IEEE TRANSACTIONS ON ELECTRON DEVICES        1

AlN Thin-Film Vacuum Ultraviolet Photodetector With High Operating Temperature and

High Rejection Ratio

Peixuan Zhang, Kewei Liu , Member, IEEE, Yongxue Zhu, Qiu Ai, Zhen Cheng, Jialin Yang, Xing Chen, Binghui Li, Lei Liu, and Dezhen Shen

Abstract — AlN thin-film vacuum ultraviolet (VUV) pho- todetector was prepared by molecular beam epitaxy (MBE) device on c-Al2O3 substrate. By a complete surface nitrida- tion of sapphire substrate in nitrogen plasma, the epitaxial preparation of high-quality thin AlN film was realized with- out any buffer layer. The AlN photodetector has an ultralow dark  current  (～40  fA  at  ;20  V),  a  high  VUV/ultraviolet-c

(UVC) (R185/R222) rejection ratio (>103), a high responsivity

(30 mA/W), and an ultrafast response (90%–10% decay time ～900 ns) at room  temperature.  More interestingly, an excellent temperature tolerance of the device can be observed, and there is no obvious degradation in the VUV/UVC rejection ratio and response speed with increas-

ing the temperature from 25 ?C to 500 ?C. Even at 500   ?C,

the dark current of the device is only 218 pA at 20 V, and the responsivity can reach to 67.3 mA/W. These results indicate that the device has excellent wavelength selective detection ability and high-temperature detection ability in the VUV band, which can be attributed to the relatively high-quality AlN thin film and the avoidance of the impact of buffer layer. Our findings provide an effective way to realize high-performance AlN VUV photodetector, which can be operated in high-temperature environment.

Index Terms— AlN, high temperature, photodetector, vacuum ultraviolet (VUV).

Manuscript received 15 May  2023;  revised  11  July  2023; accepted 2 August 2023. This work was supported in part by the National Natural Science Foundation of China under Grant 62074148, Grant 11727902, and Grant 12204474; and in part by the National Ten Thousand Talent Program for Young Top-Notch Talents and Youth Innovation Promotion Association, Chinese Academy of Sciences (CAS), under Grant 2020225. The review of this article was arranged by Editor F. Medjdoub. (Corresponding author: Kewei Liu.)

Peixuan Zhang, Kewei Liu, Xing Chen, Binghui Li, Lei Liu, and Dezhen Shen are with the State Key Laboratory of Luminescence and Appli- cations, Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China, and also with the Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing 100049, China 

Yongxue Zhu, Qiu Ai, Zhen Cheng, and Jialin Yang  are  with the State Key Laboratory of Luminescence and Applications, Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China.

Color versions of one or more figures in this article are available at

Digital Object Identifier 10.1109/TED.2023.3303126

I.       INTRODUCTION

Ga2O3 [15], [16], ZnGa2O4 [17], and diamond [18]) have the inherent advantages of high thermal stability and high radia- tion resistance, which are considered to be ideal candidates for ultraviolet (UV) photodetection  in  harsh  environments. In spite of some progress, the reported photodetectors based on WBG semiconductor cannot meet the requirements for selective and accurate detection of VUV light. This is mainly because their relatively narrow bandgaps make the devices still have strong response to the light with a wavelength longer than 200 nm. As the material with the largest bandgap (about 6.2 eV at room temperature) among III–V nitride semiconductors, AlN has good chemical stability, high mobility, high   thermal

conductivity (2.85 W/cm·K, about twice that of Si), and strong radiation resistance, which is very suitable for VUV  detection

in harsh environments [19], [20], [21], [22], [23], [24],   [25].

So far, due to the lack of high-quality AlN single-crystal substrate, the reported AlN films are commonly heteroepitaxial on other substrates, including c-Al2O3 [26], [27],  Si  [28], GaN [29], SiC [30], and so on. Among them, c-Al2O3 is currently the most widely used AlN heteroepitaxy substrate for the VUV detection application because of its low price and excellent thermal stability, and a AlN buffer layer has been commonly utilized to obtain high-quality AlN epitaxial film. For example, Li et al. [20] prepared a VUV photode- tector with metal–semiconductor–metal (MSM) structure by growing 1.5-µm-thick AlN on AlN buffer/sapphire substrate using metal–organic chemical vapor deposition (MOCVD). When the applied bias is  100  V,  the  peak  responsivity  of the device at 200 nm is 0.4 A/W. BenMoussa et al. [24] demonstrated a similar MSM VUV photodetector on AlN (thickness: 1.06 µm)/AlN buffer/sapphire by MOCVD, and the

2        IEEE TRANSACTIONS ON ELECTRON DEVICES

device has an ultralow dark current of 0.2 pA (15 V) at 25 ?C. However, due to the subbandgap absorption of low-quality AlN buffer layers, the VUV/ultraviolet-c (UVC) rejection ratio of such devices is generally not high. Nevertheless, due to the relatively large lattice mismatch and thermal mismatch between AlN and sapphire, obtaining high crystalline quality AlN films without buffer layers and their high-performance photodetectors still face significant challenges [31],  [32].

