University of Wisconsin-MEMS - Polysilicon Surface Micromachining

Polysilicon surface micromachining research started in the early 1980's at the University of Wisconsin - Madison, in an effort to create high precision, micro pressure sensors. Most recently it has been applied to even higher precision sensors using vacuum encapsulated resonant micro beams.

Polysilicon surface micromachining is an additive fabrication technique which involves the building of a device on top of the surface of a supporting substrate. This technique is relatively independent of substrate; therefore, it can be easily mixed with other fabrication techniques which modify the substrate first. An example is the fabrication of MEMS on a substrate with embedded control circuitry. This type of surface micromachining uses modified interated circuit (computer chip) fabrication techniques and materials. These materials include doped and undoped single-crystal silicon and polysilicon, silicon nitride, and silicon dioxide for the the electrical and mechanical structures; and aluminum alloys for the metal connections to the outside world. In order to create mechanical structures out of these silicon based materials, the internal stresses of each must be controlled.

The control of the internal stresses of a thin film is important for the fabrication of micro electro mechanical structures. The microelectronic fabrication industry typically grows/deposits polysilicon, silicon nitride, and silicon dioxide films with recipes that minimize time. Unfortunately a deposition process which is optimized to speed, does not always create a low internal stress film. In fact, most of these films have internal stresses which are highly compressive. Imagine creating a free standing plate of highly compressive polysilicon which is held at all edges. The plate will buckle, which is highly undesirable when creating micro mechanical structures. The solution is to modify the film deposition process to control the internal stress by making it stress free or slightly tensile.

The stress of a polysilicon film can be controlled by doping it with boron, phosphorus, or arsenic. Unfortunately a doped polysilicon film is conductive which may interfere with the electronics incorporated into the mechanical device. Another problem with doped polysilicon is that it is roughened by hydrofluoric acid which is commonly used to free sections of the final mechanical device from the substrate. Rough polysilicon has different mechanical properties than smooth polysilicon; therefore, the amount of roughing must be taken into account when designing the mechanical parts of the micro device. A better way to control the stress in polysilicon is through post annealing. In this process, pure, fine-grained polysilicon is deposited which is compressive. Annealing the polysilicon after deposition at elevated temperatures can change the film to be stress free or tensile. The annealing temperature sets the films final stress. Using this method electronics can be incorporated into polysilicon films through selective doping, and hydrofluoric acid will not change the mechanical properties of the material.

The stress of a silicon nitride film can be controlled by deposition temperature and the film's silicon to nitride ratio. Films can be deposited in compression, stress free, or in tension. (See Sekimoto, Yoshihara and Ohkubo, "Silicon Nitride Single Layer X-Ray Mask," Journal of Vacuum Science and Technology, 21(4), Nov./Dec. 1982, pp. 1017-1021.)

The stress of a silicon dioxide film can be controlled by deposition temperature and post annealing. It is difficult to control the stress in silicon dioxide accurately; therefore, silicon dioxide is typically not used as mechanical material by itself. Silicon dioxide is commonly used for electronic isolation or as a sacrificial layer under polysilicon. A sacrificial layer is a temporary layer which can be selectively removed later allowing partial or complete release of the structures above. Silicon nitride may also be used for electronic isolation and as a sacrificial layer.

Example Process

Free standing polysilicon beams can be fabricated by first oxidizing a silicon substrate and then patterning the silicon dioxide into long narrow bars. Then polysilicon may be deposited and patterned into long narrow bars positioned directly over the silicon dioxide bars but oriented 90 degrees. The silicon dioxide may then be removed by hydrofluoric acid resulting in polysilicon beams which are free from the substrate in the center. The process is shown below.

