Abstract
As a major leap in the development of micro-instrumentation systems, a miniature Fourier transform spectrometer (FTS) is presented. The FTS is predicated on the standard Michelson interferometer. Translation for the moving mirror and xyz alignment stages are enabled by dovetail MicroJoints microfabricated in (100) silicon, and the entire system assembled on a single 4" silicon wafer. The dovetail microstages allow precise actuation of the moving mirror up to 10 cm. In this report actuation mechanisms for the optomechanical components are presented and the modulation efficiency of the FTS is demonstrated by obtaining a interferogram and Fourier transformed spectrum of a narrow band interference filter.

Introduction
Infrared spectroscopy is one of the most powerful and versatile techniques available in today's chemical analysis repertoire. It permeates all aspects of chemical analysis from process control to basic research and generally serves as the "gold standard" for routine organic compound identification. Because of its increased sensitivity, speed, and reliability, Fourier transform methods[1] (FTS) have eclipsed the more traditional dispersive techniques. However, despite their utility, FT spectrometers are still rather large, expensive, and delicate instruments, which are normally reserved for the controlled environment of the laboratory. Small, reliable, and low cost FT instrumentation would not only expand FT spectroscopy applications, but also provide synergistic benefits in the realization of radically new chemical analysis concepts. Miniaturization immediately expands FT spectroscopy into remote, hostile, or previously inaccessible locations. With small size and low cost also comes the possibility of automated environmental sensors to monitor virtually any desired set of preprogrammed chemical signatures through the use of standard pattern recognition algorithms. More importantly, routine miniaturization also allows the integration of multiple instruments into a single, compact analysis system. Already miniature gas chromatagraphs (GC), micromachined using MEMS technologies, are available commercially[2], and the combination of FTIR with GC offers a powerful first generation microinstrument for automated separation and analysis of small volume samples.

FTS Design With Microfabrication Technology
Recently a miniature interferometrically based FTS spectrometeras has been proposed[3] which follows essentially the same construction as that used by Michelson in the 1890's to study the speed of light. FTS implementation with dovetail technology is shown in Figure 1 and its operation is described in Figure 2. In order to obtain high spectral resolutions, R, mirror retardations, d, must be rather large, i.e. R = 1/d. Ignoring apodization and other obligatory signal processing, to obtain a spectrum with a resolution of 0.1 cm-1, requires a minimum mirror retardation of 10 cm, i.e. a travel distance of 5 cm for a single pass instrument.

This large translational distance is enabled by precision dovetail technology[3,4]. The microstages are based on scaled-down versions of traditional dovetail joints commonly used in conventional optical positioning stages. The dovetail MicroJoints are anisotropically etched [5] from (100) single crystal silicon generating virtually perfect (111) crystal planes for precision bearing surfaces on which the translational microstages ride. The general fabrication and description of the dovetail MicroJoint is reported elsewhere [4] along with applications to selected optical systems [6-9]. Although this new microfabrication paradigm allows precision translation of microstages over large distances, actuation of those microstages is still an ongoing avenue of research.

Actuator Design
Actuation of the optomechanical component, particularly the moving mirror, is critical in white light interferometry using FTS. Not only do FTS measurements require precise knowledge of distance, but they are extremely susceptible to vibration. Standard electrostatic comb drives[7] and other commonly reported actuation mechanisms[8] lack the necessary range of motion or mechanical stability required in this application. We present a magnetic actuation mechanism which realizes the continuous, precision translation of microstages over arbitrarily large distances. It is based on linear induction motors (LIM) commonly used in magnetic levitation for high speed railway systems. Although initial work has been on DC operation (low-frequency), like LIM's AC operation offers the possibilities for frictionless translation by levitation.