Compared with the MOCVD technology, molecular beam epitaxy (MBE) has unique advantages in the preparation of AlN thin films (such as high vacuum growth environment, low growth temperature, and so on), so it is expected to achieve the epitaxial preparation of high-quality thin-layer AlN thin films  without  buffer  layer.  In  addition,  no  information can

be found about the AlN VUV photodetectors operating at temperature higher than 300 ?C. In this  work, high-quality AlN thin film was fabricated on c-Al2O3 through full nitriding technique by MBE without any buffer layer. Based on this

AlN film, MSM structure photodetector has been demon- strated and investigated at various temperatures from 25 ?C to 500 ?C. At 25 ?C, AlN photodetector has an ultralow dark current (～40 fA at 20 V), a high VUV/UVC rejection ratio (R185/R222 > 103), and an ultrafast response speed (90%–10% decay time ～900 ns). Moreover, even at 500 ?C, the dark current of the device is still only 218 pA at 20 V with a VUV/UVC rejection ratio of more than 103 and a responsivity

of 67.3 mA/W, indicating the excellent VUV photoelectric detection ability at high temperature. The relatively high crystalline quality of AlN film and the absence of low-quality buffer layer can be responsible for the high performance of our device in a wide range of temperatures. This work provides a feasible way to prepare high-performance AlN VUV detectors that can work in harsh  environment.

II.       MATERIAL EPITAXY AND DEVICE  FABRICATION

Fig. 1.  (a) RHEED pattern of AlN film. (b) (0002) plane XRC of AlN film.

force microscopy (AFM) (Bruker  Multimode-8)  were  used to investigate the optical, structural, and surface morpholog- ical properties of the AlN film. The current–voltage (I –V ) properties and time-dependent photocurrent (I –t ) properties of the device were measured using semiconductor device analyzer (Agilent B1500A), 185-nm lamp (ZW21D15W/Y), and microprobe hot platform (KT-Z1000MR4T) under differ- ent temperatures. The measurements of temporal-dependent responses were performed by an ArF excimer laser (CL-5100, 193 nm) and a digital oscilloscope (Tektronix DPO   5104).

III.       RESULTS AND DISCUSSION

AlN thin film was heteroepitaxially grown on c-plane   sap-

formula [33]: ρs  = β2

/(4.35 · |bc | ), where ρs is the screw

phire by MBE. 6N-pure Al obtained from a standard  Knudsen

effusion cell and 6N-pure N2 activated by a radio frequency (RF) plasma source were chosen to be used as the precursors. Prior to growth, the substrate was pretreated at 700 ?C for 2 h to remove gas and organic contaminants attached to the surface. After that, the substrate was nitrized completely with an RF plasma power of 350 W and a nitrogen flow rate of

1.8 sccm. Subsequently, the AlN thin film was grown at a substrate temperature  of 780 ?C and  a chamber  pressure   of 1 × 10?6 Torr. The temperature of the Al source was fixed at 1180 ?C, and the flow rate of N2  was kept at 0.7 sccm

with an RF power of 350 W. In situ monitoring of the film during nitriding and growth was carried out by reflection high-energy electron diffractometer (RHEED). AlN MSM photodetector with Pt interfinger electrodes was fabricated by lithography and magnetron sputtering techniques. Finally, the device was subjected to rapid thermal annealing for 10 min  in

Ar atmosphere at 600 ?C to improve its   stability.

UV spectrophotometer (Shimadzu UV-3101 PC scanning spectrophotometer), Raman shift spectrum, high-resolution X-ray diffraction (HRXRD) (Bruker D8 Discover), scanning electron  microscopy  (SEM)  (Hitachi  S-4800),  and    atomic

dislocation density, β(0002)  is the FWHM of the (0002)   plane,

and bc stands for the Burgers vectors of the c-axial lattice constants of the AlN (bc = 0.4978 nm). The screw dislocation density of our AlN thin film in this work was calculated to be 5.9 × 107   cm?2.

