# Fileset

[AcceptedManuscript_APL_2025_Sasama.pdf](https://mdr.nims.go.jp/filesets/4b38c078-37f2-4d72-8e8d-18afb191d17d/download)

## Creator

[Yosuke Sasama](https://orcid.org/0000-0002-8358-6101), [Takuya Iwasaki](https://orcid.org/0000-0002-1103-2433), [Masataka Imura](https://orcid.org/0000-0002-4236-9549), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Yamaguchi Takahide](https://orcid.org/0000-0003-0208-7317)

## Rights

This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Yosuke Sasama, Takuya Iwasaki, Masataka Imura, Kenji Watanabe, Takashi Taniguchi, Yamaguchi Takahide; Enhanced channel mobility of hexagonal boron nitride/hydrogen-terminated diamond heterojunction field-effect transistor. Appl. Phys. Lett. 6 October 2025; 127 (14): 143502 and may be found at https://doi.org/10.1063/5.0272041. [In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Enhanced channel mobility of hexagonal boron nitride/hydrogen-terminated diamond heterojunction field-effect transistor](https://mdr.nims.go.jp/datasets/d2f9924d-39e7-4a01-b92f-ee4209fb3779)

## Fulltext

Enhanced channel mobility of hexagonal boron nitride/hydrogen-terminated diamond heterojunction field-effect transistorEnhanced channel mobility of hexagonal boron nitride/hydrogen-terminated diamond1heterojunction field-effect transistor2Yosuke Sasama,1, a) Takuya Iwasaki,2 Masataka Imura,3 Kenji Watanabe,33Takashi Taniguchi,2 and Yamaguchi Takahide2, 441)International Center for Young Scientists, National Institute for Materials Science,5Tsukuba 305-0044, Japan62)Research Center for Materials Nanoarchitectonics, National Institute for Materials7Science, Tsukuba 305-0044, Japan83)Research Center for Electronic and Optical Materials,9National Institute for Materials Science, Tsukuba 305-0044,10Japan114)University of Tsukuba, Tsukuba 305-8571, Japan12Hydrogen-terminated diamond field-effect transistors (FETs) using a hexagonal13boron nitride (h-BN) gate insulator were fabricated on a diamond surface with re-14duced surface roughness in the direction of source/drain electrodes. The diamond15surface was prepared on a mesa structure using chemical vapor deposition (CVD)16with a low methane concentration. The hydrogen-terminated surface was laminated17with the h-BN gate insulator without air exposure to prevent the adsorption of at-18mospheric surface acceptors. The hydrogen-terminated diamond FET exhibited a19high mobility of ≈1000 cm2V−1s−1 at room temperature. We performed theoreti-20cal analysis on the temperature and carrier density dependences of mobility, which21suggested that Coulomb and surface roughness scattering were effectively reduced.22The high mobility obtained in this study indicates the high potential of diamond as23a semiconducting material. This study can contribute to the future development of24diamond devices.25a)SASAMA.Yosuke@nims.go.jp1This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041Diamond is a wide-gap semiconductor with a high breakdown electric field, high thermal26conductivity, and high mobility, making it advantageous for power device applications1,2.27Several types of diamond field-effect transistors (FETs) have been developed, including28Schottky metal-semiconductor FETs3, junction FETs4, and deep-depletion FETs5. One of29the most widely developed FETs is the hydrogen-terminated diamond FET, in which the30diamond surface is terminated with hydrogen6. This is because terminating the diamond31surface with hydrogen reduces the surface state density, and the higher energy of the valence32band maximum facilitates the induction of hole carriers7.33Hydrogen-terminated diamond FETs often use the surface conductivity induced by trans-34fer doping.8 The surface conductivity can be obtained simply by exposing the hydrogen-35terminated surface to air.9 Additionally, carrier density can be increased to >1014 cm−2 by36exposure to NO2 gas10 or the deposition of oxides with a large electron affinity11.37However, negative charges associated with the formation of surface conductivity on the38hydrogen-terminated surface (ionized surface acceptors or