The dovetail structures which have been reported elsewhere [4] provide an ideal track for a linear actuator. The maximum length of a dovetail track is determined solely by wafer diameter, and additional stability can be obtained using multiple tracks for dovetail sliders. In the process to develop a linear actuator, a literature review revealed countless macro-scale linear motors, with applications ranging from manufacturing to high speed electric rail9,10. Following conventional macro-scale designs, a diagram of a linear magnetic motor using dovetail structures is shown in Figure 4. An array of microcoils are fabricated underneath the length , 4-8 cm, of the dovetail translational track and connected in a typical three phase series configuration. An array of permanent magnets is placed on the translational microstage either by depositing / electroplating a thick film of permalloy metal, or in prototyping simply hand mounting a small permanent magnet to the dovetail stage. Each individual coil contains 6 turns and is 2mm in diameter. Energizing the microcoils with three phase timing via external feedback control circuitry provides smooth mirror translation.

Due to the relatively thin coils (~5 um), overheating induced by relatively high current requirements (~100 mA) was an issue. In attempt to minimize power with a fixed coil area and required magnetic field, coils were calculated to require a cross-sectional area of 300(m by 5(m. In addition to withstanding high currents, an electrical jumper was required to make the three phase connection. With concerns of crossing over 5(m thick coils, the electrical jumper was placed on the bottom layer. Here the jumper could be wide but thin, approximately 1.5 um, which allows the coils to easily cross over.

Fabrication
Dovetail structure fabrication has been reported elsewhere[3], but it is noted that dovetails can be made with less than 1(m of "slop" in the slider to track fit. The coil fabrication process is shown in Figure 5. The backside of dovetail wafers are first covered with deposited silicon nitride for both electrical insulation and slider coefficient of friction reduction. Crome and gold are then evaporated to be used as a seed layer for electroplating, masked by photoresist. Gold for the jumper across the coil lanes is electroplated to approximately 1.5 (m to allow for large current densities. The gold seed layer was typically 600 A and the chrome adhesion layer was 100 A. After stripping the resist, polyimide (5(m) was spun over the bottom metallization layer. Using photoresist as an etch mask (not shown), holes were etched in the polyimide. The same metallization process was repeated for the second layer. Chrome was again used as an adhesion layer to increase gold adhesion to polyimide. No polyimide degradation was observed in chrome or gold etchants as well as in photoresist stripper, after the curing cycling was completed. Figure 6 shows the completed coils. The polyimide is transparent at 5 micron thickness and the connector underneath the coil lanes is clearly visible. A view of one wafer is also included to show that coils travel across the wafer.

Preliminary Results
To demonstrate the feasibility of using magnetic actuation for translating the moving mirror, the modulation efficiency of an FTS using dovetail translation microstages was measured. A 10 cm dovetail MicroJoint translational stage with associated aluminum mirror was incorporated into one arm of a standard Twyman-Green11 interferometer. A tungsten "white" light with 10 nm interference filter whose peak was centered at 650 nm was the optical source. Interferograms were measured in the time domain and their Fourier transforms used to obtain the power spectra in the frequency domain. The Welch method of power estimation[12] was used. Measurements were restricted to the near IR to visible regime, 1,000 nm to 400 nm. Although the shorter wavelengths of visible light places more stringent requirement on the optomechanical components, i.e. mirror alignment/translation, it avoids the complications of IR optics and atmospheric absorption. Extension into the infrared region is straightforward, requiring a suitable IR beamsplitter and reflective coatings. A NeHe laser was used to clock the mirror, and a silicon p-i-n photodiode was used as the detector. The micro mirror was actuated a total distance of 8-10 cm (retardation of 16-20 cm) providing a maximum spectral resolution of 0.05 cm-1. The velocity of the mirror averaged 17 mm/sec requiring approximately 6 seconds to complete a single pass scan. The velocity of the mirror was limited not by the actuation mechanism but the response time of the silicon p-i-n diode. The interferogram and power spectrum for the interference filter is shown in Figure 7a) and compares favorably with the spectrum obtained on a commercial FTS (Bruker), Figure 7(b). The transmission peak for the miniature FTS occurs at 655.7 nm while the peak for the commercial instrument occurs at 661.2 nm.

Despite some discrepancies and jitter on the magnetically actuated mirror, the results are extremely promising considering that there was no active feedback control of the mirror alignment or position. Currently, a control strategy is being developed to generate a constant velocity drive using position feedback. Additionally, deposition of magnetic thin films are being explored which will enable a decrease in coil diameter and an increase in sensitivity.

