Realization of Oriented and Nanoporous Bismuth Chalcogenide Layers via Topochemical Heteroepitaxy for Flexible Gas Sensors.

Most van der Waals two-dimensional (2D) materials without surface dangling bonds show limited surface activities except for their edge sites. Ultrathin Bi2Se3, a topological insulator that behaves metal-like under ambient conditions, has been overlooked on its surface activities. Herein, through a topochemical conversion process, ultrathin nanoporous Bi2Se3 layers were epitaxially deposited on BiOCl nanosheets with strong electronic coupling, leading to hybrid electronic states with further bandgap narrowing. Such oriented nanoporous Bi2Se3 layers possessed largely exposed active edge sites, along with improved surface roughness and film forming ability even on inkjet-printed flexible electrodes. Superior room-temperature NO2 sensing performance was achieved compared to other 2D materials under bent conditions. Our work demonstrates that creating nanoscale features in 2D materials through topochemical heteroepitaxy is promising to achieve both favorable electronic properties and surface activity toward practical applications.

1. Introduction

Bismuth-based two-dimensional (2D) layered materials, such as bismuth chalcogenides [1, 2], bismuth oxyhalides [3, 4], and bismuth halides [5, 6], are emerging eco-friendly functional materials that find wide applications in electronics, optoelectronics and energy conversion, and storage devices [7-9]. Bi2Se3 possesses excellent electrical conductivity even under a high density of defects and dislocations, since its surface states are protected from scattering [10-12]. Nanostructures of Bi2Se3 have recently shown potential in photothermal cancer therapy [13], optical switching [14], and thermoelectric devices [15]. In addition, they have also been combined with other nanostructures, such as graphene [16], ZnO [17], and CsPbBr3 [18], for enhanced photoabsorption, stimulated surface-plasmon polaritons, and enhanced exciton transfer efficiency, respectively. However, despite their outstanding electronic and optoelectronic properties, the surface activities of Bi2Se3 seem to be overlooked and less explored.

Bismuth oxyhalides (BiOX, where X = Cl, Br, and I) are particularly promising in photocatalytic energy conversion and environmental remediation [19-22]. BiOCl is perhaps one of the most investigated bismuth oxyhalides because of its low toxicity and high stability [23, 24]. However, its relatively large bandgap (typically ~3.2 eV) has limited its electronic and optoelectronic applications [22]. To date, many strategies have been developed to improve its properties, including morphology control [25], tuning of exposed crystal facets [21], and surface modification [26]. The construction of heterostructures, by considering band level alignment, has been proven effective in modifying its electronic properties, with examples including BiOCl/TiO2 [27], BiOCl/WO3 [28], BiOCl/ZnSn(OH)6 [29], and BiOCl/BiOI [30]. Epitaxial heterostructures are important with controllable overlayer orientation and coherent heterointerfaces [31]. Till now, solution-phased methods such as the coupling reaction [32] and self-assembly process [33] have been developed for preparing epitaxial heterostructures. However, the solution phase synthesis of BiOCl-based epitaxial heterostructures has been rarely explored.

Gas sensors are needed in many applications, including medical diagnosis [34], environmental monitoring [35], food quality assessment [36], and military applications [37]. Ultrathin 2D materials as well as their heterostructures have been applied to gas sensing due to large specific surface areas and tunable electronic and mechanical properties [38-41]. Many 2D materials, especially van der Waals 2D materials without surface dangling bonds, show higher activities at edges than at basal surfaces. However, maximizing edge exposure and increasing the edge/basal surface ratio remain great challenges.

In this study, we report the epitaxial growth of nanoporous Bi2Se3 with largely exposed edge sites on BiOCl nanosheets via an anion exchange-induced topochemical conversion process. Bi2Se3/BiOCl epitaxial heterostructures were used to detect NO2 gas and demonstrated good sensing performance even on printed flexible electrodes under bent conditions. This can be attributed to the strong electronic coupling across the Bi2Se3/BiOCl heterointerface, the enhanced surface activity of Bi2Se3, and the improved film forming ability with sub-2 nm surface features.