Fig. 2(a) shows the room temperature Raman  spectroscopy

of AlN film with 532 nm laser excitation. The peak located at 657 cm?1 can be assigned to E 2(high) mode of AlN and  the Eg mode peak of sapphire at 751 cm?1 is used for wavenumber calibration. Interestingly, the position of E 2(high) peak is highly sensitive to the stress in AlN [34], and the residual

stress (σ ) in AlN film can be evaluated from the shift of Raman peak position: σ = 6ωE2/k, where 6ωE2 is the difference of E 2(high) mode Raman shift  between stressed and unstressed AlN samples and k is the stress coefficient of AlN. According to the previous reports, the values of 6ωE2

and k are 657.4 cm?1  and 2.4 cm?1GPa?1  for freestanding

AlN, respectively [34], [35], [36]. Therefore, the stress in our AlN thin film is calculated to be ～0.15 GPa. The optical transmission spectrum of the sample is shown in Fig. 2(b). The average transmittance of the grown AlN film is over 90% in the wavelength range of 210–700 nm. The bandgap of  AlN

ZHANG et al.: AlN THIN-FILM VUV PHOTODETECTOR        3

Fig. 2.  (a) Raman spectrum of AlN thin film. (b) Absorption  spectrum of AlN film in the 190–700 nm wavelength range at room temperature. The inset shows the variation of (αhν)2  versus the photoenergy (hν).

(c) SEM and (d) AFM images of AlN thin film on c-Al2O3.

film is determined to be about 6.2 eV by Tauc plot in the inset of Fig. 2(b). The optical transmission results indicate that AlN thin film is very suitable for intrinsic VUV   detection.

The top-view and cross-sectional SEM images of AlN thin film are shown in Fig. 2(c), and it can be found that the thickness of the AlN film is only about 100 nm with a uniform and smooth surface. Fig. 2(d) presents the AFM image of AlN film. A root mean square roughness (RMS) of 0.7 nm can be clearly observed within the 4 µm2  scanning area,   suggesting a very flat surface of AlN  film.

To investigate the VUV photodetection property of  AlN thin film, a Pt–AlN–Pt photodetector has been demonstrated with interdigital Pt electrodes (500 µm long, 10 µm width, and  10  µm  gap).  In  addition,  a  system  consisting  of       a

small vacuum  chamber  (～1  Pa)  and  a  heating  platform (25  ?C–500  ?C) was  used  to  study  the  performance  of the

device in high-temperature environment, as shown in Fig. 3(a). The I –V characteristic curves of the device in the dark and under 185-nm illumination were measured at different temper- atures, as shown in Fig. 3(b) and (c), respectively. When the temperature is below 200 ?C, the dark current of the device is only about tens of femtoamperes at 20 V. As the temperature rises from 200 ?C to 500 ?C, the dark current increases rapidly, and it can reach 218 pA at 500 ?C (20 V). Moreover, under 185-nm irradiation (the optical power density: ～35 µW/cm2), the current of the device shows a slight increase from 5.3 to 12 nA with the temperature increasing from 25 ?C to 500 ?C at 20 V. The increase in the semiconductor absorption coef- ficient may be the reason for the larger light current of the device at higher temperature [10], [37], [38], [39]. In addition, the I –t characteristics of the device were tested by periodically

switching ON/OFF the 185 nm lamp at different temperatures at 20 V. As  shown  in  Fig. 3(d),  the  periodic photoresponse of the device has excellent stability and repeatability at all temperatures. Even at high temperature of 500 ?C, the device still has a reliable and sensitive response to VUV   light.

Responsivity (R) is an important parameter for describing the  photoelectric  conversion  ability  of  a  device,  which  ; is

Fig. 3. (a) Variable-high-temperature photoresponse measurement system. (b) I–V characteristics of the device under 185 nm illumination at different temperatures. (c) I–V characteristics of the device in the dark state at different temperatures. (d) Time-dependent photocurrent of the device under the illumination of 185 nm light at different temperatures.

