A Prosthetic Hand with Integrated Sensing Elements for Selective Detection of Mechanical and Thermal Stimuli

Flexible electronics have gained popularity because of their capability to combine softness and functionality. Soft resistive sensors are susceptible to mechanical stimuli and detecting temperature selectively remains a challenge. In this study, soft flexible thermistors are developed for detecting temperature changes based on the positive temperature coefficient (PTC) effect, selectively. By thermomechanical analysis and differential scanning calorimetry, it is observed that thermoplastic elastomers with higher thermal expansion and semicrystalline morphology result in a sensitive thermistor response. To achieve a high sensitivity in temperature and low sensitivity in the detection of mechanical stimulus, a low‐carbon filler content is required. The opposite trend is seen for the piezoresistive sensors for mechanical strain detection. Both sensory material types are compatible with thermoplastic material extrusion‐based additive manufacturing. The method is used for the fabrication of the sensing elements and an open‐source prosthetic hand. The strain sensor detects the bending of the fingers and the temperature sensor detects the temperature when in contact with a heated surface, successfully. In addition, the temperature sensor is used as a tactile sensor to detect contact with a non‐heated surface. Combining selective multisensory capabilities will significantly affect the future development of sensorized prosthetic devices and wearable electronics.


Introduction
[3][4][5][6] While sensors developed for prosthetics usually target motion detection and proprioception capabilities, [7][8][9] sensing temperature is an additional essential sensory ability of the human skin.Especially detecting critically high levels of temperature that could potentially harm the prosthetic device is useful for long-term use. [10]However, combining multiple sensory capabilities comes with a significant challenge.Selectivity is the ability of the sensor to measure a stimulus in a presence of one or multiple interfering stimuli.In the research on soft resistors, the aspect of selectivity has been marginally explored so far. [11]Developing soft sensors that can respond selectively to the temperature or the strain stimulus will have a significant impact on sensorized prosthetic devices.
A positive temperature coefficient (PTC) thermistor is a material whose resistivity increases with an increase in temperature. [12]For materials with a negative temperature coefficient (NTC), the resistivity decreases with an increase in the temperature. [13,14]Jian and Chen have attributed two main mechanisms for carbon-filled thermoplastic elastomers (TPEs) to describe the PTC and the NTC effect. [15]or the PTC effect, conductive fillers are typically dispersed inside the soft phase of TPE materials.Due to the thermal expansion of the soft phase, the distances between the carbon black (CB) agglomerates increase with the rising temperature. [12,16,17][20] This mechanism is particularly relevant above the glass-transition temperature, as it acts as an onset for the motion Brown of the segments and molecular chains. [21][30] This is the case especially for thermoplastic-based material extrusion AM (MEX-AM), as the method is named in the standard ISO/ASTM 52 900:2015.In a previous study, Petek et al. suggested the use of MEX-AM for sensorized upper limb prosthetics, using a sensor based on polylactic acid (PLA) and acrylonitrile butadiene styrene materials.Nonetheless, using elastomers will provide a more anthropomorphic experience for the user and will also increase conformity and comfort. [31]won et al. fabricated an ankle brace with MEX-AM, using thermoplastic polyurethane.They integrated strain sensors in their prototype that were used for monitoring the motion of the ankle joint in post-stroke patients. [32]n this study, an open source prosthetic hand with flexible fingers is fabricated with integrated soft sensing elements for detecting changes in temperature and strain, using MEX-AM.For the piezo-and thermoresistive soft sensor development, CB was used as a conductive filler in a TPE matrix.The compositions were optimized to achieve a selective detection of each stimulus.There are only a few examples, where the sensor selectivity has been investigated.An et al. and Wu et al. have investigated the selectivity of their temperature sensor to flexion and torsion. [33,34]The selectivity under tension and compression (i.e., touching a non-heated surface) has not been investigated so far.For both sensing types, the sensing mechanism is based on an increase of the interparticle distance (Figure 1a-d).In addition, the mechanism of the thermoresistive sensing elements is based on the PTC effect and thus, the thermal expansion and crystallinity of the composites is explored.The selection of the soft TPE matrix and the concentration of the conductive filler are also examined.Finally, the feasibility of using the thermoreceptive sensor as a touch sensor is explored.It is expected that the external force will cause a deformation of the composite and thus, a decrease in the interparticle distance, causing the resistance to drop.

