Research Article Cite This: ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
www.acsami.org
Strategy of Constructing Light-Weight and Highly Compressible Graphene-Based Aerogels with an Ordered Unique Configuration for Wearable Piezoresistive Sensors Xiaowei He,† Qiongzhen Liu,*,†,‡ Weibing Zhong,§ Jiahui Chen,† Dengming Sun,† Haiqing Jiang,† Ke Liu,† Wenwen Wang,† Yuedan Wang,† Zhentan Lu,† Mufang Li,† Xue Liu,† Xiaojun Wang,† Gang Sun,‡ and Dong Wang*,†,§ Downloaded via GUILFORD COLG on July 21, 2019 at 04:54:46 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
Hubei Key Laboratory of Advanced Textile Materials & Application, Wuhan Textile University, Wuhan 430200, China Division of Textiles and Clothing, University of California, Davis, California 95616, United States § College of Chemistry, Chemical Engineering and Biotechnology, Donghua University, Shanghai 201620, China ‡
S Supporting Information *
ABSTRACT: Three-dimensional (3D) graphene aerogels (GAs) have attracted huge attention from researchers due to their great potential in vast applications. The hydrothermal reaction combined with freeze-drying using graphene oxide (GO) as a precursor has proven to be an effective method for obtaining relatively well-structured pure GAs. However, insufficient mechanical strength and low compressibility of the materials still limit their practical applications. Here, we report the microstructure-induced strong mechanical anisotropy of these monolithic GAs in transverse direction (TD) and longitudinal direction (LD), which has never been considered to be related to structural vulnerability. To overcome this anisotropy and enhance the structure, we hereby introduce our self-made poly(vinyl alcohol)-co-polyethylene (PVA-co-PE) nanofibers and low-molecular weight PVA as structural enhancers into the original 3D network to form a novel nanofiber−graphene composite aerogel. Intriguingly, a unique configuration is formed in the GA, in which the highly aligned stacked reduced GO sheets serve as the framework (cellular walls) and the nanofibers act as cross-linking columns anchored between the walls to support the structure along the TD, whereas the micro/nanosized PVA lamellae serve as binders. The resulting aerogel (referred to as graphene−PVA-co-PE nanofibers−PVA aerogel (GNPA)) has excellent compressive resilience along the TD and exhibits an ultrahigh gauge factor (14387%) at a very subtle strain (0.23%) in piezoresistive properties. The GNPA-TD has also been assembled into a variety of wearable sensors and demonstrates great potential for wireless human pressure sensing. In short, this study offers an extremely simple and effective method for developing graphene aerogels with a strong mechanical structure and paves the way for the application of 3D graphene in wearable sensors. KEYWORDS: graphene aerogel, anisotropy, poly(vinyl alcohol)-co-polyethylene nanofibers, mechanical properties, wearable sensors
1. INTRODUCTION Three-dimensional (3D) graphene aerogels (GAs) have attracted great attention due to the realization of a macroscopic assembly of the two-dimensional (2D) graphene nanosheets and their potential in a wide variety of applications, such as energy storage/conversion, adsorption, insulation, catalysis, damping materials, sensing, and so on.1−4 The ideal GAs are expected to have a 3D porous network and can possess excellent electrical, thermal, and mechanical properties of the 2D graphene sheets.5,6 Presently, pure GAs are mainly prepared by chemical reduction, hydrothermal reduction,7 or ice-templating8 methods using graphene oxide (GO) as a precursor followed with various drying process and posttreatments. In these processes, many of the oxygen-containing functionalities in the GO have been partially reduced and the © 2019 American Chemical Society
hydrogen bonds between the reduced graphene oxide (rGO) sheets are severely weakened when further subjected to ice sublimation upon drying.9 Therefore, the 3D porous network is mainly formed by the overlapping of the rGO sheets. This makes the structure collapse easily when undergoing deformation, due to the strong π−π stacking and van der Waals’ force between the graphene sheets.10 Obviously, the weak mechanical strength of the pure GAs greatly limits their practical applications and thus enhancing their structure strength is an urgent issue that needs to be solved. Received: February 21, 2019 Accepted: May 6, 2019 Published: May 6, 2019 19350
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 1. Schematic illustration of the fabrication process of the GNPA and all aerogels used for comparative studies.
