# Fileset

[2025 03 15 APL mauscript of H-1 desorption.docx](https://mdr.nims.go.jp/filesets/b339cb3e-467f-4eff-aa9e-cbac89957657/download)

## Creator

[Keyun Gu](https://orcid.org/0000-0002-7505-7744), [Zilong Zhang](https://orcid.org/0000-0002-9759-9253), [Jian Huang](https://orcid.org/0000-0002-1268-8899), [Yasuo Koide](https://orcid.org/0000-0001-8321-9822), [Satoshi Koizumi](https://orcid.org/0000-0003-4961-5658), [Meiyong Liao](https://orcid.org/0000-0003-1361-4266)

## 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 Keyun Gu, Zilong Zhang, Jian Huang, Yasuo Koide, Satoshi Koizumi, Meiyong Liao; Surface desorption properties of hydrogen-terminated diamond detected by micromechanical resonator. Appl. Phys. Lett. 2 June 2025; 126 (22): 221901 and may be found at https://doi.org/10.1063/5.0274650.[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Surface desorption properties of hydrogen-terminated diamond detected by micromechanical resonator](https://mdr.nims.go.jp/datasets/ba733b6a-5387-43eb-8f9b-7e4b68dd5439)

## Fulltext

Real-time monitoring of adsorbate desorption on H-terminated diamond using single-crystal diamond MEMS resonatorsKeyun Gu,1,2 Zilong Zhang,1 Jian Huang,2 Yasuo Koide,1 Satoshi Koizumi,1 and Meiyong Liao1,a)1 Research Center for Electronics and Optical Materials, National Institute for Materials Science, Namiki 1-1, Tsukuba, Ibaraki 305-0044, Japan 2 School of Materials Science and Engineering, Shanghai University, Shanghai 200444, Chinaa) Electronic mail: meiyong.liao@nims.go.jpABSTRACTThe ultimate physical and chemical properties of hygrogen (H)-terminated diamond devices are strong influenced by real-time surface adsorbates. In this study, we investigated the dynamic thermal desorption process of these adsorbates using an innovative approach—single-crystal diamond (SCD) microelectromechanical system (MEMS) resonators. By analyzing the variations in resonance performance and surface conductivity, we uncover several important findings: (i) The desorption of surface adsorbates reaches saturation around 600℃, (ii) The desorbed mass per unit area is approximately 2.3 fg/μm2, corresponding to an equivalent thickness of around 0.65 nm, (iii) Surface conductivity recovers after annealing at 600°C, indicating the stability of C-H bonds, and (iv) The impact of the adsorbate layer on mechanical energy dissipation in H-terminated diamond devices is significantly lower than that of naturally oxidized films in other semiconductors. These findings provide a reliable method for evaluating the surface properties of H-terminated diamond and offer a new approach for studying the surface characteristics of other semiconductor materialsHydrogen (H)-terminated diamond, produced by exposing diamond to hydrogen plasma or gas, exhibits negative electron affinity (NEA), attracting negative ions and water molecules. This leads to the formation of a two-dimensional hole gas (2DHG) and p-type surface conductivity via a charge-transfer doping mechanism 1,2. These properties make H-terminated diamond promising for high-performance photodetectors 3, field-effect transistors 4, and other semiconductor devices, with potential for next-generation integrated circuits. Notably, the physical and electrical properties of H-terminated diamond are strongly influenced by surface adsorbates 5. Upon heating, adsorbates gain sufficient thermal energy to desorb from the surface, affecting device performance 6,7. Understanding these desorption mechanisms is essential for optimizing H-terminated diamond-based electronic devices.Although conventional characterization techniques, such as X-ray photoelectron spectroscopy (XPS) 8 and electron energy loss spectroscopy (EELS) 9 have been employed to analyze the thermal stability of C-H bonds, real-time studies on the dynamic desorption process of adsorbates from H-terminated diamond remain limited due to the lack of high-precision measurement equipment, particularly under in situ operating conditions.Single crystal diamond (SCD)-based microelectromechanical system (MEMS) resonators, leveraging the exceptional mechanical properties including the highest known Young’s modulus (~1100 Gpa), a low friction coefficient (0.01~0.1), superior thermal stability, and high crystal quality with minimal defect density 10, can serve as highly reliable, low-noise, and