Revealing the Role of Vapor Flux in Chemical Vapor Deposition Growth of Bi₂O₂Se for Photodetectors

Abstract

Two-dimensional (2D) materials are regarded as key foundational materials for next-generation optoelectronic devices. As a promising new type of 2D layered semiconductor, Bi₂O₂Se has emerged as a strong candidate for high-performance opto-electronic devices due to its high carrier mobility, tunable bandgap, and excellent environmental stability. However, achieving precise control over Bi₂O₂Se growth to obtain high-quality Bi₂O₂Se remains a challenge in the field. In this study, we employed chemical vapor deposition (CVD) to grow thin-layer 2D Bi₂O₂Se flakes. We further used a transport model and thermodynamic Arrhenius fitting to analyze the relationship between vapor flux and the properties of the flakes. Density functional theory was used to study the electronic structure of the as-grown samples. The electrical and optoelectronic results demonstrate that Bi₂O₂Se-based FETs exhibit good performance in terms of mobility (129 cm²V⁻¹s⁻¹), on/off ratio (4.51 × 10⁵), and photoresponsivity (94.98 AW⁻¹). This work provides a new way to study the influence of vapor flux on the sizes and shapes of Bi₂O₂Se flakes for photodetectors.

1. Introduction

Two-dimensional (2D) materials have attracted significant attention due to their unique properties, including layer-dependent electronic structures, tunable photonic properties, electrocatalytic activity, and mechanical flexibility [1,2,3]. These features have enabled advancements in electronic and optoelectronic devices [4,5,6] as well as sensors [7,8]. Graphene, renowned for its high carrier mobility [9], has limitations in field-effect transistors (FETs) due to its zero bandgap, which hampers effective on/off switching. Black phosphorus, with a thickness-dependent bandgap of 0.3–2 eV and high on/off ratios, offers better potential but suffers from air instability and challenges in large-scale film growth [10,11,12]. Transition metal dichalcogenides (TMDCs) have shown promise in optoelectronics due to their favorable properties but are constrained by relatively low carrier mobility [13,14]. These limitations highlight the need to explore novel 2D materials with tunable bandgaps, high carrier mobility, and environmental stability, which are essential for next-generation optoelectronics.

In recent years, Bi₂O₂Se has gained attention in optoelectronics due to its high carrier mobility [15] (up to 28,900 cm²V⁻¹s⁻¹ at 2 K), moderate bandgap [16] (~0.8 eV), and excellent environmental stability [15,17,18]. These properties make Bi₂O₂Se a promising candidate for high-performance optoelectronic devices, though research on these applications remains limited compared to materials like MoS₂ and WSe₂. Initial studies by Oppermann et al. involved synthesizing bulk Bi₂O₂Se via hot-pressing and analyzing its semiconductor properties and phase diagram [19]. Subsequent work explored its potential in transport and thermoelectric applications [20,21]. Recently, advancements in chemical vapor deposition (CVD) have enabled the growth of 2D Bi₂O₂Se [22,23,24], revealing its high carrier mobility and low electron effective mass. The above achievements highlight the need for further research to fully realize Bi₂O₂Se’s potential in high-performance optoelectronics.

The performance of Bi₂O₂Se is strongly influenced by the preparation method. Wet chemical synthesis stands out for its low cost and high yield, making it an attractive option for large-scale production. However, it faces challenges in achieving uniform thickness and size across samples [25,26]. Physical vapor deposition (PVD) has been used for the heteroepitaxial growth of Bi₂O₂Se thin films [27], but its slow growth rates and high production costs limit its practicality. In contrast, CVD offers advantages in terms of controllability and scalability. It enables the precise regulation of critical parameters, such as thickness, size, and uniformity. Studies have demonstrated that optimizing CVD growth parameters, such as reaction temperature, precursor concentration, and pressure, can grow Bi₂O₂Se flakes with high crystallinity, uniform thickness, and tunable properties. For example, Wu et al. synthesized ultra-high-mobility Bi₂O₂Se using a controlled CVD process, achieving effective control over thickness, nucleation sites, and size by adjusting growth parameters [22]. Similarly, Khan et al. developed a salt-assisted CVD, successfully growing atomically thin Bi₂O₂Se materials [28]. They further employed a catalyst-free CVD method to grow nearly monolayer Bi₂O₂Se nanoribbons, demonstrating the ability to precisely control the morphology of Bi₂O₂Se [23]. These advancements highlight the versatility and potential of CVD as a reliable method for growing high-quality Bi₂O₂Se with tailored properties, paving the way for its integration into advanced optoelectronics.

