The Study on Mechanical Performances of Thin SiO Film by Micro-Bridge Methods WANG Yu 1, ZHANG Hai-ia 1*, ZHANG Tai-Hua 1 Institute of Microelectronics, Peking University State Key Laboratory of Nonlinear Mechanics (LNM), Institute of Mechanics, Chinese Academy of Sciences * Email: zhangh@ime.pku.edu.cn ABSTRACT In order to make use of SiO thin films in MEMS better, micro-bridge testing method was conducted to test mechanical properties of thin SiO films which was fabricated by thermal oidation. Nano Indenter XP which is the product of MTS Corporation was used to test the load-deflection curves of silicon dioide micro-bridges, and the Young s modulus and residual stress are obtained from the curve. Finite element calculating software, ANSYS, is used to provide the simulation and evaluate the precision of the calculation. At the same time, another testing method, Nano-indentation, was performed to test the mechanical properties of these thin films, and the results were compared with the micro-bridge testing method s. The result shows that Young s modulus of SiO thin films fabricated under the condition taken by the paper is 9510%GPA, and residual stress of SiO thin films is influenced by processes obviously. The requirements to the eperimental method is analyzed which should be obeyed when testing Young s modulus and residual stress by micro-bridge testing method. Keywords: MEMS, thin film, micro-bridge test, Young s modulus, residual stress 1. INTRODUCTION Thin films manufacturing is very important for MEMS (micro-electro-mechanical system) technology. To make use of thin films more sufficiently in MEMS, the performances of materials fabricated by different processes should be comprehended, as well as the mechanical and electromagnetic properties of them. In addition, the influence eerted by processes and thin films microstructures should be clarified. Based on such work, thin films performances can be optimized, the area from which materials are to be chosen can be enlarged, and the longevity and reliability of MEMS devices can be improved. Up to now, many methods for testing thin films mechanical properties have been developed, such as nano-indentation, uniaial tensile testing, beam bending, bulge testing and microstructure testing [1]. In this paper micro-bridge tests are used which was raised by Zhang TY []. It calculates the thin films mechanical properties using microstructures. In this method, two ends of the thin film are fied in the substrate. Then load at the midpoint of the rectangle film, curving and even breaking it. At the same time, record the loads and displacements of the midpoint continuously to obtain the load-displacement curve according to which the Young s Modulus and residual stress of micro-bridges can be calculated. This paper fabricated micro-bridges and calculated Young s Modulus and residual stress of thin films by such way, and correctness of the calculation was evaluated by ANSYS. Furthermore, we analyzed the scope in which the calculating model is competent, and summarized the request for methods in testing.
. CALCULATING MODULE AND TESTING METHOD Fig.1 schematic for micro-bridge test Fig. loading and unloading curves Fig.1 is the schematic of testing mechanical properties by micro-bridge test. The midpoint is the place where to be loaded, and the displacement of it can be epressed as below [] : Q tanh kl/ w N k Ql 4N M N 0 1 coskl/ 1 S PN Q N Nr SPP SPM M 0 According to formulas in [], different load-displacement curves would be obtained corresponding to different mechanical parameters guess values. The guess values of thin film s Young s Modulus and residual stress can be considered to be correct, if the curve worked out is similar to that in practice. The residual stress can be get as below: r N r t. Tests were eecuted with MTS Nano Indenter XP system. Apparatus record displacements together with the load values when loading and unloading, so as to obtain the load-displacement curves [3]. It is believed that the data are credible if the curve is repeated in si switches between loading and unloading. Fig. is an eample of a satisfactory loading and unloading curve. 3. FABRICATING PROCESS Samples were fabricated on 4 inch silicon wafer with a thickness of 555m, single polished. The SiO thin film shaped by thermal oidation under 950. The sample length are 10um, 30um, 50um, 70um, 90um, and sample width are 6um, 10um, 1um, 18um, um, respectively. The major fabrication steps are shown in Fig.3: Side view: SiO LPCVD SiO Si 3 N 4 Si Top view: a. thermal oidation b. oidation etched c. front side protected d. etching the micro-bridges Fig.3 the major fabrication steps
Micro-bridges should be fabricated at a distance to avoid effects between each other []. Fig.4 is the photo of a micro-bridge fabricated by such processes. Fig.4 micro-bridge s photo 4. RESULTS AND ANALYSIS This paper utilized ANSYS to conduct linear simulation for every Young s Modulus value worked out. The value taken as material s parameter, micro-bridge s load-displacement relation can be got. Compared with measured curve in the eperiment, it can be judged whether the calculated value is credible. 4.1 Young s modulus The relations between simulated curve and measured curve for micro-bridges with different widths and lengths are shown in Fig.5. Fig.5 comparison between measured curve and simulated curve Setting L/W<4, where L is the bridge length and W is its width, the Young s Modulus worked out isn t credible. When L/W>4, the Young s Modulus worked out is consistent with the value in practice, as shown in Fig.5. Therefore, results got from samples whose L is shorter than 4W are no longer cited in coming steps. When L/W<4, some conditions what make the model work no longer eist. The bridge has degraded into shutter, and the model based on bridge structure don t work any more. In other words, when L/W is big enough, displacement of the midpoint is barely influenced by substrate; it can reflect micro-bridge s characteristics well. The smaller L/W is, the more displacement of micro-bridge midpoint is influenced by substrate, and the more the result reflects substrate s characteristics. After getting rid of the results that were got from whose L/W<4 along with those are apparently different
from others, the calculated Young s Modulus are divided into groups in which the bridges have same length or width. The results were shown in Fig.6 by each group. Fig.6 calculated Young s Modulus values It can be seen that, the micro-bridge s Young s Modulus is independent of its length and width. It agrees with the results given by [4]. 4. Residual stress Table.1 ehibits the repetition of residual stresses which belongs to adjacent unit s micro-bridges. Table.1 repetition of residual stresses Samples Length/um Width/um Residual Stress/GPa 5-7-3-1-1 70 1-0.1717 5-7-3-1- 70 1-0.1717 5-7-3--1 70 1-0.1717 5-7-3--3 70 1-0.1717 5-7-3--8 70 1-0.1717-5-3-1-3 50 1-0.773-5-3-1-8 50 1-0.773-4-1-1-1 50 6-0.05051-4-1-1- 50 6-0.05051 Adjacent units are fabricated in same conditions. There is much probability that they are influenced by other factors simultaneously. So the repetition can be an evidence for the dependence of residual stress on fabricating process.
