ZnO–WS₂ heterostructures for enhanced ultra-violet photodetectors

Abstract

Two-dimensional (2D) materials have attracted wide attention due to their exotic properties. In particular, the lack of dangling bonds makes it possible to build highly lattice mismatched heterostructures composed of 2D materials and conventional semiconductors. Here, we report that by simply stacking a chemical vapor deposition grown monolayer WS₂ film onto the surface of a room temperature sputtered ZnO film, significant enhanced ultra-violet (UV) photoresponse can be achieved. In this heterostructure of ZnO–WS₂, the ZnO film acts as a light harvesting layer while the WS₂ monolayer functions as a carrier transport layer which facilitates the photocarrier transport and reduces its recombination. Such a mechanism was confirmed by the observation of further photoresponsivity improvement of the ZnO–WS₂ heterostructure under vacuum which removes the surface absorbates and thereby increases the carrier mobility of WS₂. The strategy presented here can be applied to other wide band-gap semiconductors, shedding light on high sensitivity and flexible UV photodetectors based on van der Waals heterostructures.

---

1. Introduction

Ultra-violet (UV) photodetectors have many applications including flame detection, space-to-space communication, missile plume detection, astronomy and biological research.¹ As a low cost UV sensing semiconductor, single crystalline ZnO nanostructures have been widely studied for UV photodetectors by virtue of their less defect-induced photocarrier recombination and high gain.^2–4 However, the integration challenge limits their application. In this regard, ZnO thin films still have their advantages for scalable UV photodetectors. Different from the efforts on increasing film crystallinity, here, we alternatively demonstrate a strategy of improving UV photoresponse by fabrication of a transition metal dichalcogenide (TMDC)-oxide thin film heterostructure, which was made by simply stacking a high quality (chemical vapor deposition) CVD-grown monolayer semiconducting WS₂ film onto the surface of a room temperature sputter-deposited ZnO film. In this heterostructure, the ZnO functions as a UV absorbing and charge carrier generating matrix, while the CVD-grown WS₂ monolayer acts as a charge transport channel. In this way, photocarriers generated in ZnO can transfer to WS₂ which has a higher mobility and thereby facilitate the UV response. Such a mechanism was further confirmed by the investigation of the device response under vacuum. We note that by means of the Schottky barrier between ZnO and graphene, photocurrent and response speed also showed significant improvement.⁵ However, this requires a special device structure and complex fabrication techniques to avoid electric short problems. By stacking graphene directly onto the surface of ZnO nanorods, the ZnO nanorod/graphene composite working in a photoconductive mode also shows enhanced photoresponsivity.⁶ However, the semi-metal nature of graphene leads to orders of magnitude increasing of the dark current, which is detrimental to the discrimination of a photodetector. Here, in our work, the ZnO–WS₂ heterostructure shows much lower dark current thanks to the semiconducting nature of monolayer TMDC material. We believe that the strategy that we demonstrate here can be applied to other wide band-gap semiconductors, shedding light towards low cost, high performance and flexible van der Waals heterostructure photodetectors.

2. Experimental section

The ZnO films with thickness of 100 nm were deposited on glass substrate at room temperature by reactive radio frequency magnetron sputtering technique using a metal target.⁷ The sputtering powder, Ar/O₂ ratio, and deposition pressure were 200 W, 4 : 1, and 0.4 Pa, respectively. No intentional optimization procedure towards improving the film crystallinity was carried out. Monolayer WS₂ films were synthesized by CVD via our previous report method.⁸ After the deposition, WS₂ films were transferred onto the surface of ZnO films by the surface-energy assisted transfer technique using polystyrene as coating material.⁹ The Al interdigital electrodes with 80 μm gap distance were deposited by thermal evaporation using a shadow mask. The detailed fabrication process is illustrated in Fig. 1.

Fig. 1 Schematic of the device fabrication process.

