Fabrication of Test Structures To Monitor Stress in SU-8 Films Used For MEMS Applications

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

2010 IEEE International Conference on Microelectronic Test Structures, March 22-25, Hiroshima, Japan.

1.2

Fabrication of Test Structures to Monitor Stress in SU-8 Films used for MEMS Applications
S. Smith , N.L. Brockie , J. Murray , C.J. Wilson , A.B. Horsfall , J.G. Terry , J.T.M. Stevenson , A. R. Mount and A.J. Walton
Institute for Integrated Micro and Nano Systems (part of the Joint Research Institute for Integrated Sytems) School of Engineering, Scottish Microelectronics Centre, The University of Edinburgh, Edinburgh, EH9 3JF, UK Email: Stewart.Smith@ed.ac.uk

School of Electrical, Electronic and Computer Engineering, Merz Court, Newcastle University, Newcastle upon Tyne, NE1 7RU, UK

IMEC, Kapeldreef 75, B-3001 Leuven, Belgium

School of Chemistry, Joseph Black Building, The University of Edinburgh, Edinburgh, EH9 3JJ, UK

Abstract SU-8, an epoxy based negative photoresist, is widely used in the manufacture of micromechanical systems but can exhibit signicant levels of stress build-up during processing. This paper describes micromechanical test structures that provide the opportunity to spatially characterise stress the in SU-8 at different stages of the process. The structures are fabricated in a thick layer of SU-8 and are subsequently released from the underlying substrate using a dry chemical vapour etch process. The initial results indicate that there is signicant tensile stress in the SU-8, and that this demonstrates a radial variation along with a dependence on the process conditions.

rotate clockwise. Similarly if there is compressive stress in the material the arms will expand and the pointer will rotate anticlockwise. These structures have been employed for use in a wide variety of metallisation processes from advanced copper damascene [8] to thick electroplated permalloy [7]. However, while these structures have been applied to many materials commonly used in IC and MEMS technologies, they have not been used in the characterisation of SU-8 lms, which are well known to have issues related to stress.

I. I NTRODUCTION SU-8 photoresist is an important material which is commonly used for the manufacture of Micro-Electro-Mechanical Systems (MEMS) [1]. This photo-active, epoxy based polymer is used in applications ranging from the production of microuidic systems [2] to the formation of moulds for electroplating of thick metal patterns [3]. However, when SU-8 is cured during post-exposure baking it tends to shrink, which introduces a tensile stress [4]. This can lead to problems with cracking [5] and adhesion, particularly on the metal layers used in electroplating [6]. MEMS stress test structures have previously been used to investigate a variety of metallisation processes [7][9]. The purpose of the present work is to adapt and characterise these structures for application to the measurement of stress in patterned SU-8 layers. II. T EST S TRUCTURES The test structure reported in this study is a micromechanical rotating stress sensor [10], [11] fabricated in SU-8. It consists of two stress or expansion arms, both xed at one end, supporting a pointer arm that is released from the substrate during the nal stage of the process. The expansion arms are offset at their connection with the pointer so that as they expand or contract they cause the pointer to twist or rotate in the plane of the layer. The layout of the structure is presented in Fig. 1. In this structure, tensile stress in the material will cause the expansion arms to contract and the pointer arm to

Fig. 1.

Schematic layout of stress test structure

There are three important dimensional parameters in the test structure that determine its sensitivity in terms of the degree of rotation observed in the pointer for a given stress. These are the length (L) and width (W ) of the expansion arms and the arm separation (Y ). The initial study of test structures fabricated in permalloy [7] suggested that the maximum rotations would be observed in structures with the parameters shown in table I.
TABLE I PARAMETERS FOR MAXIMUM STRESS SENSOR SENSITIVITY Parameter L (m) W (m) Y /W Value 850 8 2

Based on these results a new mask has been designed which

978-1-4244-6915-4/10/$26.00 2010 IEEE


8
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

includes test structures with W =8m, L=850m and three different values of Y /W : 1.5, 1.75 and 2. The layout of the test chip is shown in Fig. 2 (a) while the complete test mask layout for a 200mm (8") wafer is presented in Fig. 2 (b). Each die contains seven structures with Y /W = 1.5 and 2, and fourteen structures with Y /W = 1.75. There are also four structures on each die with expansion arms that narrow into a 5m wide hinge at the connection to the pointer arm, two of these have Y /W = 1.75 and one each of the other two values of Y /W .

