Strain sensors on water-soluble cellulose nanofibril paper by polydimethylsiloxane (PDMS) stencil lithography

Abstract

Transient electronics are of great interest in medical implants and environmentally friendly devices. These electronic devices can be partially or fully dissolved in water in a controlled period of time. Herein, we show the fabrication and characterization of silver nanoparticle-based strain sensors on the water-soluble nanofibril paper, a cellulose nanocrystal (CNC) produced flexible film, by stencil lithography. This is the first demonstration of using polydimethylsiloxane (PDMS) stencils to pattern metal electrodes on water-soluble substrates. For the large London dispersion force (LDF) between the PDMS stencil and the flexible substrate, the stencil can conformably cover the substrate and prevent metal diffusion into the area under the stencil when metal electrodes are patterned on water-soluble nanofibril papers. PDMS stencil lithography avoids damage to the cellulose nanofibril paper, which conventional photolithography processes lead to, without compromising the resolution in patterning. Our strain sensors have a high sensitivity with a gauge-factor of over 50 in strain testing, which is the highest among reported strain sensors fabricated on water-soluble substrates.

---

Introduction

Transient electronics have drawn a lot of attention in recent years for their unique property of being soluble in water, which is not feasible in modern silicon electronics. It has been demonstrated that transient electronics can be partially or fully dissolved in a controlled period of time.^1–3 This will enable potential applications in biomedical implants, drug delivery, zero-waste consumer electronics and even security devices for military applications.⁴ In this research area, Bettinger,⁵ Bao² and Irimia-Vladu⁶ have reported devices based on biocompatible and biodegradable organic materials. Rogers and his colleagues lead the development of transient electronics^1,3,7–10 using a silk substrate and a variety of materials,^11–17 such as magnesium, zinc, silicon germanium, indium-gallium-zinc oxide etc., for electrodes or conducting channels.

Wood-derived cellulose nanocrystal (CNC) based nanofibril paper is a good candidate for substrates in transient electronics because of its high-Young's modulus, low-density, low-thermal-expansion, and non-toxicity.^18,19 In addition, an acid hydrolysis method will enable mass production of CNC at low cost. Fig. 1(a) summarizes the unique properties and applications of CNC materials. Many applications that use CNC films as substrates for electronics, such as organic solar cells,^19,20 field-effect transistors,²¹ nanowire-based strain sensors,²² and organic light-emitting diodes,²³ have been demonstrated in the last two years. As an environmentally friendly nanomaterial, CNC was approved for unrestricted use in Canada and was the first nanomaterial included on Canada's Domestic Substances List (DSL). Herein, we show the fabrication and characterization of metal nanoparticle based strain sensors on CNC films. The electrical current in our devices will change when the device goes through a shape deformation, as shown in Fig. 1(b).

Fig. 1 (a) The advantages and applications of CNC materials. (b) The working principle of metal nanoparticle based strain sensors. With a strain applied on it, the output current is going to drop down sharply. (c) The fabrication process of strain sensors. From left to right: electron beam evaporation assisted by PDMS stencil to deposit silver electrodes, picture of flexible CNC film with silver electrodes, spin-coating silver nanoparticles on the electrodes shown in the middle graph.

In transient electronics, one major challenge in device fabrication concerns the substrate. Because of the high sensitivity of CNC to water and most organic solvents, a traditional photolithography process cannot be applied to pattern electrodes on CNC films. Rogers and his colleagues used the transfer-printing technique to avoid direct patterning and electrode deposition on water-soluble substrates. Nogi et al. reported direct electrode deposition on wood derived green materials by printing with a metal paste or ink.^20,24,25 Another common solution to this problem is using stencils and shadow masks. Previously reported metal²¹ and polyimide^1,26 shadow masks cannot lead to high-resolution features because there are gaps between the mask and the substrate. All physical vapor deposition methods used in metal electrode deposition are not fully directional, thus metal atoms may diffuse to substrate surfaces under the mask and result in reduced device performance. A polydimethylsiloxane (PDMS) shadow mask can solve this problem for the large London dispersion force (LDF) between the PDMS mask and flexible substrates,²⁷ providing good adhesion between the mask and the substrate. For example, a typical feature size by a metal and polyimide shadow mask is above 160 μm,²¹ whereas for the PDMS shadow mask, the feature size can be as small as 5 μm.²⁸ Up until now, no one has achieved this in device fabrication on water-soluble substrates.