2. Results and Discussion

A solvothermal method was used to synthesize BiOCl nanosheets by following a previous report [22]. A solution of Se dissolved in oleylamine (OLA) and dodecanethiol (DDT) was then hot-injected into the BiOCl nanosheet solution and heated to induce the growth of Bi2Se3 on BiOCl.

The BiOCl nanosheets and Bi2Se3/BiOCl heterostructures were characterized with transmission electron microscopy (TEM), energy dispersive X-ray spectroscopy (EDX), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS), as shown in Figure 1 and Figure S1. The original BiOCl nanosheets were square-shaped with an average edge length of ~90 nm (Figure 1(a)) and thickness of 5-8 nm (Figure S1a). After being hybridized with Bi2Se3, the square shape of the BiOCl nanosheets remained, and their surfaces became fluffier (Figure 1(b)). The dark-field scanning transmission electron microscopy (STEM) images reveal that there are pores on the surface of the nanosheets, some of which are less than 2 nm in size (Figure 1(c) and Figure S1b, c), which was further proved by the pore size distribution curve extracted from the N2 adsorption-desorption isotherm curve (Figure S2). The thickness of the hybrid nanosheets, as estimated from their side-view STEM image, was ~5 nm (Figure 1(d)). Their EDX mapping analysis confirms the presence of Se, Cl, and Bi elements (Figure 1(e)), and the XRD pattern (Figure 1(f)) reveals peaks from both BiOCl (JCPDS no. 06-0249, space group P4/nmm, a = 0.3891 nm and c = 0.7369 nm) and Bi2Se3 (JCPDS no. 33-0214, space group Rm, a = 0.4139 nm, and c = 2.8636 nm) [22, 42]. Their XPS Bi 4f spectrum shows two doublets for Bi3+ at 157.5/162.8 eV and 158.5/163.8 eV attributed to the different binding states of Bi in Bi2Se3 and BiOCl, respectively (Figure S3) [43, 44]. From the convoluted XPS peak areas, the Bi2Se3: BiOCl molar ratio can be estimated as 3 : 5 [45]. The formation of the Bi2Se3/BiOCl heterostructures is likely a result of an in situ ion-exchange reaction, as schematically shown in Figure 1(g). During this process, the Se powder dissolved in OLA and DDT was reduced to Se2- and complexed with OLA based on Equation (1) [46]. The surface layer of BiOCl might undergo an ion exchange reaction with the surrounding Se2- to produce Bi2Se3 (Equation (2)). This proposed in situ ion exchange process is also consistent with the observation that the Bi2Se3/BiOCl hybrid nanosheets did not become thicker than the original BiOCl nanosheets.

Figure 1

TEM images of (a) BiOCl nanosheets and (b) Bi2Se3/BiOCl heterostructures. (c) STEM image of a Bi2Se3/BiOCl heterostructure, revealing a nanoporous Bi2Se3 layer with a pore size of 1-3 nm. (d) Side-view STEM image of Bi2Se3/BiOCl heterostructures. (e) STEM image and EDX mapping of Bi2Se3/BiOCl heterostructures. (f) XRD patterns of BiOCl nanosheets and Bi2Se3/BiOCl heterostructures. (g) Schematic illustration of the formation process of the Bi2Se3/BiOCl heterostructure.

The microstructure of the obtained heterostructure was investigated with selected area electron diffraction (SAED) and high-resolution TEM (HRTEM), as shown in Figure 2. The SAED pattern shows three sets of patterns (Figure 2(a)). One has a fourfold symmetry (indicated by the yellow square) corresponding to BiOCl, and the other two have a sixfold symmetry (indicated by the red and green hexagons) corresponding to Bi2Se3. The presence of two hexagonal patterns suggests that the Bi2Se3 overlayer has two equivalent alignment directions with a relative rotation angle of 90o. In other words, the (110) planes of Bi2Se3 can be aligned with either the (200) or (020) planes of BiOCl, establishing an epitaxial relationship of BiOCl (001) || Bi2Se3 (001) and BiOCl [100] || Bi2Se3 [110]. In the HRTEM image in Figure 2(b), an overlap of lattice patterns of Bi2Se3 and BiOCl can be observed, and its fast Fourier transformation- (FFT-) generated diffraction pattern shows two sets of spots. By selecting the respective set of spots, the BiOCl [001]-zone and the Bi2Se3 [001]-zone lattice patterns were regenerated and are shown in Figures 2(c) and 2(d), respectively, further confirming the epitaxial relationship shown in Figure 2(e). The side-view HRTEM images of typical hybrid nanosheets show lattice spacings of 0.32 and 0.74 nm, attributable to the Bi2Se3 (009) and BiOCl (001) planes, respectively (Figures 2(f) and 2(g)). It can also be seen that the surface deposited Bi2Se3 layer contains one quintuple layer (QL) or two QLs.