In this paper, we develop an atmospheric pressure CVD method for the growth of 2D Bi₂O₂Se flakes. By controlling key parameters such as the growth temperature and gas flow rate in the furnace, the growth of Bi₂O₂Se can be precisely regulated with varying shapes and thicknesses. To investigate the underlying mechanisms of how the parameters affect the properties of as-grown Bi₂O₂Se, we used thermodynamic and transport models relating the growth temperature and gas flow rate to the sizes and shapes of as-grown Bi₂O₂Se flakes. The Bi₂O₂Se-based FETs exhibited a high on/off ratio of approximately 4.51 × 10⁵ and a high carrier mobility of 129 cm²V⁻¹s⁻¹. Furthermore, the photodetectors made of Bi₂O₂Se achieved a maximum responsivity of 94.98 AW⁻¹ and a maximum detectivity of 1.46 × 10⁹ Jones. The abovementioned excellent device performance of the Bi₂O₂Se flakes indicates their potential for applications in the fields of electronics and optoelectronics.

Figure 1a shows the experimental setup for the growth of Bi₂O₂Se flakes. The electrostatic interaction between mica and Bi₂O₂Se facilitates the lateral growth of Bi₂O₂Se flakes [22,29,30]. Therefore, freshly cleaved mica was chosen as a suitable substrate for the growth of atomically thin 2D Bi₂O₂Se (details are provided in Section 2). Dual pre-cursors, Bi₂O₃ and Bi₂Se₃, were used as the co-evaporation sources for Bi₂O₂Se growth, and the mica substrate was placed downstream for deposition. During growth, the temperature and gas flow rate inside the furnace were accurately monitored, with their temporal variations shown in Figure 1b. A typical growth temperature was 820 °C, with a gas flow rate of 40 sccm, and the growth time was maintained for 1 h, which allowed for the growth of high-quality Bi₂O₂Se flakes. Figure 1c shows a representative optical microscope (OM) image of Bi₂O₂Se flakes, which have a square or rectangular shape, with an average domain size of approximately 30 µm. As shown in Figure S1 (Supporting Information), the largest domain size of the Bi₂O₂Se flakes exceeded 220 µm. The Raman spectra in Figure 1d show a characteristic peak centered at 160.7 cm⁻¹, which is attributed to the A₁g vibration mode of Bi₂O₂Se, with the reference being the pristine mica substrate. Raman spectra of Bi₂O₂Se with different flakes are shown in Figure S2a,b (Supporting Information), showing characteristic peaks at 158.1 cm⁻¹ and 160.0 cm⁻¹, respectively. The variations in the peak can be ascribed to the effects of Bi₂O₂Se thickness. An atomic force microscopy (AFM) image shows that the thickness of Bi₂O₂Se flakes is approximately 7.7 nm (Figure 1e). Considering the thickness of a Bi₂O₂Se unit is 0.61 nm [15,31], 7.7 nm corresponds to about 13 layers. The thinnest Bi₂O₂Se flakes obtained in our growth were 4 nm, or about 6–7 layers, as shown in Figure S3 (Supporting Information). To determine the chemical state of the as-grown 2D Bi₂O₂Se flakes, X-ray photoelectron spectroscopy (XPS) was conducted. As shown in Figure 1f–h, peaks corresponding to Bi 4f_5/2 and Bi 4f_7/2 are centered at 164.38 eV and 159.08 eV, respectively, consistent with the Bi element in bismuth oxide [32]. Also, the O 1s orbit shows peaks at 531.98 eV and 529.98 eV, and the Se 3d orbit shows peaks at 54.18 eV and 53.38 eV. These binding energies of Bi, O, and Se elements are consistent with those in Bi₂O₂Se [33]. The above results confirm the successful growth of Bi₂O₂Se flakes on mica substrates.