atmospheric negative charges)39scatter carriers and reduce the mobility of FETs. Reduced mobility is undesirable for high-40speed and low-loss operation of FETs. To increase the mobility of diamond FETs, we41have fabricated FETs with a reduced surface acceptor density using a hexagonal boron42nitride (h-BN) gate insulator12–14 and an air-free process15, in which the hydrogen-terminated43surface is not exposed to air. We have obtained a high mobility of 680 cm2V−1s−1 with a44relatively high carrier density of 6.6×1012 cm−2. In addition to the high mobility, the45FETs also demonstrate normally off operation, low on-resistance, and high on-off ratio46simultaneously.1547In this study, we made two primary improvements to further increase the mobility of48diamond FETs. The first is the reduction of surface roughness on a diamond. Surface49roughness can contribute to the formation of surface states16. Charges trapped at surface50states cause Coulomb scattering, resulting in reduced mobility. Mobility is also reduced by51surface roughness scattering. Here, we grew a relatively flat diamond film on a mesa struc-52ture using chemical vapor deposition (CVD) with a low methane concentration17, and we53fabricated FETs on the diamond film. The second improvement is the lamination technique54of h-BN. Here, we use pick-up method to laminate h-BN. The use of this pick-up method18,1955effectively avoids the formation of contamination bubbles, which are often problematic when56creating van der Waals heterostructures. Through the use of a curvature polymer and slowly572This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041laminating the layers, bubbles are pushed out, resulting in the formation of a clean interface58free from contamination bubbles. Therefore, this method has been increasingly employed in59recent years to fabricate van der Waals heterostructures20, including recent h-BN/diamond60hetero devices.21,2261First, we focused on the growth of diamond films. Generally, a diamond (111) substrate62has a misorientation angle, and atomic steps exist on the substrate surface. When diamond is63grown using CVD with a low methane concentration on a (111) diamond substrate, it grows64laterally from atomic steps, and atomically flat diamond can be formed17. Following the65previous report, we grew diamond films. We used type-Ib (111) single-crystalline diamond66substrate manufactured by Sumitomo Electric Industries. After the mesa structure was67formed on the diamond substrate using oxygen plasma etching, the diamond was grown68using CVD under the following conditions: hydrogen flow rate of 1000 sccm, methane flow69rate of 0.1 sccm, microwave power of 800 W, and pressure of 80 torr. The quality of the70diamond surface was evaluated using an atomic force microscope (AFM). Measurements71were performed on an area of 20 µm square. Steps existed in the direction perpendicular72to the [111] direction, but a relatively flat diamond film with an average surface roughness73(Ra) of ≈0.15 nm was grown (Fig. 1c).74Subsequently, we fabricated an FET using a h-BN gate insulator with a Hall-bar structure75(Fig. 1a) on the CVD-grown diamond. First, Ti/Pt was deposited as electrodes. Here, the76source/drain electrodes (current electrodes in the Hall bar) were placed parallel to the step77of the diamond. An annealing was performed in a hydrogen atmosphere (hydrogen flow78rate of 500 sccm, pressure of 80 torr) at 650◦C for 35 min to form TiC, which functions as79an ohmic electrode. Thereafter, the diamond surface was exposed to hydrogen plasma to80remove resist residue and atmospheric adsorbates. The hydrogen plasma conditions were as81follows: hydrogen flow rate of 500 sccm, microwave power of 300 W, pressure of 30 torr,82and duration of 10 min. Annealing and hydrogen plasma treatment were also performed in83another CVD chamber under the same conditions as in the first chamber. The second CVD84chamber could be connected to a vacuum suitcase. The diamond was vacuum-transferred85to an Ar-filled glove box using the vacuum suitcase to avoid the adsorption of atmospheric86adsorbates