Potential Applications of Additive Manufacturing in Intervertebral Disc Replacement Using Gyroid Structures with Various TPU Filaments

1. Introduction

Disc degeneration is an increasingly common problem in modern society and often precedes disc protrusion. Causal factors include physical exertion, overuse, the natural aging process, diseases, and injuries. Over time, the fibrous ring of the intervertebral disc develops cracks and small tears, allowing fluid to escape from the gel-like core. This reduces the disc's ability to absorb shock, which can lead to protrusion or a herniated disc [1]. If conservative treatments are no longer effective, surgical intervention is necessary. The two most important procedures are spinal fusion (spondylodesis), considered the gold standard for treating disc degeneration in the cervical and lumbar spine [1,2,3]. However, spinal fusion has disadvantages, including limited spinal mobility and increased stress on adjacent vertebral segments. This can accelerate the degeneration of adjacent intervertebral discs, a condition known as adjacent segment degeneration (ASD). ASD leads to pain and nerve compression in the affected region [1,2,4,5,6]. A new alternative to spinal fusion is the use of artificial disc prostheses. These prostheses are designed to preserve the spine's natural range of motion (ROM) while reducing the risk of ASD [7,8,9,10]. According to the Federal Statistical Office, spinal fusion and artificial disc implant procedures will increase in Germany between 2020 and 2022 [11]. The number of spinal fusion procedures is expected to increase by over 2,000, rising from 67,380 in 2020 to 69,728 in 2021. In contrast, the number of artificial disc implantations will remain significantly lower: 4,415 in 2020, 4,205 in 2021, and 3,978 in 2022. From a biomechanical perspective, the human intervertebral disc must be examined in terms of the annulus fibrosus and the nucleus pulposus because it is not a homogeneous material. Cloyd et al. [12] report E-moduli of 0.1–0.5 MPa for the nucleus pulposus and 4–8 MPa for the annulus fibrosus. In contrast, artificial disc prostheses have an elastic modulus of 100–200 GPa due to their metallic titanium components and polymer components such as polyurethane or polyethylene, which have an elastic modulus between 0.5 and 3 GPa [13]. Therefore, they have no shock-absorbing properties. Since 2010, additive manufacturing has been used more and more for medical applications. It can be used to make very precise, custom-made disc prostheses [14]. Various polymeric materials have been explored for this purpose, including flexible polylactic acid (FPLA), which allows for tunable, biomimetic mechanical properties suitable for intervertebral disc tissue engineering [15,16]. These FPLA scaffolds have demonstrated enhanced viscoelasticity and mechanical compatibility with native disc tissue in vitro and in vivo. Similarly, ultra-high molecular weight polyethylene (UHMWPE) has been investigated as a durable structural component for artificial disc replacements, offering favorable wear resistance and improved tensile strength when reinforced with fibers [17]. Such materials demonstrate the potential of polymer-based solutions to bridge the gap between mechanical stability and physiological compliance in intervertebral disc prosthetics. Moreover, the continuous development of additive manufacturing materials expands the range of processable polymers, some of which are now approved for medical use [18]. A previous study [19] investigated whether flexible thermoplastic polyurethane (TPU) filaments could be used to fabricate flexible spinal disc replacements. While that work identified TPU as a highly promising candidate material due to its shock-absorbing properties, the range of printable filaments was limited. In particular, several TPU filaments could not be processed successfully because a drying unit and optimized process parameters were not available at that time. In the present study, this limitation is directly addressed. By implementing an optimized drying protocol at 55 °C before and during printing, along with improved parameter control, it was possible to reliably process those TPU filaments that had previously been unprintable. As a result, the material spectrum investigated here is broader and more representative, allowing for a systematic evaluation of different gyroid geometries and wall designs. Consequently, an extended analysis of additively manufactured TPU gyroid structures with shock-absorbing properties is conducted for their potential application in intervertebral disc replacement. To ensure comparability with the earlier work [19], standardized Ø50×10mm intervertebral disc specimens are again fabricated using fused deposition modeling (FDM). Consequently, this study can be regarded as a follow-up to the previous study by Gross et al. [19].