In this paper, we reported the first demonstration of applying PDMS shadow masks to deposit metal electrodes on water-soluble substrates. We also improved the fabrication of PDMS masks in the previous report.²⁸ We used the doctor blade method with self-drying of PDMS to replace the method using hexane to dissolve the top PDMS layer. This enables the controllable and reproducible fabrication of PDMS masks. Our strain sensors employed a thin-film of silver (Ag) nanoparticles as the sensing material, shown in Fig. 1(c), and showed a high gauge-factor of ∼50, which is the highest gauge-factor of strain sensors reported on water-soluble substrates. Our device can disintegrate in water within 30 minutes.

Experimental

Materials

Cellulose nanocrystals (CNCs) were mass produced in the CNC pilot plant at Alberta Innovates Technology Futures (AITF, Edmonton, Alberta, Canada). Trichloro(1H, 1H, 2H, 2H-perfluorooctyl)silane (97%) was purchased from Sigma Aldrich. Glycerine was purchased from Fisher Scientific. Silver nanoparticles (∼20 nm in diameter) were purchased from US Research Nanomaterials Inc. The PDMS stencils were made using a Dow Corning Sylgard 184 silicone elastomer kit.

CNC synthesis

CNC was synthesized by acid hydrolysis. First, two Pfaudler 50 gallon (189 liter) acid-resistant glass-lined reactors with a steam-heated jacket were employed to perform the sulfuric acid hydrolysis with an acid concentration of 64% and an initial reaction temperature of 45 °C. Then, a centrifuge step was performed using a GEA Westfalia SC-35 Separator. Later, the microfiltration step was performed in a GEA Filtration-Ultrafiltration Plant. In the end, the wet CNC powder was dried in a SPX-Anhydro Model 400 Spray Dryer Plant at the conditions of a 220 °C inlet temperature and an 85 °C outlet temperature.

CNC film deposition

CNC films were prepared by the following steps. First, the CNC powder (5 wt%, 2 g) and deionized water (95 wt%, 37 g) were mixed together to form a CNC solution. Then 0.8 mL glycerine (Fisher Scientific Inc.), which is a non-toxic add-on, was added to the solution to avoid cracks after drying. Finally, this mixture was poured into a plastic Petri dish and dried at room temperature for 18 hours. The dried CNC film was detached from the Petri dish for use.

PDMS stencil fabrication

As shown in Fig. 2, the fabrication process of the PDMS shadow mask can be divided into 6 steps.

Fig. 2 Process flow of the fabrication of the PDMS shadow mask.

The first two steps are commonly used in conventional photolithography. In these 2 steps, a HPR 504 photoresist was spun onto the surface of a silicon (Si) wafer, and the wafer was spun for 10 s at a spin speed of 500 rpm followed by an additional spin of 40 s using a spin speed of 4000 rpm. Before UV exposure, the wafer was put onto a hot plate baking at 115 °C for 90 s to evaporate the solvents in the photoresist so as to obtain an HPR 504 photoresist film with a thickness approximately 1 μm. The UV exposure was carried out using a mask aligner (ABM Inc. San Jose, California) for 4 s with a UV light intensity of 66.7 mW cm⁻². Finally, the wafer was merged in a Developer 354 for 40 s to get the pattern as shown in Fig. 2.

In step 3, the chromium (Cr) etching mask was fabricated by depositing a 200 nm thick chromium film onto the wafer via magnetic sputtering. After the lift-off process, the Cr etching mask showed up on the surface of the silicon wafer.

In step 4, a Bosch process (SF₆ as the etching gas and C₄F₈ as the passivation gas) was applied to get the silicon mould with an etching depth of 200 μm.

In step 5, the trichloro(1H, 1H, 2H, 2H-perfluorooctyl)silane vapor was first employed to passivate the surface of the Si mould to make it hydrophobic. Then, the PDMS base was mixed with the cross-linker curing agent at a ratio of 10 : 1 (v%), and degassed in a desiccator for 30 minutes to prepare the PDMS mixture. Later, the PDMS mixture was poured onto the surface of the Si mould. Excess PDMS was removed from the top surface of the mould using a doctor blade. The PDMS mixture was then cured at 80 °C for 30 minutes on a hotplate.

