Atomic layer deposition of crystalline Bi₂O₃ thin films and their conversion into Bi₂S₃ by thermal vapor sulfurization

Abstract

Crystalline Bi₂O₃ thin films have been grown on Si and quartz substrates at 300 °C in an atomic layer deposition (ALD) chamber using Bi(thd)₃ (thd: 2,2,6,6-tetramethyl-3,5-heptanedionato) and H₂O as precursors. Post-growth thermal sulfurization of such Bi₂O₃ films results in orthorhombic Bi₂S₃ at elevated temperatures. Atomic-force microscopy reveals that the surface roughness of Bi₂O₃ increases with growth cycles, and the roughness of Bi₂O₃/Si is generally larger than those of Bi₂O₃/quartz grown under the same conditions. However, after sulfurization, the surface morphologies became nearly similar in spite of their different substrates and/or thicknesses. The evolution of X-ray diffraction patterns as a function of growth cycles provide evidence that the growth starts with nonstoichiometric β-phase Bi₂O_2.3, and then transfers to α-Bi₂O₃ with the increase in growth cycles; the β-to-α growth transition occurs earlier on Si than that on a quartz substrate. Optical absorption spectroscopy reveals a bandgap of 3.0–3.5 eV for the as-grown Bi₂O₃ thin films, which is narrowed to 1.5 eV for the resultant Bi₂S₃ after sulfurization. X-ray photoelectron spectroscopy indicates that both the as-grown Bi₂O₃ and the sulfurized Bi₂S₃ thin films are n-type semiconductors with the valence band maxima located at about 2.1 and 1.0 eV below their Fermi levels, respectively. These observations, together with the large absorption coefficient of Bi₂S₃ in the wavelength range of visible light, suggest a significant potential for Bi₂S₃ thin films and/or Bi₂S₃/Bi₂O₃ heterojunctions in various optoelectronic devices, e.g., for solar energy conversions.

---

I. Introduction

Bi₂S₃, a direct bandgap semiconductor with the bandgap energy of 1.30 eV that locates between those of Si (1.12 eV) and GaAs (1.42 eV) and matches the potential requirement for water splitting (1.23 eV), has been attracting considerable research interest in recent years.^1–3 Such a bandgap, together with the large optical absorption coefficient (>10⁵ cm⁻¹ in the wavelength range of visible light) of Bi₂S₃, makes it a promising absorber for photovoltaic and photoelectrochemical devices.^4–6 In the nanoscale, Bi₂S₃ can form functional hybrid structures with metallic and/or semiconductor materials, e.g., carbon,^7–9 gold,^10,11 TiO₂,^12,13 ZnO,¹⁴ and In₂S₃,¹⁵ for various applications, including H- and Li-storage,^7,8,16 nonlinear optical switching,¹¹ photocatalysis,^10,14,15 gas sensing and biomolecule detection.^17,18 Although numerous optoelectronic devices and unique performances based on Bi₂S₃ and its hybrid structures have been demonstrated, the materials were generally synthesized by aqueous/solution-based methods.^4–18 As a comparison, the methods of non-aqueous/solution growth of Bi₂S₃ thin films with respect to their photovoltaic applications that can be monolithically integrated into traditional thin film based solar cell technologies are less developed and generally limited to thermal vapor deposition and chemical vapor deposition (CVD) techniques.^19–25

Atomic layer deposition (ALD) is a recently developed ultrathin film deposition technique controlled by the ‘surface saturation’ mechanism (i.e., self-limited surface reaction),^26,27 which forms single atomic layers during each reaction cycle by alternatively feeding pulsed precursors. The by-products are purged with an inert gas in between each pulse in the reaction cycles, making the growth continuous. Recently, ultrathin α-Bi₂O₃ films have been synthesized on Si substrates by means of ALD using Bi(thd)₃ (thd: 2,2,6,6-tetramethyl-3,5-heptanedionato) and H₂O as the reaction precursors.²⁸ Our previous studies show that Fe₃S₄, Mo/MoO₃, and GaAs thin layers can be converted by thermal vapor sulfurization into crystalline FeS₂ (pyrite), MoS₂ and Ga₂S₃ films, respectively, with high purity in crystal phases.²⁹ It has also been reported that Bi₂S₃ thin films can be iodinated into BiI₃ at relatively low temperatures.³⁰ These studies suggest that the ALD grown Bi₂O₃ might be able to convert into Bi₂S₃ under proper sulfurization conditions, while Bi₂S₃ can further convert back into Bi₂O₃ at elevated temperatures in the presence of oxygen.³¹ These conversion processes may have significant consequences in fabricating Bi₂S₃ thin films and Bi₂S₃/Bi₂O₃ hybrid structures for applications in various optoelectronic devices, e.g., photovoltaic solar cells,⁵ but are, unfortunately, seldom reported in the literature. In this regard, we have attempted to grow Bi₂O₃ thin films by employing ALD and to convert the ALD grown Bi₂O₃ thin films into Bi₂S₃ by thermal vapor sulfurization.

