Investigation of gray-scale technology for large area 3D silicon MEMS structures

Abstract

Micromachining arbitrary 3D silicon structures for micro-electromechanical systems can be accomplished using gray-scale lithography along with dry anisotropic etching. In this study we have investigated two important design limitations for gray-scale lithography: the minimum usable pixel size and maximum usable pitch size. Together with the resolution of the projection lithography system and the spot size used to write the optical mask, the maximum range of usable gray levels can be determined for developing 3D large area silicon structures. An approximation of the minimum pixel size is shown and experimentally confirmed. Below this minimum, gray levels will be developed away due to an excessive amount of intensity passing through the optical mask. Additionally, oscillations in the intensity are investigated by the use of large pitch sizes on the optical mask. It was found that these oscillations cause holes in the photoresist spaced corresponding to the pitch used on the gray-scale mask and penetrate the thickness of the photoresist for thin gray levels. From the holes in the photoresist, significant surface roughness results when used as a nested mask in reactive ion etching, and the very thin gray levels are lost.

Double-Exposure Grayscale Photolithography

“…Three-dimensional photolithography using a gray-scale mask [27][28][29][30][31][32][33][34][35][36] is capable of realizing structures with a higher vertical resolution than multipleexposure photolithography while still using conventional photolithography tools.…”

“…Therefore, the UV transmission pattern controls the dose delivered to the photoresist. The mechanism to control the UV intensity gradient in gray-scale photolithography varies with each technique and has been demonstrated with pixelated [27][28][29][30][31][32][33] and continuous-tone [34,35] optical masks. Pixelated gray-scale masks use diffraction through many sub-resolution pixels to control the UV dose.…”

“…This feature is convenient for maintaining process repeatability without needing to adjust the exposure time manually, but makes it very difficult to know the actual dose used for the exposure. As a solution, the entire process is calibrated using normalized values as demonstrated previously [28,[39][40][41].…”

“…The type of photoresist was selected by the thickness on the wafer and the contrast [40]. Photoresist contrast describes the development rate of the photoresist as a function of the dose.…”

Rapid assembly of multilayer microfluidic structures via 3D-printed transfer molding and bonding

“…Although rounded channels are beneficial for some microfluidic applications, few groups have developed appropriate fabrication techniques (Supplementary Material S3.4) [8][9][10] because multilevel soft lithography has historically required multiple photolithography steps 11 . Although "grayscale lithography"-whereby resists are exposed to non-binary shades of gray-can potentially generate rounded microfluidic channels [12][13][14] , the process still requires multiple exposures to obtain larger aspect ratios 15,16 . Furthermore, although multilayer PDMS-manufacturing techniques have been demonstrated by several groups 17,18 , these are even more time-consuming and labor-intensive, requiring multiple lithography steps and precision alignment, issues that are only partially addressed by dedicated PDMS-alignment tools 19 .…”