Piezo-and Thermoresistive Response of Resistive Sensor Fibers
To investigate the temperature and strain behavior of the thermoplastic polyurethane (TPU)-and polystyrene-based tri-block copolymer (TPS), sensor fibers were mixed and extruded with different carbon filler concentrations (Figure 2a).For this investigation, an MEX-AM 3D printer was used for the fiber extrusion.A fiber diameter of 0.8 mm was measured for all fibers (Table S1, Supporting Information).In the case of the TPU-based fiber sensors, above 30% w/w filler concentration, the fibers became too brittle for mechanical testing.According to Figure 2b, between 20% and 30% w/w, the resistivity decreased by two orders of magnitude for the TPU-based sensor fibers.For the TPS-based sensor fibers, a maximal filler content of 50% w/w was observed before the fibers got too brittle for mechanical testing.Between 20% and 50% w/w, the resistance decreased by five orders of magnitudes for the TPS-based sensor fibers (Figure 2c).
Looking at the stress-strain curves for TPU (Figure 2d) and TPS (Figure 2e) sensor fibers, the ultimate strength increased, with the increasing CB content, as expected.For the TPU sensor fibers, the elongation at the point of fracture decreased by increasing the CB content and this is in good agreement with what has been previously reported. [35]In polymers, a yield point can be interpreted as the transition area between elastic and plastic deformation. [36,37]For the TPS sensor fibers with 40% and 50% w/w CB, a yield point appeared at 38% and 10% strain, respectively.In Figure S1, Supporting Information, the Young's modulus of the TPU and TPS composites with the filler content is presented.For the TPU composites, a significant decrease in the Young's modulus by increasing the CB filler content was observed.This can be explained by the weak interphase between the hydrophilic TPU matrix and the hydrophobic carbon filler. [38]41][42] For all sensor fibers, a piezoresistive response was observed (Figure 2f,g).By far, the TPS sensor fiber with 50% w/w CB resulted in the highest sensitivity.[45] For the strain region below 100%, the GF for TPU fibers with a CB concentration of 20% w/w was close to 0, whereas, for the TPS fibers with CB concentration of 50% w/w, a GF of 47 could be achieved.The GF for all fibers is reported in Table S2, Supporting Information.For obtaining a sensitive response at low strains (<50%), a dense conductive network was required.For very high strains (>300% strain), the lower carbon content gave a higher sensitivity factor.However, relatively low strains are required for the application of the prosthetic finger.Therefore, the fiber with 50% w/w CB concentration had the highest sensitivity at low strains and was selected as the most suitable composition for the piezoresistive sensing element.
To investigate the thermoresistive effect, the electrical resistance was analyzed while increasing the temperature (Figure 3a).Looking at the relative resistance (Figure 3b,c), it can be observed that all fibers show a PTC effect at low temperatures.
At higher temperatures, the relative resistance decreased (NTC effect), except for the TPS fiber with 50% w/w CB content.[27] A usual method to increase the range of the PTC effect in polymers is the use of a low conductive filler content. [17,46]This is in good agreement with the TPU sensor fibers.TPU with 20% w/w CB shows the largest range of the PTC effect and the highest sensitivity (Table S3, Supporting Information).