ultrasensitive mass sensors 11. When functionalized as H-terminated diamond, these resonators enable in situ monitoring of adsorbate desorption through resonance frequency shifts, making them ideal for dynamic tracking under varying conditions. Despite these advantages, systematic studies on the dynamic desorption process using SCD MEMS resonators remain scarce, particularly in linking mechanical responses to surface chemical dynamics.Here, we present a comprehensive investigation of the dynamic desorption of surface adsorbates from H-terminated diamond using high-sensitivity SCD MEMS resonators. By monitoring resonance frequency shifts, we observe the desorption dependence on processing temperature, quantify the desorbed mass (2.3 fg/μm²), and measure the equivalent adsorbed layer thickness (0.65 nm) when desorption saturates at 600°C. Additionally, we confirm the stability of C-H bonds after annealing at 600°C. These findings provide new insights into the surface dynamics of H-terminated diamond and offer a method for studying surface properties in other semiconductor materials.High temperature high pressure (HPHT) Ib-type (100) diamond substrates were used to fabricate single-crystal diamond (SCD) microelectromechanical system (MEMS) resonators via a smart-cut process based on high-energy ion implantation (see supplemental Material FIG. S1(a)). The fabrication involved: (i) High-energy ion implantation, (ii) Epitaxial regrowth via microwave plasma chemical vapor deposition (MPCVD) at 1.5% CH₄ concentration, 1 kW microwave power, and a deposition time of 30 min, (iii) Patterning using a smart-cut process, (iv) Reactive ion etching (RIE) with 90 sccm O₂, (v) Removal of the Al mask and sacrificial layer to release the cantilevers, and (vi) Hydrogen plasma treatment for 30 min to achieve H termination.The epitaxial diamond material was characterized using atomic force microscopy (AFM) and Raman spectrometer with a 532 nm laser. The device measurement setup (Supplemental Material FIG. S1(b)) integrates a laser Doppler vibrometer (LDV, LV-1710) and a lock-in amplifier (HF2-LI) for resonance spectra and noise requency measurements. Real-time surface I-V characterization of H-terminated diamond was conducted using a Keithley 2636B source meter. This setup, combined with a vacuum chamber (10-4 Pa), temperature control system, and data acquisition software, enables comprehensive characterization of the mass changes and surface electronic properties during the desorption process of adsorbates on H-terminated diamond by heating to target processing temperatures (30℃~600℃) then cooling down to 30℃.The Raman spectrum of epitaxial diamond exhibits a sharp peak at 1332.2 cm-1 with a full width of half maximum (FWHM) of 1.90 cm-1, corresponding to the sp3-bonded carbon lattice of diamond (FIG. 1(a)) 12. No additional peaks are observed, indicating high crystalline and purity. The inset of FIG. 1(a) shows the surface morphology of a 1 μm² diamond epilayer after H plasma treatment, with etching pits attributed to hydrogen plasma interactions with defects or weak regions on the surface 13,14. Despite this, the root mean square (RMS) roughness remains low at 2.91 nm. The smooth surface and high-quality crystal structure provide a solid foundation for stable H-terminated SCD cantilever resonators. FIG. 1(b) presents a 3D microscopy image of the SCD cantilever, which exhibits upward bending due to residual stress from the sacrificial layer release. After H plasma treatment, the cantilever’s surface becomes H-terminated, attracting adsorbates that induce a two-dimensional hole gas (2DHG) and measurable current (~10⁻⁴ A) under 5 V, as shown in the inset of FIG. 1(b).FIG. 1. (a) Three-dimensional (3D) optical image and (b) Raman spectrum of the H-terminated SCD cantilever resonator.The amplitude of a 140 μm-long SCD cantilever resonator increases linearly with actuation voltage at 600 ℃, as shown in FIG. 2(a), consistent with the force being proportional to the applied voltage. Additionally, shorter cantilevers exhibit higher resonance frequencies, with a 60 μm cantilever resonating at 1067.323 kHz and a 140 μm cantilever at 210.175 kHz. The frequency is inversely proportional to the square of the cantilever’s length no matter at 30 ℃ or 600 ℃, as revealed in FIG. 2(b), which follows the Euler-Bernoulli theory 