Understanding the vapor concentration of precursors is crucial in controlling the CVD growth of 2D Bi₂O₂Se. We initially plotted the saturated vapor pressure versus temperature for both Bi₂Se₃ and Bi₂O₃ precursors, as shown in Figure 2a [34]. Then, we used it to calculate the actual vapor concentration based on a transport model based on Equation (1) [35].

$$C_i=C_i^{*}[1-\exp \left(-{\displaystyle \frac{K_i}{\overline{u}}}{\displaystyle \frac{A_h}{A}}\right)]$$

(1)

where $C_i^{*}$ represents the saturated vapor concentration, $\overline{u}$ refers to the average flow velocity of the carrier gas, and K_i stands for the mass transfer coefficient. A_h represents the area of the precursor (assumed to be 1 cm × 1 cm). A represents the cross-sectional area at the site where the precursor is located. As exhibited in Figure 2b,c, we can observe that the vapor phase concentration is in the range of 10⁻¹³–10⁻⁹ mol/m³ for Bi₂Se₃ and 10⁻²⁰–10⁻¹⁶ mol/m³ for Bi₂O₃ precursors, and this concentration increases with the temperature’s increase and gas flow rate’s decrease. Hence, the reaction concentration can be precisely controlled by modulating the growth parameters of temperature and gas flow rate. First, we investigated the effects of growth temperature. Figure 2d shows the lateral size distribution of Bi₂O₂Se flakes grown at temperatures ranging from 800 to 840 °C. The corresponding OM and individual distribution are given in Figure S4 (Supporting Information). Most Bi₂O₂Se flakes were relatively small in size. Specifically, Figure 2e shows the average size and nucleation density of the Bi₂O₂Se flakes versus the growth temperature. The increase in temperatures results in larger flakes with a lower nucleation density, while lower temperatures yield smaller flakes with higher nucleation density. To gain further insights into the growth mechanism of Bi₂O₂Se, we conducted a thermodynamic Arrhenius analysis of the average domain size versus 1000/T, as shown in Figure 2f [36]. The reaction barrier for growth was extracted to be 2.557 eV by fitting the formula in Equation (2):

$$\log (D/D_0)=C-({\displaystyle \frac{E_a^{exp}}{1000k_B\ln 10}})({\displaystyle \frac{1000}{T}})$$

(2)

where D is the average size of Bi₂O₂Se flakes, T is the growth temperature, D₀ = 1 μm is used for normalization, C is a constant, and $E_a^{exp}$ denotes the activation energy. k_B represents the Boltzmann constant. The above results indicate that the growth is a thermally activated growth mechanism. We also studied the effects of both temperature and gas flow rate on the shapes of the as-grown Bi₂O₂Se flakes. Figure 2g shows the OM images of Bi₂O₂Se flakes with different shapes, such as mixed-shape, square, nanoribbons, and nanoblocks. These variations in shape are a result of the interplay between the temperature, gas flow rates, and vapor concentration during the CVD process. As shown in Figure 2h, higher temperatures and slower gas flow rates tend to favor the growth of square and nanoblock flakes, while mixed-shape flakes are typically observed at high temperatures and gas flow rates. Nanoribbons are typically observed at lower temperatures with very high gas flow rates. The above results reveal the role of vapor flux in controlling the sizes and shapes of as-grown Bi₂O₂Se flakes.