From the thermomechanical analysis (TMA) (Figure 3d), it was seen that above 50 °C the thermal expansion coefficient was higher for the TPU compared to the TPS matrix.This is in good agreement with the higher sensitivity observed for the PTC effect for the TPU sensor fibers.A higher thermal expansion led to a larger increase in the interparticle distance for the same temperature change. [16]Especially for the profile of the response of the TPU 20% fiber, the response resembled closely the sharp increase in the resistance around the crystallization temperature that is common for semicrystalline materials.For that reason, a differential scanning calorimetry (DSC) experiment was carried out for the TPU and TPS matrix materials (Figure 3e).In the melting thermogram, it was seen that the TPU matrix exhibited a characteristic peak melting temperature (T m) around 160 °C.49] Interestingly, around this temperature, the transition from PTC to NTC effect was observed.It can be assumed that above the melting temperature, a reorganization (re-agglomeration) of the carbon filler occurs, which results in an NTC effect.The functional range of the temperature sensor was from room temperature and up to the melting point of the TPU (160 °C).This range is higher than what has been reported previous in literature with an average functional range 25-60 °C. [16]s for the cooling thermogram, a peak was observed at 67 °C.This observation agrees with the findings of Voda et al. [50] and Lin et al. [47] Rashmi et al. measured a crystallization temperature (T c ) for TPU at 70 °C. [51]A peak at T c can be typically observed in semicrystalline materials.The degree of crystallinity for TPU elastomers is closely linked with the Shore hardness value.For higher Shore hardness, more crystallites are formed during cooling, leading to a higher area under the crystallization peak of the DSC curve.This peak was expected for TPU in this Shore hardness range and has been already reported by others. [52]As for the TPS, the response of the thermograms resembled the behavior of amorphous polymeric materials. [53,54]For an amorphous material, no T m or T c are observed.57] The degree of crystallization is closely linked with the sensitivity and range of the PTC effect.60][61] Xu et al. suggest that it is difficult to get a sensitive PTC effect with amorphous materials, and semicrystalline polymers are better candidates for thermistor composites. [62]This factor in combination with the higher thermal expansion led to the higher sensor sensitivity of the TPU 20% sensor fiber.

Piezo-and Thermoresistance of TPU Strips with Integrated Sensing Elements
Based on the previous results, TPU with 20% and TPS with 50% w/w were selected for further investigations as thermo-and mechanoreceptors, respectively.The two sensing elements were integrated into the TPU substrate that was later used for the fabrication of the fingers of the prosthetic hand.A filament-based dual extruder was used for the TPU substrate and TPU-based sensor material.For the TPS-based sensor material, the TPU substrate was placed on the printing bed of the pellet-based extruder and the TPS-based conductive composite was printed on top of the TPU (Figure 4a).The two TPU strips with printed sensing elements on top of them are shown in Figure 4b.Comparing the mechanical behavior of the two strips (Figure 4c), it was observed that the strip with the TPU-based sensing element reached higher values of stress.This fact can be explained by the higher Shore hardness of the TPU (85 A for the TPU and 50 A for the TPS) and this behavior was also seen in the case of the sensor fibers.The TPU substrate had a significant effect on the stress-strain response.The profile of the stress-strain curve was different from the response of the sensor fiber with the same composition.The characteristic necking seen for the TPS element did not appear anymore.For the electrical response (Figure 4d), the strips with the TPS-based sensing element showed a sensitive piezoresistive response.The slope of the curve increased significantly at 38% strain.The TPU-based sensing element showed a small decrease in the electrical signal at low strains but above 30% strain, there was almost no piezoresistive response visible.The sensitivity factor was significantly smaller for the TPU compared to the conductive TPS element.This was in good agreement with the fiber analysis reported earlier.The TPS-based sensing element showed a sensitive response to the strain stimulus on TPU support structure, but the electrical signal of the TPU sensing element was not strongly affected by the strain stimulus (Figure 4e), as expected.Comparing the sensitivity for the temperature detection at low temperatures (50 °C) and high temperatures (100 °C), it was seen that the selectivity of the TPU sensing element was higher at higher temperatures.In a similar manner, the TPS sensing element showed higher selectivity at higher strain (50%) compared to low strain (5%).The sensitivity factor is typically reported to have values between 0.5 and 3 for piezoresistive strain sensors. [11]The strain sensor with a sensitivity factor of 4 was slightly above the average of reported values.
For the strips with TPS-based sensing elements, the sensitivity decreased compared to the fiber and the transition to the NTC effect appeared at a lower temperature.The matrix seemed to affect the temperature sensing response of the TPS-based conductive composite too.In this case, the higher filler concentration contributed to the re-agglomeration of the filler and therefore the enhancement of the NTC effect.As expected from sensor fiber investigations, the TPU strips with the integrated TPU-based sensing elements had the most sensitive response and a large range of the PTC effect, expressed by the significantly higher sensitivity factor.
In addition to the tensile testing up to the point of fracture, dynamic tensile cycling experiments were performed between 0% and 30% strain.Buckling occurred during unloading when force was not exerted on the strip.The strip did not return to the original length due to the viscoelasticity of the TPE matrix.Since the stress is a behavior dominated by the substrate, buckling was observed for both strips with the TPU and the TPS sensing element at the value of 5% strain (Table 1).
