【作者】 郑旭东；

【导师】 王跃林； 金仲和；

【作者基本信息】 浙江大学 ， 微电子学与固体电子学， 2009， 博士

【摘要】 本论文在对传统栅电容结构的惯性传感器进行研究的基础上,提出了一种新型的梳状栅电容结构,并对基于此结构的加速度计和陀螺进行了研究。具体工作包括:加速度计和陀螺器件的结构设计、制作、接口电路设计、性能测试及性能优化等。(1)与传统栅电容结构相比,本论文所提的梳状栅电容结构的栅电极不完全刻穿,而仅需刻蚀数十微米,有效的提高了加速度计和陀螺的可动质量块质量,从而使器件的机械灵敏度提高,机械热噪声下降。典型设计下,可动质量块质量可提高54%,使得加速度计和陀螺的机械灵敏度在同样的尺寸下分别提高54%和92%,机械热噪声分别下降35%和48%。另外,新的栅电容结构也使得栅电极的形成和结构的释放分开,极大的降低了深反应离子刻蚀过程中的凹缺效应和滞后效应对微结构的损伤,提高了器件的加工质量以及成品率。(2)建立了静电驱动的框架式栅电容陀螺的机械灵敏度分析模型,分析发现,在芯片面积和驱动电压一定的条件下,当驱动外框与检测质量块的面积相等时,可以使得器件的机械灵敏度达到最大,从而实现了对陀螺性能的优化。针对深反应离子刻蚀过程中的滞后效应、凹缺效应和厚胶的边缘效应造成的弹性梁尺寸误差,提出了陀螺的驱动和检测弹性梁、驱动和检测栅电极的等宽和低深宽比设计原则,从而使尺寸误差和工艺离散性造成的陀螺频率匹配偏差降到最低。对陀螺原型器件的实际测试表明:同批次陀螺谐振频率的标准差相较改进前降低了79%以上;工艺离散性导致同批次陀螺的谐振频率最大相差132Hz,但陀螺驱动模态和检测模态的频率差的均值为7.5Hz,与设计值6Hz接近,标准差为16.5Hz,保持了频率匹配。(3)对加速度计和陀螺的实际测试表明,采用单端输入-双端差分输出结构的电容检测电路,可抵消双端输出中相关性较高的共模噪声,电容检测信号经过差分放大后实测的信噪比提高了约13dB。对加速度计和陀螺的噪声进行了建模分析,分析结果均与加速度计和陀螺实测的噪声基底在同一数量级上。在根据噪声模型分析结果对微机械陀螺的性能进行了优化,实测的陀螺在1Hz处的噪声基底从0.023°/s/(?)降低到0.0079°/s/(?)。由于加速度计和陀螺均采用变面积电容检测法线性敏感位移变化,两者都获得了很高的线性度:实测加速度计在±1g内的线性度为99.997%,±30g内的线性拟合曲线的二次项和一次项系数比159ppm;陀螺在±43°/s内的线性度为99.995%,±200°/s内的线性拟合曲线的二次项和一次项系数比为203ppm。加速度计和陀螺的工作阻尼主要是滑膜阻尼,这使得器件在大气压下也获得了较小的阻尼系数和较高的品质因子,新型的梳状栅电容结构又使得器件获得了更大的可动质量块质量,因此加速度计和陀螺在大气下均实现了高性能:加速度计的偏置稳定性为0.3mg,白噪声基底为0.348mg/(?),动态范围为±30g;陀螺的偏置稳定性为21.6°/h,1Hz处的噪声等效角速度优于0.01°/s/(?)。陀螺的线性度、偏置稳定性以及噪声基底等指标,比近年来文献中报道的大气压下工作的微机械陀螺要好。更多还原