Fig. 4. (a) Responsivity of the device as a function of wavelength at different temperatures under 20 V bias. (b) Responsivity at 185 nm and rejection ratio (R185/R222) of the device as a function of temperature under 20 V bias.

expressed by the formula as [40]: R = (Ilight ? Idark)/( popt · S), where Ilight is the current under illumination, Idark is the dark current, popt is the illumination intensity, and S is the efficient area. Fig. 4(a) presents the dependence of the responsivity on the wavelength of the incident light at  25  ?C and  500 ?C. The device has the maximum responsivity at 185 nm with a VUV/UVC (R185/R222) rejection ratio of over 103  at both

25  ?C  and  500  ?C,  indicating  the  excellent  intrinsic VUV

spectral selectivity. The responsivities under 185 nm light irradiation at 25 ?C and 500 ?C are 30 and 67.3 mA/W, respectively. It should be mentioned here that the responsivity of the device to light with wavelengths greater than 310 nm is lower than the instrument limit  of  ～1  ×  10?7  A/W, which  means  the  VUV/UVB  (R185/R310)  rejection  ratio  is

higher than 105. Notably, as the temperature increases from

25 ?C to 500 ?C, the responsivity of the device increases obviously, while the VUV/VUC (R185/R222) rejection ratio remains almost unchanged, as shown in Fig.   4(b).

To further investigate the response speed of the device, the temporal-dependent response of the AlN photodetector was tested using a pulsed ArF excimer laser with a wavelength of 193 nm  (a  laser  pulsewidth  of  10  ns  and  a  frequency of 100 Hz). The response time is defined as the time inter- val between the amplitude of 90% and 10%. As shown in

TABLE I

COMPARISON TABLE FOR PERFORMANCE PARAMETERS OF VUV PHOTODETECTORS BASED ON ALN SINGLE-CRYSTALLINE FILMS

Fig. 5. (a) Transient photo-response upon a 193 nm pulsed laser excitation. (b) 90%–10% decay time of the device at different operating temperatures.

Fig. 6. Time-dependent photocurrent of the device under the illumina- tion of 185 nm light at 500 ?C.

Fig. 5(a), the 90%–10% decay time of the device is as short as around 900 ns both at 25 ?C and 500 ?C with a load resistance (Rload) of 1 kK, which is much quicker  than the most reported VUV photodetectors [20], [21], [26], [41], [43].

Fig. 5(b) shows the temperature dependence of the 90%–10% decay time of the device. It can be clearly observed that with increasing the operating temperature, the response speed of this AlN MSM VUV photodetector is almost   unchanged.

Long-term stability at high temperature is also extremely critical  for  the  practical  ;application   of   a  photodetector. As  shown  in  Fig. 6,  at  500  ?C,  the  photoelectric response

characteristics of AlN VUV photodetector did not show any change after 3600 s, indicating that the device can work stably for a long time in a high-temperature   environment.

Table I summarizes the key parameters of the reported VUV photodetectors  based  in  AlN  single-crystalline  films.  It can

be seen that our AlN thin-film device without  any  buffer layer exhibits the largest VUV/UVC rejection ratio and the highest operating temperature among the reported AlN VUV photodetectors. In addition, the dark current, responsivity, and response speed of our device are comparable with the best values obtained for AlN-based VUV photodetectors. More importantly, with increasing the operating temperature, the main performance of our AlN detector does not exhibit sig- nificant degradation. Even at high temperatures of 500 ?C, the device still shows excellent detection capabilities, and the main reason for these super VUV photodetection performance in this work (such as low dark current, large VUV/UVC rejection ratio, quick response speed, and high operating temperature) should be associated with the absence of poor-quality buffer layer and the relatively low defect density in AlN thin    film.

IV.       CONCLUSION

In summary, 100-nm-thick AlN thin film was prepared on fully nitride c-Al2O3  substrate using MBE without any  buffer,

and a low screw dislocation defect density of ～5.9 × 107 cm?2  can be estimated by XRC result. The Pt–AlN–Pt   MSM

photodetector was demonstrated and investigated at various operating temperatures. At 25 ?C, the device shows an ultralow dark current of ～40 fA and a high responsivity of 30 mA/W at 20 V. As the temperature rises to 500 ?C, the dark current

of the device only increases to 218 pA, while the responsivity increases to 67.3 mA/W. In addition, the device has excellent repeatability and stability, and no obvious degradation can be observed in the VUV/UVC rejection ratio (～103) and response

speed (90%–10% decay time ～900 ns) with increasing the

temperature from 25 ?C to 500 ?C. The relatively high crys-

talline quality of AlN thin film and the absence of any buffer during the growth should be responsible for the excellent performance of our AlN VUV photodetector. Our findings in this work indicate that AlN materials have broad application prospects for VUV photodetection in harsh  environments.