compressive graphene−carbon nanotube aerogel by embedding CNTs into the network during a hydrothermal process. Recently, Xu et al.13 achieved a significantly enhanced structure of GA that could be dried at room temperature by employing borate as cross-linkers during hydrothermal reduction. Ma et al.3 utilized MXene (Ti3C2T2) nanosheets to support the network of graphene aerogel and formed a superelastic structure. Zhou et al.14 enhanced the interaction among rGO layers via addition of cellulose nanocrystalline and low-molecular-weight carbon precursors, thus realizing a flexible and superstable structure for rGO-based carbon aerogels. Moreover, various metal ions (Fe3+, Zn2+, Nd3+, etc.) are also employed as enhancers for the graphene-based aerogels, as reviewed by Gong.15 These studies suggest that structural enhancers can act as ideal spacers to effectively prevent restacking of graphene sheets, while simultaneously interacting with GO or rGO sheets via coordination bonds, hydrogen bonds, van der Waals forces, dipole interactions, electrostatic interactions, or chemical bonds. Of course, these works are effective and enlightening and contribute greatly to improve structural strength of the GAs. However, these efforts have not fundamentally elucidated how to overcome structural vulnerability and provide both improved mechanical and electrical properties for GAs. In this work, we achieved a deep insight into the strong structural anisotropy of the monolithic pure GAs via a freeze-drying formation method. As we examined, the mechanical strength of these porous monoliths in the transverse direction (TD) is much weaker than that in the longitudinal direction (LD). When observing the microstructure, larger rGO sheets always tend to align in the LD to form cellular walls whereas smaller sized rGO sheets are spliced between the walls along the TD. Obviously, the aerogels along TD lack sufficient cross-linking reinforcement or mechanical support. Moreover, this strong mechanical anisotropy of pure GAs often exacerbates their structural fragility, especially when subjected to pressureinduced shear stress. Therefore, we need to develop a strategy
Compared with other drying methods (i.e., supercritical CO2 fluid drying), freeze-drying is considered to be the optimal method of forming high-quality aerogels with different designed shapes. Naturally, this technique is commonly used to prepare the GAs. It is worth noting that the ice-templating effect caused by the growth of ice crystals is unavoidable during freeze-drying. Thus, this will facilitate the formation of monolith-like aerogels having highly ordered cellular pores aligned along the ice growth direction.8 In addition, hydrothermal chemistry is a simple, efficient, and green method. In recent years, it has been found that in the hydrothermal process, the GO sheets can be partially reduced to rGO sheets whereas the rGO sheets can be simultaneously self-assembled into a 3D hydrogel by sheets overlapping.7 Subsequently, the graphene (rGO) aerogels can be formed by slow replacement of the liquid phase in the rGO hydrogels with gas through the freeze-drying process. Therefore, the combination of hydrothermal reduction and freeze-drying using GO as a precursor is considered as an effective method for preparing the GAs. It is also worth mentioning that ethylenediamine (EDA) or other amines are commonly used as reducing agents and crosslinkers to enhance the network of GA.10 As a result, the GAs tend to have a highly ordered 3D network while maintaining good shape integrity externally by this method. However, the insufficient mechanical strength, lack of enough elasticity, and sacrifice of electrical conductivity are the main drawbacks. In view of the above problems, considerable studies have been carried out by introducing various cross-linking agents or structure strengthening factors in the preparation of the GAs, especially during hydrothermal reactions. Zhong et al.11 developed a composite carbon nanotube (CNT)/GA aerogel via freeze-drying of the suspension of unzipped and partially exfoliated multiwalled carbon nanotubes. The microstructure showed that the carbon nanotubes (CNTs) were just like veins in supporting the graphene sheets (laminas), exhibiting good electrical conductivity and stable structural integrity upon cyclic compression. Similarly, Wan et al.12 designed a highly 19351
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 2. (a−d) Top-view SEM images of GA, GNA, GPA, and GNPA, respectively, (e, f) side-view SEM images of GNPA, (g) pore size distribution of the GNPA estimated from the side view, and (h) the porosities and (i) densities of all aerogels. 2.2. Preparation of the Elastic Graphene−Nanofiber Composite Aerogel (GNPA). The schematic illustration of the preparation process of the GNPA and all other aerogels used for comparative studies is shown in Figure 1. In brief, the GNPA was readily prepared by a combination of hydrothermal reduction and freeze-drying methods. In the preparation, fine crystalline graphite powder was used as a raw material for obtaining graphene oxide (GO) according to the modified Hummers method.20,21 First, PVA-co-PE nanofiber suspension (milky white) and a small molecule PVA solution (5 mL, 2 wt %, transparent) were added to the aqueous GO solution (5 mg/mL) to form a homogeneous mixed GO solution. Ethylenediamine (EDA) was then added to the GO solution before the subsequent hydrothermal reaction process. During the hydrothermal process (120 °C for 14 h), the rGO, nanofibers, and PVA cross-linkers formed a hydrogel with a 3D network by self-assembly. After