Conclusions
We have demonstrated the magnetic actuation of dovetail translational microstages and their applicability to the realization of a miniature Fourier transform spectrometer.

Acknowledgments
Portions of this work were supported by NASA, NSF, and Damien Associates. We would like to thank Ken P. Stewart of NASA, Goddard Space Flight Center for supplying commercial FTS spectra for comparisons.
* Joint Center for Earth Systems Technology University of Maryland Baltimore County, NASA Goddard Space Flight Center Greenbelt, Maryland

References
1.   A. E. Martin, Infrared Interferometric Spectrometers, Amsterdam ; New York : Elsevier Scientific Pub. Co., 1980.

2.  E.S. Kolesar Jr. R.R. Reston, Review and summary of a silicon micromachined gas chromatography system. IEEE Transactions on Components, Packaging and Manufacturing Technology, Part B: Advanced Packaging, vol.21, (no.4), (1997 Second Annual IEEE International Conference on Innovative Systems in Silicon, Austin, TX, USA, 8-10 Oct. 1997.) IEEE, Nov. 1998. p.324-8.

3.  C. Gonzalez, R. L. Smith, S. D. Collins, and M. Sirota, Mesoscopic Optical Instrumentation: A Miniature Fourier Transform Spectrometer, Proceedings. LEOS '97. 10th Annual Meeting IEEE Lasers and Electro-Optics Society 1997, San Francisco, CA, USA, 10-13 Nov. 1997 p. 474-5 vol.2.

4.  C. Gonzalez, R. L. Smith, D. G. Howitt, and S. D. Collins, "MicroJoinery: concept, definition, and application to microsystem development," Sensors and Actuators A-Physical, 66(1-3), 315-332 (1998).

5.  H Seidel, L. Csepregi, A. Heuberger, H Baumgartel, Anisotropic Etching of Crystalline Silicon In Alkaline Solutions .1. Orientation Dependence And Behavior Of Passivation Layers. Journal of the Electrochemical Society, Nov, 137(N11) 3612-3626, 1990. H Seidel, L. Csepregi, A. Heuberger, H Baumgartel, Anisotropic Etching of Crystalline Silicon in Alkaline Solutions .2. Influence of Dopants. Journal of the Electrochemical Society, Nov, 137(N11) 3626-3632, 1990.

6.  C. González and S. D. Collins, Micromachined 1 x n Fiber-Optic Switch, IEEE Photonics Letters, 9(5), (1997) 616

7.  W. C. Tang, M. G. Lim, and R. T. Howe, Electrostatic Comb Drimb Levitation and Control Method, Journal of Microelectromechanical Systems, vol. 1, No. 4, December 1992

8.  H. Fujita, Recent Progress in Micromachining and Applications to Microactuators, Jpn. J. Appl. Phys. Vol. 33, 1994, pp. 7163-7166.

9.  S. Yamamura, Theory of Linear Induction Motors 2nd Ed., Halsted Press, 1979

10.  I. Boldea, S. A. Nasar, Linear Motion Electromagnetic systems, John Wiley & sons, 1985

11.  F. Twyman, Prism and Lens Making, 2nd ed., Hilger, London (1988).

12.  A. V. Oppenheim and R. W. Schafer, DIGITAL SIGNAL PROCESSING, Prentice-Hall, 1975, p 556.
 

Figures

Figure 1: Photograph of an FT interferometer assembled on a 4 inch silicon wafer. An external HeNe laser beam shows the optical alignment of the components.

Figure 2: Diagram of a typical Fourier transform spectrometer with Michelson interferometer design.

As mirror moves a monochromatic light source will experience a cyclic intensity distribution at the detector due to interference.

Figure 3: Microfabricated Dovetails

Figure 4: Diagram of magnetic actuation mechanism.

Figure 5: Coil Fabrication Process

Figure 6: Microfabricated Electrical Coils, 3 Phase Arrangement With Polyimide Insulation

Figure 7: Interferogram and Fourier Transformed Spectrum Of A 650 nm Interference Filter. A) Microfabricated FTS B) Commercial