In the last step, the PDMS was peeled off from the mould to form the stencil.

Strain sensor fabrication

As shown in Fig. 1(b), our strain sensor is a two-terminal device. When strain is applied on the substrate, the current across the two terminals changes under the applied voltage. The fabrication process of our devices is briefly illustrated in Fig. 1(c).

At first, a PDMS stencil was attached onto the CNC film. The large London dispersion force between the PDMS stencil and CNC substrate enabled good adhesion between them.

In the second step, a silver (Ag) film was deposited on the aforementioned sample using e-beam evaporation. Then, the PDMS stencil was peeled off from the CNC substrate, and the obtained sample was the CNC substrate with patterned Ag electrodes.

In the last step, an Ag nanoparticle suspension was spin-coated onto the sample (CNC substrate with patterned Ag electrodes), and the sample would be ready for testing. The concentration of the Ag nanoparticle suspension is 0.5 mmol mL⁻¹ in N,N-dimethylformamide (DMF).

Strain testing setup

The strain sensors were characterized using various steel blocks with different radii commonly used in strain-gauge testing. The strain sensor tightly attached to the curved surface of the steel block was stretched to certain strains. The electrical measurement was performed by a SourceMeter (Keithley 2400).

Results and discussion

Smooth, uniform CNC films are desired in high-performance transient electronics. We produced high-quality CNC films by optimizing the film-processing conditions. In this process, we used glycerin as a plasticizer and a dried CNC/plasticizer suspension to get a flexible CNC film. In the profilometer measurement, the thickness of the CNC films was determined to be approximately 200 μm. Three randomly selected locations of the CNC film surface were imaged using atomic force microscopy (AFM), and the measured roughness values were 3.51 nm, 4.75 nm and 2.77 nm, respectively. As shown in Fig. 3(a), the AFM image provides evidence that the smooth CNC film with a variation of less than 10 nm was produced. This is further confirmed by the scanning electron microscopy (SEM) image in Fig. 3(b). We can see there are no prominent specks or dents on the surface of the CNC film.

Fig. 3 (a) AFM picture of the CNC film. (b) SEM picture of the CNC film.

The PDMS masks processed using our Si moulds and doctor blade method show high quality under SEM imaging. In Fig. 4(a), the shadow mask produced by our process shows a channel length of 150 μm and width of 3 mm between adjacent electrodes. Sharp sidewalls of the shadow mask are clearly pictured in Fig. 4(b), and this can lead to good patterns on the water-soluble substrate when metal electrode is deposited using physical vapor deposition. For the large LDF between the PDMS mask and CNC substrate, we can easily attach our shadow mask onto the CNC substrate without any glue or tape in the experiment.

Fig. 4 SEM images of the PDMS shadow mask (a) top-view (b) 60° side-view.

The operational principle of the strain sensor is based on the fact that the resistance of the conducting channel changes when the device is under shape deformation. A current–voltage characteristics measurement (inset of Fig. 5) of the device can be used to calculate the resistance value of the strain sensor at a given strain. The sensitivity of strain sensors is characterized by the gauge-factor (GF), which is extracted by taking the ratio between the electrical measurable response ΔR/R and the strain ε applied to the sensor, as expressed in eqn (1):

Fig. 5 ΔR/R–strain curves of the devices, through which GF of strain sensors can be calculated. The red line represents the device made by PDMS stencil lithography which has a conducting channel of around 150 μm. The blue line refers to the device with copper tapes as electrodes and the conducting channel is about 3 mm. Inset: I–V curves of the same device under different strains. The blue, red, and black curves show I–V curves under the strain of around 0.5%, 0.7%, and 0.8%, respectively.

Since the relation between ΔR/R and the strain ε is nearly linear, as shown in Fig. 5, we used the slope of the linear fitting lines of the scatter points on the graph to represent GF of the devices. From the plot in Fig. 5, the GF for our device is 52.44 with a standard error of 3.28, which is among the highest values in previous reports.^29–31 We also tested the device without using our PDMS stencil lithography, where copper tapes act as electrodes to make a device with a conducting channel of approximately 3 mm. As shown in Fig. 5, a GF of 19.67 with a standard error of 0.96 was extracted from the curve representing the device without PDMS stencils. According to our testing, the maximum strain applied to the device without permanent damage is approximately 1.4%. It is noted that the CNC substrate can absorb water molecules because of its high solubility in water. However, the electrical property of the device is barely influenced.

