⇦Chinese Physics Letters, 2018, Vol. 35, No. 12, Article code 127302⇨ Performance Improvement in Hydrogenated Few-Layer Black Phosphorus Field-Effect Transistors * He-Mei Zheng(郑和梅), Shun-Ming Sun(孙顺明), Hao Liu(刘浩), Ya-Wei Huan(还亚炜), Jian-Guo Yang(杨建国), Bao Zhu(朱宝), Wen-Jun Liu(刘文军)**, Shi-Jin Ding(丁士进)** Affiliations State Key Laboratory of ASIC and System, School of Microelectronics, Fudan University, Shanghai 200433 Received 20 August 2018, online 23 November 2018 ^*Supported by the National Natural Science Foundation of China under Grant Nos 61474027 and 61774041.
^**Corresponding author. Email: wjliu@fudan.edu.cn; sjding@fudan.edu.cn Citation Text: Zheng H M, Sun S M, Liu H, Hai Y W and Yang J G et al 2018 Chin. Phys. Lett. 35 127302 Abstract A capping layer for black phosphorus (BP) field-effect transistors (FETs) can provide effective isolation from the ambient air; however, this also brings inconvenience to the post-treatment for optimizing devices. We perform low-temperature hydrogenation on Al$_{2}$O$_{3}$ capped BP FETs. The hydrogenated BP devices exhibit a pronounced improvement of mobility from 69.6 to 107.7 cm$^{2}$v$^{-1}$s$^{-1}$, and a dramatic decrease of subthreshold swing from 8.4 to 2.6 V/dec. Furthermore, high/low frequency capacitance–voltage measurements suggest reduced interface defects in hydrogenated BP FETs. This could be due to the passivation of interface traps at both Al$_{2}$O$_{3}$/BP and BP/SiO$_{2}$ interfaces with hydrogen revealed by secondary ion mass spectroscopy. DOI:10.1088/0256-307X/35/12/127302 PACS:73.63.Bd, 73.22.-f, 85.30.Tv, 85.35.-p © 2018 Chinese Physics Society Article Text Black phosphorus (BP) has attracted intensive attention due to its unique properties such as the tunable direct bandgap from 0.3 to 1.8 eV, high mobility up to 5000 cm$^{2}$/V$\cdot$s and anisotropic electric and optical conductivities,[1-3] and shows great potential for applications in optoelectronic devices, photodetectors and sensors in the future.[4-8] However, the instability of BP in ambient air results in fast degradation and hinders its further applications. Methods used for protection, in general, can be summarized as capping and functionalizing, and most of them can slow down the degradation of BP and its devices. Capping layers such as atomic layer deposited Al$_{2}$O$_{3}$ or mechanically exfoliated $h$-BN were used to isolate BP from oxygen and water in its surroundings.[9-12] The functionalization of BP often aims at passivating dangling bonds on its surface.[13] In addition, post-processing was also performed for better stability and conductivity, but only a slight improvement in performance has been observed; N$_{2}$ annealing could contribute to the release of trapped charge and the improvement of contacts,[14] post-O$_{2}$ annealing degrades the channel mobility and increases the leakage current as well. Hydrogenation was commonly carried out to reduce the interface defects in semiconductor devices, which could also suppress the degradation of BP by removing the oxygen in PO$_{x}$. Nowadays, the effect of hydrogenation on device performance is scarcely reported, and the underlying mechanism is also unclear. In this work, a low-temperature supercritical hydrogenation process is performed on Al$_{2}$O$_{3}$ capped BP FETs. The BP devices with hydrogenation exhibit a pronounced performance improvement, such as subthreshold swing and field-effect mobility. Moreover, high/low frequency capacitance–voltage ($C$–$V$) measurements suggest reduced interface defects in hydrogenated BP FETs. It could be attributed to the passivation of interface traps at both Al$_{2}$O$_{3}$/BP and BP/SiO$_{2}$ interfaces with hydrogen evidenced by secondary ion mass spectroscopy (SIMS). Back-gated BP FETs were fabricated on a heavily doped silicon substrate (resistivity$ < $0.005 $\Omega\cdot$cm), and 285 nm thermally grown SiO$_{2}$ served as the gate dielectric. Few-layer BP was micromechanically exfoliated from the bulk crystals using a Scotch tape method.[15] Meanwhile, polydimethylsiloxane was used to reduce the residues of tapes during the BP transfer.[16,17] The thickness of BP flake chosen in this work for the fabrication of FETs was about 10–15 nm. Patterns of source and drain contacts (20 nm Ni/80 nm Au) were defined with e-beam lithography, followed by e-beam evaporation and a lift-off process. Subsequently, an Al$_{2}$O$_{3}$ capping layer was deposited on the backside of the BP channel, which was grown by means of atomic layer deposition at 200$^{\circ}\!$C with the precursors of trimethylaluminum and H$_{2}$O. Finally, the hydrogenation process of BP FETs was carried out in a supercritical CO$_{2}$ system under different concentrations of hydrogen. The schematic diagram of back-gated BP FETs with a 30 nm Al$_{2}$O$_{3}$ capping layer is illustrated in Fig. 1(a). Figure 1(b) shows the optical microscopy image of the fabricated BP device measured in this work. The electrical measurements were performed on a semiconductor parameter analyzer (Keysight B1500 A) with Cascade probe station in ambient air at room temperature.