II. Experiment

The Bi₂O₃ thin films were synthesized on Si (001) and quartz substrates in an ALD chamber (Picosun Oy). Bi(thd)₃ (thd: 2,2,6,6-tetramethyl-3,5-heptanedionato) and deionized water were used as bismuth and oxygen precursors, respectively. The former was evaporated from a booster bottle kept at 200 °C, while the latter was evaporated from a liquid-source bottle kept at room temperature. High-purity nitrogen (99.9995%) was used as the source carrier and purging gas. During growth, the chamber pressure was controlled at about 5–10 Torr, and the substrate temperature was maintained at 300 °C. In each ALD growth cycle, Bi(thd)₃ and H₂O vapor carried by N₂ were alternately introduced into the ALD chamber, and the precursor pulses were separated by N₂ purging. The flow rates of the carrier gas for Bi(thd)₃ and H₂O were set at 80 and 200 sccm, respectively. The pulse/purge durations for Bi(thd)₃ and H₂O were 1.6/6.0 s and 0.1/8.0 s, respectively. Before loading into the ALD chamber, the Si and quartz substrates were cleaned by acetone, methanol, and deionized water in sequence, which was followed by N₂-flow drying. No additional chemical etching was adopted. The thin film samples in this study were grown for 500 to 1500 cycles.

Post-growth thermal vapor sulfurization of the ALD grown Bi₂O₃ thin films was carried out by thermal annealing in a sulfur vapor ambient. The sulfur vapor was evaporated from sulfur powder (99.995%) and carried by N₂ in a tube-furnace. Details of the annealing setup and the general sulfurization procedures can be found in ref. 29. In this work, the sulfurization time was 20 min, and the sulfurization temperature was varied from 500 to 700 °C.

Atomic-force microscopy (AFM), X-ray diffraction (XRD, Cu-Kα₁), optical absorption spectroscopy, Raman scattering spectroscopy, X-ray photoelectron spectroscopy (XPS), and photoluminescence (PL) have been used to study the structural and optical properties of the ALD grown Bi₂O₃ thin films before and after sulfurization. The AFM was operated in an air environment using a tapping mode. The film thickness was measured by AFM addressing on an intentionally scratched area, where a step is formed between the film and the surface of the substrate. The XRD was carried out in a general-area-detector diffraction system (GADDS, Bruker-D8), which has the significant advantage of being highly sensitive to crystal phase structures. Transmittance and reflectance were measured in an UV-vis-NIR scanning spectrophotometer using a bare quartz wafer and a standard aluminum mirror as the references, respectively. Raman scattering and PL measurements were conducted using a backscattering configuration with the 532 nm line of a Nd:YAG laser as the excitation source. XPS was carried out in a VG ESCALAB 220i-XL system equipped with a monochromatic Al-Kα₁ X-ray source (1486.6 eV). The spectra were collectively shifted with the binding energy of C1s (due to contaminants, physically adsorbed on the surface, from the environment) at 285.0 eV as the reference.

III. Results and discussion

A. ALD growth of Bi₂O₃

Fig. 1(a) and (b) show the AFM images recorded from the Bi₂O₃ thin films grown for 1000 cycles (∼45 nm thick) on the Si and quartz substrates, respectively. Similarly, Fig. 1(c) and (d) show the AFM images recorded from the Bi₂O₃ thin films grown for 1500 cycles (∼110 nm thick) on the Si and quartz substrates, respectively. The insets in Fig. 1(b) and (d) are the optical absorbance spectra collected, using a bare quartz substrate as the reference, from the 1000- and 1500-cycle Bi₂O₃/quartz samples, respectively. It is clearly seen that the grain sizes in Fig. 1(c) and (d) are larger than those in Fig. 1(a) and (b), respectively. A detailed analysis revealed that the root-mean-square roughnesses are 4.9, 2.0, 9.2, and 5.8 nm as imaged by AFM in Figs. 1(a)–(d), respectively. These results indicate that the Bi₂O₃ thin films grown on quartz substrates are generally smoother than those grown on Si substrates under similar growth conditions; the grain size and the surface roughness of the Bi₂O₃ thin films increase with their growth cycles (i.e., the nominal thicknesses) on both the Si and quartz substrates.