Fig.7 calculated residual stress values On the other hand, residual stresses scatter remarkably, just as shown in Fig.7. Micro-bridges fabricating process is complicated, and structural influence may be brought into being when loading which may cause stress released. These are probably the reasons of residual stresses scattering. However, it can be summarized that residual stress highly depend on fabricating process, and it is apt to be influenced by environmental factors. When testing residual stress, it should be ensured to most degree that samples are fabricated simply, and testing method should be simplified to reduce the influences. 4.3 Comparison with other testing method Based on analysis above, micro-bridges of this paper were eperimented and calculated. Repetitive and credible results are listed below: Table. repetitive and credible results Samples Length/um Width/um Young s Module/GPa Residual Stress/GPa 5-7-3-1-1 70 1 90 0.1717 5-7-3-1- 70 1 87 0.1717 5-7-3--1 70 1 88 0.1717 5-7-3--3 70 1 97 0.1717 5-7-3--8 70 1 100 0.1717 5-7-3--4 70 1 100 0.111 5-7-3--6 70 1 10 0.07071 Young s Modulus: 9510%GPA. Residual stress: relate greatly on fabricating process and operation. The verdict cannot be reached yet. Mechanical parameters of SiO thin film are also tested by indentation method [5]. And the samples are consistent with micro-bridge testing. The results are shown in table.3: Table.3 3 tested Young s Modulus and stiffness values sample Thickness/nm indentation depth/nm Young s Modulus/GPa 1# 400 800 67.336 10# 1000 000 79.937 11# 1300 500 91.158 Comparing with results got by micro-bridge testing, when depth of indentation is small, results of indentation are apparently smaller than that of micro-bridge testing. When load is small, residual stress influences the midpoint displacement greatly; when load getting bigger, the residual stress influence becomes less. Considering this, the values of small loads were removed in this paper when calculating. As a result, the Young s Modulus worked out reflects more characters of thin film under large loading. So Young s Modulus of micro-bridge testing and indentation consistent with each other when depth of indentation is comparatively big. At the same time, the bigger the displacement of micro-bridge s midpoint is, the more serious the results are influenced by substrate. Thereby, measurement cannot be credible when the micro-bridge s midpoint
displacement is too big. 5. CONCLUSION Micro-bridge testing is one of thin film mechanical parameters testing methods. Its advantage is that probe won t slide when loading, the error is little, and Young s Modulus as well as residual stress can be obtained together. Nevertheless, sample fabricating process in micro-bridge testing is a little complicated, and every step may introduce influence into residual stress, and operation in testing also may change the micro-bridge s residual stress. As a result, although Young s Modulus and residual stress can be obtained simultaneously in theory, residual stress is hard to measure correctly when we faced with the scattered results of residual stress. When micro-bridge test is chosen to measure thin films mechanical properties, sample fabricating and testing process should be under severe control. When testing thin film s Young s Modulus, length of micro-bridge should be designed a little longer to avoid making the model inefficient to some degree, reducing substrate s influence for bridge s midpoint displacement. To reduce the influence of residual stress and substrate, values whose displacements are too big or too small need to be omitted. To test thin film s residual stress, testing method and sample fabricating process should be simplified to reduce the influence come from factors which cannot be controlled. ACKNOWLEDGEMENTS This work is supported by the National Natural Science Foundation of China (No. 60306008), China 863 project (No. 003AA404013). REFERENCES [1] Chen Longqing, Zhao Minggao, Zhang Tongyi. The Testing Method of Mechanical Properties of Thin Films. Journal of Mechanical Strength. 001,3(4),413-49. [] T. Y. Zhang, Y. J. Su, C. F. Qian, M. H. Zhao, L. Q. Chen. Microbridge Testing of Silicon Nitride Thin Films Deposited on Silicon Wafers. Acta mater. 000, 48, 843-857 [3] Zhang Taihua, Yang Yeming. Nano-hardness Tester and Its Application in MEMS. Modern Scientific Apparatus. 00,1:3-37. [4] Y. J. Su, C. F. Qian, M. H. Zhao, T. Y. Zhang. Microbridge Testing of Silicon Oide/Silicon Nitride Bilayer Films Deposited on Silicon Wafers. Acta mater. 000, 48, 4901-4915 [5] Winchester, K.J., Dell, J.M. Optoelectronic and Microelectronic Materials and Devices, 000. COMMAD 000. Proceedings Conference on, 6-8 Dec. 000, 117-10 Corresponding author: Zhang Haiia, e-mail: zhangh@ime.pku.edu.cn; phone: 010-675536; fa: 010-6751789; Address: Institute of Microelectronics, Peking University, Beijing, 100871