In addition, the drift of the sensor signal response was investigated for the dynamic cycling experiments.The drift of the sensor signal was very low for the TPS sensing element and significantly higher for the TPU.A prosthesis will not be used for dynamic movements only.To investigate the static behavior of the sensor signal, a dwell time at 0% and 30% elongation was added to the cycle tests (so called quasi-static test).The quasi-static test is relevant for the prosthetic hand for movements that involve the hand remaining in a certain position (e.g., gripping of objects).Based on the observations made during the tensile testing, the quasi-static testing was performed at the range of strains 0%-30% and the relaxation was measured at 30% and 0% strain to investigate the buckling and uncertainty of the sensor signal at low strains.The stress relaxation was similar for the strips with the TPU and the TPS sensor strip (Table 1).For the TPS strips, the relaxation of the electrical signal was 17%, 3% higher than the TPU strips.
Overall, the results of the strips with the TPS-based integrated sensors confirmed the best behavior for piezoresistive deformation sensing of the previous fiber tests.The good sensitivity, linearity, and positive piezoresistivity made them a good option for strain sensing.Especially the negligible uncertainty and low drift value made them superior to the TPU-based sensing elements.

Sensorized Prosthetic Hand Demonstrator
To demonstrate the selectivity of the mechano-and thermoresistive sensing elements, the sensing elements were tested on an Table 1.Properties of the two selected sensing elements derived from the dynamic tensile testing between 0% to 30% strains.Mechanical and electrical relaxation derived from the quasi-static tensile test.For the calculation, the Figure 2  open source prosthetic device (Figure 5).The fingers were printed from TPU and clipped into a metacarpal prosthesis made from PLA (Figure 5a).Temperature sensing elements were placed on the dorsal and volar side of the finger to be able to detect temperature changes and investigate the effect of mechanical deformation on the sensor response (Figure 5b,c).The strain sensing elements were only placed on the dorsal side of the finger due to the limited space on the volar side of the finger.
To assess the performance of the strain and temperature sensing elements on the dorsal side finger, an object recognition test was performed where three balls with different diameters (small, medium, large) were used.The goal was to be able to recognize the size of the ball based on the relative resistance signal value (Figure 5d).Using the relative signal of the TPU-based sensor element, it was not possible to detect the size of the balls, the sensor showed significantly lower sensitivity than the TPS element.Looking on the TPS-based sensing element the relative signal can be used for posture recognition of the hand.The values of the relative resistance change in the different positions for all fingers are summarized in Table S4 and S5, Supporting Information.Despite some small differences between the different fingers, the trend was similar for all positions.The example of the index finger is used further on.First, the hand moved from position open to close, picking up the large ball (5 cm).The motion caused an increase in the relative resistance of 0.38% and 11% relaxation during the holding phase.After opening the hand, the relative resistance returned to the initial value.Due to the low geometrical stiffness of the open source flexible fingers some noise of the electrical signal was present while the hand moved to pick up the smallest ball (3 cm).The change in relative resistance was 0.52 and the relaxation at this position was 6%.When opening the hand, the relative resistance returned to the initial value after dropping the object and by grapping the medium size ball (4 cm), the change in relative resistance was 0.41 and a relaxation of 10% could be detected.Finally, the hand opened to intermediate position (half-closed hand posture) without holding a ball and the relative resistance decreased from 0.43 to 0.3.