Fig. 1 AFM images recorded from ALD grown Bi₂O₃ thin films: (a) 1000 cycles on Si, (b) 1000 cycles on quartz, (c) 1500 cycles on Si, and (d) 1500 cycles on quartz. The insets in (b) and (d) are the absorbance spectra measured from the 1000- and 1500-cycle Bi₂O₃/quartz samples, respectively.

The absorbance spectra in Fig. 1(b) and (d) show that the onsets of apparent increment, i.e., the optical bandgap E_Opt, locate at about 3.0 and 3.5 eV, respectively. These values are consistent with those reported for α-Bi₂O₃ at 3.04–3.89 eV, depending on the growth conditions.³² It is also seen that the absorbance ratio between the 1500- and 1000-cycle samples, for example, at the photon energy of 6.0 eV is about 3.3, which is considerably larger than the nominal thickness ratio (i.e., 1.5) of the two Bi₂O₃ thin film samples. This deviation is mainly caused by the variations in the effect of light reflections due to the significantly changed surface morphologies [see Fig. 1(b) and (d)].^32,33 These morphology change induced variations in light reflection were not taken into account in the absorbance measurements (see a later section for further discussions).

Fig. 2(a) presents typical XRD patterns collected from the as-grown Bi₂O₃/Si samples. In terms of the angles, the diffraction peaks well match those of α-Bi₂O₃ (JCPDS 71-2274) and β-Bi₂O_2.3 (JCPDS 76-2477) with the former dominant in the thicker films. The intensity and the linewidth correlations between the diffraction peaks of α-Bi₂O₃ (012) and α-Bi₂O₃ (024), together with their spotty distributions in the two-dimensional imaging frames (see Fig. S1, ESI†), indicate that the α-Bi₂O₃ grains have a collective growth orientation. These results, except for the incorporation of the nonstoichiometric β-phase, are consistent with those reported by Shen et al. in ref. 28. In fact, the coexistence of α- and β-phase crystals has already been observed in Bi₂O₃ nanostructures grown by oxidative metal vapor deposition,³⁴ where the β-phase is, however, stoichiometric and largely dominant over the α-phase in the nanostructures. It is worth to note that although the α-Bi₂O₃ related diffractions increased with the growth cycles, the β-phase Bi₂O_2.3 related diffractions do not exhibit any significant variation. Such an evolution of XRD patterns indicates that the nonstoichiometric β-phase mainly locates at the initial grown region, i.e., in the area close to the Bi₂O₃/Si interface, probably due to the presence of the native amorphous SiO₂ and the non-optimal ALD growth conditions.²⁷ In comparison, the onset of β-to-α growth transition occurs later on the quartz substrate, which is only observed in the 1500-cycle Bi₂O₃/quartz sample (see Fig. S2, ESI†). These results are consistent with the morphological comparisons shown in Fig. 1.

Fig. 2 (a) XRD (Cu-Kα₁) patterns of the ALD Bi₂O₃/Si thin film samples as a function of growth cycles and (b) Raman spectra collected at room temperature from a Bi₂O₃/quartz sample and a bare quartz substrate. Spectral fittings using Lorentzian functions are also presented for comparison.

Fig. 2(b) presents the typical Raman spectrum taken from a bare quartz substrate and a 1500-cycle Bi₂O₃/quartz sample along with its spectral fittings. The use of a quartz substrate is to avoid the complex photon features of the Si substrate in the studied frequency range. By fitting to the experimental spectrum using Lorentzian functions and comparing to the spectrum of the quartz substrate, we have unambiguously deconvoluted five additional features as indicated by P₁–P₄ and P_a in Fig. 2(b) that originated from the Bi₂O₃ thin film. The P_a feature (at about 205 cm⁻¹), when compared to P₁–P₄, is much larger in linewidth (∼130 cm⁻¹), which is probably a PL-emission feature due to defects and/or Bi-ions/clusters rather than a phonon mode.^35,36 The features of P₁–P₄ are well consistent with those of α-Bi₂O₃ at 101, 151, 314, and 598 cm⁻¹, respectively.³⁶ In contrast, the typical modes of β-Bi₂O₃ at 124, 311, and 462 cm⁻¹ are not distinguishable in Fig. 2(b),³⁷ probably due to the nonstoichiometry-induced disordering and/or the low crystal quality of β-Bi₂O_2.3 that was grown at the initial stage directly on the surface of the quartz substrate. A further comparison with those experimental observations and theoretical calculations reported by Narang et al. in ref. 36 reveals that only P₁, P₂, and P₃ are the Raman active phonon modes out of the 14/22 observed/predicted ones in the frequency range of 100–600 cm⁻¹. This is possibly due to the structural disorders and/or impurity defects of the α-Bi₂O₃ crystals; moreover, the collective growth orientation of the thin film as revealed by XRD discussed above could also result in the limitation of phonon detections in the backscattering configuration.³⁸ The feature of P₄ at ∼598 cm⁻¹ is an infrared active and Raman silent mode; its presence in the Raman spectrum can be attributed to the disorder- and/or defect-induced phonon activation, i.e., the infrared-active to Raman-active transition.^3,39

