Manufacturing of 8 million Q-factor micro hemispherical resonator gyroscopes via patterned coating technology Authors: Feng Zhu, Xuezhong Wu, Yan Shi, Mengxue Li, Jiangkun Sun, Kun Lu, Yongmeng Zhang, Dingbang Xiao, Xiang Xi Published: 2026 ABSTRACT As the core sensitive element of high-precision inertial navigation systems, the Q-factor and frequency splitting of the micro hemispherical resonator gyroscope (mHRG) are critical determine the performance of the inertial measurement unit (IMU). However, the continuous films formed by traditional coating processes introduce significant damping, leading to a substantial decrease in the Q-factor and severely restricting the improvement of the mHRG’s accuracy. In this paper, a patterned coating solution based on magnetron sputtering physical vapor deposition is proposed. High-precision patterned deposition of Ti/Pt thin films on the three - dimensional complex curved surface of the shell has been successfully achieved. The performance of the packaged device was characterized. The results indicate that the Q-factor loss rate of the mHRG after patterned coating is less than 14%, and the measured Q-factor is as high as over 8 ×10⁶. The circumferential Q-factor uniformity is better than 1%, and the frequency splitting is controlled within 1 mHz. All indicators outperform those of the traditional coating process. Mechanism analysis indicates that the patterned design significantly reduces the additional damping caused by the friction at the grain boundaries of the thin film and at the film-substrate interface by greatly reducing the coverage area of the thin film, which is the key to achieving a high Q-factor. Based on the harmonic error analysis theory, this paper establishes a correlation model between the film pattern and the circumferential uniformity of the device as well as frequency splitting, ensuring the uniformity of various performance parameters of the gyroscope in its working mode. Model calculations and experimental results have verified that this patterned film can significantly enhance the Q-factor, and is effective in suppressing the circumferential Q-factor non-uniformity and frequency splitting, providing key process support and theoretical basis for high-performance mHRG. We proposed a patterned coating process using magnetron sputtering and 3D masks. -Prepared mHRG has Q-factor>8M, outperforming traditional full-coat devices. -Clarified the principle by which patterned thin films reduce thin film damping. -The thin film patterns based on the harmonic error theory ensure uniformity Introduction Quality factor (Q-factor) is a key characteristic of micro/nano-electromechanical systems (MEMS/NEMS) resonators, which are the core components of resonator gyroscopes, accelerometers, mass sensors, and other devices1–6. The micro hemispherical resonator gyroscope (mHRG) has significant advantages such as small size and weight, as well as strong batch manufacturing capabilities, and has been a key focus of research in the industry in recent years7–9. The hemispherical resonator gyroscope (HRG) is a high-precision and highly reliable gyroscope and is a relatively mature product among resonant gyroscope10–12. Compared with HRG, although mHRG has the prominent advantage of structural miniaturization, it still has deficiencies in Q-factor performance13–15. Based on the constraint of the Q-factor on the performance of mHRG, we have studied many process methods to improve the Q-factor of mHRG7,16–18. For instance, process research on the thermally induced elastic damping of fused silica can reduce its thermally induced elastic Q-factor loss. Laser modification combined with wet etching can minimize the Q-factor loss caused by release damage. Moreover, multiple harmonic tuning can reduce the energy dissipation due to support imbalance. Although these studies have achieved fused silica resonators with relatively high Q-factor, manufacturing a high Q-factor mHRG remains a significant challenge based on current research. This is mainly due to the fact that fused silica resonators require coating to enable electrical conduction and ensure functional operation, and energy dissipation induced by the thin film severely restricts the Q-factor of the mHRG. The common approach to achieving conductivity in fused silica resonators is to deposit adhesion layers and conductive layers, such as Cr/Au, Ti/Pt, etc. film combinations19–21, and some research has been conducted to reduce the impact of film damping on the Q-factor of the resonator. For instance, the Q-factor can be enhanced by reducing the film thickness22, improving the annealing process23, and regulating the uniformity of the film distribution4, among other technological approaches. Although some effective results have been achieved, thin film damping remains a key issue restricting the production of high Q-factor mHRG. An effective process method and the dissipation mechanism of thin film damping have yet to be clearly defined. Exploring the application of different types of thin film materials in resonators is another approach to reducing thin film damping. For example, growing graphene (a two-dimensional carbon material) on the HRG24, Depositing a metal Ir thin film on a mHRG15, And the application of niobium nitride material films in high Q-factor resonators25. However, most applications of new thin film materials for resonators remain in