Process Variations in Microsystems Manufacturing

Chapter 2 provides a foundation for future chapters and begins with a review of the transduction mechanisms most commonly used in the implementation of MEMS microsensors and microactuators. Among the transduction effects covered includes piezoresistive; capacitive; piezoelectric; tunneling; magnetic; photoconduction; thermoelectric; electrostatic; thermal; shape-memory alloy; and others. The differences between a processing step; a process module; a process sequence; and a process technology are then described. The concept of batch fabrication is explained including the significant benefits that are derived from the use of this manufacturing method. Some of the important distinguishing characteristics of MEMS fabrication compared to IC manufacturing are reviewed, and reasons why MEMS is both more interesting and challenging compared to IC manufacturing are given. Specifically it is noted that MEMS design is very interesting since the number of device types and potential application areas is enormous. Nevertheless, MEMS implementation usually involves significant customization of both the design and the process sequence, and therefore the device designer often does not have much prior knowledge to leverage from. Lastly, a review of some of the basics about semiconductor materials, which are heavily used in MEMS manufacturing, is provided. Miller indices are explained as part of this discussion.

A general overview of the processing steps commonly used in integrated circuit (IC) manufacturing is provided in Chap. 3. How each processing step is performed, the equipment commonly used, and guidance on the expected dimensional variations when performing the processing step are given. The subsequent chapter will focus on specialized processing steps used in MEMS fabrication. The major categories of processing steps used in IC fabrication include depositions or growths; lithography; etching; impurity doping; and metrology. Depending on the process sequence involved, there may also be other types of processing steps in the sequence as well such as planarization, rapid thermal anneals, and others. A number of these processing steps will be performed sequentially to implement the ICs, and some will be repeated multiple times. Once the fabrication is completed, the wafers will usually go through a series of tests to determine their functionality and performance. This is discussed in more detail in Chaps. 7, 8, and 9. Table 3.4 provides a compilation of the expected “best-case” dimensional variations for each of the processing steps reviewed in this chapter as a quick reference.

Chapter 4 gives an overview of the processing steps and process modules used in MEMS manufacturing. Like IC fabrication, MEMS processing steps can be lumped into major categories including depositions; lithography; etching; impurity doping; etc. While MEMS fabrication shares a number of attributes with IC processing steps reviewed in Chap. 3, there are also a number of differences. In some cases, MEMS uses the same equipment as IC fabrication, with the distinguishing feature either that the material processed is not something used in IC manufacturing or some other attribute, such as the thickness of the film deposited is exclusive to MEMS. Additionally, there are some MEMS processing steps that are unique (i.e., are not performed in IC fabrication) and may use specialized equipment. Some MEMS fabrication methods, such as bulk micromachining, are better labeled as process modules rather than processing steps, and these are also described. The substrates used in MEMS manufacturing are also far more diverse than those used in IC manufacturing and are also reviewed. Lastly, as in Chap. 3, general guidance as the best-case expected dimensional variations that can be obtained in performing these MEMS processes is summarized in Table 4.2.

Microsystems fabrication utilizes a number of metrology techniques during development and manufacturing that are reviewed in Chap. 5. These techniques are used in development after processing steps are performed (reviewed in Chaps. 3 and 4) to find and diagnose problems that may be present. In manufacturing, metrology is employed to maintain quality control and thereby increase production yields. Metrology also enables the process engineers to monitor the status of the processing equipment used in production. There are a wide variety of metrology tools that are available, including inspection of the devices during and after fabrication; measurement of the dimensions of various important elements of the microsystems; and chemical analysis of materials used in fabrication. Some of the basic metrology techniques that have been developed for the IC industry are discussed, followed by a review of several more specialized metrology techniques specific for MEMS fabrication. Guidance as the accuracy of each of the metrology methods is given in Tables 5.2 through 5.5 providing a quick look-up summary of these metrology methods along with their resolution, precision, and accuracy.