B. Sulfurizing Bi₂O₃ into Bi₂S₃

Figures 3(a)–(c) present the AFM images recorded after sulfurization at 500 °C from the 500-, 1000-, and 1500-cycle Bi₂O₃/Si samples, respectively. For comparison, an AFM image recorded from the 1500-cycle Bi₂O₃/Si sample after sulfurization at 600 °C is presented in Fig. 3(d). Unlike the as-grown samples (see Fig. 1), after sulfurization, the surface morphologies of the samples on the quartz substrates (see Fig. S3, ESI†) are nearly similar to those of the samples on the Si substrates grown and sulfurized under identical conditions. For the samples sulfurized at 500 °C, it can be observed in Figs. 3(a)–(c) that their morphologies are also quite similar to one another. However, in Fig. 3(d), it is seen that the grain sizes are apparently increased by increasing the sulfurization temperature to 600 °C; the morphology clearly exhibits an onset of surface evaporation. In fact, we have observed that the original Bi₂O₃ films were completely evaporated when the sulfurization temperature was further increased to 700 °C, even under a sulfur-rich condition. This is because of the sublimation behavior and the high vapor pressure of Bi₂S₃ at temperatures higher than 550 °C.⁴⁰

Fig. 3 AFM images recorded from the sulfurized Bi₂S₃ samples, i.e., the Bi₂O₃ thin film samples after sulfurization on Si substrates: (a) 500 cycles sulfurized at 500 °C, (b) 1000 cycles sulfurized at 500 °C, (c) 1500 cycles sulfurized at 500 °C, and (d) 1500 cycles sulfurized at 600 °C.

Fig. 4(a) shows the XRD patterns collected from the sulfurized samples on Si together with that of JCPDS 75-1036 (orthorhombic Bi₂S₃). The excellent match between the experimental spectra and JCPDS 75-1036, in terms of peak positions and intensity ratios, provides clear-cut evidence that orthorhombic Bi₂S₃ were obtained with high phase purities from the ALD grown Bi₂O₃ thin films. It is also seen in Fig. 4(a) that the diffraction intensities increase with the nominal film thickness after sulfurization at 500 °C. However, an increase in the sulfurization temperature to 600 °C turns to reduce the diffraction intensities, which is consistent with the AFM observations (see Fig. 3) and confirms the onset of surface evaporation of Bi₂S₃ at 600 °C. It has to be noted that, after sulfurization, identical XRD patterns of the thin films (i.e., Bi₂S₃) on the Si and quartz substrates are observed (see Fig. S4, ESI†). This is consistent with the morphological similarities observed in Fig. 3 (on Si substrates) and Fig. S2† (on quartz substrates) after sulfurization (see discussions above).

Fig. 4 (a) XRD (Cu-Kα₁) patterns collected from the Bi₂S₃ thin film samples obtained by thermal vapor sulfurizing the ALD grown Bi₂O₃ on Si substrates. The patterns of orthorhombic Bi₂S₃ (JCPDS-ICDD 75-1306) are also presented for comparison. (b) Raman spectra collected at room temperature from the sulfurized Bi₂S₃ thin films grown on quartz substrates using the 532 nm line of a Nd:YAG laser as the excitation source. The spectra of an as-grown Bi₂O₃ sample and a bare quartz substrate are also presented for easier comparison.