the stage of technical exploration to date, and there are no reports of sensors that exceed industry standard levels. In this paper, we have deposited Ti/Pt patterned films on the three-dimensional complex curved surface of the resonator shell through the magnetron sputtering physical vapor deposition process. The Q-factor, Q-factor uniformity and frequency splitting of the packaging mHRG were tested. The test results show that the Q-factor loss rate of the mHRG after the patterned coating is less than 14%, its Q-factor is higher than 8 million, the Q-factor uniformity is less than 1%, and the frequency splitting is less than 1 mHz. The realization of high Q-factor mHRG is mainly attributed to the significant reduction in the film area by the patterned thin film, which ultimately greatly reduces the film damping by lowering the energy dissipation caused by the friction at the grain boundaries of the film and the interface between the film and the fused silica substrate. Based on the harmonic error analysis theory, we designed the thin film pattern and correlated it with the circumferential Q-factor uniformity and frequency splitting. Based on the model calculation results, it has been proven that this patterned film can control the uniformity in the working mode of the gyroscope, the non-uniformity of the circumferential Q-factor, and the frequency splitting. Result Patterned thin films mHRG We implemented patterned coating through the preparation of three-dimensional masks. Ultimately, by reducing the coating area of the mHRG, the film damping of the gyroscope was decreased, thus enabling the manufacture of high Q-factor mHRG. A brief introduction to the structure and working principle of mHRG: The main structure and dimensions of the resonator are shown in Fig. 1a, where r = 6 mm, h = 4 mm, and d = 200 μm. An AC voltage (500 mV) is applied to the electrode substrate of the mHRG, and a DC bias voltage (100 V) is applied to the anchor point. It mainly consists of a resonator and an electrode substrate, both of which are made of fused silica material, the assembled mHRG is shown in Fig. 1e. After the resonator and electrode substrate are coated with a thin film, they can conduct electricity. This mHRG adopts electrostatic drive and capacitive detection technology. The assembly gap between the electrode substrate and the teeth of the resonator forms a planar capacitor. Thus, the thin film on the resonator must establish a connection from the resonator anchor points to the mass teeth for the mHRG to operate properly.Fig. 1Schematic design and experimental characterization of a patterned thin-film mHRG.a Integrating structural dimensions with circuit principles. b Physical image of the 3D-printed patterned thin-film mask. c Schematic of the magnetron sputtering thin-film process apparatus, showing key deposition equipment and operating conditions. d Local structure image of the patterned thin-film resonator, revealing the resonator after mask-assisted patterning. e Fully-covered thin-film mHRG. f Patterned thin-film mHRG, corresponding to the final device fabricated via the mask and sputtering processes Schematic design and experimental characterization of a patterned thin-film mHRG. a Integrating structural dimensions with circuit principles. b Physical image of the 3D-printed patterned thin-film mask. c Schematic of the magnetron sputtering thin-film process apparatus, showing key deposition equipment and operating conditions. d Local structure image of the patterned thin-film resonator, revealing the resonator after mask-assisted patterning. e Fully-covered thin-film mHRG. f Patterned thin-film mHRG, corresponding to the final device fabricated via the mask and sputtering processes Masking is a crucial process for realizing the patterned thin film of mHRG. The manufacturing accuracy of the mask determines the dimensional accuracy of the thin film and the performance of mHRG. Three-dimensional masks are fabricated through 3D printing technology26–28, with a layer precision of 2 μm, as shown in Fig. 1b. To ensure excellent adhesion and conductivity, we deposited a 10 nm Ti layer followed by a 20 nm Pt layer on the surface of the micro-hemispherical resonator. This design represents a comprehensive balance between adhesion, conductivity, and film uniformity22. The deposition of the metal thin film is achieved via the magnetron sputtering coating process, as depicted in Fig. 1c. Keeping the initial vacuum level prior to coating below 10⁻⁴ Pa is conducive to ensuring the stability of plasma discharge, enhancing the sputtering efficiency of the target material, and minimizing the impurity content within the thin film29–31. Heating the fused silica resonator to a high temperature of 100 °C facilitates the formation of large-grained thin films and the stabilization of the mechanical properties of the thin films32–34. The high-speed rotation of the coating pedestal of the fused silica resonator is instrumental in enhancing the circumferential uniformity of the deposited thin film. In turn, this mitigates the circumferential non-uniformity of the Q-factor that is attributable to the damping effect of the thin film22. The resonator with patterned films deposited on the magnetron sputtering platform is shown in Fig. 1d, and it is fixed on the electrode by mechanical means. By comparing it with the traditional full coverage coated mHRG, it can be