natural cooling to the room temperature, the hydrogel was dialyzed in an ethanol−water solution (w/w = 1:10) for 24 h to remove residues and prevent possible cracks during freeze-drying. The hydrogel was then freeze-dried to obtain the highly porous aerogel with ordered cellular pores. Afterward, postreduction by the gaseous phase of hydrazine hydrate (N2H4) was employed to obtain the final GNPA aerogel. As schemed, the GNPA eventually forms a unique configuration; that is, nanofiber bundles are connected between the rGO cellular walls, whereas the PVA cross-linkers consolidate the anchoring between the rGO sheets and the nanofiber bundles. This configuration further proves beneficial to the mechanical properties and piezoresistive performance of the GNPA. Also note that in Figure 1, the pure graphene-based aerogels (GAs), graphene−PVA composite aerogels (GPAs), and graphene−nanofiber composite aerogels (GNAs) were prepared by the same method as control samples. 2.3. Assembly and Test of the GNPA-Based Pressure Sensors. The GNPAs were first carefully cut into smaller pieces with different designed dimensions. Then, an aluminum foil strip was made and one end of a copper wire was firmly adhered to the adhesive side of an insulating tape to form one piece. The other piece was prepared in the same way. Thereafter, the two pieces were sandwiched with the smaller GNPA aerogel while leaving two ends of copper wires for measurement. As such, the change in electrical resistance under varying deformation can be measured by Autolab system (PGSTAT302, Switzerland). The external pressure was applied by a motorized test stand for tension and compression testing (Mark10,
to improve the mechanical strength of the GAs in both TD and LD directions while maintaining excellent electrical conductivity. On the basis of the above understanding, poly(vinyl alcohol)-co-polyethylene (PVA-co-PE) polymeric nanofibers and low-molecular PVA cross-linkers were introduced into the network as structural enhancers to form novel graphene− nanofiber composite aerogels. As expected, a unique cellular porous network was formed during the hydrothermal process and the mechanical strength of the novel GA in the transverse direction (TD) was significantly enhanced. More interestingly, the highly aligned rGO sheets serve the cellular walls, nanofibers act like columns, and PVA lamellae act as binders. Especially along the TD, the nanofibers function like elastic springs that support the cellular walls, avoiding structure collapse in a case of stress or shear. Delightingly, this configuration also imparts excellent compressive resilience and superior cyclic performance to the graphene−nanofiber composite aerogel (designated as graphene−PVA-co-PE nanofibers−PVA aerogel (GNPA)). It is worth noting that this microstructure provides a very large resistance variation due to the geometric effects and high compressibility. Taking into account the above advantages, we further evaluated the piezoresistive performance of the GNPA and finally proved the great potential of the aerogel for wearable pressure sensors, demonstrating its ultrahigh sensitivity to subtle strain or pressure.
2. MATERIALS AND METHODS 2.1. Materials. Fine graphite powder (D50 < 400 nm) was purchased from Macklin Biochemical Co., Ltd. The poly(vinyl alcohol-co-ethylene) (PVA-co-PE) nanofibers (Mn = 78 495 g/mol) were massively produced and supplied by our lab.16−19 Lowmolecular weight PVA (1799) was purchased from Aladdin. Chemicals including concentrated sulfuric acid, potassium permanganate, phosphorus pentoxide, potassium persulfate, acetone, hexane, toluene, and hydrochloric acid were obtained from Sinopharm Group Co., Ltd. (China). Deionized water was made in our lab. All chemicals were used as received without further purification. 19352
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 3. (a, b) TEM and AFM images of as-prepared GO, (c) XRD patterns, (d) Raman spectra, (e) Fourier transform infrared (FT-IR) spectra, (f) overall XPS spectra, and (g−i) the respective high-resolution XPS images of C 1s, O 1s, and N 1s spectra of all testing samples. ESM301). For the assembly of wireless wearable sensors, a commercial Bluetooth module (BIAZE D13) was fixed to a badge for signal transmission. The badge was connected to a headband or a wristband with an embedded thin GNPA aerogel and fixed by interconnecting sewed copper wires. The signals thus could be recorded by a mobile using our wireless sensors. For a better understanding, the assembly details of the headband wireless pressure sensor are illustrated in Figure S1 in the Supporting Information. 2.4. Characterizations. The as-prepared GO sheets were characterized by high-resolution transmission electron microscopy (TEM, FEI TF20) and an atomic force microscope (NT-MDT, Prima). The porous structures of all rGO aerogels were observed by a scanning electron microscope (JEOL, IT-300). A Fourier transform infrared (FT-IR) spectroscope (TENSOR-27) in the attenuated total reflection mode at a range of 4000−400 cm−1 was used for all test samples. Raman spectroscopy (LabRAM HR800, JY) at an excitation wavelength of 514 nm was also employed for chemical characterization. Besides, an X-ray diffractometer (XRD, Bruker D8 ADVANCE) and X-ray photoelectron spectroscopy (XPS) (ESCALAB 250Xi, Thermo Fisher Scientific) were also employed. The apparent density of the aerogels was calculated as mass divided by the geometric volume (Va). The porosity of all aerogels is defined as
displacement of 30 mm/min. Young’s modulus was evaluated from stress−strain curves in the elastic portion (ε < 10%).