Furthermore, to avoid the complicated photon features of Si in the studied frequency range, Raman spectroscopy studies were carried out for the sulfurized thin film samples on the quartz substrates, and the results are shown in Fig. 4(b). For simple comparisons, the Raman spectra collected from an as-grown sample and a bare substrate are also presented. It is seen that, after sulfurization, the typical modes of α-Bi₂O₃ at 314 and 598 cm⁻¹, corresponding to the displacements of oxygen atoms with respect to the bismuth atoms,³⁶ disappeared; instead, five phonon modes (indicated by P^′₁–P^′₅) emerged in the Raman spectra of the samples sulfurized at 500 °C, and an additional mode (indicated by P^′₆) appeared in the spectrum of the sample sulfurized at 600 °C. The modes P^′₁–P^′₅ are consistent with the Raman active phonons of Bi₂S₃ observed by Zhao et al. at 100.0, 168.7, 186.0, 237.1, and 254.5 cm⁻¹, respectively.³ The mode of P^′₆ at 117.0 cm⁻¹ does not correspond to any of the theoretically predicted Raman active modes but well matches the infrared active mode of Bi₂S₃ at 119.3 cm⁻¹.^2,3 We have already mentioned above that structural disorders and/or impurity defects could induce such infrared-active to Raman-active transitions. Since the transition only occurs at a higher sulfurization temperature, it is reasonable that impurity atoms, e.g., Si, were doped into the Bi₂S₃ crystals from the substrate due to the increased sulfurization temperature. Nevertheless, both the XRD and Raman spectroscopy studies show that crystalline orthorhombic Bi₂S₃ thin films with high phase purity have been obtained by sulfurizing the ALD grown Bi₂O₃.

C. Optical properties of Bi₂S₃

Fig. 5(a) shows the absorbance spectra collected from the sulfurized Bi₂S₃ thin films on the quartz substrates. The spectra exhibit a knee-like feature with an exciton resonance absorption peak at 2.96 eV.⁴¹ The values of absorbance, e.g., at the photo energy of 2.96 eV, have a strong linear correlation with the nominal thickness of the thin films. This observation indicates that the surface and the internal light reflections have similar effects on the absorbance measurements due to the similar morphologies after sulfurization, different from those of the as-grown thin films discussed above. In general, the absorption coefficient of a thin film can be determined via measuring the transmittance T, reflectance R, and the film thickness d. In the case of negligible light reflections, the absorption coefficient reads α = ln[1/T]/d;³² however, when only the surface reflection is taken into account, the absorption coefficient reads α = ln[(1 − R)/T]/d; furthermore, when both the surface and the internal reflections are taken into account, the absorption coefficient reads ,^33,42 which is obviously more accurate. To verify the effect of reflections and to measure the absorption coefficient of the resultant Bi₂S₃ with an increased accuracy, we have measured and shown, in Fig. 5(b), the T and R for the Bi₂S₃ thin film obtained by sulfurizing the 1500-cycle Bi₂O₃/quartz sample at 500 °C. The film thickness, d = 70 nm, was measured by AFM from an intentionally scratched area (see Fig. S5, ESI†). In consequence, the absorption coefficient spectra were obtained, as shown in Fig. 5(c), with and without counting the internal light reflections. It is seen that the E_Opt is reduced from 3.0 to 3.5 eV of the as-grown Bi₂O₃ to 1.50 eV after sulfurization. The overestimation of the absorption coefficients in the range of photon energies larger than E_Opt due to the uncalculated internal reflections is less than 10%. The most important observation in Fig. 5(c) is that the optical absorption coefficients in the wavelength range of 400–700 nm are larger than 1.5 × 10⁵ cm⁻¹, which confirms that the resultant Bi₂S₃ could have significant potentials for solar energy conversion applications.

Fig. 5 (a) Optical absorbance spectra measured from the sulfurized Bi₂S₃/quartz samples, (b) reflectance, R, and transmittance, T, of a sulfurized Bi₂S₃/quartz sample with the initial Bi₂O₃ grown for 1500 cycles, and (c) optical absorption coefficients derived from the measured R and T with and without counting the internal light reflections.

Fig. 6 shows the PL spectra collected at low-temperatures from a sulfurized Bi₂S₃ thin film sample. Basically, the decrease in PL emission intensity with the increase in sample temperature is due to the activation of nonradiative recombination centers. The PL peaks are asymmetric and can be deconvoluted into two distinct emissions with the peak splitting of ∼23 meV. In general, these PL emission energies are a few tens of meV smaller than the bandgap energy of E_Opt = 1.50 eV obtained at room temperature [see Fig. 5(c)], which clearly indicate that the PL emissions are primarily transitions associated with shallow defects rather than the band-to-band emissions reported by Sträter et al. in ref. 43. However, the intact PL emission energies as the sample temperature increased from 10 to 60 K and the redshift of ∼10 meV induced by a further temperature increase to 110 K are consistent with the reported temperature-dependent bandgap shift of Bi₂S₃ (see Fig. S6, ESI†).⁴³ It is evident that more work is needed to identify the detailed shallow defects and the recombination mechanisms involved in the PL emissions. Nevertheless, the PL spectroscopy, together with the optical absorption measurements, provides clear evidence that the bandgap energy of the crystalline orthorhombic Bi₂S₃ thin films is about 1.5 eV, which is slightly larger than the predicted value of 1.3 eV.