2. Materials and Methods

2.1. Materials

For this study, five different flexible TPU filaments were selected: FlexMed and FlexSemiSoft (extrudr, Lustenau, Austria), SUNLU TPU 95A (SUNLU, Zhuhai, China), OVERTURE TPU (Overture 3D Technologies LLC, Houston, USA), and GEEETECH TPU (HK GETECH Co., Shenzhen, China). To clearly differentiate them during handling, the filaments were used in distinct colors: FlexSemiSoft (transparent), SUNLU (orange), OVERTURE (purple), and GEEETECH (light brown, semi-transparent). An overview of their physical and mechanical properties is given in Table 1.

2.2. Methods

2.2.1. Filament Preparation

Before processing, all filaments were conditioned to reduce moisture uptake. Drying was performed in a Creality Space Pi Drying Box (Creality 3D Technology Co., Ltd., Shenzhen, China) at 55 °C for at least five hours. The dried filament was then directly guided via a 2 mm PTFE tube from the drying chamber to the extruder, preventing reabsorption of humidity during printing. Two different printers were used for fabrication: a Prusa MK3S+ (Prusa Research, Prague, Czech Republic) and a Sovol SV04 (Sovol, Shenzhen, China). Throughout the printing process, the spools remained inside the heated drying chamber.

2.2.2. 3D Printing Parameter Optimization

To establish suitable process parameters, temperature towers were produced for each filament type to identify the best printing temperature and extrusion speed. Representative examples for FlexSemiSoft, SUNLU, GEEETECH, and OVERTURE are shown in Figure 1. The trials demonstrated that variations in temperature strongly influence surface quality, ranging from stringing effects to incomplete structures. Based on these results, optimized parameter settings were determined. The final values applied in the experiments are highlighted in bold and blue in Table 2.

2.2.3. Additive Manufacturing

Three-dimensional models were designed in Creo Parametric 6.0.6.0 (PTC, Boston, Massachusetts, USA). To maintain consistency across all experiments while facilitating reproducibility, the gyroid structures were modeled in the form of discs with a diameter of 50 mm and a height of 10 mm [19]. Four Gyroid unit cell sizes (10, 8, 6, and 4 mm) were selected to represent different levels of structural coarseness. Each of these designs was further varied by applying three wall thicknesses (0.5, 0.75, and 1.0 mm) as well as three outer wall thicknesses (0.4, 0.6, and 0.8 mm), resulting in 36 unique design variants (Table A1 in the Appendix). These design variations were realized for each of the four TPU filaments under investigation, yielding a total of 144 disc specimens. The digital models were exported in STL format and subsequently processed in PrusaSlicer 2.7.4 (PRUSA Research, Prague, Czech Republic). Printing was performed on either a Prusa i3 MK3S+ (PRUSA Research) or a SOVOL SV04 printer. Prior to slicing, filament-specific extrusion parameters were adjusted in the software (see Table 2), while other settings such as 100% infill and a constant print speed were applied uniformly. Final G-code files were generated and transferred to the printers. All specimens were printed on a textured spring steel build plate, with adhesion enhanced by a thin layer of glue stick (Kores, Vienna, Austria). In addition to the disc geometries, standardized tensile test specimens (DIN EN ISO 527, Type 5A) were produced from each TPU filament at 100% infill. These samples were reserved for subsequent mechanical characterization.

2.3. Sample Characterization

2.3.1. Microscopy

Structural evaluation of the printed specimens was carried out using a stereomicroscope (Olympus SZ61, Olympus Inc., Tokyo, Japan) equipped with an SC30 digital camera. For this purpose, discs were produced at half the height of the standard test specimens, and the printing process was deliberately paused at 50% completion in order to provide unobstructed insight into the internal gyroid geometry. In addition, selected full-height samples were bisected using razor and microtome blades. Dimensional features of the gyroid structures, including wall thickness and pore size, were quantified at multiple locations (≥5 per sample) and compared with the corresponding design specifications from the STL models. Image analysis was performed in ImageJ (FIJI version 1.53t). Scale calibration was established via Stream Motion (Olympus Inc.), which allowed measurements to be reported in metric units rather than pixel lengths. For each variant, at least 20 measurement points were collected per wall type and averaged for subsequent comparison.