seen that the mHRG fabricated by the patterned film process significantly reduces its coating area. The following will discuss the reasons for the patterned coating and the design ideas of the pattern thin film. Patterned thin films with low damping In order to explore the impact of patterned thin films on the Q-factor of the mHRG, it is of necessity to analyze the energy dissipation mechanism of thin film damping in the resonator. A clear discussion of this issue contributes to understanding the mechanism by which the Q-factor of the resonator is enhanced through the patterned thin films. Existing studies have explored the thermoelastic damping of thin films. However, relevant research reveals a significant order-of-magnitude difference in thickness between the thin film and the substrate (micron-scale and nanometer-scale respectively). Therefore, the thermoelastic damping of the thin film is not the crucial factor influencing the Q-factor of the mHRG17. Research on thin film friction also indicates that the internal friction of metal films mainly stems from two aspects: one is caused by residual stress during film deposition, and the other is related to dislocation movement. Due to structural defects in metal films and unreleased internal stress during sputtering and deposition, their internal friction is relatively high, which will increase the energy loss of resonators35–37. However, these studies have not yet clarified the loss mechanism of thin film damping, and thus it is difficult to directly use them as theoretical evidence for the claim that patterned thin films can enhance the Q-factor of mHRG by reducing the thin film area. Therefore, we will analyze the relationship between frictional losses and the area of the resonator coating from the perspectives of grain boundary friction and interface friction in polycrystalline thin films, and then explain how patterned thin films can reduce the Q-factor loss of the resonator by minimizing frictional losses. Grain boundary friction Metal films prepared by magnetron sputtering often exhibit a polycrystalline structure38. The atoms in the grain boundary regions are in a metastable state of structural disorder, and local lattice distortion causes the formation of defect-rich zones at the interfaces, as evidenced by the atomic-scale disorder and defect aggregation shown in Fig. 2b. Obvious grains exist in Pt and other metal films. When the resonator vibrates, cyclic deformation induces a non-uniform strain field within the film material, and this non-uniform mechanical response leads to sliding friction at the grain boundaries, as illustrated in Fig. 2a.Fig. 2Friction and energy dissipation at grain boundaries and interfaces of the resonator film.a Schematic diagram of frictional heat dissipation at the grain boundaries of the resonator film. b TEM image of the grain boundaries of the film. c Schematic diagram of friction at the film-substrate interface. d TEM image of the interface between the film and the fused silica substrate. e TEM image of the interface between the films Friction and energy dissipation at grain boundaries and interfaces of the resonator film. a Schematic diagram of frictional heat dissipation at the grain boundaries of the resonator film. b TEM image of the grain boundaries of the film. c Schematic diagram of friction at the film-substrate interface. d TEM image of the interface between the film and the fused silica substrate. e TEM image of the interface between the films The Prandtl-Tomlinson (PT) model is a foundational framework for atomic-scale friction analysis. This model represents the upper surface of the frictional interface as a driving block moving at constant velocity. Driven periodically by a resonator, this system yields the total potential energy during sliding. The PT model is applied to analyze the internal friction of the resonator thin film, where VPT represents the total potential energy of atoms along the tangential sliding direction at the film’s grain boundary, as shown in Eq. 139.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{PT}(x,t)=-\frac{U}{2}\,\cos \left(\frac{2{\rm{\pi }}x}{a}\right)+\frac{1}{2}{k}_{PT}{[vt-x]}^{2}$$\end{document}VPT(x,t)=−U2cos2πxa+12kPT[vt−x]2 VPT denotes the total potential energy of the system (J); x denotes the tangential relative displacement between adjacent grains at the grain boundaries (m); U represents the energy amplitude of grain boundary corrugation at the quasi-static interface during thin film (J);a is the lattice period (m); v is the sliding velocity of the equivalent mass block of the grain (m/s); t is its sliding time (s); kPT is the stiffness of the grain structure of the thin film (N/m). In the context of the thin film grain boundary region, this study defines three core physical quantities: the damping force Ff, the equivalent damping coefficient Cf, and the relative frictional velocity dx/dt. The mathematical relationship among these three physical quantities is given by Eq. 22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{f}={c}_{f}\cdot \frac{{\rm{d}}x}{{\rm{d}}t}$$\end{document}Ff=cf⋅dxdt Let the frictional heat conversion coefficient of the resonator thin film be α, the surface area of the thin film be S (m²), and the grain boundary density be ρgb (m⁻²). An alternative expression for the damping force is then derived, as given by Eq. 3.