(1) Grow silicon dioxide: Silicon dioxide is first grown thermally on a silicon substrate. For example a growth could be done in a water vapor ambient at 1000 degrees Celsius for 1 hour. The result would be that 0.3 microns of the silicon surface would be converted into 0.6 of silicon dioxide. Thermal oxide thickness is limited to a few microns due to the diffusion of water vapor through silicon dioxide. Silicon dioxide can be deposited without modifying the surface of the substrate, but this process is slow and the film is highly stressed. The result are films limited to near 4 microns. Silicon nitride may also be deposited, but its thickness is limited to the same range.
(2) Apply photoresist: A photo-sensitive material, commonly called photoresist, is next applied to the surface of the silicon dioxide. This is typically done by spin coating the photo-sensitive material suspended in a solvent. The result after spinning and driving off the solvent is a photo-sensitive material that is 0.3 to 1.5 microns in thickness. Of course these are just typical values of thickness. The photoresist is then soft baked to drive off the solvents inside.
(3) Expose and Develop: The photoresist is then exposed to light patterned by a mask. This mask blocks the light and controls the pattern. The photoresist is next developed. The drawing above shows a positive photoresist where the exposed areas are removed in the developer. In a positive photoresist the light will decrease the molecular weight of the photoresist, then the developer will selectively remove (or etch) the lower molecular weight material.
(4) Etch silicon dioxide: The silicon dioxide may then be etched. The remaining photoresist will be used as a hard mask which protects sections of the silicon dioxide. Etchants may be classified as wet (i.e. hydrofluoric acid) or dry (i.e. plasma etch, NF3). The photoresist may then be removed by a wet process (i.e. piranha = sulfuric acid and hydrogen peroxide) or dry (i.e. ash = oxygen plasma). The result is a silicon dioxide bar on the silicon substrate.
(5) Deposit polysilicon: Polysilicon is next deposited over the silicon dioxide bars. Polysilicon is typically deposited in an LPCVD system (Low Pressure Chemical Vapor Deposition system) at temperatures near 600 degrees Celsius in a silane (SiH4) ambient. Deposition rates are slow, near 70 angstroms per minute. The slow deposition rates and internal stress prevents films from being deposited over 4 microns in thickness. The polysilicon must be stress free or have a tensile internal stress. A compressive polysilicon beam will buckle.
(6) Apply photoresist: Photoresist is again applied, but planarization is now a problem. The patterned silicon dioxide bars changes the topology of the substrate surface. It is difficult to apply a uniform coat of photoresist over a surface with different heights. The result is a photoresist film which varies in thickness and may not cover the corners or edges of the patterns. In our case, a 0.6 micron step height is not that significant; but for thicker films and multiple layers, replanarization may be necessary.
(7) Expose and Develop: A photolithography mask containing the definition of the polysilicon bars must then be aligned to the silicon dioxide bars already on the silicon substrate. Alignment tolerances can be quite great because the silicon dioxide and polysilicon bars are large. Bars in the hundreds of microns in length are common allowing for alignment in the micron range. This is easily achieved with older VLSI fabrication equipment. Thus many of todays older VLSI fabrication facilities are being converted to run MEMS, and universities can easily get into the MEMS game by acquiring old VLSI equipment.
(8) Etch polysilicon: The polysilicon film is next etched with the photoresist protecting the polysilicon bars. It is difficult to find a wet etch for polysilicon which does not attack photoresist; therefore, a dry etch using plasma is common. Selectivity of the plasma between polysilicon and silicon dioxide is not necessary here because the silicon dioxide will be removed later. This allows some process latitude, since the polysilicon can be overetched by etching it longer than necessary. The result is an etch which can have higher yield in creating isolated polysilicon bars.
(9) Remove photoresist: The photoresist protecting the polysilicon bars is next removed.
(10) Remove silicon nitride: The silicon nitride is next removed. A common technique is to perform a wet etch since plasma cannot easily remove the silicon dioxide in the confined space under the polysilicon beam. A common wet etch is hydrofluoric acid. Since hydrofluoric acid does not attack pure silicon, the polysilicon bar and silicon substrate will not be etched. After the silicon dioxide is completely removed, free beams of polysilicon will be formed. These beams can bend down and stick to the surface of the substrate during drying after the wet etch. This can be a problem. Others have solved this problem by creating rough polysilicon which does not stick or designing the internal stress of the polysilicon so that it springs up during drying. Unfortunately both of these procedures change the mechanical properties of the polysilicon beam. A more generic technique is freeze dry sublimation which was created at here at the University of Wisconsin.

Hopefully the above discussion has explained the polysilicon surface micromachining technique and has exposed some of the concerns in structure design and fabrication. Polysilicon surface micromachining is a thin film process. Each film is limited to thicknesses near 4 microns. The result is a rather planar micro device. The advantage of polysilicon surface micromachining is the use of standard VLSI equipment and fabrication techniques; therefore, cost-effectiveness can be derived from the highly successful semiconductor industry.