The surface morphologies of the heterostructures were investigated by a atomic force microscopy (AFM, Veeco Multimode). The absorbance spectra were measured by a modular spectrometer (BLK-C-SR, Stellar Net Inc.) using a deuterium halogen light source (SL50CUV, Stellar Net Inc.). The electric characteristics were measured in a home-made vacuum probe station with an Agilent B2902A source unit. Monochromatic illumination was provided by a Zolix Omni-300 monochrometer with a Xenon lamp (150 W) light source. The illumination light power was measured using a Newport 1935C power meter.

3. Results and discussion

The heterostructure can be clearly seen from AFM image shown in Fig. 2a as indicated in the figure. Since the electrode contact property greatly affects the photocarrier transport of a photodetector, we investigate the current vs. voltage (I–V) response of the as-fabricated ZnO and ZnO–WS₂ heterostructure devices before applying light illumination. As shown in the inset of Fig. 2b, the near linear dependence of the current on the applied voltage indicates Ohm contacts are formed for both devices, which means our photodetectors work in a photoconductive mode. In addition, ZnO–WS₂ heterostructure shows a slightly higher dark current due to the relatively higher conductivity of the heterostructure than the as-deposited pure ZnO film. Under UV light illumination (340 nm), both devices show apparent photoresponse. However, the current increase of the ZnO–WS₂ heterostructure is much larger than that of pure ZnO film. The enhanced photoresponse can be also confirmed by the light-intensity-dependent photocurrent (I_photocurrent = I_current − I_dark) measurement, as shown in Fig. 2c. For example, at a bias voltage of 8 V and light intensity of 18.2 μW cm⁻², the photocurrent of the ZnO–WS₂ device reaches to 810 pA (corresponding to a responsivity of 2.42 mA W⁻¹), while for the ZnO device the photocurrent is only 110 pA (corresponding to a responsivity of 0.34 mA W⁻¹), almost 8 times increase after simply stacking a monolayer WS₂ onto the surface of ZnO film.

Fig. 2 Photoresponse performance of the devices in air. (a) AFM image of the heterostructure. (b) Current vs. voltage (I–V) plots of the devices under dark and light illumination conditions. The wavelength and intensity of the incident light are 340 nm and 18.2 μW cm⁻², respectively. The inset shows the enlarged I–V curves in dark condition. (c) Photocurrent as a function of light intensity. The wavelength of incident light is 340 nm and the bias voltage is 8 V. The inset of (c) shows a photo image of the ZnO–WS₂ device on glass substrate. (d) Wavelength-dependent photoresponse of the devices. The bias voltage is 8 V and the light intensity for each wavelength is kept at 5.2 μW cm⁻². The inset shows the enlarged part of the photocurrent at visible range.

To get more detailed photoresponse properties of the devices, the wavelength-dependent photoresponse spectra of the devices were also measured (Fig. 2d). We can clearly see that near seven times UV response enhancement can be obtained by forming ZnO–WS₂ heterostructure. For the typical heterostructure device, the photocurrent ratio between UV (340 nm) and visible (520 nm) is about 37, which suggests heterostructure is more UV sensitive. We also find that the ZnO film shows only UV response with a band-gap cutoff wavelength around 380 nm. In contrast, for the ZnO–WS₂ heterostructure, the photoresponse extend to visible region with a cutoff wavelength at 650 nm corresponding to the direct band-gap absorption of monolayer WS₂ (1.9 eV). In addition, except the main photoresponse peaks around 380 nm, two additional peaks located at 620 and 520 nm can be also clearly seen in the visible region as shown in the inset of Fig. 2d. This visible photoresponse is originated from monolayer WS₂ because ZnO does not show any visible light photoresponse. The positions of the two peaks are in good agreement with the A- and B-excitons of monolayer WS₂ as a result of valence band splitting at the K point caused by spin–orbit splitting and the absence of inversion symmetry.¹⁰ The observation of these exciton peaks in photocurrent spectrum suggests the high quality of the CVD-grown monolayer WS₂.¹¹ Considering the practical using as a visible-blind UV photodetector, adding an optical filter as the case in commercial available silicon UV photodetector can remove the visible response.