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{f}=\frac{\partial (S\cdot {\rho }_{gb}\cdot \alpha \cdot {V}_{PT})}{\partial x}=S\cdot {\rho }_{gb}\cdot \alpha \left(\frac{U{\rm{\pi }}}{a}\,\sin \left(\frac{2{\rm{\pi }}x}{a}\right)+{k}_{PT}x-{k}_{PT}vt\right)$$\end{document}Ff=∂(S⋅ρgb⋅α⋅VPT)∂x=S⋅ρgb⋅αUπasin2πxa+kPTx−kPTvt By combining Eqs. 2 and 3, the equivalent damping coefficient Cf for intergranular friction within the thin film is derived as Eq. 4. An increase in grain boundary density thus results in a proportional increase in Cf.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{f}=S\cdot {\rho }_{gb}\cdot \alpha \left(\frac{U{\rm{\pi }}}{av}\,\sin \left(\frac{2{\rm{\pi }}x}{a}\right)+{k}_{PT}\frac{x}{v}-{k}_{PT}t\right)$$\end{document}Cf=S⋅ρgb⋅αUπavsin2πxa+kPTxv−kPTt Interface friction In MEMS resonators, the thin film interface is one of the critical regions for energy dissipation. Relevant studies have shown that during thin film deposition, defects tend to form at the interface and residual stress is generated. In regions of stress concentration, the vibration damping significantly increases, thereby intensifying the conversion of energy into thermal energy. Additionally, some scholars have pointed out that lattice mismatch between the substrate and the thin film introduces interface defects, which cause extra energy dissipation40,41. Moreover, the interface dissipation of thin films can also be explained from the perspective of continuum mechanics. For fused silica mHRG, the amorphous material properties make it possible to explain such losses more directly from a mechanical perspective. After coating, the fused silica resonator has interfaces between the thin film and the fused silica substrate, as well as between the thin film conductive layer and the thin film adhesion layer, as shown in Fig. 2c. During operation of the resonator, periodic elastic deformations occur at various positions42, and there are differences in the elastic moduli of the metal thin film and the fused silica substrate43,44. For instance, the elastic modulus of fused silica is 73 GPa, that of Ti, often used as the adhesion layer, is 110 GPa, and that of Pt, commonly used as the conductive layer, is 168 GPa. Ti exhibits superior adhesion layer properties compared to Cr (245 GPa) and forms an elastic modulus gradient with platinum. However, the modulus difference may still induce stress jumps under cyclic vibration. Let the elastic strain in the x-direction of the resonator during vibration be ∂u/∂x, then the stress of the metal film and substrate can be expressed by Eq. 5. Specifically, u is the displacement of the resonator along the x direction during vibration. E1 and E2 represent the elastic moduli of the thin film and the substrate, respectively.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left(\begin{array}{l}{\sigma }_{1}={E}_{1}\cdot \frac{\partial u}{\partial x}\\ {\sigma }_{2}={E}_{2}\cdot \frac{\partial u}{\partial x}\\ \Delta E={E}_{1}-{E}_{2}\\ \Delta \sigma ={\sigma }_{1}-{\sigma }_{2}=\Delta E\cdot \frac{\partial u}{\partial x}\end{array}\right)$$\end{document}σ1=E1⋅∂u∂xσ2=E2⋅∂u∂xΔE=E1−E2Δσ=σ1−σ2=ΔE⋅∂u∂x According to the theory of continuum mechanics45–48, the stress difference will cause a slip uslip at the interface between the film and the substrate. ω is the resonant frequency of the mHRG. λ is a coefficient (m³/Ns) that is only related to the interface properties.6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${u}_{slip}=\lambda \frac{\Delta E}{\omega }\frac{\partial u}{\partial x}$$\end{document}uslip=λΔEω∂u∂x The energy loss, denoted as ΔW, resulting from the friction at the interface between the thin film and fused silica is determined by S. Here, S represents the area of the thin film (i.e., the interface area), which is dictated by the area of the thin film deposited on the resonator. Therefore, by reducing the film coverage area on the resonator, the energy dissipation caused by the friction at the film interface can be decreased, thereby enhancing the Q-factor.7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta W={\rm{\pi }}{\int }_{S}\Delta \sigma \cdot {u}_{slip}{{\rm{d}}}_{s}$$\end{document}ΔW=π∫SΔσ⋅uslipds Based on the theoretical analyses and derivations presented in the Grain Boundary Friction and Interface Friction sections, From the perspective of grain boundaries, by combining the expression of the equivalent damping coefficient (Eq. 4) derived from the PT model, the film coverage area S and the grain boundary density ρgb jointly determine the overall effect of grain boundary friction. An increase in S will directly enlarge the effective region of grain boundary friction. As a result, the equivalent damping coefficient Cf increases in a positive correlation with S, thereby exacerbating the energy dissipation of the resonator. Viewed from the interface level, the formula for interface friction energy loss (Eq. 7) deduced from continuum mechanics shows that the energy loss ΔW between the thin film and the substrate or between the conductive layer and the adhesive layer is directly proportional to the thin film coverage area S. Additionally, an increase in S will expand the stress jump region induced by the difference in elastic moduli between the thin film and the substrate. This further promotes the interface slip uslip (Eq. 6), thereby leading to additional energy dissipation. Although grain boundary friction exerts a