Fig. 1. (a) Schematic diagram and (b) optical microscopy image of the fabricated few-layer BP FET.

Figure 2 shows the transfer characteristics of BP FETs with and without hydrogenation. A pronounced improvement in device performance of BP FETs with hydrogenation treatment is observed. Compared to the control sample, the field-effect mobility of the BP device increases from 20.6 to 28.2 cm$^{2}$v$^{-1}$s$^{-1}$ ($\sim$40%) when treated with low concentrated hydrogenation, while it increases from 69.6 to 107.7 cm$^{2}$v$^{-1}$s$^{-1}$ ($\sim$55%) for the high concentration case, as presented in Figs. 2(a) and 2(b), respectively. In addition, after hydrogenation, the on/off current ratio increases by more than one time, and the subthreshold swing decreases from 8.4 to 2.6 V/dev, almost one quarter of the initial value without any optimization. The output characteristics of BP FETs presented in Figs. 2(a) and 2(b) are shown in Figs. 2(c) and 2(d), respectively.

Fig. 2. The $I_{\rm d}$–$V_{\rm g}$ characteristics of the BP FETs before and after hydrogenation for one hour with (a) low and (b) high hydrogen concentrations. (c) and (d) $I_{\rm d}$–$V_{\rm d}$ characteristics of BP FETs in (a) and (b), respectively. All the hydrogenation processes are carried out in the same supercritical CO$_{2}$ system.

Fig. 3. Capacitance–voltage characteristics of BP FETs (a) before and (b) after hydrogenation under different frequencies from 50 to 1000 kHz.

Figure 3 depicts the $C$–$V$ characteristics of BP devices before and after hydrogenation with the frequency varying from 50 kHz to 1 MHz. Usually, the conventionally unipolar transistors show a zig-zag of $C$–$V$ curves at high frequency, while the ambipolar BP devices exhibit uniform V shape curves at either high or low frequency. The difference between high and low frequency capacitance is related to the interface defect. According to the high-low frequency capacity method, the density of interface states can be written as[18,19] $$\begin{alignat}{1} D_{\rm it} =\frac{\Delta C}{q^{2}}\Big( {1-\frac{C_{\rm HF} +\Delta C}{C_{\rm ox}}}\Big)^{-1}\Big( {1-\frac{C_{\rm HF}}{C_{\rm ox}}}\Big)^{-1},~~ \tag {1} \end{alignat} $$ where $C_{\rm LF}$ and $C_{\rm HF}$ are the minimum capacitance at low and high frequency, respectively, $\Delta C$ is the difference between $C_{\rm LF}$ and $C_{\rm HF}$, and $C_{\rm ox}$ is the capacitance density of the insulator. Figure 4 presents the variation of minimum capacitance ($\Delta C$) in the range of 50–1000 kHz with and without hydrogenation, which was calculated with the same data from Fig. 3. At most frequencies, such as the frequency range of 50–100 kHz, 100–300 kHz and 300–500 kHz, $\Delta C$ of BP devices before hydrogenation is larger than that after hydrogen processing, indicating the reduction of interface defects after hydrogen treatment. Notably, in the range of 500–1000 kHz, $\Delta C$ of the hydrogenated BP device is slightly larger than that of the initial one. This is because the minimum capacitance for ambipolar FETs could be affected not only by the capacitance of the depletion layer, but also by the carrier situation when the conduction polarity changes. After hydrogenation, the off current of BP FETs decreases, resulting in a lower bottom of the $C$–$V$ curve, thus $\Delta C$ is the sum of $\Delta C_{1}$ caused by interface defects, and $\Delta C_{2}$ stemmed from the reduction of the off current. In other words, for the initial device without hydrogenation, $\Delta C$ is only associated with $\Delta C_{1}$, where $\Delta C_{2}$ is negligible; while with hydrogenation $\Delta C$ consists of both $\Delta C_{1}$ and $\Delta C_{2}$. Note that, in the frequency range of 500–1000 kHz, these two parts are comparable so that the enlargement from $\Delta C_{2}$ offsets the decreased $\Delta C_{1}$, causing the aforementioned observations.