For investigating the thermoresistive sensing behavior, the fingertip (volar side) touched a heated surface at a temperature of 170 °C (Figure 5e).From the electrical response of the TPUbased sensor, it was seen that there was a rapid increase in the relative resistance (PTC effect) of the sensing element.After removing the finger from the heated plate, the signal decreased.Significant relaxation was observed after returning to 25 °C, which was attributed to the slow heat transfer.The finger didn't return to room temperature instantly, but several seconds were required.Nonetheless, because of the sharp increase in relative resistance, it was possible to detect when the temperature would rise close to critical values within 10 ms.This threshold value (Rel.Resistance = 6) indicated that the temperature had risen to a critical level (170 °C) because of the melting point of TPU at 180 °C.This threshold will be critical for the lifetime of the prostheses.In a similar manner, nociceptors in the human skin can indicate the presence of eminent thermal damage on the skin surface.In addition to the heating test, the thermoresistive sensor on the volar side touched a surface with lower temperature than the environmental (25 °C) and a negative relative resistive signal was detected (Figure 5f ).As discussed previously (Figure 4), a small mechanical sensitivity can be expected and due to the finger design used in this study.[65] Therefore, a force gauge device was used to detect the mechanical sensitivity of the temperature on the fingertip.It has been reported that the force for typing a keyboard by human finger results in a force between 1.5 and 5 N, respectively. [66]Based on the performed experiments, we can see that applying a force up to 2.5 N and a relative resistance of 0.17 is still lower in comparison to the relative resistance signal change by temperature.The sensing element in that case responded with the relative resistance decreasing when the contact occurred.With this finding, it was confirmed that the attached sensing element could distinguish between heated and non-heated surfaces, showing selectivity in the temperature and the tactile stimulus.

Conclusion
In a study, a prosthetic hand with integrated selective sensing elements was produced with MEX-AM.Two different elastomers, TPU and TPS, were combined with different concentrations of CB to form selective resistive sensor materials for either mechanical or thermal stimulus.The TPS-based sensor material with the highest amount of CB (50 wt%) showed good sensitivity, even at low strains.The same composition was integrated in TPU strips and the results could be confirmed.In addition, the sensor on the TPU strip showed good dynamic response with negligible uncertainty and small drift.For the thermoresistive response, the sensitive was very poor.
The composition based on TPU with the smallest concentration of CB filler (20%) showed good sensitivity and large range for the PTC effect.Therefore, the material was selected as selective thermoresistive sensor material.This behavior was attributed to the higher thermal expansion and semicrystalline morphology of the TPU matrix compared to the amorphous TPS.When the same composition was integrated in the TPU strips, the NTC effect disappeared and that was considered an effect of the thermal expansion of the TPU substrate.
Finally, the selected sensing elements were integrated in a prosthetic hand structure.Every finger included one deformation and two temperature-sensing elements.The assessment showed that the TPS-based element could recognize the size of the ball gripping by the hand.The signal was linear with small drift but there was the presence of relaxation during holding time.As expected, the thermoresistive sensor could not be used to detect the gripping of balls different sizes.However, the thermoresistive element on the volar side could indicate when the finger was touching a heated surface.Overall, with the open source demonstrator, we could prove the applicability of MEX-AM fabrication of sensorized prosthetics.New commercial MEX-AM printers allow multi-material printing with more than two materials.MEX-AM in general is an interesting low-cost technology to customize prosthetics for patients from different financial and social backgrounds.With this study, it was showed that by integrating sensing elements inside MEX-AM prosthetics bio-inspired sensing abilities (thermoreception, tactility, and proprioception) can be integrated.

Experimental Section
TPE-Based Resistive Sensor Fibers and Composites: In this study, a styrenic TPE with a shore hardness of 50 A (Kraiburg TPE GmbH, Waldkraiburg, Germany) and a thermoplastic polyurethane with a Shore hardness of 85 A (BASF, Ludwigshafen am Rhein, Germany) were used.The production of soft TPE sensors with conductive fillers has been described in detail in previous studies. [67,68]In short, Carbon Black Ensaco 250G (Timcal, Bodeo, Switzerland) was mixed with a TPE material.The material was mixed using a torque rheometer (Thermofisher, Karlsruhe, Germany.The resulted composites were extruded into filaments with a capillary rheometer (Netzsch Gerätebau, Selb, Germany) and manually cut into pellets.For the different composites, torque as well as temperature analysis for the filament fabrication are summarized in Table S1, Supporting Information.In a second step, fibers were extruded with the pellet-based fused deposition modeling (FDM) printer Voladora Nxþ (International Technology 3D Printers S.L., Valencia, Spain).Both the TPS and TPU composites were extruded in the form of sensor fibers.An extrusion speed of 10 mm s À1 and a nozzle of size 0.8 mm were used for the fiber extrusion.