significant influence, annealing can increase the grain size of thin films and reduce the area of grain boundary friction, thereby substantially mitigating its adverse effects. While interfacial friction accounts for a smaller proportion, it remains difficult to effectively suppress through existing processes. Therefore, reducing the film area S offers a targeted and effective solution: on one hand, it shrinks the effective area for grain boundary friction, suppressing the increase in the equivalent damping coefficient Cf; on the other hand, it minimizes stress gradients and slip ranges at interfaces, reducing the interfacial friction energy loss ΔW. The combined effect of these two aspects effectively suppresses energy loss caused by grain boundary and interface friction, ultimately mitigating the adverse impact of thin-film friction damping on the Q-factor of mHRG. The theory of thin film uniformity error When reducing the thin film Q-loss by decreasing the area of the thin film covering the resonator, it is also necessary to take into account the influence of the thin film distribution uniformity on the performance of the gyroscope. The mHRG is a circumferentially symmetric gyroscope featuring periodic vibration characteristics. In the case of a circumferentially non-uniform distribution of the metal thin film on the resonator, specific harmonic components will emerge in the vibration signal, thereby affecting the overall performance10,49,50. These harmonic components are closely associated with factors such as the position, size, and shape of the non-uniform thin film distribution. Since the Fourier series has the ability to decompose periodic signals into a series of harmonic components, it is highly applicable to the harmonic decomposition of circumferential non-uniformity errors in the metal thin films of resonators11. By means of Fourier series analysis, the vibration signal of the resonator can be decomposed into harmonic components of various frequencies, and the specific harmonic components induced by the non-uniform metal film can be identified. These harmonic components encompass the harmonic errors of the non-uniform mass m and the harmonic errors of the non-uniform damping c. When a circumferentially non-uniform metal film is deposited at the angular position φ of the resonator, a non-uniform mass distribution m(φ) and a non-uniform damping distribution c(φ) will arise at the angular position φ. Among these, expanding m(φ) in a Fourier series can yield the harmonic distribution functions of the circumferentially non-uniform mass m as shown in Eq. 811. Herein, θn represents the phase angle of the n-th harmonic unbalanced mass.8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m(\varphi )={m}_{0}+\mathop{\sum }\limits_{n=1}^{\infty }\{\partial {m}_{n}\,\cos [n(\varphi -{\theta }_{n})]\}$$\end{document}m(φ)=m0+∑n=1∞{∂mncos[n(φ−θn)]} Owing to the circumferential non-uniformity of the thin film damping, the value of ξ(φ) varies with the circumferential angle. Likewise, through Fourier series decomposition, the harmonic distribution Eq. 9 of the circumferential non-uniformity of the film attenuation ξ can be derived as follows:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi (\varphi )={\xi }_{0}+\mathop{\sum }\limits_{n=1}^{\infty }\{\partial {\xi }_{n}\,\cos [n(\varphi -{\theta }_{n})]\}$$\end{document}ξ(φ)=ξ0+∑n=1∞{∂ξncos[n(φ−θn)]} One of the primary causes of the frequency splitting in the mHRG is the fourth-harmonic error ε4 resulting from mass non-uniformity. The magnitude of the frequency splitting Δω can be expressed by Eq. 10. The resulting frequency splitting is proportional to the relative amplitude of the fourth-order harmonic mass distribution in the circumferential direction51,5210\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta \omega ={\omega }_{1}-{\omega }_{2}={\mu }_{4}{\varepsilon }_{4}{\omega }_{12}$$\end{document}Δω=ω1−ω2=μ4ε4ω12 The magnitudes of the first, second, and third harmonic errors ε1, ε2, and ε3 caused by mass non-uniformity and the frequency splitting can be expressed by Eq. 11. The square of the relative amplitudes of the first three harmonic errors due to mass non-uniformity is proportional to the magnitude of the frequency splitting.11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{\begin{array}{l}\Delta \omega ={\omega }_{1}-{\omega }_{2}={\mu }_{1}{\varepsilon }_{1}^{2}{\omega }_{12}\\ \Delta \omega ={\omega }_{1}-{\omega }_{2}={\mu }_{2}{\varepsilon }_{2}^{2}{\omega }_{12}\\ \Delta \omega ={\omega }_{1}-{\omega }_{2}={\mu }_{3}{\varepsilon }_{3}^{2}{\omega }_{12}\end{array}\right.