2.3.2. Mechanical Properties

To assess filament-specific behavior, tensile testing was performed on dogbone specimens (DIN EN ISO 527, Type 5A) printed from four different TPU filaments (extrudr FlexSemiSoft, SUNLU TPU 95A, OVERTURE TPU 95A, GEETECH TPU 95A). Testing was carried out on a universal testing machine (LTM10, Zwick-Roell GmbH & Co. KG, Ulm, Germany), with a minimum of five replicates per filament. The mechanical response of the disc-shaped samples was examined in uniaxial compression (Zmart.Pro, Zwick-Roell GmbH & Co. KG). At least three discs per parameter set were tested. Following preload application (1 N), force–displacement data were continuously recorded at a sampling frequency of 1000 Hz while the specimens were compressed at 5 mm/min until 50% strain or a maximum load of 20 kN was reached. The software automatically terminated the test once either limit condition was met. Thickness measurements were taken with a digital caliper at five points before, immediately after, and 24 h post-compression to monitor deformation recovery. Force–displacement curves were averaged from three replicates per condition. Target force ranges for testing were derived from in vivo spinal loading data: vertebral endplates typically withstand 4000–6000 N [20], and intervertebral discs can experience up to 200–250% of body weight during daily activity or exertion [21], corresponding to ~2100 N in an 85 kg individual. To simulate possible peak loads, e.g. during trauma, a broader working window of 4000–7500 N was defined. Samples exceeding 7500 N or compressive strengths above 3.81 MPa (normalized to 50 mm disc diameter) were excluded. This threshold was chosen to reflect the larger contact area of the printed discs relative to native discs, while still aligning with previously reported physiological values (≤3.5 MPa) for intact human lumbar segments [22].

2.4. Statistics

All data are reported as mean values accompanied by their respective standard deviations. Prior to statistical testing, the distribution of the data was examined using the Shapiro–Wilk test. For comparisons between groups, a one-way ANOVA with Tukey’s post hoc test was applied, considering differences significant at p < 0.05. Statistical processing was carried out with OriginPro 2025 SR1 (OriginLab, Northampton, USA).

3. Results

3.1. Sample Dimensions

With the temperatures optimized by the temperature towers (see Table 2), the discs with the different gyroids of varying volumes and wall thicknesses could be printed consistently. In addition, additional drying of the filaments at 55°C in the filament drying box further improved the printing result. The following Figure 2 shows examples of slices with different wall thicknesses (inner vs. outer) for the different filaments for a Gyroid volume of 8 mm³. Due to space limitations, the figures for SemiSoft and Overture can be found in the Appendix (Figure A1 and Figure A2). Table 3 below compares the measured wall thicknesses with the target values from the CAD models for the various TPU filaments.

With the exception of minor deviations, all filaments were within the specified target range, both for the Gyroid and for the outer wall. However, there are differences between the filaments. This can be seen quite clearly in Figure 2 and Figure A1 and Figure A2 in the Appendix in terms of print defects. In particular, Overture experienced an increased number of print defects at the lower wall thicknesses. A substantial reduction was achieved through further optimizations in printing speed and retraction. However, some drop formation and slight stringing was observed. With the other filaments, these printing errors were not as pronounced.

3.2. Mechanical Properties

3.2.1. Tensile Tests

The tensile tests were carried out in accordance with ISO EN 527 with 5A specimens. The measurements were repeated at least 5 times. Due to the nature of the machine, the samples could only be stretched up to a maximum of 250%. The measurement setup on the Zwick/Roell LTM10 is shown in the following Figure 3. The specimens were pre-tensioned to 5 N and then pulled at a rate of 1 mm/s.