To understand the mechanism of this enhancement, we further measured the absorption spectrum of ZnO film and ZnO–WS₂ heterostructure respectively, as shown in Fig. 3a. We can see that compared with the pure ZnO film, the UV absorbance increases a little after adding a monolayer WS₂ onto the ZnO film while the visible absorbance of the heterostructure get apparently more increase. The excitonic peaks of monolayer WS₂ can be also clearly observed with peak positions centered at 519 nm and 624 nm as shown in the inset of Fig. 3a. Over all, those features well match the photocurrent spectra shown in Fig. 2d. We find that although both ZnO and monolayer WS₂ are direct band-gap UV sensing semiconductors, the enhancement cannot be attributed to the sum of photocurrent from ZnO and from monolayer WS₂ respectively (ESI, Fig. S3†). Additional photocurrent enhancement mechanism must be involved after forming ZnO–WS₂ heterostructure.

Fig. 3 (a) Absorption spectra of ZnO film and ZnO–WS₂ heterostructure. (b) Schematic of the band diagram of the ZnO–WS₂ heterostructure. E_c and E_v denote the conduction band and valence band, respectively. (c) Schematic drawing shows charge generation and transfer process upon light illumination on the ZnO–WS₂ heterostructure.

We find that the photoresponse enhancement can be well explained with the classic gain generation mechanism inside a photoconductor. Generally, the photocurrent of a photoconductor is proportional to photo-gain (G), and the G can be written as:¹²

G = τμ_eE/w (1)

where τ is the excess-carrier recombination lifetime, μ_e is the mobility of electron, E is the external electric field, w is the width between two electrodes. According to eqn (1), long recombination time and large electron mobility lead to large gain. Upon the incidence of photons above the band-gap of ZnO, most of the photons are absorbed by ZnO, considering the ultrathin nature of monolayer WS₂ and the ignorable light absorption compared with ZnO as shown in Fig. 3a. The valence band electrons are excited to the conduction band, generating electron–hole pairs in ZnO. Because the conduction band of WS₂ is lower than that of ZnO as shown in Fig. 3b,^13,14 photo-generated electrons can roll down to the conduction band of WS₂ as depicted. Therefore, the photocarrier can be energetically transferred to WS₂. The ZnO film used in our experiment has very low electron mobility, which shows no field effect based on our fabricated field effect transistor. While monolayer WS₂ has an electron mobility of 0.91 cm² V⁻¹·s⁻¹ derived from the transfer curve of our reported WS₂ field effect transistor.⁸ Therefore, the electron mobility of monolayer WS₂ is much larger than that of ZnO film. Due to the larger electron mobility for monolayer WS₂ compared with as-deposited ZnO, the electrons drift faster in monolayer WS₂ than that in ZnO (v_e = μ_eE, where v_e is the velocity of electron), and can be rapidly collected by the electrodes. In addition, substantial defects in the room temperature deposited ZnO film can act as recombination centers for photocarriers (photo-generated electron–hole pairs), resulting in short recombination lifetime in ZnO. In contrast, for the monolayer WS₂ prepared at a high temperature, which has a better crystallinity and less defects (see Characterization section in the ESI†), the number of recombination centers significantly reduced compared with ZnO, resulting in a longer recombination lifetime in monolayer WS₂. Therefore, the photo-generated electron can be transported more efficiently in monolayer WS₂. In other words, that is to say stacking a monolayer WS₂ onto the surface of ZnO film increases the gain according to eqn (1). Eventually, the photocurrent gets much enhanced as shown in Fig. 2c. Here, we can see that ZnO acts as photon absorption layer while monolayer WS₂ functions as charge transport layer as depicted in Fig. 3c. In this heterostructure, the separation of photon absorption and charge carrier collection successfully increases the detecting performance by fully utilizing the key functionality of each individual layer. Conducted by this mechanism, we find that similar enhancement can be also obtained on a sputter-deposited MgZnO film, which is an n-type wide band-gap materials with a deeper UV response than that of ZnO, as indicated in the ESI (Fig. S4†).