Fig. 4. The values of $\Delta C$ in different ranges of measuring frequencies before and after hydrogenation. Here $\Delta C$ (50 kHz-100 kHz) is calculated with $C_{\min}$ (50 kHz)–$C_{\min}$ (100 kHz) extracted from Figs. 3(a) and 3(b), and the others are calculated in the same manner.

Fig. 5. SIMS for typical elements of the hydrogenated sample. The relative concentration of oxygen, aluminum and phosphorus was used for depth judgement, and the absolute concentration of hydrogen for judgment of the hydrogenation effect.

The MOS area can be roughly calculated by ${C_{\max}}/{C_{\rm ox}}$, where $C_{\max}$ is the maximum value of the $C$–$V$ curve. The sums of all $\Delta C$ in the frequency range of 50–1000 kHz with and without hydrogenation are displayed as follows: $\Delta C_{\rm initial}=3.45\times10^{-9}$ F$\cdot$cm$^{-2}$, $\Delta C_{\rm after H}=2.57\times10^{-9}$ F$\cdot$cm$^{-2}$. According to $D_{\rm it}={\Delta C}/{q^{2}}$, the $D_{\rm it}$ values before and after hydrogenation are $2.16\times10^{10}$ cm$^{-2}\cdot$eV$^{-1}$ and $1.61\times10^{10}$ cm$^{-2}\cdot$eV$^{-1}$, respectively. Thus the reduced defect density after hydrogenation is $5.5\times10^{9}$ cm$^{-2}\cdot$eV$^{-1}$. We further examine the hydrogen profile at the Al$_{2}$O$_{3}$/BP interface. Figure 5 shows the SIMS results of the BP device with hydrogenation. Four typical elements were detected, among which the relative concentrations of Al, P and O (refer to the right axis) were used to confirm the interface of different layers. The location of rapid changes indicates the interfaces of Al$_{2}$O$_{3}$/BP/SiO$_{2}$, denoted as dashed lines. The absolute concentration of hydrogen (refer to the left axis) exhibits that the concentration of H in the Al$_{2}$O$_{3}$ passivation is about 10$^{22}$ atom/cm$^{3}$, which is higher than that of Al$_{2}$O$_{3}$ without hydrogen treatment.[20] It is revealed that the hydrogenation process increases the hydrogen concentration in Al$_{2}$O$_{3}$, and at both Al$_{2}$O$_{3}$/BP and BP/SiO$_{2}$ interfaces. The hydrogenation is believed to passivate the interface defect, thus resulting in the improved performance of BP FETs. In summary, we have performed hydrogenation treatment on Al$_{2}$O$_{3}$ capped few-layer BP FETs. A 55% increase of field-effect mobility from 69.6 to 107.7 cm$^{2}$v$^{-1}$s$^{-1}$ and a remarkably reduced subthreshold swing from 8.4 to 2.6 V/dec, as well as an improved on/off ratio are observed. The interface defects in BP FETs can be significantly reduced after hydrogenation, which could be attributed to the passivation of defects at the interfaces. Our findings provide useful insights into the process optimization of BP device performance. We thank Professor Ting-Chang Chang for the use of the hydrogenation equipment and fruitful discussions. References Rediscovering black phosphorus as an anisotropic layered material for optoelectronics and electronicsBlack phosphorus field-effect transistorsHighly anisotropic and robust excitons in monolayer black phosphorusOptical modulators with 2D layered materialsBlack PhosphorusâMonolayer MoS ₂ van der Waals Heterojunction pân DiodeAir-Stable Black Phosphorus Devices for Ion SensingStable and Selective Humidity Sensing Using Stacked Black Phosphorus FlakesIsolation and characterization of few-layer black phosphorusEffective Passivation of Exfoliated Black Phosphorus Transistors against Ambient DegradationTransport properties of pristine few-layer black phosphorus by van der Waals passivation in an inert atmosphereAl ₂ O ₃ on Black Phosphorus by Atomic Layer Deposition: An in Situ Interface StudyLong-Term Stability and Reliability of Black Phosphorus Field-Effect TransistorsSurface Coordination of Black Phosphorus for Robust Air and Water StabilityTemporal and Thermal Stability of Al ₂ O ₃ -Passivated Phosphorene MOSFETsA roadmap for grapheneDirect Growth of Al2O3 on Black Phosphorus by Plasma-Enhanced Atomic Layer DepositionDeterministic transfer of two-dimensional materials by all-dry viscoelastic stampingDescription of the SiO2î¸Si interface properties by means of very low frequency MOS capacitance measurementsExamining the role of hydrogen in the electrical performance of in situ fabricated metal-insulator-metal trilayers using an atomic layer deposited Al ₂ O ₃ dielectric