(a)

(b) Fig. 2. Schematic layouts of stress sensor chip (a) and wafer (b)

III. FABRICATION The SU-8 stress test structures were fabricated on 200mm silicon substrates. The wafers were prepared for coating with the photoresist by cleaning for 10 minutes in Piranha solution (3:1 H2 SO4 :H2 O2 ) before rinsing in deionised water. The wafers were then dried before the surface was dehydrated by baking on a hot plate at 200C for 20 minutes followed by an oxygen plasma in a reactive ion etch tool to promote adhesion. The wafers were then placed on a spin coating tool and SU-8 3035 was poured slowly by hand to prevent bubble formation. Each wafer was spun up to 500rpm for 10 seconds to spread the resist then at 3,000rpm for 30s, which creates a layer thickness of around 35m. The SU-8 3035 formulation used in this work is produced by the MicroChem Corporation and is patterned by exposure to UV light, which begins the process of cross-linking the polymer. It can be spin coated onto

silicon substrates up to a thickness of around 80m in a single coating [12]. The wafers were subsequently soft baked for 5 minutes at 65C followed by 15 minutes at 95C. Following soft baking the wafers are allowed to cool before an automated Edge Bead Removal (EBR) process is performed using a stream of Propylene Glycol Methyl Ether Acetate (PGMEA) on a spin coating tool. This removes a 10mm ring of the resist from the wafer perimeter to ensure even contact between the wafer and mask during photolithography. A further short baking step at 65C for 2 minutes is performed before exposure to ensure the wafer is dry following EBR. The SU-8 was exposed with the test pattern using a UV mask aligner through an i-line long pass lter which attenuates light below 350nm wavelength that can cause problems with the sidewall angle of printed features. Two test wafers (W1 and W2) were exposed with an approximate dose of 400mJ-cm2 and two (W4 and W5) with a dose of 320mJ-cm2 . In addition, one wafer (W3) was blanket exposed with a dose of 400mJ-cm2 for use in later optical lm stress measurements. The next step was a post-exposure bake, initially at 65C for 1 minute followed by 5 minutes at 95C and a further 1 minute at 65C before cooling to room temperature. The resist is developed in a spin/spray developer tool using PGMEA for at least 6 minutes before rinsing in IsoPropyl Alcohol (IPA) and drying. Subsequently, two of the wafers (W2 and W5) were subjected to a nal hard-baking process on a hotplate for 10 minutes at 150C followed by 10 minutes at 200C. At this stage the wafers are ready for release etching with a dry XeF2 vapour in a Memsstar SVR XF etch tool. This isotropically etches the silicon substrate and releases the suspended test structures. Vapour etching is used so that there are no problems with stiction that can occur with wet release processes [13]. The release etch was somewhat uneven on the initial test wafer but the use of a very short oxygen plasma prior to vapour etching appeared to greatly improve the etch uniformity. A short dip in 48% HF solution was also investigated in order to determine if the release etch problems were related to the native silicon oxide on the wafer. Unfortunately this appeared to encourage partial delamination of the structures from the substrate and will not be used in future processes. Wafers W1 and W4 were successfully released with a uniform silicon etch while W2 was only partially released as the etch was performed without the short oxygen plasma. Unfortunately it was not possible to complete the processing of wafer W5 and so no results from it are available. IV. M EASUREMENTS AND R ESULTS Fig. 3 shows a test structure before release with Fig. 4 showing a similar structure after XeF2 etching. The clockwise rotation of the structure clearly indicates signicant tensile stress in the exposed and developed SU-8 lm. Wafer mapping software has been developed in LabView, which is capable of controlling a SSS MicroTec PA200 semiautomatic 200mm wafer prober along with a digital camera

9
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

(a)

(a)

(b) Fig. 3. (a) Unreleased stress sensor structure, (b) magnied image of the same structure

