Chip Of Polysilicon Nanofilm Pressure Sensor Based On Sacrificial Layer Technology

Silicon piezoresistive pressure sensors are widly used, which have a very important position in the sensors field. The development direction of these sensors is miniaturization, high sensitivity, good temperature characteristics and integration, so semiconductor piezoresistive materials and sensor structures have been researched in depth. Researches show that the polysilicon nanofilms(PSNFs) have good piezoresistive properties, which are applied on bulk silicon pressure sensors availably. However, its existing algorithm for piezoresistance coefficients has a slight theoretical derivation deficiency, and field of its application should be widened. In order to improve this algorithm, an algorithm for piezoresistance coefficients of p-type PNSFs is presented in this dissertation, which three fundamental piezoresistance coefficients are established. The calculated gauge factor(GF) is in agreement with the measured GF. And then, in order to utilize excellent piezoresistive properties of polysilicon nanofilm effectively, a sacrificial layer pressure sensor chip with PSNFs piezoresistors is designed and developed, it has the advantages of small volume, high full scale output, strong overload capacity and easy integration, and has good application prospects.In term of the tunneling piezoresistive theory, the mechanism of tunneling piezoresistive effect is illuminated using quantum tunneling effect and energy band split decoupling theory. Consequently, a new model of polysilicon piezoresistive properties –– tunneling piezoresistive model(TPM) is established. The theory better explains the phenomenon that the heavily doped p-type PSNFs have higher GF. However,for the existing algorithm for piezoresistance coefficients of PSNFs based on this theory, the model of impurity concentration dependent piezoresistance coefficient is based on the measured result fitting curve of piezoresistance coefficient as a function of impurity concentration in p- type silicon, only piezoresistance coefficient π44 is presented, Therefore, the algorithm should be modified.In this dissertation, according to the mechanism that silicon valence band and hole conductivity effective masses vary with the change of stress, using the TPM, an algorithm for piezoresistance coefficients of p-type PNSFs is presented. For this algorithm, the piezoresistance coefficients as a function of impurity concentration is presented; it includes the PSNFs fundamental piezoresistance coefficient π11, π12 and π44, which the GFs with any crystal direction combination can be calculated. With this algorithm, the curves of the relationship between the polysilicon nanofilm GFs and impurity concentration are drawn,which is consistent with the measured curve. This result shows that the improved model can reasonably explain the relationship between PSNFs GFs and impurity concentration, which improves the piezoresistive theory.In order to effectively apply the piezoresistive properties of the PSNFs and the advantages of sacrificial layer pressure sensor, such as small size and easy integration, a sacrificial layer piezoresistive pressure sensor chip with p-type PSNF piezoresistors is proposed in this dissertation. The proposed sensor chip includes silicon substrate, a step type bent polysilicon diaphragm that forms a vacuum cavity with the silicon substrate, four PSNF piezoresistors on the diaphragm that are connected with metal to form a Wheatstone bridge converting pressure to output voltage and eight sealed etch holes around the diaphragm. Using finite element analysis software, the structure of proposed sensor is optimized with large displacement static and nonlinear contact analysis. Utilizing the advantage that polysilicon has high tensile strength, the design method for dimensions of the proposed sensor by measuring range is presented. And its overload capacity is improved by adjusting the cavity height properly. According to the proposed sensor structure, the process sequences of the proposed sensor chip are designed. In this sequences, silicon dioxide layers are deposited by Plasma Enhanced Chemical Vapor Deposition(PECVD), polysilicon layers are deposited by low-pressure chemical vapor deposition(LPCVD), the sacrificial layer is removed by wet etching, etch holes are opened by plasma etching, boron is doped by ion implantation, An aluminum layer is deposited by sputtering, the residual stress of the polysilicon diaphragm is reduced and boron impurity is activated by annealing process, t-butyl alcohol cooling and drying method is adopted to prevent the adhesion with diaphragm and the bottom of the cavity.According to the design structure and process sequences, the proposed sensor chips are fabricated for four batches. The first batch of chips is failure due to the cavity leakage. For the second batch of chips, the leakage problem is resolves with the improved process, but this batch has less sensitivity because of the adhesive with the diaphragm and the bottom of the cavity under a pressure of zero. Using the modified processes, the third batch of chips is fabricated. However, the sacrificial layer etching is not completely clean due to low concentration corrosive liquid. After the concentration is modified, the fourth batch achieves the design requirement. The sample’s measurement results show that a full scale pressure of 2.5MPa, a full scale output voltage of 362 mV at 25℃ with a supply voltage of 5 V, an overpressure of 7 times higher than the full scale pressure at 25℃, a thermal zero drift of-0.01% FS/ ℃ and a thermal sensitivity drift of-0.1% FS/ ℃ in a temperature range of-55℃~150 ℃ are achieved. Compared with sacrificial layer piezoresistive pressure sensors reported earlier, the proposed sensor has wide working temperature range, high full scale output voltage. Its linearity and hysteresis are good. But its thermal sensitivity drift is slightly poor. Compared with silicon pressure sensors with similar full scale pressure that are produced by several famous companies, the proposed sensor has advantages of small volume, high full scale output and strong overload capacity. Specifically, Its full scale output voltage is 3 times more than that one, and it has good linearity, its overload capacity is 7 times more than the full scale pressure, which is improved obviously compared with a typical 2~5 times full scale overload capacity. But the proposed sensor has slightly poor repeatability and thermal sensitivity drift, which should be improved in further research.