Fig. 6 Low-temperature PL spectra collected from a sulfurized Bi₂S₃ thin film sample with the initial Bi₂O₃ grown for 1500 cycles on Si substrate.

D. Electronic structures of Bi₂O₃ and Bi₂S₃

Electronic states of the as-grown Bi₂O₃ thin films and the sulfurized Bi₂S₃ ones were further compared using XPS. Fig. 7(a) shows the XPS survey spectra which revealed only O and Bi along with C (due to contaminants from the environment) for the as-grown Bi₂O₃ sample, while S presented in the sulfurized Bi₂S₃ samples together with largely reduced O. The high-resolution XPS spectra of the Bi4f and O1s/S2s core levels are presented in Fig. 7(b) and (c), respectively. It is clearly seen in Fig. 7(b) that the core levels of Bi4f were shifted to lower binding energies by ∼1.1 eV after sulfurization due to the smaller electronegativity of S than that of O. However, it can be observed that, after sulfurization, small shoulders emerged at the higher energy sides of both the Bi4f_5/2 and Bi4f_7/2 peaks. The peak energies of the small Bi4f shoulders of the Bi₂S₃ samples are nearly the same as those of Bi4f of the as-grown Bi₂O₃ sample. Similarly, O1s is also detected in the sulfurized Bi₂S₃ samples and can be deconvoluted into three distinct peaks with one located at 530 eV, i.e., the same binding energy of the lattice O²⁻ of the as-grown Bi₂O₃ [see Fig. 7(c)]. Similar to Bi₂O₃ grown by a solution method,⁵ the presence of O–H (∼531.2 eV) in the as-grown Bi₂O₃, which is significantly reduced in intensity after sulfurization, could be due to the use of a water precursor and the relatively low substrate temperature during the ALD growth in this study. The other O1s peak (∼533 eV) of the sulfurized Bi₂S₃ samples is possibly due to adsorbates on the surface, e.g., chemically adsorbed oxygen, O_ad, at 532.5 eV and H₂O at 533 eV.⁵ On the other hand, the symmetric line shape of the S2s peaks in Fig. 7(c), as well as their binding energy at 225.5 eV, indicates the absence of elemental sulfur in the sulfurized Bi₂S₃ samples. The combination of Bi4f and O1s spectra of the sulfurized samples in Fig. 7(b) and (c) provides evidence for the incorporation of Bi₂O₃ in the Bi₂S₃ samples. It is well known that XPS is a surface sensitive technique, which can only detect the information from a depth of a few nanometers below the film surface. It is thus unclear, whether Bi₂O₃ was formed after sulfurization due to a native oxidation of Bi₂S₃ or it remained in Bi₂S₃ due to a limited sulfurization time. However, as we have discussed above that both XRD and Raman scattering measurements could not detect the presence of Bi₂O₃ in the sulfurized Bi₂S₃ thin films, it is more reasonable to suggest that Bi₂O₃ was formed due to the native surface oxidation of Bi₂S₃.

Fig. 7 XPS spectra collected from the ALD grown Bi₂O₃/Si thin films samples before and after sulfurization: (a) survey spectra, (b) high-resolution spectra of Bi4f and S2p core levels, (c) high-resolution spectra of O1s and S2s core levels, and (d) high-resolution spectra of the valence bands.

Fig. 7(d) shows the valence band spectra collected from the as-grown Bi₂O₃ and the sulfurized Bi₂S₃ thin films. The sulfurization induced increase in the sharpness of the valence band maximum, together with the reduced linewidths of Bi4f in Fig. 7(b), indicates the increased crystallinity and the reduced density of defect states in the bandgap. Linear fittings to the valence band maxima reveal that they have been raised from 2.1 to 1.0 eV below the Fermi level after sulfurization. Taking into account the optical bandgaps of 3.0–3.5 eV [see Fig. 1(b) and (d) and Fig. S7, ESI†] and 1.5 eV for the as-grown Bi₂O₃ and the sulfurized Bi₂S₃ thin films, respectively, we may conclude that they are both n-type semiconductors. These observations clearly indicate that type-I Bi₂S₃/Bi₂O₃ heterojunctions can be formed by partially sulfurizing Bi₂O₃ for effective photocatalyst and/or photoelectrochemical applications.