To further confirm the above-proposed mechanism, the response speeds of photodetector working in air and vacuum are measured as shown in Fig. 4. As can be seen from Fig. 4a, both ZnO film and ZnO–WS₂ show repeatable switch behavior upon chopped light. Usually, the time needed for the current to increase from 10% to 90% of the peak value and vice versa is defined as the rise time and decay time, respectively. The rise and decay time of ZnO film is 0.4 and 1 s, respectively, which is much shorter than that of ZnO–WS₂ heterostructure with rise time of 5.7 s and decay time of 2.6 s. It is known that the photocurrent decay time of a photoconductive detector is approximately equal to the recombination time τ. Therefore, these results suggest that the photo-carrier recombination time in ZnO–WS₂ film is longer than that in ZnO film, which is consistent with the above-proposed mechanism. Because the electronic properties of ZnO and WS₂ are both very sensitive to ambient, the time-dependent current with chopped light in vacuum were also measured as shown in Fig. 4d–f. The rise time and decay time (6 and 4.2 s, respectively) of ZnO–WS₂ film are longer than that of ZnO film (1 and 1.4 s, respectively), which is consistent with the case in air. Note that the rise time and decay time in vacuum for both of the devices are longer than those in air. The increase of rise and decay time suggests that the recombination time in vacuum is prolonged compared with that in air. It is known that surface adsorbed gas molecules can act as recombination centers for photo-generated electron–hole pairs. So the longer rise and decay time in vacuum can be ascribed to the reduced adsorbed gases in vacuum. The longer recombination time should lead to a larger photocurrent according to eqn (1). The photocurrent of ZnO device in air and in vacuum is 114 and 147 pA, respectively, which is consistent with theory. The photocurrent of ZnO–WS₂ heterostructure device in vacuum is 2.3 nA, which is much larger than that in air (0.8 nA). The photo-detecting ability of the heterostructure is much enhanced in vacuum, about 16 times of that of ZnO film. Except the longer recombination time, the increased electron mobility should also contribute to the increase of photocurrent according to eqn (1) since WS₂ has a larger electron mobility in vacuum than that in air.⁸ Considering the practical using, a surface passivation technique such as dielectric coating by atomic layer deposition can be employed to eliminate the surface adsorbate-sensing photo-detecting effect and thereby obtain a stable photoresponse.

Fig. 4 Time-dependent current with chopped light (340 nm). (a–c) in air; (d–f) in vacuum (0.1 Pa). The bias voltage is 8 V and the light intensity is 18.2 μW cm⁻².

4. Conclusions

By simply stacking a CVD-grown monolayer WS₂ film onto the surface of a polycrystalline ZnO film, we achieved nearly seven times enhancement of the UV photoresponse in air sixteen times enhancement in vacuum compared to the bare ZnO film. In this ZnO–WS₂ heterostructure, ZnO acts as a photo absorption layer and the monolayer WS₂ functions as an electron transport layer. The photo-generated electron was transferred to a high quality WS₂ and collected more efficiently by virtue of its high electron mobility. This mechanism was further confirmed by the response time measurement. We believe the strategy can be extended to other photo-sensing materials for the implementation of enhanced photoresponse.

Acknowledgements

This work was supported by the Natural Science Foundation of China (grant No. 61421002, No. 61106040 and No. 61475030), the Program for New Century Excellent Talents in University (grant No. NCET-13-0092), the State Key Laboratory of Electronic Thin Film and Integrated Device Program (No. KFJJ201408), and the Central University Basic Scientific Research Business Expenses (No. ZYGX2013J060 and No. ZYGX2013J061).