on the hydrogen-terminated diamond surface. In the glove box, the hydrogen-87terminated diamond was laminated with a cleaved h-BN thin film. Here, we used the pick-up88method18,19,21,22 for the h-BN transfer. A single-crystalline h-BN was cleaved with scotch893This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041tape and transferred onto a Si substrate. The cleaved h-BN film was then picked up from90the Si substrate using a polymer stamp and transferred onto the diamond. The lamination91of diamond with h-BN was performed at room temperature and then the diamond was92heated to 110◦C to release the h-BN from the stamp. The thickness of the h-BN film used93in this study was 16 nm. After the h-BN lamination, the diamond substrate was annealed94in the glove box at 200◦C for 3 h. The diamond substrate was then removed from the glove95box. The h-BN gate insulator was etched into a Hall-bar shape using capacitively coupled96plasma reactive ion etching (RIE) with N2, CHF3, and O2 gases. The diamond surface,97except for the area underneath the h-BN, was converted to oxygen termination during this98process, providing device isolation. Subsequently, Ti/Au/Ti films were deposited on the h-99BN as a gate electrode. The hydrogen-terminated diamond may attract negatively charged100impurities, resulting in the intercalation of impurities at the interface between the diamond101and h-BN, because the hydrogen-terminated diamond surface is slightly positively charged102owing to the difference in electronegativity between carbon and hydrogen atoms8. To prevent103impurities from contaminating the interface, we deposited 30-nm Al2O3 using atomic layer104deposition. We consider that the Al2O3 passivation contributes to the long-term stability105of the device. Thereafter, through holes were opened in Al2O3 using wet etching with106dilute tetramethylammonium hydroxide. Finally, the gate contact was fabricated through107the sputter deposition of Ti/Au. The optical microscope image of the fabricated device is108shown in Fig. 1b.109We evaluated the electrical characteristics of the FET. Figure 2a shows the transfer110characteristics. (See also Fig. S1 of the supplementary material.) The FET exhibits p-type111FET characteristics. The maximum sheet conductance is 1.1 mS, which corresponds to a112sheet resistance of 0.93 kΩ. The sheet conductance (σ) is obtained as σ = (Lp/WG)(1/R),113where R is the resistance measured using the four-terminal method, WG (= 0.97 µm) is the114gate width, and Lp (= 3.0 µm) is the distance between the voltage probes. This device115has a fringe region in which the diamond surface is hydrogen-terminated and covered by116h-BN, but it is not under the gate electrode. This region may be conductive owing to hole117carriers generated by surface acceptors originating from residual gases in the glove box. The118conductance in the fringe region may result in an overestimate of the sheet conductance.119However, if we assume that the sheet conductance in the fringe region is equal to that in the120gate-overlapped region at zero gate voltage, the impact of the fringe region on the maximum1214This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041sheet conductance is estimated to be only 7%. This value is obtained as ∆σ = σ(VGS =1220) × (Wh−BN − WG)/Wh−BN using the h-BN width (Wh−BN = 3.97 µm) and gate width.123The threshold voltage evaluated from the transfer curve is 0.67 V. Unlike our previous124study, the threshold voltage is positive, which indicates normally on operation. We consider125that the reason for this positive threshold voltage is a lower concentration of nitrogen23126acting as a donor unlike in the previous one15. The diamond substrates used in our previous127study were type-IIa high pressure, high temperature(HPHT)-grown diamond purchased from128the Technological Institute for Superhard and Novel Carbon Materials (TISNCM), and129secondary ion mass spectrometry (SIMS) on a similar substrate showed that the nitrogen130concentration in the substrate was ≈500 ppb. In contrast, the nitrogen concentration in131diamond