IV. Conclusions

In conclusion, Bi₂O₃ crystalline thin films were grown on Si and quartz substrates by atomic layer deposition at 300 °C. The films in smaller thickness are dominated by nonstoichiometric β-Bi₂O_2.3, but the component of stoichiometric α-Bi₂O₃ increases with the grown thickness and becomes dominant when the growth period is larger than 1000 cycles on Si. Such a β-to-α growth transition also occurs on a quartz substrate but later than that on the Si substrate. Morphological evolutions of the thin films show that clustering of Bi₂O₃ crystallites occurred when the growth period is increased over 1000 cycles, leading to the larger surface roughnesses. As a result, the effect of surface and internal light reflections on the absorbance measurements significantly varied by the increase in film thickness. Optical bandgaps of 3.0–3.5 eV, depending on the film thickness, were obtained for the as-grown Bi₂O₃ thin films. After sulfurization at elevated temperatures, the Bi₂O₃ thin films were converted into orthorhombic Bi₂S₃, and the surface morphologies became similar in spite of different substrates and/or thicknesses. The optical bandgap was reduced to 1.5 eV after sulfurization, and this value is fairly constant among the samples with different film thicknesses. Absorption coefficients of larger than 10⁵ cm⁻¹ in the wavelength range of visible light are observed for the sulfurized Bi₂S₃ thin films, confirming their significant potentials in solar energy conversion applications. Electronic structural studies, together with the absorption spectroscopies, provide evidence that both the as-grown Bi₂O₃ and the sulfurized Bi₂S₃ are n-type semiconductors with the valence band maxima located at 2.1 and 1.0 eV below their Fermi levels, respectively. These results suggest that type-I Bi₂S₃/Bi₂O₃ heterojunctions can be formed via partially sulfurizing Bi₂O₃ for effective photocatalyst and/or photoelectrochemical applications.

Acknowledgements

The authors would like to thank Lim Poh Chong for his help in XRD data collection and Dr Dong Zhaogang for his help in the Raman scattering experiments.

