We present preliminary results for the analysis of shrinkage of 3D carbon posts obtained by the pyrolysis of SU-8 precursor structures, in a technique known as Carbon MEMS. SU-8 is a negative photoresist that has become a workhorse in micro and nanofabrication¹. Previous reports showed a near isometric shrinkage of the precursor structure during carbonization. Although this shrinkage is highly reproducible for a given structure and carbonization protocol, there seems to be a strong dependence of shrinkage on the initial dimensions of the pillars.  Here we take methodical steps towards understanding this phenomenon. The aim is at obtaining a model to predict the shrinkage of the precursor structure during carbonization and facilitate the design of carbon geometries. Our hypothesis is that shrinkage depends on the area available for degassing during carbonization. In the specific case of cylindrical posts, shrinkage would depend on the ratio between the surface area at the top of the cylinder and its lateral surface area. We have evaluated the shrinkage of cylindrical posts with height of 10, 50 and 100 µm and diameters of 10, 20,30,40,60, and 80 µm. The final heating temperature included values of 650°C, 900°C, 1150°C and 1300°C while the heating atmosphere was either nitrogen or vacuum. The dimensions of the structures before and after pyrolysis were measured at different positions in the sample, and used to calculate percent shrinkage with respect to the aspect ratio and surface area of the structures.

Initial results are presented in Fig. 1.  Structures with aspect ratios (height/diameter) lower than 1 feature more shrinkage in height than in diameter. The top surface area in this case is much higher than the lateral surface area, resulting in more degassing through the top surface of the post. As the aspect ratio increases beyond 1, the values for shrinkage in height and diameter start to be common. In other words, shrinkage tends to become isometric as the height of the posts increases for a given diameter.  Shrinkage does not seem to be dependent on temperature after 800°C, as also observed in previous works². However, the shrinkage observed when using a vacuum atmosphere is higher than in the case of nitrogen as seen from Fig.2.

These results reinforce our hypothesis that the shrinkage is dependent on the surface available for degassing, or the ratio of lateral to top surface area. A faster rate of degassing through the lateral area results in higher lateral shrinkage and same is true for the top surface. The choice of atmosphere is likely to play a role on determining the speed of diffusion of the species being generated during pyrolysis.

Ongoing work is 1) exploring the shrinkage of different geometries featuring the same surface area as those explored here; 2) relating this shrinkage to the ratio of volume/surface area for different structures across different length scales; and 3) quantifying the rate of shrinkage based on the byproducts of carbonization at different temperatures. The ultimate goal is developing a mathematical model to explain the shrinkage phenomena and incorporate it as a tool in the design of carbon MEMS structures. 

References:

1. R. Martinez-duarte and M. J. Madou, in Microfluidics and Nanofluidics Handbook: Fabrication, Implementation and Application, S. Chakraborty and S. Mitra, Editors, p. 231–268, CRC Press (2011).

2. B. Y. Park, L. Taherabadi, C. Wang, J. Zoval, and M. J. Madou, J. Electrochem. Soc., 152, J136 (2005).

3. L. Amato et al., Carbon N. Y., 94, 792–803 (2015) http://linkinghub.elsevier.com/retrieve/pii/S0008622315005291.