grown under the similar conditions as in the present study was below the detection132limit of SIMS measurement (<20 ppb).133Figure 2b shows the output characteristics. The drain-source length is 8.8 µm. The134maximum drain current density is ≈240 mA mm−1. The drain current density is obtained135by dividing the drain current by the gate width. Based on a simple calculation considering136the gate and h-BN width, the current flowing through the fringe region (I(VGS = 0) ×137((Wh−BN − WG)/Wh−BN)) is 15 mA mm−1, which is as low as 6% of the maximum drain138current density. The FET is not completely turned off because of the fringe current. Because139the gate electrode does not cover the fringe region, the hole carriers do not deplete even when140a positive gate voltage is applied. Therefore, a small drain current flows even when a positive141gate voltage is applied. To completely turn off the FET, we believe that device isolation142techniques should be improved.143Carrier density and mobility were evaluated using Hall-effect measurements. Figure 2c144shows the gate voltage dependence of the carrier density. The carrier density increased145monotonically, with a maximum carrier density of 4.7×1012 cm−2. Figure 2d shows the Hall146mobility. Field-effect mobility (µFE = thBNϵhBN∣∣∣∂σ∂VGS∣∣∣) and effective mobility (µeff = thBNϵhBNσ|VGS−Vth|)147estimated from the slope of the transfer characteristic (Fig. 2a) are also shown. σ is the sheet148conductance shown in Fig. 2a, VGS is the gate voltage, thBN is the thickness of h-BN，and149ϵhBN = 3 is the dielectric constant of h-BN. For the estimation of Hall mobility, the conduc-150tance of the side paths was subtracted using the same method as in our previous work12. The151Hall mobility (µHall) is calculated as µHall = (σ− σ(VGS = 0)(Wh−BN −WG)/Wh−BN)/(nHe).152σ is the sheet conductance, nH is the Hall carrier density, and e is the elementary charge1535This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041(Figure S2 of the supplementary material). A high mobility of ≈1000 cm2V−1s−1 was ob-154served in the Hall, field-effect, and effective mobilities. The Hall mobility is slightly higher155than effective and field-effect mobilities, which might be related to the current paths besides156the channel. We note that we can rule out the possibility that the high mobility originates157from boron contamination in the diamond film. The density of carriers induced by boron158does not increase with increasing negative gate voltage; hence, at least the field-effect mo-159bility is considered to be the mobility of a two-dimensional hole gas induced by applying a160gate voltage on a hydrogen-terminated diamond.161We also evaluated the temperature dependence of the mobility. Figure 3 shows the tem-162perature dependence of the mobility evaluated using Hall-effect measurements. The mobility163of our present FETs fabricated on a CVD-grown diamond increases with decreasing temper-164ature more rapidly than our previous FETs fabricated on a diamond surface without CVD165growth. The maximum mobility is ≈2300 cm2V−1s−1 at 100 K. Below 100 K, the mobility166could not be evaluated accurately owing to the large contact resistance at low temperature.167The large contact resistance at low temperatures may originate from the gap formed at the168h-BN/diamond interface near the Hall-bar electrodes, as discussed in our recent paper22.169The cleaner surface compared with before may increase the electrical resistance of diamond170underneath the gap near electrodes and, therefore, the contact resistance at low tempera-171tures. We consider that the use of gate electrode self-aligned with source/drain electrodes22172can eliminate the gap formed at the h-BN/diamond interface near the electrodes and reduce173the contact resistance.174The reduction of Coulomb and surface roughness scattering was supported by the the-175oretical analysis of mobility. Figure 4 shows the theoretical analysis of (a) carrier density176dependence and (b) temperature dependence of mobility. (See also Fig. S3 of the supple-177mentary material.) The