Figure 2

(a) SAED pattern of a Bi2Se3/BiOCl heterostructure. The pattern for BiOCl with a fourfold symmetry is indicated by a yellow square; two patterns for Bi2Se3 with a sixfold symmetry are indicated by red and green hexagons. (b) HRTEM image of an area with overlapping lattices of BiOCl and Bi2Se3. Inset: the corresponding FFT diffraction pattern. (c) BiOCl [001]-zone and (d) Bi2Se3 [001]-zone lattice patterns generated by performing inverse-FFT of the spots forming the red hexagon and the yellow square in (b), respectively. Insets in (c and d): the selected spots. (e) Schematic top-view and side-view models indicating the BiOCl (001) || Bi2Se3 (001) and BiOCl [100] || Bi2Se3 [110] epitaxial relationship. Side-view HRTEM images of typical Bi2Se3/BiOCl heterostructures with (f) 1 QL and (g) 2 QLs Bi2Se3 grown on a BiOCl nanosheet.

The energy levels of BiOCl and Bi2Se3 (Figure S4, 5) were determined with UV-vis absorption spectra (Figure S6) and UV photoelectron spectroscopy (UPS, Figure S7). The results show that while Bi2Se3 shows n-type semiconducting behavior, BiOCl is a p-type semiconductor, consistent with previous reports [47, 48]. The electronic properties of the heterostructure were further studied with first-principle calculations based on density functional theory (DFT). A supercell based on 2 L Bi2Se3 and 2 L BiOCl was built, its geometric structure was optimized, and the projected local density of states (PLDOS) was calculated (Figures 3(a) and 3(b) and Figure S8). The Fermi level of BiOCl shifts closer to its conduction band edge after forming a heterojunction with Bi2Se3 (Figure 3(a)), suggesting electron transfer from Bi2Se3 to BiOCl across the interface. A zoomed-in PLDOS plot for BiOCl in the heterostructure (inset in the 3rd panel in Figure 3(a)) shows new states that follow the projected states of Bi2Se3 (2nd panel in Figure 3(a)), resulting in bandgap narrowing at the interface. The charge transfer across the interface can also be seen from the mapping of the charge density difference, in which electron deficient regions are observed on the Bi2Se3 side and electron rich regions appear toward the BiOCl side (Figure 3(b)). Such charge transfer across the interface and the change of the band structure of BiOCl were further demonstrated by the absence of the characteristic Raman peak for BiOCl (A1g mode at 144 cm−1) in the Bi2Se3/BiOCl heterostructure (Figure 3(c)) [49, 50].

Figure 3

(a) Calculated PLDOS of Bi2Se3 (the top 1st panel) and BiOCl (bottom panel) before and after (the 2nd and 3rd panels) contact. (b) Calculated structure model of the Bi2Se3-BiOCl interface superimposed with the charge density difference (cyan: positive; pink: negative). (c) Raman spectra of BiOCl, Bi2Se3, and Bi2Se3/BiOCl heterostructures.

