Screen-Printable p- and n‑Type Functionalized Graphene Inks for Flexible Textile Thermoelectric Generators

Abstract

Wearable thermoelectric (TE) generators offer a sustainable solution for powering low-energy electronics by harvesting body heat. In this study, we report the development of fully screen-printable, solution-processed graphene inks exhibiting p- and n-type thermoelectric behavior for flexible textile applications. The inks were printed onto woven cotton substrates by using scalable screen-printing methods, forming all-graphene thermoelectric modules that retain textile flexibility and breathability. Electrical characterization demonstrated Seebeck coefficients of +34 μV/K. Under a modest temperature gradient of 40 °C, the TE textile device generated open-circuit voltages up to 5.24 mV with stable operation. The materials and device exhibited good adhesion, flexibility, and thermal response without requiring postprocessing or high-temperature annealing. This work presents a cost-effective, scalable, and environmentally benign approach to fabricating wearable thermoelectric systems, offering strong potential for powering flexible electronics, health monitoring sensors, and future energy-autonomous garments.

Introduction

Wearable smart devices are an important part of future medical, fitness, and protective service industries as they allow garments or wearable accessories to act as sensors, actuators, or monitoring devices without requiring bulky, rigid structures. − To make these devices autonomous from external energy sources, a thermoelectric system can be used. Thermoelectric (TE) systems have the ability to use the human body’s waste thermal heat to generate electricity and require positively charged (p-type) and negatively charged (n-type), electrically conductive materials. The imbalance in electrons between the two materials creates an electrical current by moving electrons from one side of the device to the other by heating one side of the device. Metals, polymers, and graphene can be used for both types of materials and have been successfully used in textile-based thermoelectric systems. Because graphene is a neutrally charged material, it can also be doped to be positively or negatively charged. − Additionally, using graphene as the principal element for both sides of the p–n junction is not common practice as the electron imbalance has not been great enough to produce thermoelectric capabilities. −

For wearable systems, the material needs to withstand transfer to a human, maintain its flexibility during use, and be washed. One fabrication method, screen printing, allows a smooth, even coat of ink to be applied onto a fiber, yarn, or fabric substrate in a simple, replicable process. ,− Screen printing requires a viscosity between 1 and 500 Pa·s, making it a promising technique for a wide variety of conductive inks. Adding protective polymer, such as polyurethane, to the graphene can increase the device’s flexibility, reduce its brittleness, and reduce transference. , Therefore, this research investigates a wearable, washable, screen-printed TE system comprised from the novel use of doped graphene ink, which is both positively and negatively charged through additive manufacturing methods. The resulting device proves that screen-printing graphene onto a woven textile can create a wearable, woven, and functional TE garment.

Results and Discussion

The n-type graphene ink was made from amine [−NH₂] functionalized graphene with a flake size of 5–10 μm (purchased from Ad-Nano, U.K.) and the p-type graphene ink was made from metal-doped graphene with a flake size of 10–20 μm (purchased from National Technology, CH). Dimethyl sulfoxide (DMSO) (purchased from Sigma-Aldrich, U.K.) was used as solvent to prepare the graphene ink, and Poly­[4,4′-methylenebis­(phenyl isocyanate)-alt-1,4-butanediol/di­(propylene glycol)/polycaprolactone] (PU) (purchased from Sigma-Aldrich, U.K.) was used as a multifunctional agent which works as a stabilizer, provides sufficient adhesion between the graphene coating and the fabric to protect the graphene from washing, and forms a protective hydrophobic layer to provide durable performance. 100% cotton fabric was used for the substrate (purchased from Abakhan, U.K.).