References

1. L. Sang, M. Liao and M. Sumiya, A Comprehensive Review of Semiconductor Ultraviolet Photodetectors: From thin film to one-dimensional nanostructures, Sensors, 2013, 13, 10482.
2. K. Liu, M. Sakurai and M. Aono, ZnO-Based Ultraviolet Photodetectors, Sensors, 2010, 10, 8604.
3. C. Soci, A. Zhang, B. Xiang, S. A. Dayeh, D. Aplin, J. Park, X. Bao, Y. H. Lo and D. Wang, ZnO Nanowire UV Photodetectors with High Internal Gain, Nano Lett., 2007, 7, 1003.
4. Z. Guo, S. Wei, B. Shedd, R. Scaffaro, T. Pereira and H. T. Hahn, Particle Surface Engineering Effect on the Mechanical, Optical, and Photoluminescence Properties of ZnO/Vinyl-Ester Resin Nanocomposites, J. Mater. Chem., 2007, 17, 806.
5. X. W. Fu, Z. M. Liao, Y. B. Zhou, H. C. Wu, Y. Q. Bie, J. Xu and D. P. Yu, Graphene/ZnO Nanowire/Graphene Vertical Structure Based Fast-Response Ultraviolet Photodetector, Appl. Phys. Lett., 2012, 100, 223114.
6. H. Chang, Z. Sun, K. Y. F. Ho, X. Tao, F. Yan, W. M. Kwok and Z. Zheng, A Highly Sensitive Ultraviolet Sensor Based on a Facile in situ Solution-Grown ZnO Nanorod/Graphene Heterostructure, Nanoscale, 2011, 3, 258.
7. C. Lan, B. Lin, Y. Jiang and C. Li, Optical Properties of (100) Oriented ZnO: Gd Films Deposited by Reactive Radio Frequency Magnetron Sputtering, Mater. Lett., 2014, 132, 116.
8. C. Lan, C. Li, Y. Yin and Y. Liu, Large-Area Synthesis of Monolayer WS₂ and Its Ambient-Sensitive Photo-Detecting Performance, Nanoscale, 2015, 7, 5974.
9. A. Gurarslan, Y. Yu, L. Su, Y. Yu, F. Suarez, S. Yao, Y. Zhu, M. Ozturk, Y. Zhang and L. Cao, Surface-Energy-Assisted Perfect Transfer of Centimeter-Scale Monolayer and Few-Layer MoS₂ Films onto Arbitrary Substrates, ACS Nano, 2014, 8, 11522.
10. Z. Zhu, Y. Cheng and U. Schwingenschlögl, Giant Spin–Orbit-Induced Spin Splitting in Two-Dimensional Transition-Metal Dichalcogenide Semiconductors, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 153402.
11. I. Bilgin, F. Liu, A. Vargas, A. Winchester, M. K. L. Man, M. Upmanyu, K. M. Dani, G. Gupta, S. Talapatra, A. D. Mohite and S. Kar, Chemical Vapor Deposition Synthesized Atomically Thin Molybdenum Disulfide with Optoelectronic-Grade Crystalline Quality, ACS Nano, 2015, 9, 8822.
12. B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, Wiley, New York, NY, 2007.
13. C. Gong, H. Zhang, W. Wang, L. Colombo, R. M. Wallace and K. Cho, Band Alignment of Two-Dimensional Transition Metal Dichalcogenides: Application in Tunnel Field Effect Transistors, Appl. Phys. Lett., 2013, 103, 053513.
14. Y. I. Alivov, E. V. Kalinina, A. E. Cherenkov, D. C. Look, B. M. Ataev, A. K. Omaev, M. V. Chukichev and D. M. Bagnall, Fabrication and Characterization of n-ZnO/p-AlGaN Heterojunction Light-Emitting Diodes on 6H-SiC Substrates, Appl. Phys. Lett., 2003, 83, 4719.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra12643j

---