3. RESULTS AND DISCUSSION 3.1. Microstructure. Figure 2a−d shows the crosssectional scanning electron microscopy (SEM) images of the cylindrical GA, GNA, GPA, and GNPA samples in the top view. The insets are their magnified microstructures near pores. As seen in Figure 2a, the pure GA mainly exhibits irregular cellular walls stacked by rGO sheets. The pore size ranges from 5 to 15 μm. For GNA (Figure 2b), some nanofiber bundles are clearly observed in cellular pores. It is also noted that the pore size notably decreases to 5 μm after the introduction of PVA-co-PE nanofibers. For GPA (Figure 2c), a more regular cellular structure is formed compared with GA, indicating the positive role of PVA in the construction of a 3D network. For GNPA, the SEM images in cross-sectional and longitudinal section views are presented in Figure 2d−f, respectively. Interestingly, the rGO sheets serve as skeletons (walls and roofs) whereas the PVA-co-PE nanofibers act as supporting columns (or beams) and the PVA lamellae as binders, extremely similar to the frame structure of a building. It appears that the larger sized GO sheets grow along the longitudinal direction of the cylinder to form aligned and continuous cellular walls whereas the smaller sized rGO sheets are stacked laterally to form the top and bottom layers of the cell. Perhaps most interesting of all, the PVA-co-PE nanofibers are entangled in the form of bundles and confined in the cellular pores. They are almost vertically connected to the cellular walls while parallel to the cellular bottoms. Meanwhile, the PVA lamellae consolidate the connection between the cellular walls and the nanofibrous bundles. The average size of
porosity = (Va − Vs)/Va × 100% wherein Vs is the skeleton volume or true volume, which is measured by a true density tester (AccuPyc II 1340, Micromeritics) using He gas replacement method. The compressive stress−strain measurements of all aerogels were carried out using an Instron model 5965 equipped with two testing plates and a load cell of 1000 N. The cylindrical aerogel sample for the compression test has a dimension of about ⌀10 mm × 8 mm. A preload of 0.2 N was applied in advance to make sure of a close contact with the two plates for test samples. The strain rate was controlled at 5 mm/min, and the cyclic loading was carried out at a 19353
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 4. (a) Proposed chemical structure evolution of the GNPA during preparation and (b) the possible microstructural formation mechanism from the perspective of side and top views.