(b) Fig. 4. (a) Released stress sensor structure, (b) magnied image of the same structure

attached to the probers microscope. The software enables the user to select any or all of the structures on the test wafers for measurement, including the exclusion of areas occupied by electrical linewidth test structures, which are obviously not designed for measuring stress. These structures are not used on the SU-8 wafers anyway, as they require fabrication in a conducting material. A ow chart representing the wafer mapping software is presented in Fig. 5 Pattern recognition is performed using a template image le of one of the test structures with a resolution better than that available from the camera. One challenge encountered was the camera losing focus as the wafer is scanned, which was caused either by non-uniformities in the thickness of the wafer or the SU-8 resist. This was addressed by developing a software routine that is capable of autofocusing the camera wherever required. This involves taking an average SNR (Signal-to-noise-ratio) of the image after edge detection has been performed. The SNR for the image is the ratio of the mean pixel value to the standard deviation of the pixel values, and can be used to the evaluate the contrast of the edges. The camera is then focused by adjusting the chuck height until the value is within a set SNR threshold. A Hough edge detection transform available in LabView is used to locate the edges of the rotated pointer arm and also the edges of the surrounding SU-8 structure, which are used to obtain a reference angle. Lines of best t are applied to these rough edges and used to calculate the pointer arm angle which is subsequently saved to a spreadsheet for further analysis.

Fig. 5. Flowchart representing the stress measurement wafer mapping software routine

Visual inspection of these structures identies a signicant difference compared with previously reported metallic structures. This is that their rotation is considerably greater, with many of the pointer arms having hit the end stops so that a signicant number have bent or broken pointers. Wafer maps of the full results from wafers W1 and W4 are presented in Fig. 6 and Fig. 7 respectively. In both gures, the areas without stress structures or where the structures

10
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

Pointer Rotation (degrees)

Pointer Rotation (degrees)

6 100 80 60 Vertical Position (mm)

6.5

7.5

8.5

9.5

10 10.5 11 100 80 60 Vertical Position (mm) 40 20 0 -20 -40 -60 -80

6.5

7.5

8.5

9.5

10 10.5 11

40 20 0 -20 -40 -60 -80 -100 -100 -80 -60 -40 -20 0 20 40 Horizontal Position (mm)

60

80 100

-100 -100 -80 -60 -40 -20 0 20 40 Horizontal Position (mm)

60

80 100

Fig. 6. Wafer map of pointer rotation for a 200mm wafer with SU-8 stress test structures (Wafer W1)

Fig. 7. Wafer map of pointer rotation for a 200mm wafer with SU-8 stress test structures (Wafer W4)

have been damaged have been assigned a rotation value of 0 and appear black. This includes the area at the centre of each die where the electrical test structures are located, see Fig. 2(a). The vertical black lines on the right side of W4 indicate columns of structures that have all failed, probably due to a consistent error on the mask which has left these structures prone to failure. A similar but less clear pattern is also observed on W1 where some, but not all, of the structures in these columns have failed. These results show that there is much more variation in the pointer rotation on W4 than on W1, mainly between the centre and edge of the wafer but there is also a difference from left to right across the W4 wafer. This wafer also shows some clear variation within each die with larger rotations in the top right corner of each die where the structures with the narrowed expansion arms are located. The overall spatial variation can be seen more clearly by plotting the average rotation for each die and these results are presented in Fig. 8 (W1) and Fig. 9 (W4). These results again show there is signicantly more variation in stress across W4 than across W1. The pattern on W4 suggests an effect caused by the initial pour of the SU-8 resist. SU-8 3035 is extremely viscous and unexposed wafers tend to show a pattern where the outline of the poured resist prior to spinning can be clearly seen. This is likely to be slightly thicker than the resist across the rest of the wafer and may be enough to alter the level of stress. Measurements of the thickness are difcult to perform after the release of the test structures and so there is no thickness data available for either wafer. Each die contains structures with different values of the expansion arm separation ratio Y /W as detailed in section II. The average rotations for each value of Y /W have been