15,                                                         (1)where k is a constant of 0.162 for the first mode vibration, t and L represent cantilever’s thickness and length, respectively, while Eeff and ρeff denote the effective Young’s modulus and mass densityy. These results confirm the stability and reliability of SCD cantilever resonators at both 30 °C and 600 °C, establishing them as a robust platform for studying the dynamic desorption of adsorbates on H-terminated surfaces.The real-time monitoring of surface adsorbate desorption relies on mass sensing via cantilever frequency shifts. However, frequency fluctuations from thermal noise, mechanical noise, and environmental disturbances can constrain the resolution of small mass changes. 16,17. A smaller frequency fluctuation corresponds to a higher mass resolution of the cantilever resonator. To quantitatively assess frequency stability, Allan variance analysis is employed. Allan variance, defined as the average of the squared differences between consecutive frequency measurements over a time interval τ 18-20, is crucial for evaluating resonator performance, particularly in mass sensing. The Allan deviation, the square root of the Allan variance, directly measures frequency fluctuation. The measured frequency stability over a 5-minute period with a sampling rate of 1.6k Sa/s is shown in FIG.S2. The entire frequency noise dataset is divided into N-1 parts, and each averaged over an integration time τ. Thus, the Allan deviation can be expressed as equation (2) 19.                                       (2)After calculation, the Allan deviation as a function of integration time for cantilevers of different lengths is presented in FIG. 2(c). The minimum Allan deviation of 140 μm-long, 120 μm-long, 80 μm-long, 60 μm-long cantilevers are 0.0028 Hz, 0.013 Hz, 0.024 Hz, and 0.068 Hz, occurring at integration times of 2.89 s, 2.67 s, 9.02 s, and 8.63 s, respectively. The minimum Allan deviation increase as cantilever length decreases, as displayed in FIG. 2(d). This trend arises because shorter cantilevers, with higher resonance frequencies, are more susceptible to external perturbations such as thermal, mechanical, and environmental noise, leading to greater frequency fluctuations under identical measurement conditions 21-23. The minimum detectable mass for cantilevers of different lengths is determined based on the minimum Allan deviation using equation (3) 24,25,                                                           (3)wherein f0 is the fundamental resonance frequency, m0 is the mass of cantilever, Δf is the frequency shift, and Δm is the mass change. As revealed in FIG. 2(d), the minimum detectable mass of 60 μm-, 80 μm-, 120 μm-, and 140 μm-long cantilevers are 4.72×10-16 g, 3.98×10-16 g, 7.34×10-16 g, 2.45×10-16 g, respectively. Despite differences in length, all cantilevers exhibit a comparable mass resolution on the order of ~10-16 g. Since the cantilevers are fabricated from the same SCD substrate and share identical cross-sectional geometry (e.g., width and thickness). From the equation (3), we can know that the minimum detectable mass Δmmin∝Δf/f0. Although shorter cantilever resonators have higher resonance frequency f0, they also experience larger frequency fluctuation Δf due to increased noise sensitivity. Conversely, longer cantilever resonators exhibit lower resonance frequencies but reduced frequency fluctuations caused by noise. These opposing effects effectively balance each other, resulting in similar Δmmin values across different cantilever lengths. This demonstrates that SCD cantilever resonators, ranging from 60 μm to 140 μm in length, possess the capability to detect mass changes at the molecular layer level. FIG. 2. (a) Resonance spectra of a 140 μm cantilever at different actuation voltages. (b) Resonance frequency vs. cantilever length. (c) Allan deviation vs. integration time for various cantilever lengths. (d) Frequency fluctuation and mass resolution for different cantilever lengths.The resonance spectra of H plasma-treated diamond were measured at 30℃ as a reference. As shown in FIG. 3(a), the initial resonance frequency and Q factor of a 120 μm-long cantilever were 285101.4 Hz and 6135.7, respectively. By heating the H-terminated SCD cantilever to target processing temperatures (30℃-600℃) and then cooling down to 