the pores (or the small rectangle frames) is about 10−20 μm. It should be noted that this configuration is significantly different from those reported in graphene-based aerogels. The PVA-co-PE nanofiber bundles in this network function as springs to support the rGO sheets, thus endowing the aerogel with high elasticity. Figure 2g summarizes the “frame or pore” size distribution of the GNPA in the side view from the analysis of the Figure 2e. The “frame” size is mainly concentrated at 9−18 μm. Combining with the top-view SEM image (Figure 2d), it can be inferred that the cell pores in GNPA are cylindrical macropores with an average diameter of about 5 μm and a depth (or height) of 9−18 μm. Furthermore, Figure 2h shows that the porosity of all aerogels is between 96 and 99%. GNPA has a high porosity of 96.8%. In addition, the apparent density of GNPA (59.3 mg/cm3) is higher than that of the GA (37.1 mg/cm3) due to the addition of nanofibers and PVA lamellae, as shown in Figure 2i. On the basis of the above results, it is confirmed that the PVA-co-PE nanofiber bundles and PVA lamellae are both involved in the construction of the 3D network of the GNPA, thus forming a unique reinforced porous structure. 3.2. Chemical Structure. To better understand the origin of this unique configuration in the GNPA, the morphologies, crystal structures, as well as chemical structures of the raw GO sheets and the resulting graphene aerogels were characterized. Figure 3a is the bright-field TEM image of the self-made GO sheets used for this work. The GO sheets have a lot of wrinkles with distinct curled edges. Figure 3b is the atomic force
microscopy (AFM) image of our self-made GO sheets or flake showing a rough dimension of 2.8 μm × 1.428 μm × 1.22 nm. Figure 3c shows a comparison of the XRD patterns of the raw graphite powder, GO powder, and the crushed as-prepared GA powder, respectively. It shows that the fine graphite powder has a very strong and sharp diffraction peak (2θ = 26.5°), corresponding to the (002) crystal plane with an interplanar spacing (d value) of 0.34 nm. For the GO powder, the 2θ diffraction angle shifts to 11° with an interplanar spacing of 0.804 nm. This can be attributed to the presence of abundant oxygen-containing functional groups on the surface of the GO sheets after chemical exfoliation. Here, the as-prepared GA is defined as the product of the GO after hydrothermal reduction with the aid of EDA. Its pulverized powder shows a strong 2θ angle around 21° corresponding to an interplanar spacing of 0.39 nm. Apparently, this value is much lower than that of the GO (0.804 nm) but still slightly higher than that of the graphite (0.34 nm). This indicates there still exist persistent residual oxygen-containing functionalities in the rGO sheets even after hydrothermal process. Figure 3d presents the Raman spectra of graphite, GO, pulverized GA, and GNPA powders, respectively. All samples exhibit typical characteristic peaks of graphite showing G band and D band.22−24 As well known, the G band near 1580 cm−1 corresponds to the in-plane vibrations of carbon atoms, whereas the D band around 1350 cm−1 is associated with the disordered regions of the carbon rings.6,23 The intensity ratio of the D band to G band is so called ID/IG, which has been used to determine the defect in graphene or 19354
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces graphite materials.24 Compared with pure graphite (0.13), the ID/IG ratio is much higher in GO (1.21), showing severe structural defects in GO. However, the ID/IG ratio of the asprepared GA is still as high as 1.22, which indicates that postreduction for deoxygenation is very necessary. Therefore, we choose the hydrazine hydrate in the gas phase for further reduction of the as-prepared GNPA. As expected, the ID/IG ratio for GNPA decreases to 1.06, indicating a partially restored conjugated carbon structure, but is still far below the structural perfection of graphite. Figure 3e shows the FT-IR transmittance spectra of the GO, as-prepared GA, GA, GNA, GPA, and GNPA powders, respectively. The GO powders show characteristic absorption peaks, including the −O−H stretching vibration (3216 cm−1), the CO stretching vibration (1720 cm−1), CC stretching vibration (1622 cm−1), C−H in-plane bending vibration (1387 cm−1), as well as the C−O stretching vibration (1251, 1062 cm−1). This demonstrates that the hydroxyls, epoxies, and carbonyls are the main defects or functionalities of the GO sheets. In contrast, the vibration peaks of CO, C−OH, and C−O−C are noticeably weakened for as-prepared GA and further reduced for GA. Additionally, a new peak around 1557 cm−1 from N− H vibration appears, indicating the introduction of amino groups. Moreover, GNA, GPA, and GNPA have a common feature in the FT-IR spectra, that is, an absorption peak around 1670 cm−1 derived from the −CO−NH− amide stretching vibration. This indicates that in addition to some residual oxygen-containing functional groups, additional amide and amino groups are introduced into GNA, GPA, and GNPA during hydrothermal and postreduction process. Furthermore, X-ray photoelectron spectroscopy (XPS) analysis of the samples is shown in Figure 3f. The peak intensity ratio of C/O in GO powder is 2.74, whereas it increases to about 9.59 after hydrothermal reduction, indicating that considerable oxygen-containing functionalities have been removed from rGO sheets. In addition, an evident N 1s peak appears in GA, implying the introduction of amides and amines in GA after the reduction process. Also note that the C/O ratio of GNPA decreases to 2.99, which can be ascribed to the introduction of oxygen-containing hydroxyls from PVA-co-PE and PVA. From the C 1s spectra of the GO (shown in Figure 3g), four characteristic peaks can be distinguished, that is, C−C or C C bond at 284.6 eV, C−O−C bond at 285.4 eV, CO bond at 287.4 eV, and O−CO bond at 288.5 eV.25 It is obvious that the peak intensity of the CO and C−O bonds is dramatically reduced in GA, GNA, GPA, and GNPA compared to that of GO. Combined with the O 1s and N 1s spectra (Figure 3h,i), it can be confirmed that reduction of CO groups and formation of amino (C−NH2) (∼286 eV) and amide groups (OC−N) groups are the main reactions during hydrothermal reduction and postreduction process. 