$$\end{document}Δω=ω1−ω2=μ1ε12ω12Δω=ω1−ω2=μ2ε22ω12Δω=ω1−ω2=μ3ε32ω12 In summary, we can assess the influence of the circumferential uniformity of the film on the performance of the mHRG by analyzing the harmonic errors induced in the film. Given the effects of the first, second, third, and fourth circumferential harmonic errors of the film on the frequency splitting and drift rate of the mHRG, when fabricating patterned films, the magnitudes of the first four harmonic errors they generate are also the key considerations. Normally, the conventional film coating of an mHRG covers the entire inner surface of the resonator. However, according to the operating principle of the mHRG described above, it is only necessary to connect the anchor points and mass teeth on the shell. It is necessary to design a thin film connection pattern with smaller errors in the first four harmonics. This thin film layout method can reduce its coverage area on the resonator surface while ensuring uniformity, thereby reducing energy loss caused by damping and improving the Q-factor of the mHRG. Thin film pattern Based on the correlative analysis between the thin film uniformity and the performance of mHRG, a circumferential harmonic-error decomposition analysis of the mass and damping harmonics of the patterned thin film is required. The prerequisite for the algorithm calculation is that the energy dissipation resulting from the internal friction within the thin film grain boundaries and the friction at the film/substrate interface is macroscopically homogeneous. The damping-harmonic error is solely determined by the deposition position distribution and thickness of the thin film. Thus, when designing patterned thin films, it suffices to analyze the harmonic error arising from the thin film area, without differentiating between the mass-harmonic error and the damping-harmonic error. In addition, according to the distribution of the proportion of the deformation energy of the mHRG, it is evident that the thin film on the resonator anchor has little impact on its Q-factor22. Consequently, a more straightforward process solution of full-film coating can be employed in the anchor region. Based on the foregoing discussion, the results of the harmonic-error decomposition of different thin film patterns are presented in Fig. 3. The thin film patterns of lines 1, 2, 3, and 4 generate the first four harmonic errors to varying extents, which can induce frequency splitting and standing-wave drift of the resonator. For the thin film patterns of lines 5, 6, 7, and 8, the first four harmonic errors are less than 2% of the amplitude, and no significantly influential first four harmonic errors are produced. Compared with the thin film patterns of lines 6, 7, and 8, the thin film area covered by line 5 is smaller, rendering it a relatively ideal option among these patterns.Fig. 3Film patterns with varying wire counts and their harmonic decomposition.a Single-wire pattern & decomposition. b Two-wire pattern & decomposition. c Three-wire pattern & decomposition. d Four-wire pattern & decomposition. e Five-wire pattern & decomposition. f Six-wire pattern & decomposition. g Seven-wire pattern & decomposition. h Eight-wire pattern & decomposition Film patterns with varying wire counts and their harmonic decomposition. a Single-wire pattern & decomposition. b Two-wire pattern & decomposition. c Three-wire pattern & decomposition. d Four-wire pattern & decomposition. e Five-wire pattern & decomposition. f Six-wire pattern & decomposition. g Seven-wire pattern & decomposition. h Eight-wire pattern & decomposition When determining the line width, it is necessary to consider the influence of the magnetron sputtering coating process. In the magnetron sputtering process, the randomness arising from collisions with plasma particles can result in deviations in the film thickness at the edges of the lines53. A wider line width design can effectively mitigate the non-uniformity of film thickness caused by edge effects, thereby ensuring the geometric integrity of the electrode pattern. By comprehensively considering the design redundancy that has no impact on structural symmetry and the constraints of process stability, the line width is finally determined to be 600 μm. Experiments verification As depicted in Fig. 3e, the parameters of the patterned thin film were configured as five lines, with a line width of 600 μm and a total thin film thickness of 30 nm. The performance of the mHRG with the patterned thin film was characterized subsequent to packaging. The performance metrics under investigation included frequency splitting, Q-factor, and Q-factor non-uniformity. Prior to thin film deposition, the Q-factor of the device was measured to be 9.3×10⁶. After coating, it decreased to 8×10⁶, resulting in a Q-factor loss rate of merely 14%, as depicted in Fig. 4a. The Q-factor non-uniformity, measured at intervals of 15°, was found to be less than 1%, as shown in Fig. 4b. In conjunction with laser-assisted mass removal trimming, the post-deposition frequency splitting can be effectively constrained within 1 mHz, as illustrated in Fig. 4c.Fig. 4Performance comparison of patterned and fully coated mHRGs.a Comparison of the Q-factor loss between the patterned and fully coated mHRG. b Frequency measurement of the mHRG with a patterned thin film. c Frequency measurement of the fully coated mHRG. d Circumferential Q-factor uniformity of the mHRG with a patterned thin film. e Circumferential Q-factor uniformity of the fully coated mHRG Performance comparison of patterned and fully coated mHRGs. a Comparison of the Q-factor loss between the patterned and fully coated mHRG. b Frequency measurement of the mHRG with a patterned thin film. c Frequency measurement of the fully coated mHRG. d Circumferential Q-factor uniformity of the mHRG with a patterned thin film. e Circumferential Q-factor uniformity of the fully coated mHRG For the purpose of comparison, the packaging performance tests of the mHRG with a fully coated inner surface were carried out under identical conditions. Prior to coating, the Q-factor of this mHRG was 8.5×10⁶, and it decreased to 4.2×10⁶ after coating, resulting in a