[1]	Xia F, Wang H and Jia Y 2014 Nat. Commun. 5 4458
[2]	Li L, Yu Y, Ye G J, Ge Q, Ou X, Wu H, Feng D, Chen X H and Zhang Y 2014 Nat. Nanotechnol. 9 372
[3]	Wang X, Jones A M, Seyler K L, Tran V, Jia Y, Zhao H, Wang H, Yang L, Xu X and Xia F 2015 Nat. Nanotechnol. 10 517
[4]	Sun Z, Martinez A and Wang F 2016 Nat. Photon. 10 227
[5]	Deng Y, Luo Z, Conrad N J, Liu H, Gong Y, Najmaei S, Ajayan P M, Lou J, Xu X and Ye P D 2014 ACS Nano 8 8292
[6]	Li P, Zhang D, Liu J, Chang H, Sun Y and Yin N 2015 ACS Appl. Mater. Interfaces 7 24396
[7]	Yasaei P, Behranginia A, Foroozan T, Asadi M, Kim K, Khaliliaraghi F and Salehikhojin A 2015 ACS Nano 9 9898
[8]	Castellanosgomez A, Vicarelli L, Prada E et al 2014 2D Mater. 1 025001
[9]	Wood J D, Wells S A, Jariwala D et al 2014 Nano Lett. 14 6964
[10]	Doganov R A, O'Farrell E C T, Koenig S P et al 2015 Nat. Commun. 6 6647
[11]	Zhu H, Mcdonnell S, Qin X et al 2015 ACS Appl. Mater. Interfaces 7 13038
[12]	Illarionov Y Y, Waltl M, Rzepa G et al 2016 ACS Nano 10 9543
[13]	Zhao Y, Wang H, Huang H et al 2016 Angew. Chem. 55 5003
[14]	Luo X, Rahbarihagh Y, Hwang J C M et al 2014 IEEE Electron Device Lett. 35 1314
[15]	Novoselov K S, Fal'Ko V I, Colombo L et al 2012 Nature 490 192
[16]	Wu B B, Zheng H M, Ding Y Q et al 2017 Nanoscale Res. Lett. 12 282
[17]	Castellanos-Gomez A, Buscema M, Molenaar R et al 2014 2D Mater. 1 011002
[18]	Sze S M and Ng Kwok K 2007 Physics of Semiconductor Devices 3rd edn (Hoboken: John Wiley & Sons Inc) p 215
[19]	Castagné R and Vapaille A 1971 Surf. Sci. 28 157
[20]	Kozen A C, Schroeder M A, Osborn K D et al 2013 Appl. Phys. Lett. 102 173501