The large specific surface area enabled by the formation of a nanoporous surface layer and the additional charge transfer channels enabled by the new electronic states upon heterojunction formation suggest that our epitaxial heterostructures are promising for sensing applications. As a proof-of-concept demonstration, the NO2 gas sensing properties of BiOCl nanosheets, Bi2Se3 nanosheets, and Bi2Se3/BiOCl heterostructures were studied at room temperature (Figures 4(a) and 4(b) and Figure S9). To preclude the influence of the baseline shift, baseline correction was implemented for the sensing responses (Figure S10) [51-53]. Upon exposure to NO2 (an oxidizing gas), the BiOCl film and Bi2Se3 film showed decreased and increased resistance, respectively, consistent with their p-type and n-type semiconducting properties [54-57]. However, their sensitivities are poor (e.g., 1.8% response at 1 ppm for BiOCl and 5.4% response at 1 ppm for Bi2Se3). The sensor based on Bi2Se3/BiOCl, which also showed an increased resistance upon NO2 exposure, exhibited ~35 and ~12 times higher responses than BiOCl and Bi2Se3 at 1 ppm, respectively (Figures 4(a) and 4(b)). In addition, the porous structure of the surface layer of the heterostructure resulted in a large specific surface area, which allowed more gas molecules to interact with the sensing material, but might cause a problem of slow gas desorption [58, 59]. This was also reflected in the slower recovery times than response times (Figure S11). The sensor was also capable of providing a 9.4% response at 100 ppb, along with good reproducibility (Figure S12). Its good selectivity toward NO2 was proven by exposing it to different gases at 10 ppm, including NO2, H2S, C7H8, C2H5OH, NH3, (CH3)2CO, CO2, and HCHO, at room temperature (Figure 4(c)). The sensing response of Bi2Se3/BiOCl could be further improved by light irradiation (Figure 4(d) and Figure S13). This is consistent with the charge transfer and carrier modulation-based sensing mechanism of this type of chemiresistive sensors.

Figure 4

(a) Dynamic response-recovery curve of the sensor fabricated from Bi2Se3/BiOCl heterostructures in response to NO2 gas with increasing gas concentration at room temperature. (b) Sensing response vs. NO2 concentration for BiOCl-, Bi2Se3-, and Bi2Se3/BiOCl-based sensors. (c) Response of Bi2Se3/BiOCl heterostructures upon exposure to 10 ppm NO2, H2S, C7H8, C2H5OH, NH3, (CH3)2CO, CO2, and HCHO at room temperature. (d) Sensing response vs. NO2 concentration for Bi2Se3/BiOCl-based sensor with and without a 365 nm light irradiation (8.31 μW).

The enhanced sensing performance of the Bi2Se3/BiOCl gas sensor at room temperature can be explained from the following aspects. First, since the sensing mechanism of our sensors is mainly based on charge transfer [60], DFT calculations were performed to shed light on how effective the NO2 adsorption can take away electrons from the heterostructures (Figure 5(a)). The adsorption energies of NO2 on the (110), (100), and (001) planes of Bi2Se3 were calculated to be -0.73 eV, -1.18 eV, and -0.006 eV, respectively, which are more negative than that on BiOCl (001) (-0.005 eV), indicating the higher affinity of NO2 toward Bi2Se3. Note that the edges of Bi2Se3 layers, i.e., the (110) and (100) facets are particularly more effective in NO2 adsorption and charge transfer compared to those on the (001) basal plane (Figure 5(b)). Indeed, the nanoporous Bi2Se3 layer epitaxially deposited on BiOCl was (001)-oriented, showing exposed (110) and (100) facets (Figures 5(c) and 5(d)), thus, favoring NO2 adsorption and electron transfer.

Figure 5

(a) Calculated structural models of NO2 molecules absorbed on the BiOCl (001), Bi2Se3 (001), (100), and (110) surfaces. (b) Calculated adsorption energy (eV) and electron transferred (e) toward NO2 on different sensing surfaces. HRTEM images showing the (c) (110) edge plane and (d) (100) edge plane of Bi2Se3.