【Abstract】 A novel comb-bar capacitor is proposed on the basis of the study of traditional bar capacitor based inertial sensors, and the study of comb-bar capacitor based MEMS accelerometers and gyroscopes is presented in this paper, specific work including: the structural design for an accelerometer and gyroscope, the process flow design for the fabrication, the design of the interface circuits, performance tests of the prototypes, and the analysis and optimization of the accelerometer and gyroscope.（1） Compared to the traditional bar capacitor structure, the bar electrodes of the comb-bar structure is only etched for a few tens of micrometers instead of being cut though, which increases the proof masses of the accelerometer and gyroscope efficiently, resulting in higher mechanical sensitivities and lower mechanical-thermal noises. For this typical design, the proof masses are increased by 54% both for the accelerometer and gyroscope. As a result, the mechanical sensitivities have been increased by 54% and 92%, and the mechanical-thermal noises have been reduced by 35% and 48% respectively within the same geometry. Furthermore, the bar electrodes and the structure are released separately, which lowers the damages caused by the lag effect and notching effect during DRIE greatly, resulting in higher fabrication quality and yield rates.（2） An analysis model for the mechanical sensitivity of the electrostatically driven bar capacitor based frame gyroscope is built, which reveals that, given fixed chip area and driving voltages, the mechanical sensitivity is maxed when the areas of the outer frame and the inner proof mass are equal, thus the performance of the gyroscope is optimized. An equal width, low aspect-ratio design principle for both spring beams and bar electrodes of the driving motion and sensing motion is proposed to deal with the large dimension errors caused by the lag effect, notching effect and edge effect of the thick photoresist during DRIE, thus the deviation of the difference between the driving and sensing resonant frequencies, which is caused by the the relatively large dimension error and fabrication ununiformity, is minimized. Test results show that: the standard deviation for resonant frequencies of gyroscopes from the same batch has been reduced by more than 79%; With the maximal frequency difference 132Hz of the tested gyroscope resonant frequencies from the same batch caused by process tolerances, the average value for the difference between the driving and sensing resonant frequencies is 7.5Hz, which is close to the designed value 6Hz, and the standart deviation for the difference is 16.5Hz, which implies that the driving and sensing resonant frequencies keep matched after fabrication.（3） Test results of the accelerometer and gyroscope show that the SNR of the capacitor sensing carrier has been increased by around 13dB after passing by a single input—double differential output CV circuit, which is because the correlated common noises are canceled after the differtial amplifier. Noise analysises for the accelerometer and gyroscope are presented and analysis results are of the same order of magnitude as the tested values. An optimization for the gyroscope is carried out according to the noise analysis results, and the noise floor of the gyroscope at 1Hz hasbeen decreased from 0.023 o/s/Hz^1/2 to 0.0079 o/s/Hz^1/2 .The accelerometer and gyroscope both achieve high linearity as a result of sensing displacement by the varying overlapped area: the linearity for the accelerometer is 99.997% within±lg input acceleration, and the ratio of the quadratic term coefficient and the linear term coefficient of the quadratic fit line is 159ppm for±30g input accelerotion; the linearity for the gyroscope is 99.995% within±43°/s input angular rate, and the ratio of the quadratic term coefficient and the linearterm coefficient of the quadratic fit line is 203ppm for±200°/s input angular rate. Themain system damping of the accelerometer and gyroscope is slide-film damping, thus low damping coefficients and high quality factors are achieved at atmospheric pressure, and the novel comb-bar capacitor have ensured larger proof masses for the inertial sensors, as a result, the accelerometer and gyroscope have achieved high performances at atmospheric pressure: The bias instability is 0.3mg, the white noise floor is 0.348 mg/Hz^1/2 and the dynamic range is±30g for the accelerometer; Thebias instability is 21.6°/h, and the noise floor at 1Hz is better than 0.01°/s/Hz^1/2 forthe gyroscope. The linearity, bias instability and noise floor of the comb-bar structure gyroscope are better than the MEMS gyroscopes working at atmospheric pressure reported in recent years.更多还原