We attributed the high GF to the quantum transport between silver nanoparticles and carried out an analysis using the Arrhenius model.³² According to the former reports, electron transport in the metal nanoparticle matrix granular metals system could be described as thermally activated tunneling.^33–37 The conduction is determined by several parameters such as the particle radius, interparticle distance and so on. In the Arrhenius model, the curve should be nearly linear in the small strain section.

As shown in Fig. 5, our device works at small strains showing a linear relation for ΔR/R and ε, which means that it can fit to the Arrhenius model. The decreased performance in devices made without PDMS stencils is due to more micro-cracks in long conducting channels. For a small channel length, it is expected that nanoparticles will separate in a relatively homogenous way with more or less the same distance when the strain is applied on the device. Whereas, in long channel devices patterned without PDMS stencils, more micro-cracks appear in the conducting channel and divide nanoparticles into domains. While in each domain, nanoparticles still separate with similar distances under a certain strain, where most of the applied strain is shared by the opening of these micro-cracks. Thus, the actual strain within the nanoparticle domain may be much smaller than the applied strain.

To verify the reproducibility and reliability of our technique, we produced more than 25 stencils from the same Si mould. The recorded SEM images of PDMS stencils made after 15, 20 and 25 times using the same Si mold are provided in ESI (Fig. S1†), and strain sensors made by these PDMS stencils were also tested. According to the experimental data, the PDMS stencils and strain sensors can be reproduced within a reasonable variance. We also fabricated several devices from the same PDMS stencil, and a similar performance was achieved in these devices (ESI Fig. S2†). After repeatable bending of 40 times, our sensors still show a high GF of about 42 (ESI Fig. S3†).

In order to know how this device deforms and partially dissolves in water, we put the substrate with silver electrodes into a beaker with deionized (DI) water. As shown in Fig. 6, after 3 minutes, the silver electrodes began to peel off. At 10 minutes, there is no silver on the substrate which has a strong deformation. In 30 minutes, the entire substrate dissolved in water. After the substrate was fully dissolved, we collected the supernatant of the solution only for recycling CNC materials. Finally, a new CNC film was produced as shown in the last graph in Fig. 6 when the solution was fully dried.

Fig. 6 The dissolving process of CNC film with silver electrodes in DI water and recycled CNC film.

In reality, after the whole device dissolves in water, silver electrodes can be filtered out by a normal filter, and silver nanoparticles can be isolated by a hyperfiltration membrane or ultracentrifuge. The metals can be filtered out and recycled by some chemical process. Our strain sensors only contain silver and CNC materials. Silver is a biocompatible material, and has been used in surgical implants. As mentioned before, CNC is an environmentally friendly material. Therefore, our strain sensors have a minimal environmental impact from the perspective of materials. Inhaled nanoparticles are still huge concerns for human safety, and the medical use of our devices should be carefully investigated before real applications.

Conclusions

In conclusion, we reported on the fabrication and characterization of a strain sensor employing silver nanoparticles and water-soluble CNC substrates, which is the first demonstration of using polydimethylsiloxane (PDMS) stencils to pattern metal electrodes on water-soluble substrates. The stencil was fabricated using a silicon mould and PDMS stencils, and has the potential to be widely used in the field of transient electronics and produce high-resolution features with further optimization. Our strain sensors show a high gauge-factor of 52.44, which is higher than any other strain sensors on water-soluble substrates and comparable with the best metal nanoparticle based strain sensors on other types of substrates.

Acknowledgements

We thank the staff in nanoFAB at the University of Alberta, for useful discussions regarding photolithography and material deposition methods. We thank Shuhuan Hu in the City University of Hong Kong for discussion about the photolithography process. This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), Alberta Innovates-Bio Solutions (AI-Bio) and the University of Alberta Start-up Fund.