30 ℃ subsequently, surface adsorbates such as H2O, H3O+, OH-, and HCO3- 26,27 gradually gained energy and desorbed form the diamond surface, leading to a frequency shift. After 600℃ annealing, the resonance frequency increased to 285172.1 Hz, while the Q factor improved to 6705.1. This enhancement in Q factor is attributed to the reduced surface energy damping caused by adsorbate desorption. Similar effects have been observed in single-crystal silicon cantilevers by Takahito Ono et al. 28 and Jinling Yang et al. 29. However, in this study, the effect on the H-terminated diamond surface was minimal. This difference arises because silicon surfaces consist of a solid oxide film, whereas the H-terminated diamond surface hosts a molecular adsorbate layer. FIG. 3(b) presents the frequency shift of the 120 μm-long cantilever as a function of temperature, revealing three distinct regions: I (30℃–200℃), II (200℃–450℃), and III (450℃–600℃). In region I, the frequency shift of ~21.5 Hz corresponds to a 1.2 pg mass desorption, primarily from water molecules. In region II, the shift of ~27.3 Hz (~1.5 pg) is attributed to the desorption of weakly bound anions, such as HCO₃⁻, which become mobile above 300℃ 1. Between 420℃ and 450℃, the frequency stabilizes, indicating near-complete desorption of these anions. As the temperature increases, the anions with stronger C-H bonding force on the H-terminated diamond surface gradually escape, such as OH-. The desorption process tends to be saturated at 600 ℃, and the desorption mass in region Ⅲ is around 1.2 pg. The total mass of desorbed adsorbates including water moleculars and anions after 600 ℃ annealing is approximately 3.9 pg. The consistent frequency shift and Q factor trends observed for 60 μm-, 80 μm-, and 140 μm-long cantilevers (FIG. S3) further validate the reliability of diamond MEMS cantilevers for dynamic surface desorption monitoring.The surface conductivity of H-terminated diamond is directly determined by surface adsorbates, which act as electron acceptors, drawing electrons from the diamond valence band and generating hole carriers, leading to p-type surface conductivity. To verify the adsorbate desorption process, I-V characteristics were recorded at 30°C after annealing at different processing temperatures, as shown in FIG. 3(c). The initial current is ~10-5 A measured in vacuum at 30 ℃ and under 5 V applied voltage and degrade gradually with increasing temperature. Dynamic desorption can alter the surface hole carrier density, resulting in the change of surface conductivity. Thus, after 600 ℃ annealing, the desorption of water molecules and anions led to a drastic current decrease to ~10-10 A at 5 V. At lower processing temperatures below 350°C, the metal and H-terminated diamond interface maintained an ohmic contact due to high surface carrier density on the H-terminated diamond surface. However, as the temperature increased, the contact gradually transitioned into a Schottky contact due to reduced carrier density and an increased barrier 6,30,31. The current trend in FIG. 3(d) aligns with the frequency shift in FIG. 3(b), with three distinct regions. In region Ⅰ, the desorption of 1.2 pg water molecules caused the surface current to drop from 4.57×10-5 A to 4.64 ×10-6 A. For region Ⅱ between 200 ℃ to 400 ℃, the current drops two orders of magnitude to 1.13×10-8 A, and almost keep unchanged from 400 ℃ to 450 ℃, corresponding to the frequency changing trend in this temperature region. In region Ⅲ, further desorption at 600℃ reduced the current to ~10⁻¹⁰ A, suggesting only a few strongly bound adsorbates remained. To assess whether 600°C annealing disrupted the C-H bond structure of the H-terminated diamond, the SCD MEMS device was re-exposed to air for 12 hours. The surface current returned to the same magnitude as the initial state~ 10-4 A, indicating H-terminated diamond almost maintain structural integrity and attract new adsorbates from air to form hole carriers. This finding aligns with prior studies by V. Serpente et al. 32 and further confirms that breaking C-H bonds in a vacuum requires temperatures exceeding 600℃ 33,34. These results reaffirm the dynamic desorption process of adsorbates on H-terminated diamond surfaces.FIG. 3. (a) Resonance spectra of a 140 μm-long cantilever at 30 °C and after cooling from 600 °C. (b) Frequency shift