The n-type ink (G_N) was manufactured by sonicating the amine functionalized graphene into DMSO at 0.1 g/mL for 9 h. The p-type ink (G_P) was manufactured by sonicating the metal-doped graphene into DMSO at 0.08 g/mL for 9 h. To ensure adherence to the substrate and washability, a DMSO/PU solution was created by adding PU to DMSO in a 1:9 ratio and mechanically stirring it for 24 h at 70 °C. Finally, 10 mL of this solution was added to both inks and mechanically stirred for 4 h at 70 °C. Once the inks were manufactured, the inks were screen printed in a series of p–n junctions on a substrate. The cotton fabric was mounted on an inflexible base, and the printing screen (110 polyester mesh) was prepared with a predefined rectangular block design (2 cm × 1 cm), allowing for tailored placement of the inks. Each ink is deposited in a 1 cm × 2 cm long strip with a 0.5 cm overlap between the two types of inks. A single layer of p-type ink was deposited onto the cotton fabric via screen printing and cured at 80 °C for 1 h. This process was repeated once. Next, a single layer of negative-type ink was printed with an overlap of 0.5 cm on the p-type ink and cured at 80 °C for 1 h. This was repeated two times. A total of four strips, containing 9 junctions, were completed, with the process shown in Figure a,b, with a total of 36 junctions used for testing. A final schematic of the electrical layout of the device can be seen in Figure c.

1.

(a) Schematic of fabrication of TE strips, (b) image of the completed TE strip, and (c) schematic of electrical layout for testing.

To analyze the ink, several tests were undertaken. A Bruker α FTIR spectrometer captured the transmittance of both the p- and n-type inks, as shown in Figure a. The FTIR spectra are used to understand the covalent and noncovalent polymerization of the two inks. The peaks at 2996 and 2913 cm^–1 are C–H stretch vibration bands indicative of DMSO. The three sharp peaks at 1436, 1407, and 1300 cm^–1 show the deformation bands of the C–H bonds in DMSO. The large peak at 1015 cm^–1 linked to the smaller peaks at 950 and 930 cm^–1 is the Si–O deformation bands. The rounded peak at 3494 and 3424 cm^–1 shows the C–OH and C–H stretching, respectively. The functionalized graphene peak, 1725 cm^–1, is characteristic of the C0 stretch for the COOH group. The peaks at 1658 cm^–1 are the CC stretching. The large peak at 1040 cm^–1 is a C–N stretch. The larger peaks found at 1041 cm^–1 for the G_N ink show additional ammonia in the functionalized graphene.

2.

Characterization of both p-type and n-type inks in (a) FTIR, (b) RAMAN spectroscopy, and (c) SEM imaging of the n-type, p-type, and junction of graphene inks.

Raman Spectrometry was used to determine the composition of two types of graphene and if there were any defects. A Horiba LabRAM instrument was used with a coherent 488 nm 100 mW laser. As seen in Figure b, the spectra for G_P and G_N were taken. G_P has its D peak around 1355 cm^–1, its G peak at 1575 cm^–1, and its 2D peak at 2719 cm^–1. G_N has its D peak at 1344 cm^–1, its G peak at 1574 cm^–1, and its 2D peak at 2700 cm^–1. The G band denotes bonded carbon atoms, these inks confirm a carbon makeup. , The 2D band peaks show the multiple number of layers, which is expected from the number of coatings applied to the substrate. The D band is not found in pristine graphene and it being shown in the n-type graphene reflects the ammonia doping.

The morphologies of the two inks were investigated through a FEI Quanta 250 FEG-SEM + Gatan 3view EDX. As shown in Figure c, the n-type and p-type inks have different flake morphologies. G_N shows small flake size with clear ridges of the fabric’s weave seen and G_P shows a larger flake size with a smooth, solid covering of graphene flakes on the cotton fabric. The junction between the two inks shows the flake difference and the impregnation of G_N into the fabric while G_P remains on the surface. The elemental analysis was conducted with a Gatan 3view EDX system, shown in Table . The results show that G_N was derived from pure graphene while G_P had impurities such as magnesium, silicon, and iron.

1. Elemental Makeup of G_P and G_N .

GP		GN
element	% found	element	% found
C	66.82	C	83.44
O	8.35	O	16.56
Mg	5.74
Si	17.86
Fe	1.23

Because the inks were screen printed onto the cotton fabric, an AR-G2 rheometer was used to measure the ink’s average viscosity under atmospheric temperature. Figure a shows that G_N has a lower viscosity than G_P but both inks featured high enough viscosity to not bleed when deposited on the cotton substrate.

3.