References

1. H. Koc, H. Ozisik, E. Deligöz, A. M. Mamedov and E. Ozbay, J. Mol. Model., 2014, 20, 2180.
2. I. Efthimiopoulos, J. Kemichick, X. Zhou, S. V. Khare, D. Ikuta and Y. Wang, J. Phys. Chem. A, 2014, 118, 1713.
3. Y. Zhao, K. T. E. Chua, C. K. Gan, J. Zhang, B. Peng, Z. Peng and Q. Xiong, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 205330.
4. D. Becerra, M. T. S. Nair and P. K. Nair, J. Electrochem. Soc., 2011, 158, H741.
5. X. Lu, F. Pu, Y. Xia, W. Huang and Z. Li, Appl. Surf. Sci., 2014, 299, 131.
6. Y. Lu, J. Jia and G. Yi, CrystEngComm, 2012, 14, 3433.
7. Y. Zhao, D. Gao, J. Ni, L. Gao, J. Yang and Y. Li, Nano Res., 2014, 7, 765.
8. Y. Zhao, T. Liu, H. Xia, L. Zhang, J. Jiang, M. Shen, J. Ni and L. Gao, J. Mater. Chem. A, 2014, 2, 13854.
9. G. Li, X. Chen and G. Gao, Nanoscale, 2014, 6, 3283.
10. G. Manna, R. Bose and N. Pradhan, Angew. Chem., 2014, 126, 6861.
11. J. L. T. Chen, V. Nalla, G. Kannaiyan, V. Mamidala, W. Ji and J. J. Vittal, New J. Chem., 2014, 38, 985.
12. I. Zumeta-Dubé, V. Ruiz-Ruiz, D. Díaz, S. Rodil-Posadas and A. Zeinert, J. Phys. Chem. C, 2014, 118, 11495.
13. H. Yu, J. Huang, H. Zhang, Q. Zhao and X. Zhong, Nanotechnology, 2014, 25, 215702.
14. W. Xitao, L. Rong and W. Kang, J. Mater. Chem. A, 2014, 2, 8304.
15. J. Zhou, G. Tian, Y. Chen, Y. Shi, C. Tian, K. Pan and H. Fu, Sci. Rep., 2014, 4, 4027.
16. B. Zhang, X. Ye, W. Hou, Y. Zhao and Y. Xie, J. Phys. Chem. B, 2006, 110, 8978.
17. K. Yao, Z. Zhang, X. Liang, Q. Chen, L. Peng and Y. Yu, J. Phys. Chem. B, 2006, 110, 21408.
18. L. Cademartiri, F. Scotognella, P. G. O'Brien, B. V. Lotsch, J. Thomson, S. Petrov, N. P. Kherani and G. A. Ozin, Nano Lett., 2009, 9, 1482.
19. S. T. Haaf, H. Sträter, R. Brüggemann, G. H. Bauer, C. Felser and G. Jakob, Thin Solid Films, 2013, 535, 394; S. T. Haaf, B. Balke, C. Felser and G. Jakob, J. Appl. Phys., 2012, 112, 053705.
20. X. Yu and C. Cao, Cryst. Growth Des., 2008, 8, 3951.
21. M. E. Rincón, M. Sánchez, P. J. George, A. Sánchez and P. K. Nair, J. Solid State Chem., 1998, 136, 167.
22. J. Lukose and B. Pradeep, Solid State Commun., 1991, 78, 535.
23. H. Zhang, J. Ge and Y. Li, Chem. Vap. Deposition, 2005, 11, 147.
24. J. Waters, D. Crouch, J. Raftery and P. O'Brien, Chem. Mater., 2004, 16, 3289.
25. A. A. Tahir, M. A. Ehsan, M. Mazhar, K. G. U. Wijayantha, M. Zeller and A. D. Hunter, Chem. Mater., 2010, 22, 5084.
26. C. Marichy, M. Bechelany and N. Pinna, Adv. Mater., 2012, 24, 1017.
27. S. M. George, Chem. Rev., 2010, 110, 111.
28. Y. D. Shen, Y. W. Li, W. M. Li, J. Z. Zhang, Z. G. Hu and J. H. Chu, J. Phys. Chem. C, 2012, 116, 3449.
29. H. F. Liu and D. Z. Chi, J. Vac. Sci. Technol., A, 2012, 30, 04D102; H. F. Liu, K. K. Ansah Antwi, S. J. Chua and D. Z. Chi, Nanoscale, 2014, 6, 624; H. F. Liu, K. K. Ansah Antwi, J. F. Ying, S. J. Chua and D. Z. Chi, Nanotechnology, 2014, 25, 405702; H. F. Liu, K. K. Ansah Antwi, N. L. Yakovlev, H. R. Tan, L. T. Ong, S. J. Chua and D. Z. Chi, ACS Appl. Mater. Interfaces, 2014, 6, 3501.
30. B. B. Nayak, S. K. Singh and B. S. Acharya, Phys. Status Solidi A, 1988, 105, 455.
31. H. F. Liu, A. Huang and D. Z. Chi, J. Phys. D: Appl. Phys., 2010, 43, 455405; H. F. Liu, D. Z. Chi and W. Liu, CrystEngComm, 2012, 14, 7140.
32. L. Leontie, M. Caraman, M. Delibaş and G. I. Rusu, Mater. Res. Bull., 2001, 36, 1629.
33. J. I. Pankove, Optical Processes in Semiconductors, Dover, New York, 1971.
34. L. Kumari, J. Lin and Y. Ma, J. Phys.: Condens. Matter, 2007, 19, 406204.
35. B. Ling, X. W. Sun, J. L. Zhao, Y. Q. Shen, Z. L. Dong, L. D. Sun, S. F. Li and S. Zhang, J. Nanosci. Nanotechnol., 2010, 10, 8322.
36. S. N. Narang, N. D. Patel and V. B. Kartha, J. Mol. Struct., 1994, 327, 221.
37. F. D. Hardcastle and I. E. Wachs, J. Solid State Chem., 1992, 97, 319.
38. V. N. Denisov, A. N. Ivlev, A. S. Lipin, B. N. Mavrin and V. G. Orlov, J. Phys.: Condens. Matter, 1997, 9, 4967.
39. H. F. Liu and S. J. Chua, Appl. Phys. Lett., 2010, 96, 091902.
40. V. Piacente, V. D. Gianfreda and G. Bardi, J. Chem. Thermodyn., 1983, 15, 7.
41. H. F. Liu and S. J. Chua, J. Appl. Phys., 2009, 106, 023511.
42. J. I. Pankove, Phys. Rev., 1965, 140, A2059.
43. H. Sträter, S. T. Haaf, R. Brüggemann, G. Jakob, N. Nilius and G. H. Bauer, Phys. Status Solidi B, 2014, 251, 2247–2256.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra09896j

---