calculated for the whole wafer, a block of 618 dice in the centre of the wafer and for the remaining dice around the edge. These results are plotted against Y /W for both the standard structures and those with expansion arms that narrow where they connect to the pointer in Fig. 10 (for W1) and Fig. 11 (for W4). Each point on the graph represents the mean value while the error bars show the standard deviation (1). The number of structures that were measured to produce each point varies but is at least 100 in every case. The results from wafer W1 for standard test structures suggest that the level of rotation is independent of both the arm separation ratio and the position on the wafer. There does appear to be a small difference between the centre and edge but this is not statistically signicant. The results from the structures on W1 with narrowed expansion arms appear to increase slightly for Y /W = 1.75 but the variability of these results is also larger and so this is again of questionable signicance. The results from wafer W4 are quite different, with a rotation that increases with Y /W and a signicant difference between the centre and the edge. This would certainly be expected from the wafer maps. The slope, coupled with the signicant difference between the standard structures and those with narrowed expansion arms also suggests a signicant difference in stress between the two wafers. Simulations of similar structures, such as those detailed in [14], show that the position of the peak value of the curve showing the relation between the pointer angle and Y /W changes with strain. The difference between the results from the two wafers suggests that they are operating at different positions on the curve indicating a real difference in the stress in the material. This is most likely to be due to the fact that wafer W4 was printed with a lower exposure dose than W1. Results from the

11
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

Pointer Rotation (degrees)

Pointer Rotation (degrees)

6 100 80 60 Vertical Position (mm)

6.5

7.5

8.5

9.5

10 10.5 11 100 80 60 Vertical Position (mm) 40 20 0 -20 -40 -60 -80

6.5

7.5

8.5

9.5

10 10.5 11

40 20 0 -20 -40 -60 -80 -100 -100 -80 -60 -40 -20 0 20 40 Horizontal Position (mm)

60

80 100

-100 -100 -80 -60 -40 -20 0 20 40 Horizontal Position (mm)

60

80 100

Fig. 8. Wafer map of average stress structure pointer rotation for each die on a 200mm wafer (Wafer W1)

Fig. 9. Wafer map of average stress structure pointer rotation for each die on a 200mm wafer (Wafer W4)

literature suggest that a lower exposure leads to a lower density of cross-linking in the SU-8 which subsequently means that there is more shrinkage of the material during post exposure baking and therefore a greater level of stress [15]. As mentioned in section III, wafer W2 was only partially released. In fact only structures near the centre of the wafer were successfully released from the substrate. Fig. 12 presents the average rotations for the central area of W2 plotted against Y /W . The results from structures with narrowed expansion arms are quite similar to those from W1 but with much larger variability. The results from the standard structures show much lower rotations and an increase in rotation with Y /W . This suggests a lower stress in W2 related to the hard baking process which was somewhat unexpected as other studies have found increased shrinkage during hard bake [4]. However, the low yield of structures on W2 means that further samples and measurements are required to be sure of this result. V. C ONCLUSIONS AND F UTURE W ORK For the rst time, microrotating stress sensor test structures have been fabricated for the spatial characterisation of SU-8 photoresist. The test structures were produced by patterning a 35m thick layer of SU-8 3035 coated on a 200mm silicon wafer. The test structures were released from the silicon substrate using a XeF2 vapour etch and were subsequently observed to display clockwise rotation indicating signicant tensile stress. Measurements of the pointer arm rotation have been made on three wafers with different process parameters. The results from two wafers patterned with different exposures suggest that a larger exposure dose during lithography can reduce the level of tensile stress in the material. Initial results from a wafer subjected to a high temperature hard bake process after development suggest a reduction in tensile stress, but a

non-optimised release process led to a low yield of the test structures on this wafer. Further wafers need to be processed in order to conrm this result, which is contrary to other results from the literature. The micromechanical test structures used in this study have demonstrated their utility in the measurement and wafer mapping of the stress in patterned SU-8. This includes the identication of cross wafer variation which may be introduced at the spin coating stage of processing. Further work involving lm thickness measurements are required to conrm these results. ACKNOWLEDGEMENTS The authors would like to acknowledge the support of the Edinburgh Research Partnership in Engineering and Mathematics. They would also like to thank Dr. Anthony OHara and Graeme Pringle of Memsstar for help with processing. Jeremy Murray would like to acknowledge the support of the SFC funded SPIRIT studentship scheme. R EFERENCES
[1] E. Conradie and D. Moore, Su-8 thick photoresist processing as a functional material for mems applications, J. Micromech. Microeng., vol. 12, no. 4, pp. 368374, 2002. [2] L. Yu, F. E. H. Tay, G. Xu, B. Chen, M. Avram, and C. Iliescu, Adhesive bonding with SU-8 at wafer level for microuidic devices, Journal of Physics: Conference Series, International MEMS Conference 2006, vol. 34, pp. 776781, 2006. [3] C. Kourouklis, T. Kohlmeier, and H. H. Gatzen, The application of chemicalmechanical polishing for planarizing a SU-8/permalloy combination used in MEMS devices, Sensors & Actuators: A. Physical, vol. 106, pp. 263266, Jan 2003. [4] R. Feng and R. J. Farris, Inuence of processing conditions on the thermal and mechanical properties of SU8 negative photoresist coatings, J. Micromech. Microeng., vol. 13, no. 1, pp. 8088, 2003.