TPS-based sensing elements were printed with the FDM printer Voladora Nxþ on top of printed thermoplastic polyurethane TPU strips.The TPU strips were 3D printed with the printer Raise Pro 2 (Raise 3D, Irvin, USA). [69]To print the TPU strips (130 Â 10 Â 0.4 mm), Spectrum Flex-S 90A (Spectrum Filaments, Piece, Poland) filaments were used.For the TPU-based sensing elements, the TPU strips and the TPU sensing elements were printed with the Raise Pro 2 printer.For all 3D printing experiments, an extrusion speed of 10 mm s À1 and a nozzle of size 0.8 mm were used.
Mechanoreception Behavior: The electromechanical behavior was investigated on single sensor fibers and TPU strips with integrated TPS-and TPU-based sensing elements.The tensile tests were performed with a Zwick Roell Z005 universal testing machine (Zwick Roell, Ulm, Germany) with a strain rate of 200 mm min À1 .Pneumatic clamps applied the pressure of 4 bar to the sample to avoid slipping.Tensile testing was performed up to the point of fracture, under dynamic (cyclic) and quasi-static conditions that involved a dwell time of 60 s at maximum and minimum strain.The electrical response of the sensor signal was in situ monitored during the tensile testing using Keithley 2450 multimeter (Keithley Instruments, Solon, Ohio, USA).
The electrical drift was calculated from the dynamic testing as the percentage difference of the value at maximal strain between the second and the tenth cycle.The relaxation behavior was determined by the quasi-static tests as the percentage difference of the value at the beginning and end of the dwell time.
The relative resistance (R rel ) was calculated using the following equation The sensitivity factor was calculated with the following formula for the strain sensor sensitivity factor ðstrainÞ ¼ ΔR rel Δε with ΔR rel being the change in the relative resistance and Δε the change in strain.
The sensitivity factor was calculated with the following formula for the temperature sensor sensitivity factor ðtemperatureÞ ¼ ΔR rel ΔT with ΔR rel being the change in the relative resistance and ΔT the temperature change.
The resistivity ρ was calculated by the equation with R being the resistance, A the area of the cross-section of the fiber cross section and l the length between the electrodes.The dimensions of the printed structures (Table S1, Supporting Information) were investigated with a light microscope Zeiss SteREO Discovery.V8 (Carl Zeiss AG, Jena, Germany).Before the imaging assessment, the samples were immersed in liquid nitrogen and then cut to reveal the cross-section area.
Thermoreception Behavior: For the characterization of the thermistor response, the heating plate Digital Hot (Thermo Fisher Scientific, Waltham, Massachusetts, USA) was used.Single sensor fibers, strips, and flexible fingers with integrated sensor elements were investigated using a thermocouple type K (Fluke Corporation, Everett, Washington, USA) and the Keithley 2450 multimeter.
TMA: The TMA measurement was carried out with the device TMA 402 (Netzsch, Selb, Germany) from 30 to 130 °C.The heating rate was 5 °C min À1 .The sample had a cylindrical shape with a height of 10 mm and a diameter of 8 mm.The samples were fabricated with pellet-based AM as described in Section 2.1.
DSC: The DSC measurement was carried out with the device DSC 8000 (PerkinElmer, Waltham, MA, USA) from 20 to 240 °C.The heating rate was 20 °C min À1 .The sample weight was 8 mg.
3D Printing of Prosthetic Hand with Integrated Sensing Elements: For the design of the prosthetic hand, an open source design called "osprey hand" was used. [70]The design of the hand was based on tendon-based soft bending actuator modules for the flexible fingers and a rigid base (back of the hand).The flexible fingers were clipped into the rigid base mechanically.The base was printed with antibacterial PLA (PLACTIVE antibacterial filaments, Santiago, Chile) with the Raise Pro 2 3D printer, using an extrusion multiplier of 0.9, speed of 3000 mm min À1 , nozzle size of 0.8 mm, and a layer height of 0.2 mm.The temperature used was 215 °C for the nozzle and 60 °C for the printing bed.Similar to the strips with integrated sensing elements, the flexible fingers were printed from the TPU filament Spectrum Flex-S 90 A; an extrusion multiplier of 1.1, speed of 1500 mm min À1 , nozzle size of 0.6 mm, and a layer height of 0.2 mm were used.The temperature used was 230 °C for the nozzle and 45 °C for the printing bed.