Based on the above CVD-grown Bi₂O₂Se flakes, we studied their electrical properties by fabricating Bi₂O₂Se-based FETs. Bi₂O₂Se is a layered material composed of alternating [Bi₂O₂]n²ⁿ⁺ and [Se]n²ⁿ⁻ layers stacked along the c-axis, which form covalent bonds through electrostatic interactions, as visualized in Figure 3a. Previous studies have extensively employed density functional theory (DFT) to calculate the band structure of 2D semiconductors [37,38]. Hence, we used density functional theory (DFT) to calculate the band structure and density of states of Bi₂O₂Se, revealing that it has an indirect bandgap of approximately 0.34 eV, with the valence band maximum at the X point and the conduction band minimum at the Γ point (Figure 3b,c). It should be noted that the bandgap of bulk Bi₂O₂Se was measured to be ~0.85 ± 0.05 eV using scanning tunneling spectroscopy (STS) [17]. We then fabricate the back-gated device by transferring Bi₂O₂Se flakes from mica to a 300 nm thick SiO₂/Si substrate. Figure 3d shows a schematic of Bi₂O₂Se flake-based FETs. Figure 3e shows the output characteristics (Ids-Vds) of the Bi₂O₂Se device measured at room temperature under zero gate bias, with a current reaching up to approximately 5.75 × 10⁻⁶ A at a bias of 1 V, indicating excellent electrical performance. The asymmetry of this curve demonstrates a Schottky barrier existing at the electrode–channel contact interface. The inset of Figure 3e shows an OM image of a Bi₂O₂Se FET device with a channel length of 18.28 µm and a width of 12 µm. The effective area is 219.36 µm². When the back-gate voltage was adjusted from −40 V to 40 V in a 10 V step, it was observed that the drain current (Ids) strongly depended on Vgs, demonstrating its back-gate controllability (Figure 3f). The nonlinear features of these Ids-Vds curves further suggests the presence of a Schottky barrier, which may be due to the electrode’s oxidation during the deposition process. Figure 3g shows the transfer characteristics (Ids-Vgs) at a bias voltage Vds of 1 V, displayed in both linear and logarithmic scales. The n-type behavior of the Bi₂O₂Se device causes Ids to increase (decrease) with a positive (negative) Vgs. The on/off ratio at a bias of 1 V is approximately 4.51 × 10⁵, and the field-effect mobility (µ) extracted from this curve is 129 cm²V⁻¹s⁻¹. Furthermore, Figure 3h,i show the Ids-Vgs curves at various bias voltages Vds ranging from −1 V to 1 V with a step of 0.25 V in both linear and logarithmic scales, further confirming the n-type characteristics of the Bi₂O₂Se-based FETs. The above results demonstrate the relatively high electronic performance of as-grown Bi₂O₂Se-based devices.

The above Bi₂O₂Se-based FET with its high mobility and on/off ratio demonstrates its potential in optoelectronic devices. Hence, we systematically studied its optoelectronic behaviors by using different incident light wavelengths and powers, as well as applying different drain and gate voltages. Figure 4a shows the output curves (Ids-Vds) with 447 nm laser excitation at different light powers, which was tested at Vgs = 0 V to elucidate the intrinsic behavior of the Bi₂O₂Se photodetector. The device reached its maximum photocurrent of 5.45 × 10⁻⁵ A at a bias voltage of 1 V and an incident laser power of 19.20 µW. Figure 4b shows the Ids-Vgs curves at a bias voltage of 1 V with different laser powers. The gate voltages range from −40 V to 40 V. Figure S5 (Supporting Information) presents the results of the I_ph-P relationship curve. It can be observed that I_ph increases with an increasing laser power as the enhancement in incident light power leads to a higher carrier concentration in the material, resulting in a stronger photocurrent. The relationship between I_ph and P can be fitted using the following power-law equation [39]:

I_ph~P^α

(3)

where α is the power-law exponent, obtained from data fitting as 0.736.