The stress-strain curves of the four distinct TPU filaments (GEEETECH, OVERTURE, SUNLU, and SemiSoft) manifest conspicuous viscoelastic behavior, accompanied by elevated ductility (Figure 4). It has been demonstrated that all samples achieve maximum elongations of 250% without brittle fracture, depending on the testing machine. The curves are characterized by an initially steep rise in the elastic range, followed by plastic-viscous flow with moderate stress hardening. GEEETECH and SUNLU demonstrated the highest tensile strengths, at 10.6 ± 0.2 MPa and 10.9 ± 0.15 MPa, respectively, and exhibited substantial hardening with increasing elongation. These materials are well suited for functional applications that require both strength and elastic recovery. In contrast, OVERTURE exhibited the lowest overall load values (9.1 ± 0.05 MPa) and reduced stiffness in the initial range, suggesting a softer material behavior. SemiSoft exhibits a balanced profile between stiffness and elongation capacity, with a reading of 10.0 ± 0.07 MPa. The tensile modulus of SUNLU TPU was the highest of all filaments tested at 29.3 ± 2.2 MPa. The tensile modulus results of Sunlu, Geeetech and SemiSoft were not significantly different. A similar result was found for the compressive strength of the four filaments. Figure 5 and Table 4 summarize the results for tensile modulus and compressive strength.

There is no significant difference between the filaments of SemiSoft, SUNLU, and GEEETECH in terms of tensile modulus (p > 0.05). However, OVERTURE´s filament differs significantly from the other three manufacturers in terms of tensile modulus, with p<0.0001. The filaments of SUNLU and GEEETECH do not differ significantly from each other in terms of tensile strength, with p = 0.83. But the filament of SemiSoft differs significantly from SUNLU (p < 0.01), GEEETECH (p < 0.01), and OVERTURE (p < 0.001) in terms of tensile strength. The same applies to the significant differences in tensile strength between OVERTURE and SUNLU (p < 0.001) and between OVERTURE and GEEETECH (p < 0.001).

3.2.2. Compression Tests

3.2.2.1. SemiSoft

As can be seen in Figure 6, only two samples are within the specified range. These are samples FS23 with F_max of 6918 ± 66 N and compressive strength of 3.52 ± 0.03 MPa) and FS33 with F_max of 4171 ± 38 N and compressive strength of 2.13 ± 0.02 MPa.

3.2.2.2. SUNLU

For SUNLU filament, three samples: SL11; SL21 and SL33 were within the specified range. SL11 achieved F_max of 6607 ± 267 N and compressive strength of 3.4 ± 0.16 MPa; SL21 achieved F_max of 5718 ± 222 N and 2.9 ± 0.11 MPa and SL33 achieved F_max of 6123 ± 207 N and compressive strength of 3.1 ± 0.08 MPa (see Figure 7).

3.2.2.3. GEEETECH

Samples GS07, GS21 and GS33 were within the specification. GS07 had a maximum force of 5533 ± 38 N and a compressive strength of 2.81 ± 0.06 MPa. GS 21 was in a similar range of the maximum tolerated force with 5935 ± 221 N and a compressive strength of 3.02 ± 0.09 MPa. GS33 was also in the same range with 5542 ± 190 N and 2.82 ± 0.08 MPa (see Figure 8).

3.2.2.4. OVERTURE

The samples OT09, OT21 and OT33 met the specified requirements. OT09 had a maximum load of 4585 ± 117 N and a compressive strength of 2.29 ± 0.06 MPa. OT21 showed comparable values with a maximum load of 5171 ± 45 N and a compressive strength of 2.55 ± 0.02 MPa. Similarly, OT33 achieved 4922 ± 69 N and 2.38 ± 0.03 MPa (see Figure 9).

The actual values achieved in the maximum tolerated force for the tested samples are summarized in Table 5. The dimensions of all samples that were within the specification were summarized in the following Table 6.

3.2.3. Sample Height

Sample height was measured using a digital caliper before, immediately after, and 24 hours after the mechanical tests. It was found that after compression to 50% of the height, the specimens required a period of time to return to their original height. After 24 hours, all specimens had returned to their original height. The following Figure 10 summarize the specimen height distribution for the specimens in the 4000-7500N target range.