mobility was calculated theoretically using the same method as178in our previous report (Supplementary Information in Ref.15). As shown in Fig. 4, the179theoretical calculations of mobility adequately explain the experimental results for both the180carrier density and the temperature dependences of mobility. Here, the boron and nitrogen181concentrations were set to 1 and 5 ppb, respectively, and the average roughness (∆) was set182to 0.15 nm based on AFM measurements. We have tuned the interfacial charge density (nic)183and distance between the interfacial charges and hole carriers (d). The values of d and nic184were determined by minimizing the total sum of squared residuals between the calculated1856This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041and experimental mobilities at VGS = −2,−4 and −6 V. Consequently, the optimal values186were found to be d = 0.0 nm and nic = 2.4 × 1011 cm−2. The obtained value of d is con-187sistent with those assumed in references14,15, and25, and is similar to the value reported in188reference24. The experimental results of our previous study15 are well explained when the189calculation parameter values are nic = 6 × 1011 cm−2, ∆ = 0.18 nm, whereas the results190of this study are well explained when the calculation parameter values are nic = 2.4× 1011191cm−2, ∆ = 0.15 nm. The reduction in nic is considered to be owing to both the decrease in192the surface roughness and improvement in the h-BN lamination technique. The reduction in193surface roughness is expected to decrease the density of the interface states. The use of the194pick-up method for the h-BN lamination has also contributed to the improvement in mo-195bility by reducing the charged impurity density at the hydrogen-terminated diamond/h-BN196interface.197In summary, we fabricated diamond FETs using an h-BN gate insulator on a diamond198film grown using CVD with a low methane concentration. Through the CVD with a low199methane concentration, a diamond film with reduced surface roughness in the direction of200source/drain electrodes was grown. The h-BN gate insulator was formed via the cleavage of201h-BN and a pick-up method using a curvature polymer stamp. We demonstrated that a high202hole mobility of ≈1000 cm2V−1s−1 can be achieved at room temperature in the hydrogen-203terminated diamond FET by improving the quality of the interface between the hydrogen-204terminated diamond and gate insulator. The theoretical analysis of the carrier density and205temperature dependences of mobility supported that this high mobility was achieved by re-206ducing surface roughness and interface charge density. This mobility approaches that of the207electron mobility in GaN high electron mobility transistors (HEMTs), suggesting that com-208bining our diamond FET with a GaN-HEMT can enable high-performance complementary209circuits based on wide-bandgap semiconductors.210Supplementary Material211See the supplementary material for additional analysis on the FET’s long-term electrical212stability, calculation of Hall mobility accounting for fringe conductance, and the theoretical213and experimental exploration of temperature-dependent mobility at various gate voltages.2147This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041Data availability215The data supporting the findings of this study are available from the corresponding author216upon reasonable request.217REFERENCES1N. Donato, N. Rouger, J. Pernot, G. Longobardi, and F. Udrea, “Diamond power devices:state of the art, modelling, figures of merit and future perspective,” Journal of Physics D:Applied Physics 53, 093001 (2020).2D. Eon, “Diamonds in the current: Navigating challenges for the integration of diamondin power electronics,” physica status solidi (a) 221, 2400085 (2024).3H. Kawarada, M. Aoki, and M. Ito, “Enhancement mode metal‐semiconductor field effecttransistors using homoepitaxial diamonds,” Applied Physics Letters 65, 1563–1565 (1994).4T. Iwasaki, Y. Hoshino, K. Tsuzuki, H. Kato, T. Makino, M. Ogura, D. Takeuchi, T. Mat-sumoto, H. Okushi, S. Yamasaki, and M. Hatano, “Diamond