As the laser intensity increases, the photocurrents of both the on state and off state increase a lot, which may be due to the photogenerated electrons in the Bi₂O₂Se channel. Since the response time is a key parameter for photodetectors, we measured the time-resolved photo response of the photodetectors by periodically turning the laser on and off under different laser powers ranging from 1.843 µW to 19.20 µW. As shown in Figure 4c, the photodetector exhibited good on/off characteristics with excellent repeatability. Additionally, the photoinduced polarization enhancement in Bi₂O₂Se modulates the Schottky barrier and reduces resistance, causing the dark current to increase with increasing light power. Meanwhile, photogenerated carriers fill trap states, with some carriers remaining trapped, making it difficult for the dark current to recover. Furthermore, the magnified response curves shown in Figure 4d,e were used to calculate the rise and decay times. The rise time is defined as the time taken for the photocurrent to increase from 10% to 90% of Ids (max), and the decay time is the reverse, resulting in a rise time and a decay time of 18.32 ms and 19.97 ms, respectively, for the Bi₂O₂Se photodetector. For photodetectors, responsivity (R) and detectivity (D*) are important parameters for evaluating performance. Hence, we calculated the R and D* values of the photodetector versus incident laser powers, as shown in Figure 4f. The R value, external quantum efficiency (EQE), noise equivalent power (NEP), and detectivity (D*) were determined using the following equations [40]:

$$R={\displaystyle \frac{I_{ph}}{P_{in}\times A_0}}$$

(4)

$A_0$ and $P_{in}$ represent the active area of the device and the incident light power density, respectively. The external quantum efficiency (EQE) is determined using the following equation:

$$EQE={\displaystyle \frac{h\times c\times R}{e\times \lambda }}$$

(5)

where $h$, $e$, $c$, and $\lambda$ represent Planck’s constant, elementary charge, the speed of light, and the wavelength of the illumination source, respectively. The noise equivalent power (NEP) is the minimum detectable power density of the photodetector (P_in) with a 1 Hz bandwidth and can be defined as follows:

$$NEP={\displaystyle \frac{\sqrt{2e\times I_d}}{R}}$$

(6)

D* is determined by the following equation:

$$D^{*}={\displaystyle \frac{R\times \sqrt{A_0}}{NEP}}$$

(7)

At V_gs = 0 V, the Bi₂O₂Se photodetector achieved a maximum R of 72.52 AW⁻¹ and a maximum D* of 8.04 × 10⁸ Jones. The R value decreases with increasing incident light power (P_in), primarily influenced by changes in photoconductive gain (G) [41,42]. The GG is given by the following equation:

$$G=({\displaystyle \frac{T_t}{T_r}})(1+{\displaystyle \frac{{\mu }_h}{{\mu }_e}})$$

(8)

where ${\mu }_h$ and ${\mu }_e$ represent the carrier mobility of holes and electrons, respectively, and $T_t$ and $T_r$ correspond to the hole transit time and recombination lifetime. A higher h⁺ trap concentration suppresses e⁻-h⁺ recombination, resulting in a recombination time that is much longer than the transit time, leading to an increased G value and, consequently, a higher R value at low P_in. Additionally, the detectivity (D*) follows the same trend as R [43]. At a wavelength of 447 nm, the EQE and NEP values were calculated, as shown in Figure S6 (Supporting Information). The NEP increases with incident light power, whereas the EQE exhibits the opposite trend. The calculated minimum NEP and maximum EQE values are 1.84 × 10⁻¹⁴ WHz^−1/2 and 2011.89%, respectively.

Additionally, we tested the optoelectronic performance of the Bi₂O₂Se photodetector at 520 nm and 637 nm wavelengths. As shown in Figure S7a–d (Supporting Information), the maximum R of the Bi₂O₂Se photodetector reached 94.98 AW⁻¹, and the maximum D* was 1.46 × 10⁹ Jones under a 520 nm laser. Figure S8a–d (Supporting Information) show that the maximum R and D* values of the Bi₂O₂Se photodetector are 64 AW⁻¹ and 1.5 × 10⁹ Jones, respectively, with a 637 nm laser. Taken together, the Bi₂O₂Se photodetector achieved the highest R and D* values with 520 nm laser illumination. As seen in the comparision in Table S1 (Supporting Information), Bi₂O₂Se exhibits a higher responsivity and considerable detectivity compared to other photodetector materials.