12
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

(a)

(a)

(b) Fig. 10. Average pointer arm rotations from W1 plotted against arm separation ratio for (a) standard structures and (b) structures with narrowed expansion arms

(b) Fig. 11. Average pointer arm rotations from W4 plotted against arm separation ratio for (a) standard structures and (b) structures with narrowed expansion arms

[5] S. Bystrova, R. Luttge, and A. van den Berg, Study of crack formation in high-aspect ratio SU-8 structures on silicon, Microelectronic Engineering, vol. 84, pp. 11131116, 2007. [6] W. Dai, K. Lian, and W. Wang, A quantitative study on the adhesion property of cured SU-8 on various metallic surfaces, Microsystem Technologies, vol. 11, no. 7, pp. 526534, 2005. [7] S. Smith, N. L. Brockie, J. G. Terry, N. Wang, A. B. Horsfall, and A. J. Walton, Application of a Micromechanical Test Structure to the Measurement of Stress in an Electroplated Permalloy Film, in Proceedings of the IEEE International Conference on Microelectronic Test Structures (ICMTS09), Oxnard, CA, US, March-April 2009, pp. 7580. [8] C. J. Wilson, K. Croes, Z. Tokei, G. P. Beyer, A. B. Horsfall, and A. G. ONeill, Demonstration of a Sub-micron Damascene Cu/Lowk Mechanical Sensor to Monitor Stress in BEOL Metallization, in Proceedings of the IEEE International Conference on Microelectronic Test Structures (ICMTS09), Oxnard, CA, US, March-April 2009, pp. 3135. [9] J. G. Terry, S. Smith, A. J. Walton, A. M. Gundlach, J. T. M. Stevenson, A. B. Horsfall, K. Wang, J. M. M. dos Santos, S. M. Soare, N. G. Wright, A. G. O. Neill, and S. J. Bull, Test Chip for the Development and Evaluation of Sensors for Measuring Stress in Metal Interconnect, IEEE Transactions on Semiconductor Manufacturing, vol. 18, no. 2, pp. 225261, May 2005. [10] B. P. van Drieenhuizen, J. F. L. Goosen, P. J. French, and R. F. Wolffenbuttel, Comparison of techniques for measuring both compressive and tensile stress in thin lms, Sensors and Actuators A (Physical), vol. 37-38, pp. 756765, 1993. [11] X. Zhang, T. Y. Zhang, and Y. Zohar, Measurement of Residual Stresses in Thin Films Using Micro-Rotating Structures, Thin Solid Films, vol. 335, no. 1, pp. 97105, 1998.

Fig. 12. Average rotations from structures in the centre of wafer W2 plotted against arm separation ratio

[12] SU-8 3000 Permanent Epoxy Negative Photoresist, Data Sheet, MicroChem Corp. [13] A. J. Walton and S. Smith, A Review of Test Structures for Characterising Microelectronic and MEMS Technology, Advances in Science and Technology, vol. 54, pp. 356365, Sep 2008. [14] C. J. Wilson, Stress Measurements in Deep Sub-micron Damascene Copper Interconnects, Ph.D. dissertation, Newcastle University, 2009. [15] D. Sameoto, S.-H. Tsang, I. G. Foulds, S.-W. Lee, and M. Parameswaran, Control of the out-of-plane curvature in SU-8 compliant microstructures, Journal of Micromechanics and Microengineering, vol. 17, pp. 10931098, 2007.

13
Authorized licensed use limited to: IEEE Xplore. Downloaded on November 24,2011 at 13:56:00 UTC from IEEE Xplore. Restrictions apply.

You might also like