Both inks analyzed for (a) viscosity, (b) average particle sizes, and (c) coatings vs resistance.

To measure the graphene flake sizes, a Malvern MasterSizer 3000 was used, and the flake sizes were averaged from five samples from 5 different runs. The results, seen in Figure b, show that G_P had an average particle size of 8.68 μm with some aggregation while G_N has no aggregation and an average particle size of 7.64 μm. The aggregation of the G_P ink may be because it is positively charged and the flakes do not repel each other as with G_N.

The resistance of the inks impacts the performance of the device. As shown in Figure c, each coating has an error bar which reveals how each coating is different depending on the sample, therefore impacting the overall resistance and of the inks. A lower resistance increases the conductivity of the device, and the more coatings applied, the lower the resistance. The electrical conductivity of the inks and device was calculated from the resistance of each coating with G_P increasing from 0.115 to 0.134 S·cm^–1 and G_N increased from 0.012 to 0.028 S·cm^–1.

Once the inks were analyzed, the device was created via weaving the p–n strips through a cotton fabric so that the junctions alternated sides, as shown in Figure a. The woven strips, made up of 9 junctions, were connected in series to each other with copper connectors to create a total of 36 junctions. The device was placed on a Benchmark 230 V hot plate that was heated to three specific, steady-state temperatures with the voltages and currents taken using T-type thermocouples to determine the temperatures, an OWON OW18E multimeter to test the electrical resistance and current of the device, and a 10-bit (±1 V) Voltage Input Phidget to measure the voltage output of the system. The device was first heated from an ambient, cold temperature (T _C) of 21.5 °C to a heated temperature (T _H) of 40 °C to understand how long the device took to respond. The long length of time, 10 min, may be due to low conductivity in the material and thermal conductivity. From that, the device was tested at a T _H of 36, 38, and 41 °C, shown in Figure b. For each of these tests, T _C was kept at 21.5 °C. The results show the device responding to the temperature differences because the steady-state voltages were 0.4 mV at 36 °C, 0.55 mV at 38 °C, and 0.91 mV at 41 °C.

4.

TE capabilities including (a) temperature ramp up to 41 °C with device image, (b) voltages at all tested temperatures, (c) Seebeck coefficients of device compared against other research, and (d) the I–V curves at 38 and 41 °C.

The Seebeck coefficient (α) was calculated from the equation α = ΔV/ΔT and the results are seen in Figure c, with the system being most efficient at ΔT 16.5 °C at 28.68 μV·K^–1. The efficiency drops at higher ΔT because the system fails to produce an effective voltage. This is acceptable because this device is for wearable applications, requiring low ΔTs. The Seebeck coefficient from this work is also compared to other work, ,,, showing that using graphene for both the p- and n-type material does not negatively impact the overall Seebeck coefficient.

The current–voltage (I–V) curve was created with a voltage reader, a resistor, and a current reader. The I–V curve was successfully created at 38 and 41 °C, as seen in Figure d, but for the T _H of 36 °C, there was no viable current as the results fell in the error range of the instruments. From this, the maximum output power (P _max) can be generated using the equation P _max = (1/4)*V _oc*I _sc. This resulted in P _max of 31.7 and 62.4 μW for ΔT 13.5 K and ΔT 16.5 K, respectively.

In conclusion, a woven, advanced-material-based thermoelectric device, fabricated from novel functionalized graphene and a flexible, cotton substrate, can perform as a TE energy harvesting device. The 36-junction device produced a voltage of 0.91 mV, a current of 0.2 μA, a maximum power generation of 62.4 μW at a ΔT of 16.5 K, and a calculated Seebeck of 28.68 μV·K^–1. The low ΔT used to test the device shows promise in the future of wearable, energy harvesting textiles that function without metallic parts, which restrict the ductile nature of a garment.

The data underlying this study is available in the published article.

Samantha Newby: conceptualization, methodology, data curation, writingoriginal draft preparation, investigation, validation. Md Raju Ahmed: methodology, writing-reviewing and editing. Wajira Mirihanage: supervision, writing-reviewing and editing. Anura Fernando: supervision, writing-reviewing and editing, methodology.

The authors declare no competing financial interest.