3.3. Chemical Structure Evolution and Microstructure Formation Mechanism. On the basis of above characterizations, the evolution of chemical structure and the microstructure formation mechanism of the GNPA during preparation have been elucidated, as shown in Figure 4a,b. It is widely accepted that GO has massive oxygen-containing functional groups, although its precise chemical structure has not yet been determined. According to the most commonly used Lerf−Klinowski model,26,27 hydroxyls and epoxies are the main functionalities. Besides, it is also believed that carbonyls likely exist along the edges of the holes whereas the carboxyl groups may exist in the edge region of the GO sheets.10 Both
PVA-co-PE nanofibers and PVA lamellae contain plenty of hydroxyls, which can form a uniform dispersion with GO in aqueous solution. When EDA is added into the system and hydrothermal reaction is carried out at 120 °C, both EDA and the supercritical water have strong reducibility due to the hightemperature and high-pressure environment. In the hydrothermal reaction, the chemically reactive epoxy groups evolve to hydroxyl groups via ring-opening reactions, involving the nucleophilic attack at the α-carbon by the amine from the EDA.10 Simultaneously, amine groups can be reacted with carboxylic acid groups of the GO sheets through amidation reaction. To the best of our knowledge, there remains an open question for the GO chemistry up to now; that is, which are the exact reaction pathways of deoxygenation from the epoxide, hydroxyl, amide, or carbonyl groups? According to the molecular dynamic simulation of GO sheets during thermal annealing by Bagri,28 the carbonyls at defect sites can be converted into a phenol, then into a hydroxyl, finally leaving the basal plane in the form of a water molecule. Therefore, dehydration may occur in GO, which is driven by the restoration of the sp2-conjugated structure of graphene. At the same time, the rGO sheets can self-assemble into a hydrogel with a 3D network due to the regional hydrophobicity and the partial overlapping of the sheets.29 The PVA-co-PE nanofibers and nano/microsized PVA lamellae are also involved in this assembly to form the GNPA hydrogel. They are connected to adjacent rGO sheets via the hydrogen bonding between the hydroxyl−hydrogel groups and hydroxyl−amino/amine groups. The chemical structure of the as-prepared GNPA aerogel continues to evolve during the postreduction process. As reported, the reduction mechanism of GO by hydrazine hydrate in the gas phase is very similar to that of the EDA, that is, to further remove the oxygen-containing functionalities and form amide groups in the rGO sheets, which has been confirmed by the above XPS analysis (the sharply decreased C/O ratio). The microstructure formation mechanism can be elucidated from the perspective of side and top views upon freezing. In the beginning, the solution of GO sheets, PVA-co-PE nanofibers, and PVA lamellae is very homogeneous due to their high dispersion in water. After the hydrothermal process, hydrogel with a unique configuration is formed due to the selfassembly of rGO, nanofibers, and PVA lamellae. Interestingly, the rGO sheets partially overlap to form the cellular walls (skeleton) whereas the nanofiber bundles act as springs supporting the walls with the PVA as binders to strengthen the configuration. During freezing, water in the pores of the hydrogel begins to freeze and form ice crystals driven by the huge temperature gradient. The growth of ice crystals typically results in expansion of the skeleton and the forced alignment of the larger sized rGO sheets along each cell wall located at the crystal boundaries.8,30,31 Meanwhile, pushed by the monolithic ice crystal, the 2D rGO sheets at the cell top go upward along the moving solidification front. Worth noticing is that this microstructural formation mechanism is quite different from that of the direct freezing of the rGO or GO solution. The biggest difference is that it is a hydrogel before freezing, which means it has a pre-existing 3D scaffold. Therefore, even freezing could not dramatically change its original structure; otherwise, it would cause structural damage. However, the alignment of the cellular walls as structural adjustment is sure to occur since the growth force of ice crystals is extremely strong. Consequently, a highly ordered cellular network is 19355
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 5. (a) Schematic diagram of sampling instructions for the compression test along the transverse direction (TD) and longitudinal direction (LD), respectively, (b, c) the strain−stress curves of all testing samples in LD and TD, (d, e) loading−unloading stress−strain curves for GNPA-LD and GNPA-TD under different strains.