and desorbed mass per unit area as a function of temperature. (c) Surface current-voltage (I-V) characteristics of H-terminated diamond at different temperatures. (d) Surface current at −5 V vs. processing temperature. The inset shows the temperature dependence of the surface current at -5 V.To assess the uniformity of adsorbates on the surface of H-terminated diamond, we recorded the frequency shifts of cantilevers of different lengths after 600°C annealing, as shown in FIG. 4(a). The frequency shift of the 60 μm-, 80 μm-, 120 μm-, and 140 μm-long cantilevers are 249.0 Hz, 148.0 Hz, 70.7 Hz, and 47.8 Hz, corresponding to the mass change of 1.8 pg, 2.4 pg, 3.9 pg, and 4.3 pg, respectively. Since all cantilevers were fabricated on the same diamond substrate with identical cross-sectional geometries (width and thickness), the desorbed mass exhibited an approximately linear relationship with cantilever length. The measurement errors of 0.1~0.3 pg exists among cantilever beams of different lengths, which may be attributed to variations in residual stress 35,36. Regardless of cantilever length, the mass per unit area remained around 2.3 fg/μm², indicating a uniform distribution of surface adsorbates on the H-terminated diamond. Due to the diverse nature of adsorbate species, determining the exact effective density is challenging. However, assuming a density of 3.5 g/cm³ (similar to diamond), the estimated adsorbate layer thickness is approximately 0.65 nm.FIG. 4. (a) Resonance frequency shift and corresponding desorbed mass for cantilevers of different lengths. (b) Mass per unit area and estimated desorption layer thickness for cantilevers of varying lengths.In conclusion, single-crystal diamond MEMS has been employed for the first time to investigate the dynamic desorption process of adsorbates on H-terminated diamond by analyzing changes in vibration characteristics and surface conductivity with respect to temperature. The desorption process reaches saturation at approximately 600℃, with a desorbed mass per unit area of 2.3 fg/μm² for 30-minute H plasma-treated diamond. It is confirmed that the C-H bonds maintain structural stability after 600℃ annealing, with the desorbed species consisting primarily of water molecules and anions adsorbed on the surface. These findings provide valuable insights into surface chemistry, aid in optimizing surface properties, and contribute to the development of advanced electronic devices based on H-terminated diamond. This work demonstrates that MEMS technology offers a precise and detailed method for analyzing the surface properties of semiconductors.This work was supported by JSPS KAKENHI (Grant Number 20H02212, 22K18957, 15H03999), Bilateral joint research between JSPS/CAS, and Advanced Research Infrastructure for Materials and Nanotechnology in Japan (ARIM) of MEXT (JPMXP1223NM5297). Keyun Gu thanked to the financial support from China Scholarship Council (No. 202306890007).1 F. Maier, M. Riedel, B. Mantel, J. Ristein, and L. Ley, Physical review letters 85 (16), 3472 (2000).2 C. Nebel, C. Sauerer, F. Ertl, M. Stutzmann, C. Graeff, P. Bergonzo, O. A. Williams, and R. Jackman, Applied physics letters 79 (27), 4541 (2001).3 K. Gu, K. Wu, Z. Zhang, T. Ohsawa, J. Huang, Y. Koide, M. Toda, and M. Liao, Advanced Functional Materials, 2420238 (2024).4 Y. Sasama, T. Kageura, M. Imura, K. Watanabe, T. Taniguchi, T. Uchihashi, and Y. Takahide, Nature Electronics 5 (1), 37 (2022).5 H. Sato and M. Kasu, Diamond and related materials 24, 99 (2012).6 H. Kawarada, Surface Science Reports 26 (7), 205 (1996).7 S. Matsumoto and N. Setaka, Carbon 17 (6), 485 (1979).8 S. Kumaragurubaran, T. Yamada, and S. Shikata, Diamond and related materials 17 (4-5), 472 (2008).9 S. Michaelson, R. Akhvlediani, and A. Hoffman, Physical Chemistry Chemical Physics 13 (24), 11471 (2011).10 M. Liao, Functional Diamond 1 (1), 29 (2022).11 M. Possas, L. Rousseau, F. Ghassemi, G. Lissorgues, P. Gonzales, E. Scorsone, and P. Bergonzo, Procedia engineering 120, 1115 (2015).12 R. Brunetto, G. Baratta, and G. Strazzulla, Journal of Applied Physics 96 (1), 380 (2004).13 X. Jiang and C. Rickers, Applied Physics Letters 75 (25), 3935 (1999).14 N. Tsubouchi, Y. Mokuno, and S. Shikata, Diamond and Related Materials 63, 43 (2016).15 W. Weaver Jr, S. P. Timoshenko, and D. H. Young, Vibration problems in