References

1. S.-W. Hwang, H. Tao, D.-H. Kim, H. Cheng, J.-K. Song, E. Rill, M. A. Brenckle, B. Panilaitis, S. M. Won, Y.-S. Kim, Y. M. Song, K. J. Yu, A. Ameen, R. Li, Y. Su, M. Yang, D. L. Kaplan, M. R. Zakin, M. J. Slepian, Y. Huang, F. G. Omenetto and J. A. Rogers, Science, 2012, 337, 1640–1644.
2. C. J. Bettinger and Z. Bao, Adv. Mater., 2010, 22, 651–655.
3. X. Huang, Y. Liu, S.-W. Hwang, S.-K. Kang, D. Patnaik, J. F. Cortes and J. A. Rogers, Adv. Mater., 2014, 26, 7371–7377.
4. S. Greengard, Commun. ACM, 2013, 56, 20.
5. A. Mahdavi, L. Ferreira, C. Sundback, J. W. Nichol, E. P. Chan, D. J. D. Carter, C. J. Bettinger, S. Patanavanich, L. Chignozha, E. Ben-Joseph, A. Galakatos, H. Pryor, I. Pomerantseva, P. T. Masiakos, W. Faquin, A. Zumbuehl, S. Hong, J. Borenstein, J. Vacanti, R. Langer and J. M. Karp, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 2307–2312.
6. M. Irimia-Vladu, P. A. Troshin, M. Reisinger, L. Shmygleva, Y. Kanbur, G. Schwabegger, M. Bodea, R. Schwödiauer, A. Mumyatov, J. W. Fergus, V. F. Razumov, H. Sitter, N. S. Sariciftci and S. Bauer, Adv. Funct. Mater., 2010, 20, 4069–4076.
7. C. Dagdeviren, S. W. Hwang, Y. Su, S. Kim, H. Cheng, O. Gur, R. Haney, F. G. Omenetto, Y. Huang and J. A. Rogers, Small, 2013, 9, 3398–3404.
8. C. H. Lee, J.-W. Jeong, Y. Liu, Y. Zhang, Y. Shi, S.-K. Kang, J. Kim, J. S. Kim, N. Y. Lee, B. H. Kim, K.-I. Jang, L. Yin, M. K. Kim, A. Banks, U. Paik, Y. Huang and J. A. Rogers, Adv. Funct. Mater., 2014, 25, 1338–1343.
9. S. W. Hwang, X. Huang, J. H. Seo, J. K. Song, S. Kim, S. Hage-Ali, H. J. Chung, H. Tao, F. G. Omenetto, Z. Ma and J. A. Rogers, Adv. Mater., 2013, 25, 3526–3531.
10. S. H. Jin, S.-K. Kang, I.-T. Cho, S. Y. Han, H. U. Chung, D. J. Lee, J. Shin, G. W. Baek, T. Kim, J.-H. Lee and J. A. Rogers, ACS Appl. Mater. Interfaces, 2015, 7, 8268–8274.
11. S.-K. Kang, G. Park, K. Kim, S.-W. Hwang, H. Cheng, J. Shin, S. Chung, M. Kim, L. Yin, J. C. Lee, K.-M. Lee and J. A. Rogers, ACS Appl. Mater. Interfaces, 2015, 7, 9297–9305.
12. S. K. Kang, S. W. Hwang, H. Cheng, S. Yu, B. H. Kim, J. H. Kim, Y. Huang and J. A. Rogers, Adv. Funct. Mater., 2014, 24, 4427–4434.
13. L. Yin, H. Cheng, S. Mao, R. Haasch, Y. Liu, X. Xie, S. W. Hwang, H. Jain, S. K. Kang, Y. Su, R. Li, Y. Huang and J. A. Rogers, Adv. Funct. Mater., 2014, 24, 645–658.
14. S.-K. Kang, S.-W. Hwang, S. Yu, J.-H. Seo, E. A. Corbin, J. Shin, D. S. Wie, R. Bashir, Z. Ma and J. A. Rogers, Adv. Funct. Mater., 2015, 25, 1789–1797.
15. H. L. Hernandez, S.-K. Kang, O. P. Lee, S.-W. Hwang, J. A Kaitz, B. Inci, C. W. Park, S. Chung, N. R. Sottos, J. S. Moore, J. A Rogers and S. R. White, Adv. Mater., 2014, 26, 7637–7642.