The sensing elements based on TPU were also printed with the Raise Pro 2 3D printer, using an extrusion multiplier of 1.5, speed of 1500 mm min À1 , nozzle size of 0.6 mm, and a layer height of 0.2 mm.The temperature used was 250 °C for the nozzle and 45 °C for the printing bed.For the TPS-based sensing elements because of the low Shore hardness (50 A) of the TPS, a pellet-based MEX-AM 3D printer Voladora Nxþ was used.An extrusion multiplier of 1.1, speed of 3000 mm min À1 , nozzle size of 0.8 mm, and a layer height of 0.2 mm.The temperature used was 230 °C for the nozzle and 60 °C for the printing bed.
The printed strain sensing element was fixed on the dorsal side of the fingers using thermal heat treatment.The fingers were placed in the oven at 170 °C for 1 hour, with the sensing elements below the finger structures.The temperature sensing element was attached with thermal treatment on the volar side of the finger.The temperature sensing element was 3D printed on the dorsal side of the finger.Finally, the fingers were equipped with metal wire tendons and clipped onto the base part of the hand.The placing of the temperature sensor on the fingertips of osprey hand design had to be done manually.Due to these reasons, the sensor elements were printed on the print bed directly and laminated later on the flexible fingers using the thermal fusing method.
Finally, the fingers were equipped with metal wire tendons (Habia Cable, Söderfors, Sweden) and clipped onto the base part of the hand.The tendons acted at the same time as wires for the electrical measurements.The connection between the wires and the sensing elements was made with a conductive paste Bare Conductive (Bare Conductive, London, England) at the two ends of the sensing element.The contact force measurements for the fingers when in contact with a surface were performed using a handheld force gauge FGV-1XY from Shimpo instruments (Cedarhurst, USA).

Figure 1 .
Figure 1.Schematic of the sensing mechanism of the different resistive sensors integrated into the prosthetic hand: a) temperature sensing, and b) touch sensing on the volar side of the finger, c) temperature sensing, and d) proprioceptive (strain) sensing on the dorsal side of the finger.

Figure 2 .
Figure 2. a) The methods of thermoplastic processing for thermoplastic elastomer (TPE)-based sensor fibers.Optical image of fibers with a filler content of 30% w/w and the change in the resistivity by varying the carbon black (CB) concentration in b) thermoplastic polyurethane (TPU) and c) thermoplastic polystyrene-based tri-block copolymer (TPS) material.Stress-strain response during the tensile test to the point of fracture for sensor fibers based on d) TPU and e) TPS.The change in relative resistance during the tensile test to the point of fracture for sensor fibers is based on f ) TPU and g) TPS.

Figure 3 .
Figure 3. a) The setup for the measurement of the thermistor fibers.b) Resistive-temperature response for the b) TPU and c) TPS sensor fibers with different CB concentrations, d) thermomechanical analysis (TMA) results, e) thermal expansion coefficient, and f ) differential scanning calorimetry (DSC) thermograms for the TPU and TPS fibers.

Figure 4 .
Figure 4. a) Schematic of the fabrication of the sensor strips with integrated TPU or TPS sensing elements.b) Photograph of the sensing elements integrated in the TPU substrate.c) Stress-strain response during the tensile test to the point of fracture.d) The change in relative resistance during the tensile test to the point of fracture.e) Comparison of the sensitivity factor at 50 and 100 °C, and 5% strain and 50% strain.f ) Change of the resistance with the temperature for resistive sensor fibers and strips based on TPU and TPS.

Figure 5 .
Figure 5. a) The prosthetic hand including flexible fingers with integrated mechano-and thermoresistive sensing elements at the back of the fingers (dorsal side).b) Thermoresistive sensing element were placed also on the inside of the hand (volar side) to detect heated and non-heated surfaces c) Connection between electrodes and sensing elements, facilitated by conductive paste.d) The signal from the deformation sensing element of the index finger of the prosthetic hand when the prosthetic hand grips balls of different diameters (5, 3, 4 cm) to the position of rest.During the last cycle, the hand remained in intermediate position (half-open) before returning to the position rest.The signal from the TPU-based sensing element integrated into the prosthetic hand, e) when the hand was touching a heated (170 °C) and f ) a non-heated (room temperature) surface.
and 3 were used.