Q-factor loss rate of 51%, as presented in Fig. 4a. The non - uniformity of the Q-factor was 3%, and the frequency splitting was 2 mHz, as shown in Fig. 4c, e respectively. These results demonstrate that, in contrast to the mHRG with a fully coated surface, the device incorporating patterned thin films exhibits a lower Q-factor loss and more excellent Q-factor uniformity. After laser trimming, the frequency splitting of the fully metallized coated gyroscope stabilized at 2 mHz, while that of the patterned coated gyroscope stabilized at 1 mHz. This difference primarily depends on the precision of laser trimming and is unrelated to whether the coating process employs a patterned technique. Frequency splitting cannot be characterized by microcapacitance detection prior to coating. A combination of laser vibrometry and a phase-locked amplifier is required to test frequency splitting before and after both patterned and fully metallized coating using a sweep method. Test results indicate no significant change in frequency splitting between the two coating methods. Conclusion Through the development of a patterned coating process for mHRG and the integration of theoretical model calculations and analyses, our research on patterned coating has provided a crucial solution to the bottleneck of thin film damping in the fabrication of high Q-factor mHRG. Consequently, this study not only unveils the mechanism by which thin films affect the Q-factor of mHRG but also offers valuable insights into the innovative thin film design for resonators. For example, the Q-factor can be reduced by designing different thin films, and the modal balance can be fine tuned using appropriate thin film patterns. Furthermore, this thin film process can be adapted and applied to other types of resonators, such as HRG and ring resonator gyroscopes, by optimizing the relevant processes. However, it should be noted that extending this process to different resonator systems presents potential challenges: First, the mask design and alignment accuracy must be re-optimized for resonators with varying structural dimensions. Second, the compatibility between the film thickness specified in the paper and the resonator materials requires targeted verification. Methods Measurement of resonator dimensions Although fused silica mHRG possess high hardness and low loss characteristics, their micron-scale thin-walled structure is highly sensitive to external stress. Traditional contact measurement methods (e.g., probe profilers) tend to cause sample deformation, and destructive testing methods (e.g., cross-sectional cutting) will directly damage the device. In contrast, industrial computed tomography (CT) employs non-contact scanning54. Throughout the process, no mechanical force is exerted, enabling precise measurement while preserving the original state of the device and ensuring the authenticity of the interface morphology data, as depicted in Fig. 5.Fig. 5Test results of the CT morphology of the resonator Test results of the CT morphology of the resonator Furthermore, the three-dimensional reconstruction capabilities of industrial CT allow for a comprehensive quantitative analysis of the interface morphology. Critical parameters such as sphericity, concentricity, and tilt angle can be accurately extracted from the three-dimensional model. With the assistance of software, multi-dimensional statistical analyses, including the standard deviation of the roughness spatial distribution and the probability density function of characteristic dimensions, can be conducted. This provides a quantitative basis for optimizing process parameters such as thin-film deposition. In comparison, traditional two-dimensional detection methods can only acquire local cross-sectional information. Due to sampling biases, these methods are prone to misinterpret the overall interface morphology and thus struggle to meet the analytical requirements for the uniformity of the resonator interface, which directly influences the splitting of the resonant frequency and the circumferential uniformity of the Q-factor. Mask design and manufacturing 3D printing technology55,56 has demonstrated significant advantages in the manufacturing of masks (Fig. 6a), with its core value lying in the high-precision and high-degree-of-freedom construction capability for complex three-dimensional structures, effectively breaking through the limitations of traditional mask manufacturing processes57. The manufacturing of traditional masks often relies on methods like photolithography and mechanical processing58. These techniques have obvious limitations when dealing with complex configurations: not only is it difficult to efficiently achieve the integrated formation of micrometer structures and macroscopic three-dimensional shapes, but also cumulative errors may occur due to multi-step processing, ultimately compromising the dimensional accuracy and structural consistency of the masks. The edge design of the mask plays a dual critical role in the thin film deposition process: on the one hand, the edge, by forming a continuous structural border, significantly enhances the overall mechanical stability of the mask. During high-energy deposition processes such as sputtering, the mask requires the rigid support of the edge to effectively suppress mask warping, shifting, or vibration, ensuring precise alignment with the resonator and preventing