formed. From the side view, the porous configuration is like a rectangular frame showing high order and anisotropy. In contrast, from the top view, it is more like a hexagonal honeycomb. Fortunately, this network maintains a high structural integrity during thawing due to solid intersheet π−π attractions and strong cross-linking among the rGO sheets, nanofibers, and PVA lamellae. 3.4. Mechanical Properties. The schematic diagram of sampling instructions for the compression test along the transverse direction (TD) and longitudinal direction (LD) is presented in Figure 5a. As seen, the cylinder sample with compression direction along the LD is referred to as sampleLD whereas the cuboid sample cut from its original cylinder with the compression direction along the original TD is referred to as the sample-TD. Therefore, mechanical properties of the two tested samples can represent the mechanical properties of the original cylinder along LD and TD, respectively. As seen from Figure 5b, the maximum compressive strength of GA-LD is 4 kPa, which is 10 times that of the GA-TD (∼0.4 kPa), indicating that pure GA has significant mechanical anisotropy. Assuming that GA is subjected to a bending, the mechanical properties of the GALD can represent its ability to withstand normal stress whereas the mechanical properties of the GA-TD can reflect its ability to withstand transverse shear stress. Obviously, the pure GA can withstand much less shear stress than normal compressive stress. Unfortunately, pressure-induced bending and shear stress in the transverse direction for nonrigid aerogel materials are often unavoidable. As a result, the shear stress tends to
cause sliding and cracking of the cellular walls composed of rGO sheets, eventually leading to the structural collapse of the GA. It is noticeable that when the strain reaches 25%, the structure of GA-LD begins to collapse and its compressive strength decreases sharply. In contrast, GA-TD collapses dramatically when the strain reaches only 15%. The results clearly show that the compressive strength of GA-TD is much lower than that of the GA-LD. Therefore, the microstructural and mechanical anisotropy of the GA prepared by hydrothermal reaction and freeze-drying are worth noting. Figure 5c shows the stress−strain curves of GNA, GPA, and GNPA in LD and TD, respectively. Compared with GA, the maximum compressive strength of all aerogels in the LD and TD is increased by more than 10 times (10−50 kPa). In addition, the mechanical anisotropy of all aerogels is significantly weakened but still exists; that is, the compressive strength in TD is slightly lower than that in LD. It is also worth mentioning that GNPA-LD and GNPA-TD have the highest compressive strength in all aerogels thanks to the unique “frame-column-binder” configuration. Figure 5d,e compares the dynamic loading−unloading stress−strain curves of GNPA-LD and GNPA-TD under different strains (20, 40, 60, and 80%). Both aerogels exhibit a viscoelastic behavior similar to rubber elastomers, which was also observed in other graphene foams or graphene aerogels.14,32 It is seen that there exists an obvious hysteresis loop when the strain is larger than 40%. Moreover, the viscoelasticity in TD is more pronounced than that in LD. However, GNPA still maintains excellent elastic resilience and structural integrity in the TD and LD 19356
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 6. (a, b) Cyclic compression performance of GNPA-TD under the loading−unloading test and (c−e) the proposed mechanical mechanism related to microstructure of the GNPA-TD.
Figure 7. (a) Current−voltage (I−V) curve of all test samples with the inset I−V curve of the GA-LD, (b) the relative change in resistance (ΔR/ R0) as a function of the compressive strain, (c) the estimated gauge factor (GF) of the GNPA relative to the compressive strain, (d) the ΔR/R0 relative to the pressure for estimation of sensitivity (S) for the GNPA (in TD and LD), (e) the instant current and (f) response time of the GNPA (in TD and LD) relative to the varied pressure.