The improved sensing ability can also be attributed, in part, to the formation of a p-n junction at the epitaxial interface between BiOCl and Bi2Se3 (Figure S14), where the interdiffusion of electrons and holes across the interface created a charge depletion region along with a built-in potential barrier [61-64]. The height of the energy barrier was mainly determined by the hole concentration in BiOCl because of the much wider bandgap of BiOCl than that of Bi2Se3 [65, 66]. Exposure of Bi2Se3/BiOCl to NO2 gas increased the hole concentration in BiOCl, leading to a wider depletion region and thus a higher energy barrier. Because the conductance of the sensing material changes exponentially with the energy barrier at the heterojunction [67, 68], a much improved gas sensing response can be obtained. A higher carrier concentration in BiOCl could be achieved by light irradiation, as shown in Figure 4(d), where the sensing response of the Bi2Se3/BiOCl sensor doubled under 365 nm light excitation (8.31 μW).

Moreover, the Bi2Se3/BiOCl heterostructure-based sensor showed a faster response time than both Bi2Se3 and BiOCl sensors (Figure S11). This can be attributed to the newly generated hybrid electronic states upon formation of epitaxial interface, as shown in the calculated PLDOS in Figure 3(a). These hybrid states could provide additional and faster channels to lose electrons to NO2, and therefore, a shorter response time was observed for the Bi2Se3/BiOCl-based sensing material.

Last, flexible gas sensors based on Bi2Se3/BiOCl heterostructures were fabricated by depositing them on a plastic substrate with inkjet-printed interdigitated electrodes (Figure 6). It is worth mentioning that flexible electrodes fabricated with facile printing techniques such as inkjet printing suffer from poor resolution and large gaps between electrodes. This can cause reduced device performance compared to using electrodes fabricated by standard lithography techniques. The growth of the nanoporous Bi2Se3 layer on the BiOCl nanosheet increased the surface roughness and might help enhance the interlocking between the stacked hybrid nanosheets (Figure 6(a)). This ensured the formation of a continuous sensing film across the printed electrodes with a large gap of ~200 μm (Figure S15), even though the average lateral size of the Bi2Se3/BiOCl nanosheets was less than 100 nm (Figure 1(b)). In sharp contrast, pristine BiOCl nanosheets deposited on inkjet-printed electrodes failed to form conductive films (Figure S16). The response of the flexible gas sensor under bent conditions (bending radius: 7.5 mm) showed a slightly enhanced response at sub-ppm levels compared to that under flat conditions, with a calculated limit of detection down to 1.6 ppb (Figure 6(b) and Figure S17), outperforming previously reported flexible sensors based on 2D materials under bent conditions (Figure 6(c) and Table S1) [69-75]. In addition, the bending radius also influenced the sensing performance. A further reduced bending radius to 5 mm could induce more exposed active edge sites and thus higher responses at sub-ppm levels (Figure S18). Last, the mechanical durability of our flexible sensors was tested by repeatedly bending a sensor for 100 times. The sensor maintained approximately 95% of its original response toward 5 ppm NO2 (Figure S19).

Figure 6

(a) Schematic diagram of the Bi2Se3/BiOCl heterostructures on flat and bent substrates. (b) Responses of Bi2Se3/BiOCl-based sensors under flat and bent states at various NO2 concentrations. Inset: a photograph of the printed electrode array. (c) Comparison of the sensing performance between our flexible gas sensor and other 2D material-based flexible sensors under bending conditions.

3. Conclusion

In this work, we demonstrated the in situ topochemical conversion of layered materials, which is capable of maximizing the active edge sites of the deposited overlayers and the formation of coherent heterointerfaces. Using Bi2Se3/BiOCl as a demonstration, onto BiOCl (001) basal surfaces with fourfold symmetry, hexagonal Bi2Se3 (001) layers were epitaxially deposited with largely exposed (100) and (110) edge sites. Compared to BiOCl or Bi2Se3 alone, the Bi2Se3/BiOCl heterostructure showed a much enhanced sensing response toward NO2 gas. According to the theoretical calculation results, the Bi2Se3 edge surfaces are generally more active than the basal surfaces of Bi2Se3 and BiOCl in NO2 adsorption and charge transfer. Flexible gas sensors based on Bi2Se3/BiOCl heterostructures possess good sensing properties with a limit of detection down to 1.6 ppb at room temperature, demonstrating their potential for wearable and portable devices in the future. Our strategy of generating nanopores in van der Waals layered materials via topochemical heteroepitaxy will provide more opportunities to tailor both their electronic and chemical properties.