2. Experimental Section

2.1. CVD Growth of Bi₂O₂Se

Two-dimensional Bi₂O₂Se flakes were grown using the CVD method. High-purity bismuth selenide powder (Bi₂Se₃, 48 mg) and bismuth oxide powder (Bi₂O₃, 12 mg) were placed in separate boats inside the tube furnace, with Bi₂Se₃ positioned at the center and Bi₂O₃ placed 4 cm away from the center. Mica substrates (1 × 2 cm²) were used as the growth substrate and positioned 12–16 cm downstream from the center of the furnace. The center of the furnace was heated to 820 °C with a heating rate of 40 °C/min, and this temperature was maintained for 1 h to conduct the growth. Additionally., 40 sccm Ar gas was used as carrier gas to transport reaction vapors. After the growth, the furnace was naturally cooled to room temperature in the presence of Ar.

2.2. Density Functional Theory Calculation

The first-principles calculations were implemented in the Vienna Ab initio Simulation Package (VASP). The generalized gradient approximation (GGA) with Perdew–Burke–Ernzerhof (PBE) were adopted for exchange and correlation interactions. Bulk Bi₂O₂Se was chosen as crystal structure. For the calculation, the Hellman–Feynman forces were set to less than 0.02 eV Å, and the energy cutoff was set to 400 eV. The K points were sampled by Monkhorst-Pack with a 7 × 7 × 7 mesh. The single electronic step was converged to 1 × 10⁻⁹ eV.

2.3. Device Fabrication

Bi₂O₂Se flakes on a mica substrate were transferred onto a Si/SiO₂ substrate using a 1.5% polystyrene (PS) solution. The electrodes were patterned via photolithography. The pattern was developed using a 25% aqueous solution of tetrame-thylammonium hydroxide, which was diluted with deionized water at a ratio of 3:25. Subsequently, a 5 nm Cr layer and a 50 nm Au layer were deposited via thermal evaporation.

2.4. Characterization

The sizes and shapes of Bi₂O₂Se flakes were characterized by using optical microscopy (Sunny Optical CX40M, Yuyao, China). The thickness of Bi₂O₂Se flakes was measured by AFM (Cypher ES, Asylum Research, Santa Barbara, CA, USA). Raman spectra were collected with a confocal microscope spectrometer (Renishaw LabRAM Invia, Wotton-under-Edge, UK). The element chemical states of Bi₂O₂Se flakes were tested using XPS (Thermo Scientific Escalab 250Xi, Waltham, MA, USA) with Al-ka (1486.6 eV) as a source. The electronic performance of FET and photodetectors was tested using a probe station and a semiconductor analyzer (Keithley 2634B, Cleveland, OH, USA), with the laser power calibrated using a power meter.

3. Conclusions

In summary, we developed an atmospheric pressure CVD method for the growth of 2D thin-layer Bi₂O₂Se. By adjusting the growth parameters, we achieved precise control over the sizes and shapes of Bi₂O₂Se flakes. The thinnest Bi₂O₂Se flake obtained was 4 nm, and the largest Bi₂O₂Se flake was 220 µm. The role of vapor flux of precursors was studied to bridge the growth parameters and vapor concentration. The FETs fabricated from Bi₂O₂Se flakes exhibited an n-type transport behavior, with an electron mobility of 129 cm²V⁻¹s⁻¹ and an on/off ratio of 4.51 × 10⁵. Additionally, the corresponding photodetectors showed a responsivity of 94.98 AW⁻¹ and a detectivity of 1.46 × 10⁹ Jones. Our study demonstrates that 2D Bi₂O₂Se flakes are promising for high-performance optoelectronics.