_{i}^{2}+\sum {b}_{iii}{x}_{i}^{3}+\sum {b}_{ij}{x}_{i}{x}_{j}+\sum {b}_{iij}{x}_{i}^{2}{x}_{j}+\sum {b}_{ijj}{x}_{i}{x}_{j}^{2}+{e}_{r}$$
(7)

where \(Y\) is the response of the model; \({b}_{0}\), \({b}_{i}\), \({b}_{ii}\), and \({b}_{iii}\) are the offset constant coefficient, the constant coefficients of linear terms, the constant coefficients of quadratic terms, and the constant coefficients of cubic terms, respectively. Also, \({b}_{ij}\), \({b}_{iij}\), and \({b}_{ijj}\) are constant coefficients interaction variables. In addition, \({x}_{i}\) and \({x}_{j}\) are independent variables and \({e}_{r}\) is the unpredicted error.

As mentioned earlier, the analysis of the RSM was done by considering the effect of printing parameters on shape memory properties. Table 4 shows the input and output parameters used in this study. Based on this, shell, infill density, and nozzle temperature (quantitative parameters) are considered input variables; and loading and recovery stress, shape fixity, and shape recovery are defined as related responses.

The ANOVA is used to estimate the importance of each printing parameter on the output response and assess the fitness of the presented model. Figure 7 is the equality graphs of the experimental data against the predicted values of shape fixity, shape recovery, loading stress, and recovery stress. It shows that there is a good agreement between the values predicted by the proposed model and the experimental ones.

Fig. 7
figure 7

Equality graphs of predicted values against experimental data: (a) shape fixity, (b) shape recovery, (c) loading stress, and (d) recovery stress

The results of the analysis of variance (ANOVA) are done and displayed in Tables 5, 6, 7 and 8 in the Appendix. They show the accuracy of the presented model and the importance of the parameters. Most of the probability values (P value) are below 0.05, which confirms the accuracy of the model [38]. The P value below 0.0001 for the answers of the current research indicates the significance of the proposed model. In addition, all individual and interactive parameters are significant because their corresponding P value is below 0.05 [38].

3.4 Applied stress

ANOVA results of applied stress are shown in Table 5 in the Appendix. According to the results, the p value of the model is much lower than 0.05 indicating the accuracy and significance of the model. Also, the P value for all coefficients is less than 0.05. The interaction effect of printing parameters on the applied stress is presented in Fig. 8. According to the F-value and coefficient of printing parameters in Eq. 8, the infill density has the most linear effect on applied stress. Also, the second-order influence of the shell is greater than the other two parameters. Among the second-order interaction, the effect of shell and infill density is much greater. As expected, the loading force under compression in PLA-TPU rubber state increases with the increase of the infill density and the number of shell. With the increase of infill density, the areas resistant to deformation increase. Also, shells act as constraints that resist deformation. In fact, more force should be applied to overcome them and absence of them makes it easier deformation with less force. The highest value of applied force is for sample 1, which has the highest number of loose and dense.

Fig. 8
figure 8

Interaction effect of printing parameters on the applied stress: (a) shell and infill density, (b) shell and temperature, and (c) infill density and temperature

$$\mathrm{Applied\;stress}=2.1667+0.6175\mathrm{A}+1.625\mathrm{B}+0.42\mathrm{C}+1.4875\mathrm{AB}+0.6925\mathrm{AC}+0.365\mathrm{BC}+2.13417{\mathrm{A}}^{2}+1.13667{\mathrm{B}}^{2}-1.0733{\mathrm{C}}^{2}+1.2625{\mathrm{A}}^{2}\mathrm{B}+0.8725{\mathrm{A}}^{2}\mathrm{C}+1.26{\mathrm{AB}}^{2}$$
(8)

3.5 Recovery stress

The amount of recovery stress is strongly dependent on the amount of applied stress and its fixity. As mentioned in the previous section, infill density has the greatest effect on the amount of applied stress, and thus, it also affects the recovery process. With the increase in infill density, a larger volume of material participates in the recovery process, and this factor increases the recovery stress. On the other hand, the sample with higher infill density can recover more stress due to more applied stress. The negative effect of shells on the stress recovery can be due to their plastic buckling, which, in addition to changing the deformation mode, is also effective in reducing stress recovery (they do not participate in recovery and even prevent it). In fact, with shell buckling, more instability is created in the system, and this factor causes the reduction of the recovery values. Interaction effect of printing parameters on the stress recovery and the fitted equation are presented in Fig. 9 and Eq. 9.