4. Discussion

Although all of the filaments tested in this work were specified by the manufacturer to have a Shore hardness of 85A or 95A, the filaments differed significantly in their properties in both tensile and compression tests, as well as in their 3D printability. The TPU filaments tested in this work were significantly softer than the FlexMed and FlexHard filaments we had previously tested [19] with a Shore hardness of 95A as specified by the manufacturer. Nevertheless, Sunlu and Geeetech exhibited analogous values for tensile strength and tensile modulus. SemiSoft exhibited a similar trend but demonstrated a substantial disparity in tensile strength. The TPU from Overture exhibited marked differences from the other three filaments in all cases. The tensile strength of SemiSoft was only 1/5 that of FlexHard and 1/3 that of FlexMed from our previous work [19]. A parallel situation was observed for the SUNLU and GEEETECH TPU filaments. In fact, the OVERTURE filament was only 1/10 the strength of FlexHard and 1/6 the strength of FlexMed. Shin et al. [23] described a similarly strong TPU filament with 19.9 MPa, which they also 3D printed using FDM at 180~220°C. Harynska et al. [24] also printed their TPU for medical applications using FDM, but at 200°C, achieving a comparable tensile strength of 26 ± 2 MPa. However, Shin et al [25] printed their TPU as pellets, not filaments, at 190°C and achieved tensile strengths between 7.42 and 26.93 MPa, depending on the composition. By blending TPU with ABS, Soltanmohammadi et al. [26] achieved compressive strengths in over 150 MPa.

Bayati et al [27] argue that in extrusion-based 3D printing, the orifice through which the material is extruded is circular, resulting in cylindrical print paths. When these paths are layered and joined together to form the final object, gaps or spaces are often created between the layers [28]. This is illustrated in Figure 4, which shows a cross-section of a printed part. The size of these spaces, also known as voids, depends on the print quality. Optimal printing parameters and high printability reduce these voids, resulting in a more uniform microstructure of the printed part [29]. Improved overlap, material flow, and wettability minimize voids, thereby improving the mechanical properties of the part [30]. However, the presence of voids can contribute to damage in printed samples. Stresses can develop at the layer interfaces, which can promote crack growth or delamination [31]. These defects, combined with external stresses or environmental factors, can compromise the structural integrity of the printed object and lead to failure or reduced performance over time. This is exactly what we observed with the flexible TPU filaments. Because the Prusa MK3S+ extruder had difficulty with the TPU filament, there was some stringing and drooling on the prints, but they were otherwise flawless.

Regarding the recovery of the tested discs, we were able to determine, analogous to Feki et al. [32] or Goode and Teodore [33], that our discs behave like intervertebral discs and return to their original thickness overnight. How is this effect achieved? The osmotic pressure of the nucleus pulposus counteracts this stress; however, it is often exceeded during physical activity, resulting in the displacement of water from the disc [34,35].This process leads to a reduction in disc height and volume [36,37,38]. During overnight rest, the decrease in external loading allows the osmotic pressure of the nucleus to reabsorb water, facilitating rehydration of the disc [39,40,41].While some research suggests that loss of disc height is due to radial expansion of the annulus fibrosus [42,43]. Botsford et al [38] argue in their study of diurnal disc changes that such radial bulging is minimal during typical daily activities. They claim that the main factor contributing to the reduction in disc height is the loss of fluid from the disc itself. It is noteworthy that a healthy intervertebral disc can fully recover its original pressure, height, and volume despite being loaded twice as long during the day as during the resting phase at night.

5. Conclusion

In this study, a comprehensive test matrix comprising four different Gyroid sizes, each combined with three wall thicknesses and three outer wall configurations, resulting in 36 distinct specimens per filament, was evaluated across four TPU filaments. By applying an optimized drying protocol at 55 °C prior to and during the printing process, filaments that had been unprintable in a previous study [19] were successfully processed, with substantial reductions in printing errors such as stringing and droplet formation. The TPU filaments with Shore hardness 95A displayed increased flexibility compared to previously reported 95A materials, with 2–3 Gyroid structures per filament achieving peak loads within the target range of 4000–7500 N. These findings indicate that structural variation in Gyroid geometry and wall design can significantly influence the mechanical performance of printed samples. Compared with rigid UHMWPE and titanium implants, the additively manufactured TPU Gyroid structures produced via FDM demonstrate promising shock-absorbing capabilities, reinforcing their potential as intervertebral disc substitutes. Clinically, these results emphasize the importance of tailoring internal architecture to optimize biomechanical properties. Future work should extend to intervertebral disc–shaped specimens with and without outer walls, under both static and dynamic loading conditions, while also accounting for the natural three-dimensional curvature of discs.