engineering. (John Wiley & Sons, 1991).16 A. Cleland and M. Roukes, Journal of applied physics 92 (5), 2758 (2002).17 J. Mouro, R. Pinto, P. Paoletti, and B. Tiribilli, Sensors 21 (1), 115 (2020).18 D. Jin, X. Li, J. Liu, G. Zuo, Y. Wang, M. Liu, and H. Yu, Journal of Micromechanics and Microengineering 16 (5), 1017 (2006).19 D. W. Allan, Proceedings of the IEEE 54 (2), 221 (1966).20 P. Sadeghi, A. Demir, L. G. Villanueva, H. Kähler, and S. Schmid, Physical Review B 102 (21), 214106 (2020).21 M. Li, H. X. Tang, and M. L. Roukes, Nature nanotechnology 2 (2), 114 (2007).22 B. D. Wiltshire, M. Alijani, S. Mohammadi, A. Hosseini, J. M. Macak, and M. H. Zarifi, ACS Applied Materials & Interfaces 14 (4), 6203 (2022).23 A. Husain, J. Hone, H. W. C. Postma, X. Huang, T. Drake, M. Barbic, A. Scherer, and M. Roukes, Applied Physics Letters 83 (6), 1240 (2003).24 K. Jensen, K. Kim, and A. Zettl, Nature nanotechnology 3 (9), 533 (2008).25 L. Wei, X. Kuai, Y. Bao, J. Wei, L. Yang, P. Song, M. Zhang, F. Yang, and X. Wang, Micromachines 12 (6), 724 (2021).26 H. Kawarada, Japanese Journal of Applied Physics 51 (9R), 090111 (2012).27 M. Kubovic, M. Kasu, and H. Kageshima, Applied Physics Letters 96 (5) (2010).28 T. Ono, D. F. Wang, and M. Esashi, Applied physics letters 83 (10), 1950 (2003).29 J. Yang, T. Ono, and M. Esashi, Applied Physics Letters 77 (23), 3860 (2000).30 T. Teraji, S. Koizumi, and Y. Koide, Journal of Applied Physics 104 (1) (2008).31 H. Kawarada, in The Physics of Diamond (IOS Press, 1997), pp. 505.32 V. Serpente, A. Bellucci, M. Girolami, M. Mastellone, S. Iacobucci, A. Ruocco, and D. Trucchi, Diamond and Related Materials 114, 108290 (2021).33 R. Thomas, R. Rudder, and R. Markunas, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 10 (4), 2451 (1992).34 C. Su and J.-C. Lin, Surface science 406 (1-3), 149 (1998).35 S. Massl, J. Keckes, and R. Pippan, Scripta materialia 59 (5), 503 (2008).36 B. Wang and K. Wang, Journal of Applied Physics 113 (15) (2013).2image2.emf209.0 209.1 209.2 209.30.000.040.080.12Amplitude (Vrms)Frequency (kHz) 3 V 4 V 5 V 6 V 7 V 8 V140 μm-longMeasured at 600 ℃(a)(b)(c)(d)image3.emf284.8285.0 285.2 285.4Amplitude (a.u.)Frequency (kHz) 30 ℃ base 600 ℃ back to 30 ℃f: 285101.4 HzQ: 6135.7f: 285172.1 HzQ: 6705.1120 μm-long(a) (b)(c)(d)image4.emf6080100120 14050100150200250 Frequency shift Δf Desorption mass ΔmLength (μm)Frequency shift (Hz)1.52.02.53.03.54.04.5Deaorption mass (pg)(a)(b)image1.emf12001300 1400 1500Intensity (a.u.)Raman shift (cm-1)Peak position: 1332.2 cm-1FWHM: 1.90 cm-1 Measurement Lorentz Fitting(a)(b)AdsorbatesRMS=2.91 nm1     Real - time monitoring of adsorbate desorption on H - terminated diamond using single - crystal diamond MEMS resonators   Keyun Gu, 1,2   Zilong Zhang, 1   Jian Huang, 2   Yasuo Koide, 1   Satoshi Koizumi, 1   and Meiyong  Liao 1, a)     1   Research Center for Electronics and Optical Materials, National Institute for Materials  Science, Namiki 1 - 1, Tsukuba, Ibaraki 305 - 0044, Japan    2   School of Materials Science and Engineering, Shanghai University, Shanghai 200444, China   a)   Electronic mail :  meiyong.liao@nims.go.jp     ABSTRACT   The ultimate  physical and chemical   properties of hygrogen (H) - terminated diamond  devices are  strong influenced   by real - time surface adsorbates. In this study, we investigated the  dynamic thermal desorption process of these adsorbates using an innovative approach — single - crystal diamond (SCD) microelectromechanical system (MEMS) resonators. By analyzing the  variatio ns in resonance performance and surface conductivity, we uncover several important  findings: (i) The desorption of surface adsorbates reaches satu ration around 600℃, (ii) The  desorbed mass per unit area is approximately 2.3 fg/ μ m 2 , corresponding to an equivalent  thickness of  around   0.65 nm ,   (ii i )   Surface conductivity recovers after annealing at 600°C,  indicating the stability of C - H bonds, and (iv) The impact of the adsorbate layer on mechanical  energy dissipation  in   H - terminated diamond devices is significantly lower than that of naturally  oxidized films in other semiconductors. These findings provide a reliable method for   evaluating  the surface properties of H - terminated diamond and offer a new approach for studying the  surface ch aracteristics of other semiconductor materials