Fig. 9
figure 9

Interaction effect of printing parameters on the stress recovery: (a) shell and infill density, (b) shell and temperature, and (c) infill density and temperature

$$\mathrm{Recovery\;stress}=1.21667+0.4375\mathrm{A}+1.04\mathrm{B}+0.415\mathrm{C}+0.6725\mathrm{AB}+0.4675\mathrm{AC}+0.38\mathrm{BC}+1.1491{\mathrm{A}}^{2}+0.75167{\mathrm{B}}^{2}-0.5133{\mathrm{C}}^{2}+0.5075{\mathrm{A}}^{2}\mathrm{B}+0.1575{\mathrm{A}}^{2}\mathrm{C}+0.475{\mathrm{AB}}^{2}$$
(9)

3.6 Shape fixity

Figure 10 shows the effect of printing parameters on the shape fixity ratio. According to Eq. 10, the infill density has the most linear and second-order effects on shape fixity ratio. Also, among the second-order interaction, the effect of shell and infill density is much greater. The non-linear and irregular behavior of all three parameters on shape fixity is significant. Based on the results, sample 10 has 99.23% shape fixity, which is the highest value among all samples. This sample has a shell and 70% infill. In addition, two samples 16 and 11 have a shape fixity of over 95%. These samples have 40% infill density and 2 and 1 shells, respectively. The fixity value for the solid sample with two shells is the lowest value (58%) by a large difference compared to the rest of the samples.

Fig. 10
figure 10

Interaction effect of printing parameters on the shape fixity: (a) shell and infill density, (b) shell and temperature, and (c) infill density and temperature

$$\mathrm{Shape\;fixity\;ratio}=9796-1.295\mathrm{A}-4.4025\mathrm{B}-1.9775\mathrm{C}-5.67\mathrm{AB}-3.04\mathrm{AC}-0.3775\mathrm{BC}-7.95125{\mathrm{A}}^{2}-8.82375{\mathrm{B}}^{2}-0.32125{\mathrm{C}}^{2}-5.4875{\mathrm{A}}^{2}\mathrm{B}-3.4675{\mathrm{A}}^{2}\mathrm{C}-5.37{\mathrm{AB}}^{2}$$
(10)

3.7 Shape recovery

The interaction effect of printing parameters on the shape recovery is presented in Fig. 11. Based on Eq. 11, temperature has the most linear effect on shape recovery. Shape recovery values for samples are in the range of 53.24% and 90.61%, the trend of shape recovery changes is consistent with stress recovery, and samples 1 and 9 have the highest and lowest shape recovery values, respectively (Fig. 11).

Fig. 11
figure 11

Interaction effect of printing parameters on the shape recovery: (a) shell and infill density, (b) shell and temperature, and (c) infill density and temperature

$$\mathrm{Shape\;Recovery\;ratio}=57.176+4.375\mathrm{A}+0.7175\mathrm{B}+5.9875\mathrm{C}+3.635\mathrm{AB}+4.475\mathrm{AC}+5.2225\mathrm{BC}+4.7979{\mathrm{A}}^{2}+129345{\mathrm{B}}^{2}-3.6904{\mathrm{C}}^{2}+6.8225{\mathrm{A}}^{2}\mathrm{B}-9.5625{\mathrm{A}}^{2}\mathrm{C}+0.15{\mathrm{B}}^{2}$$
(11)

4 Conclusion

In this paper, for the first time, the Box-Behnken design method was used to investigate the effect of printing parameters on shape memory properties. Compared to the previous researches in this article, functional shape memory polymer was used and the effect of printing parameters on the SME was investigated quantitatively and comprehensively. Based on the design of the experiment, three FDM printing parameters including shell, temperature, and infill density were considered in three levels; 17 sets of samples were printed; and the shape memory effect in two recovery mechanisms, constrained and free, was comprehensively investigated.

  • • The loss diagram of PLA-TPU had two mild and intense peaks at temperatures of -20 and 70 °C, which are related to the role of TPU and PLA, respectively. The presence of these two peaks, which are related to the components, indicates the immiscibility of the PLA-TPU compound. Due to the higher contribution of PLA, the corresponding peak was more intense and the glass transition temperature for the PLA-TPU was considered 67 °C

  • • SEM images showed that TPU droplets were dispersed in the PLA matrix and the immiscible matrix-droplet morphology for PLA70TPU composition was quite evident. By applying two steps of molten extrusion and rapid cooling to prepare the filament and perform printing, the uniformity of the droplets in terms of distribution and size had been lost, and their geometry changed from circular to elliptical

  • • In previous research, the brittleness of PLA and its poor performance in recovery and high stress relaxation rate caused its practical limitations. The addition of TPU to PLA increases its formability and shape memory properties by strengthening the elastic part

  • • The ANOVA was used to estimate the importance of each printing parameter on the output response and assess the fitness of the presented model. The equality graphs of the experimental data against the predicted values of shape fixity, shape recovery, loading stress, and recovery stress showed that the good agreement between the predicted values by the model and the experimental results.

  • • The results of the analysis of variance (ANOVA) showed the accuracy of the model and the importance of the parameters. Most of the probability values (P value) were below 0.05, which confirms the accuracy of the model. In addition, all individual and interactive parameters were significant because their corresponding P value was below 0.05

  • • Among the examined parameters, infill density and nozzle temperature had the greatest and least roles on shape memory properties, respectively. Also, the values of shape fixity and shape recovery were obtained in the ranges of 58 to 100% and 53 to 91%, respectively

5 Ethics approval

Not applicable.

6 Consent to participate

Not applicable.

7 Consent for publication

All authors read and approved the final manuscript for publication.

8 Competing interests

The authors declare no competing interests.