16. S. Hwang, C. H. Lee, H. Cheng, J. Jeong, S.-K. Kang, J. Kim, J. Shin, J. Yang, Z. Liu, G. A. Ameer, Y. Huang and J. A. Rogers, Nano Lett., 2015, 15, 2801–2808.
17. S.-W. Hwang, S.-K. Kang, X. Huang, M. A. Brenckle, F. G. Omenetto and J. A. Rogers, Adv. Mater., 2015, 27, 47–52.
18. R. J. Moon, A. Martini, J. Nairn, J. Simonsen and J. Youngblood, Chem. Soc. Rev., 2011, 40, 3941–3994.
19. Y. Zhou, C. Fuentes-Hernandez, T. M. Khan, J.-C. Liu, J. Hsu, J. W. Shim, A. Dindar, J. P. Youngblood, R. J. Moon and B. Kippelen, Sci. Rep., 2013, 3, 1536.
20. M. Nogi, M. Karakawa, N. Komoda, H. Yagyu and T. T. Nge, Sci. Rep., 2015, 5, 1–7.
21. C.-Y. Wang, C. Fuentes-Hernandez, J.-C. Liu, A. Dindar, S. Choi, J. P. Youngblood, R. J. Moon and B. Kippelen, ACS Appl. Mater. Interfaces, 2015, 7, 4804–4808.
22. S. Mun, L. Zhai, S.-K. Min, Y. Yun and J. Kim, J. Intell. Mater. Syst. Struct., 2016, 27, 1011–1018.
23. E. Najafabadi, Y. H. Zhou, K. A. Knauer, C. Fuentes-Hernandez and B. Kippelen, Appl. Phys. Lett., 2014, 105, 063305.
24. H. Yagyu, T. Saito, A. Isogai, H. Koga and M. Nogi, ACS Appl. Mater. Interfaces, 2015, 7, 22012–22017.
25. M. Hsieh, C. Kim, M. Nogi and K. Suganuma, Nanoscale, 2013, 5, 9289–9295.
26. S. Selvarasah, S. H. Chao, C. L. Chen, S. Sridhar, A. Busnaina, A. Khademhosseini and M. R. Dokmeci, Sens. Actuators, A, 2008, 145–146, 306–315.
27. X. Huang, Y. Liu, H. Cheng, W. J. Shin, J. A. Fan, Z. Liu, C. J. Lu, G. W. Kong, K. Chen, D. Patnaik, S. H. Lee, S. Hage-Ali, Y. Huang and J. A. Rogers, Adv. Funct. Mater., 2014, 24, 3846–3854.
28. H. Wang, A. Chakraborty and C. Luo, J. Micromech. Microeng., 2010, 20, 127001.
29. M. Zheng, W. Li, M. Xu, N. Xu, P. Chen, M. Han and B. Xie, Nanoscale, 2013, 6, 3930–3933.
30. T. Yamada, Y. Hayamizu, Y. Yamamoto, Y. Yomogida, A. Izadi-Najafabadi, D. N. Futaba and K. Hata, Nat. Nanotechnol., 2011, 6, 296–301.
31. O. Kanoun, C. Müller, A. Benchirouf, A. Sanli, T. Dinh, A. Al-Hamry, L. Bu, C. Gerlach and A. Bouhamed, Sensors, 2014, 14, 10042–10071.
32. M. Huth, J. Appl. Phys., 2010, 107, 5–12.
33. I. S. Beloborodov, A. V. Lopatin, V. M. Vinokur and K. B. Efetov, Rev. Mod. Phys., 2007, 79, 469–518.
34. J. Herrmann, K. H. Müller, T. Reda, G. R. Baxter, B. Raguse, G. J. J. B. De Groot, R. Chai, M. Roberts and L. Wieczorek, Appl. Phys. Lett., 2007, 91, 89–92.
35. B. Abeles, P. Sheng, M. D. Coutts and Y. Arie, Adv. Phys., 1975, 24, 407–461.
36. J. L. Tanner, D. Mousadakos, K. Giannakopoulos, E. Skotadis and D. Tsoukalas, Nanotechnology, 2012, 23, 285501.
37. J. Yin, P. Hu, J. Luo, L. Wang, M. F. Cohen and C. J. Zhong, ACS Nano, 2011, 5, 6516–6526.

---

Footnote

† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c6ra10069d

---