junction field-effect transis-tors with selectively grown n+-side gates,” Applied Physics Express 5, 091301 (2012).5T. T. Pham, J. Pernot, G. Perez, D. Eon, E. Gheeraert, and N. Rouger, “Deep-depletionmode boron-doped monocrystalline diamond metal oxide semiconductor field effect tran-sistor,” IEEE Electron Device Letters 38, 1571–1574 (2017).6H. Kawarada, “Hydrogen-terminated diamond surfaces and interfaces,” Surface ScienceReports 26, 205–259 (1996).7S. J. Sque, R. Jones, and P. R. Briddon, “Structure, electronics, and interaction of hy-drogen and oxygen on diamond surfaces,” Physical Review B 73, 085313 (2006).8H. Kawarada, “Diamond p-FETs using two-dimensional hole gas for high frequency andhigh voltage complementary circuits,” Journal of Physics D: Applied Physics 56, 053001(2022).9F. Maier, M. Riedel, B. Mantel, J. Ristein, and L. Ley, “Origin of surface conductivity indiamond,” Physical Review Letters 85, 3472–3475 (2000).10M. Kasu, “Diamond field-effect transistors for RF power electronics: Novel NO2 holedoping and low-temperature deposited Al2O3 passivation,” Japanese Journal of AppliedPhysics 56, 01AA01 (2017).8This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.027204111S. A. O. Russell, L. Cao, D. Qi, A. Tallaire, K. G. Crawford, A. T. S. Wee, and D. A. J.Moran, “Surface transfer doping of diamond by MoO3: A combined spectroscopic and Hallmeasurement study,” Applied Physics Letters 103, 202112 (2013).12Y. Sasama, K. Komatsu, S. Moriyama, M. Imura, T. Teraji, K. Watanabe, T. Taniguchi,T. Uchihashi, and Y. Takahide, “High-mobility diamond field effect transistor with amonocrystalline h-BN gate dielectric,” APL Materials 6, 111105 (2018).13Y. Sasama, K. Komatsu, S. Moriyama, M. Imura, S. Sugiura, T. Terashima, S. Uji,K. Watanabe, T. Taniguchi, T. Uchihashi, and Y. Takahide, “Quantum oscillations indiamond field-effect transistors with a h-BN gate dielectric,” Physical Review Materials 3,121601(R) (2019).14Y. Sasama, T. Kageura, K. Komatsu, S. Moriyama, J.-i. Inoue, M. Imura, K. Watanabe,T. Taniguchi, T. Uchihashi, and Y. Takahide, “Charge-carrier mobility in hydrogen-terminated diamond field-effect transistors,” Journal of Applied Physics 127, 185707(2020).15Y. Sasama, T. Kageura, M. Imura, K. Watanabe, T. Taniguchi, T. Uchihashi, andY. Takahide, “High-mobility p-channel wide-bandgap transistors based on hydrogen-terminated diamond/hexagonal boron nitride heterostructures,” Nature Electronics 5, 37–44 (2022).16M. Nagai, R. Yoshida, T. Yamada, T. Tabakoya, C. E. Nebel, S. Yamasaki, T. Makino,T. Matsumoto, T. Inokuma, and N. Tokuda, “Conductive-probe atomic force microscopyand kelvin-probe force microscopy characterization of OH-terminated diamond (111) sur-faces with step-terrace structures,” Japanese Journal of Applied Physics 58, SIIB08 (2019).17N. Tokuda, H. Umezawa, S.-G. Ri, M. Ogura, K. Yamabe, H. Okushi, and S. Yamasaki,“Atomically flat diamond (111) surface formation by homoepitaxial lateral growth,” Dia-mond and Related Materials 17, 1051–1054 (2008).18F. Pizzocchero, L. Gammelgaard, B. S. Jessen, J. M. Caridad, L. Wang, J. Hone, P. Bog-gild, and T. J. Booth, “The hot pick-up technique for batch assembly of van der waalsheterostructures,” Nature Communications 7, 11894 (2016).19T. Iwasaki, K. Endo, E. Watanabe, D. Tsuya, Y. Morita, S. Nakaharai, Y. Noguchi,Y. Wakayama, K. Watanabe, T. Taniguchi, and S. Moriyama, “Bubble-free transfer tech-nique for high-quality graphene/hexagonal boron nitride van der waals heterostructures,”ACS Appl Mater Interfaces 12, 8533–8538 (2020).9This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.027204120W. Dong, Z. Dai, L. Liu, and Z. Zhang, “Toward clean 2D materials and devices: Recentprogress in transfer and cleaning methods,” Adv Mater , e2303014 (2023).21Y. Huang, J. Xiao, R. Tao, Z. Liu, Y. Mo, X. Yu, Z. Cao, Y. Wu, Z. Li, H. Wang, andL. Wang, “High mobility hydrogen-terminated diamond FET with h-BN gate dielectricusing pickup method,” Applied Physics Letters 123, 112103 (2023).22Y. Sasama, T. Iwasaki, M. Monish, K. Watanabe, T. Taniguchi, and Y. Takahide, “Self-aligned gate electrode for hydrogen-terminated diamond field-effect transistors with ahexagonal boron nitride gate insulator,” Applied Physics Letters 125, 092103 (2024).23T. Matsumoto, T. Yamakawa, H. Kato, T. Makino, M. Ogura, X. Zhang, T. Inokuma,S. Yamasaki, and N. Tokuda, “Fabrication of inversion p-channel MOSFET with anitrogen-doped diamond body,” Applied Physics Letters 119, 242105 (2021).24Y. Li, J.