pattern misalignment or deposition deviation caused by mask deformation. On the other hand, the geometric parameters of the edge (such as width, inclination angle γ, etc.) are the core variables for regulating the uniformity of the thin film in the shell and mass tooth regions. The structure and size of the mask are shown in Fig. 6b.Fig. 6Fabrication platform and patterned thin-film mHRG characterization.a A 3D printing platform with a fabricated mask. b Dimensions of the mask and its assembly drawing. c Optical photograph of the patterned thin-film mHRG. d Detailed magnification of a local region (Region 1 in Fig. c) of the patterned thin film Fabrication platform and patterned thin-film mHRG characterization. a A 3D printing platform with a fabricated mask. b Dimensions of the mask and its assembly drawing. c Optical photograph of the patterned thin-film mHRG. d Detailed magnification of a local region (Region 1 in Fig. c) of the patterned thin film The 3D-printed mask is aligned and secured to the resonator through dimensional fit, maintaining surface contact. Experimental results confirm that the material and assembly method of this mask do not affect the surface quality of the fused silica resonator. Regarding alignment control: In the Z-axis direction, correct positioning is achieved when the upper surface of the mask is parallel to the planar surface of the resonator’s serrated structure. Due to the resonator’s circumferential symmetry, no strict tolerance requirements are imposed on the mask’s rotational orientation. Figure 6c shows an actual photograph of the patterned mHRG thin film fabricated using a 3D printed mask process, presenting the overall device structure. Figure 6d displays an enlarged view of Region 1, clearly revealing the microstructure of the thin film and the edge transition state. Film morphology test The microstructure characterization of the cross-section of thin films requires precise sample preparation and high-resolution measurement: Firstly, the cross-section of the thin film is embedded with resin to fix the sample with embedding materials such as epoxy resin, to prevent cross-sectional damage or structural deformation caused by mechanical stress during subsequent processing, and to provide a stable support substrate for subsequent focused ion beam (FIB) thinning, as shown in Fig. 7a59,60. The FIB thinning process must strictly follow a step-by-step fine processing logic: The embedded sample is fixed on the FIB stage, and the target cross-sectional area is located through the scanning electron microscope (SEM) mode. Firstly, a low acceleration voltage of 10 keV is used to clean the surface of the cross-section to remove surface contaminants; then, it is switched to a 30 keV Ga⁺ high-energy ion beam for rough thinning, gradually stripping the material layer by layer along the cross-sectional direction until a thin sheet with a thickness of about 1-2 μm is obtained; to reduce ion beam damage and meet the thickness requirements for TEM observation, it is thinned to 80 nm61. This thickness is chosen to balance TEM imaging quality and sample structural stability: it enables sufficient electron beam penetration for clear microstructural characterization, while preventing cracking or wrinkling of overly thin samples. Ultimately, an ultra-thin cross-sectional sample with a flat surface and uniform thickness is obtained, as shown in Fig. 7b.Fig. 7TEM characterization of the thin film cross-section.a Thinned and fixed sample. b TEM observation of the film and its substrate. c EDS elemental map. d High-resolution TEM image of the film TEM characterization of the thin film cross-section. a Thinned and fixed sample. b TEM observation of the film and its substrate. c EDS elemental map. d High-resolution TEM image of the film Transmission electron microscopy (TEM) measurement is based on the above-prepared ultra-thin samples: The FIB-thinned thin sheet is transferred to a copper grid through a micromanipulator, placed in the TEM sample chamber and evacuated to a high vacuum environment. According to the analysis requirements, an appropriate acceleration voltage is selected. First, low-magnification imaging is used to determine the cross-sectional observation area, and then it is switched to high-resolution mode (HRTEM) to characterize the lattice structure and interface atomic arrangement of the thin film. Combined with EDS, the elemental distribution of the thin film is synchronously analyzed through surface scanning (as shown in Fig. 7c), obtaining the gradient distribution of elements in the cross-sectional direction62. The entire process, through the precise thinning of FIB and the high-resolution imaging of TEM, can achieve multi-dimensional characterization of the cross-section of thin films from the nanoscale microstructure to the elemental distribution, as shown in Fig. 7d. In this study, the film parameters are a width w = 600 μm and a thickness d = 30 nm (d « w). The selection of film width is primarily based on considerations of deposition process uniformity and stability. Control over film resistance is achieved through precise regulation of film thickness, which correlates with the grain size of the metal, the mean free path of electrons, and the film dimensions. The current 600 μm width design meets process requirements without imposing restrictive effects on the resonator’s electrical performance.