Chapter 6 reviews some of the important properties of the most commonly used materials in microsystems manufacturing. It is explained that the material properties are dependent on the processing conditions, and since many process sequences are customized, there is often insufficient knowledge of the properties during development. Most attention is given to two specific material properties, namely, Young’s modulus and residual stress, due to the fact that these usually have an important impact on the behavior of MEMS devices and the fact that these properties can vary quite significantly depending on the processing conditions. The use of test structures, including both mechanical and electrical, for measuring various material properties is explained. A review of the material properties for some of the most commonly used materials in microsystems manufacturing is then provided including semiconductors; dielectrics; and metals. The purpose of providing information about reported values of Young’s modulus and residual stress in deposited thin-film layers is to give an appreciation of the amount that these properties can vary with processing conditions and some guidance about the ranges that these properties may span.

This chapter covers the important topic process integration where a number of individual processing steps (covered in Chaps. 3 and 4) are combined into a process sequence for the implementation of MEMS devices. A general outline of yield improvement activities is given. Some of the challenges of process integration for MEMS are discussed including the significant time, cost, and risk that are commonly encountered. Integrated MEMS process sequences are defined as the merging of microelectronics onto the same substrate as the MEMS devices. Reasons why attempting to develop integrated MEMS is so difficult are also explained. Two notable examples of process technologies are reviewed; the first is a generic MEMS surface micromachining process sequence, and the second is an integrated MEMS process technology. Parameter variations that arise in process sequences are discussed, and an example of the parameter variations of a well-known MEMS process technology called PolyMUMPS™ is reviewed. The concept of design rules in microsystems manufacturing is then covered along with a review of the design rules for the PolyMUMPS™ process sequence. This chapter also briefly discusses the testing and packaging of MEMS, including device trimming and calibration.

This chapter focuses on a major theme of this volume, namely, how to analyze the variations in device parameters that occur when using microsystems fabrication technologies. It is explained that parameter variations are important since they result in the device output differing from the expected device output behavior that is based on the design. Two different types of parameter variations are discussed: systematic (bias) variations and random variations. Bias variations are fixed amounts of offsets that occur in the device parameters, while random parameter variations are caused by non-systematic process variations. It is discussed that the magnitude of these parameter variations can significantly vary depending on the specific details of the equipment, process being performed, and the aggressiveness of the device dimensions. This chapter also covers the important concepts of precision and accuracy. Both are important for a well-controlled manufacturing process. The tools of statistical analysis are covered for both continuous and discrete probability distributions. Various examples are used to reinforce how these statistical methods can be effectively employed in analyzing the variations of the device output behavior using microsystems manufacturing. The material covered in this chapter will be used in the next chapter in explaining parametric yield analysis.

This chapter covers the importance of ensuring quality and reviews methods for performing yield analysis in microsystems manufacturing. Methods for determining device functional and parametric yields are both reviewed. The types of defects that result in nonfunctional devices and various analytical techniques used for functional yield modeling based on point defects are explained. Parametric yield is then reviewed wherein a manufacturing function that incorporates all of the statistical distributions of the parameters describing the microsystems device’s output response is overlaid by an acceptance region to determine the yield. Methods to ensure quality during manufacturing using statistical process control (SPC) are covered. Control charts are used to show how processing steps can be monitored for whether they are in control or not. Process capability is shown to be an excellent gauge for whether a manufacturing process is able to produce devices having acceptable yields. Lastly, techniques for sampling of the measurements on wafers during and after manufacturing so as to obtain data that is rationally subgrouped so as to identify non-random changes are discussed.

The information covered from the previous chapters is brought together in this chapter to explain various techniques used in the microsystems design to manage the parameter variations resulting from use of microsystems manufacturing. Design for manufacturability (DfM) of microsystems is covered followed by some general recommendations for developing microsystems designs that adhere to DfM principles for MEMS devices. A review of the design techniques to manage device parameter variations is then provided including design centering: device parameter variation reduction; device size scaling; acceptance region increase; and best practices for layout. These techniques allow the variation region to be better aligned with the acceptance region. Each of these techniques is substantiated with examples in a one-dimensional parameter space, followed by how these techniques are used in multidimensional space. The use of Monte Carlo analysis techniques for design methods is then discussed including specific methods such as the centers of gravity algorithm; correlated sampling; and the common points method. The confidence of correct yield ranking is included in this discussion. Subsequently, sensitivity analysis for manufacturing or performance function improvement is outlined in both one- and multidimensional spaces. Lastly, a method for optimization of the manufacturing cost function is given.