-F. Zhang, G.-P. Liu, Z.-Y. Ren, J.-C. Zhang, and Y. Hao, “Mobility of two-dimensional hole gas in H-terminated diamond,” physica status solidi (RRL) - RapidResearch Letters 12, 1700401 (2018).25G. Daligou and J. Pernot, “2D hole gas mobility at diamond/insulator interface,” AppliedPhysics Letters 116, 162105 (2020).Acknowledgments218We thank H. Osato, E. Watanabe, D. Tsuya, N. Ikeda, and A. Ohi for technical sup-219port. We thank T. Ando, S. Koizumi, and T. Teraji for their assistance with this study.220This study was financially supported by the NEDO Uncharted Territory Challenge 2050221(Project No. JPNP14004), JSPS KAKENHI (Grant No. JP22H01962) and Advanced Re-222search Infrastructure for Materials and Nanotechnology in Japan (ARIM) (Proposal No.223JPMXP1224NM5213).224Competing Interests225The authors declare that they have no competing financial interests.22610This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041Data availability227The data that support the findings of this study are available from the corresponding228author upon reasonable request.22911This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041FIG. 1. (a) Schematic and (b) optical microscope image of the diamond FET fabricated in thisstudy. (c) AFM image of the diamond surface after CVD growth. The arrow represents theprojection of the [111] direction onto a two-dimensional plane.12This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041FIG. 2. Device characteristics of the diamond FET (Sample B1) at 300 K. (a) Transfer charac-teristics, (b) output characteristics, (c)(d) gate voltage dependence of (c) carrier density and (d)mobility.13This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041FIG. 3. Temperature dependence of Hall mobility. The points connected by lines are the mea-surement results of the same device measured at different temperatures. Samples B1 and B2 werefabricated on a CVD-grown diamond surface, whereas samples C1, C2 (ref.15) and D1-D3 (ref.12)were fabricated on a diamond surface without CVD growth. Samples B1, B2, C1, and C2 werefabricated without air exposure to hydrogen-terminated surface, whereas samples D1-D3 were fab-ricated on an air-exposed hydrogen-terminated surface. Measurements were performed with theapplication of a gate voltage (−6 V for B1, −15 V for B2, −8 V for C1, −10 V for C2, −4 V forD1, −7 V for D2, −5 V for D3, depending on the thickness of h-BN). Sample B2 exhibited fluctu-ations in electrical measurements at low temperatures owing to a high contact resistance. SampleB2 was fabricated using a process in which the gate electrode was self-aligned with source/drainelectrodes. Some of the hydrogen-terminated surface in the access region was damaged during theh-BN etching process, resulting in high contact resistance in sample B2.14This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041FIG. 4. Theoretical analysis of mobility. (a) Carrier density dependence and (b) temperaturedependence of mobility. Red circles show experimental results for sample B1. The carrier densitydependence was measured at 300 K. The temperature dependence was measured with an appliedgate voltage of −6 V. Solid lines show the results of the theoretical calculation of mobility. Thelabels “ic,” “sr,” “ac,” and “op” indicate the calculated mobilities limited by interface charges,surface roughness, acoustic phonon, and optical phonon, respectively. The mobility limited bybackground impurity scattering has values much higher than the plot range. The label “tot”indicates the calculated mobility considering all the above scattering.15This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.PLEASE CITE THIS ARTICLE AS DOI: 10.1063/5.0272041