even under 80% compression. On the basis of the above results, it is confirmed that this unique microstructure can greatly enhance the compressive mechanical properties in TD. Figure 6a,b shows the cyclic compression performance of GNPA-TD under the loading−unloading test. As seen, GNPATD exhibits a very stable cyclic compression performance under a constant strain of 60%, showing exceptional compression resilience. The proposed mechanical mechanism related to the microstructure of the pure GA, GNPA-LD, and GNPA-TD is shown in the schematic diagram in Figure 6c−e, respectively. For pure GA, the cellular walls or skeletons composed of stacking rGO sheets are more likely to bend and slide when compressed. Due to the weak joints and inevitable shear stress, the structure collapse is highly prone to occur. In GNPA-LD, the joints are greatly strengthened by PVA-co-PE nanofibers and PVA cross-linkers. As shown, the nanofiber bundles function as spacers or cushions between the walls when compressed to prevent restacking of the rGO sheets,
thereby imparting high compressibility to GNPA-LD. As for the GNPA-TD, in addition to the increased anchoring points in the network, the PVA-co-PE nanofiber bundles presumably act as elastic springs between the cellular walls, providing energy dissipation when compressed or sheared. Due to the high flexibility and toughness of the nanofibers, GNPA-TD can possess excellent elastic resilience and outstanding cyclic compressive performance. 3.5. Piezoresistive Performance. Figure 7a compares the current−voltage (I−V) curve of GNA-LD, GPA-LD, GNPALD, and GNPA-TD with the inset showing the I−V curve of GA-LD. Here, the LD and TD samples for testing were cut into rectangular thin pieces with the same dimension of 10 mm (length) × 10 mm (width) × 11 mm (height). From the I−V curves, the initial bulk resistance (R0) of GA-LD is estimated to be 10 kΩ. However, the overall resistance of GNA-LD, GPA-LD, and GNPA-LD is significantly larger than that of pristine GA-LD, in the range of 200−400 kΩ due to the 19357
DOI: 10.1021/acsami.9b02591 ACS Appl. Mater. Interfaces 2019, 11, 19350−19362
Research Article
ACS Applied Materials & Interfaces
Figure 8. Schematic representation of (a) the assembly of the GNPA-TD pressure sensor, (b) the basic sensing principles, and (c) the potential application scenarios for human body monitoring.
insulative nature of the PVA-co-PE nanofibers and PVA lamellae. However, it is worth noting that GNPA-LD possesses a relatively lower resistance when compared to that of GNALD and GPA-LD, ascribed to its relatively smaller pores and lower porosity. Interestingly, the resistance of GNA-LD is slightly higher than that of GPA-TD exhibiting anisotropy in electrical conduction. It is also noticed that these aerogels possess a modestly larger resistance (∼300 kΩ). However, this resistance range further proves beneficial for piezoresistive performance due to the large resistance variations for sensing resulting from excellent compressibility and ultrahigh porosity of the aerogels. Figure 7b shows the relative change in resistance (ΔR/R0) as a function of the compressive strain. It is worth noting that the ΔR/R0 of GA-LD shows a dramatical drop in strain to 40%, indicating severe structural damage. In contrast, GNA-LD, GPA-LD, GNPA-LD, and GNPA-TD still maintain a more integral structure when the compressive strain reaches 80%. Clearly, GNA shows relatively smaller ΔR/R0 under the same compression compared to that of GPA-LD, GNPA-LD, and GNPA-TD. Also note that ΔR/R0 of all aerogels (GNA-LD, GPA-LD, GNPA-LD, and GNPA-TD) increases significantly with a strain range of 0−40%. With the exception of GNA-LD, the maximum ΔR/R0 of the other aerogels at 40% strain is greater than 90%. In addition, when the strain is increased from 40 to 80%, ΔR/R0 is no longer sensitive to strain changes and tends to a stable value (∼99%). It is worth noting that GNPA exhibits a greater ΔR/R0 in all aerogels at the same compressive strain. Moreover, GNPA-TD shows a larger resistance response to strain in comparison to that of GNPA-LD, especially when the strain is less than 10%. Hereafter, we mainly focus on the pressure-sensing performance of GNPA-LD and GNPA-TD for a comparative study. Gauge factor (GF) and S values are important parameters for evaluating the sensitivity of a pressure or strain sensor.33−36 GF and S are defined as33,37
GF =
S=
∂(ΔR /R 0) ∂ε
∂(ΔR /R 0) ∂P
(1)
(2)
wherein R0, ΔR, ε, and P represent the initial resistance, the change of resistance, the strain, and the applied pressure, respectively. Obviously, the higher the value of GF or S, the more sensitive the sensor is to the tiny strain or subtle stress. The gauge factors of GNPA-LD and GNPA-TD as a function of strain are shown in Figure 7c. It can be seen that at low strain (
