IMPROVING HOLOGRAPHIC LITHOGRAPHY
The present invention relates to holographic lithography. In holographic lithography a periodic or guasi periodic pattern is exposed in a photosensitive film (usually called a resist) by overlapping two beams, typically with mirrors, from a laser or other coherent source. In one particular implementation of holographic lithography, termed "achromatic holographic lithography", gratings are used to split and recombine the beams. (see E. Anderson, K. Komatsu, and H. I. Smith, "Achromatic Holographic Lithography in the Deep UV, " J. Vac . Sci .
Technol . B6 , 216 , ( 1988 ) ) . As a result, the source need not have the high degree of temporal coherence (i.e., narrow bandwidth) or spatial coherence commonly seen in laser sources. The minimum period, p, (i.e., center-to-center distance between adjacent lines) obtainable in holographic lithography is given by
p = λ/2 sin θ , (1) where λ is the wavelength of the exposing radiation and is half the angle between the intersecting beams. It is relatively easy to make this angle as large as 62 degrees, in which case p = 0.57 λ. The limit is, of course, p = λ/2.
Holographic lithography has been known for many years (H. I. Smith, "Fabrication Technigues for Surface-Acoustic wave and Thin-Film Optical Devices," IEEE Proc 62. 1361-1387 (1974)). It is used commercially to
produce large-area periodic diffraction gratings for spectroscopy. Prior to the advent of the laser, which made holographic lithography more practical, large area gratings were made by mechanical ruling engines, which are expensive, costly to operate, and produce well-known systematic errors, called ghosts.
Holographic lithography is also used in research, in areas ranging from distributed feedback lasers, and
quantum-effect devices, to x-ray imaging and spectroscopy. Competitive means of creating periodic and quasi-periodic patterns in a resist, such as scanning-electron-beam lithography, suffer from two main
shortcomings: extremely long writing times and small field-of-view. It is generally agreed that in those applications where the area of the grating exceeds about one square millimeter, and the period is finer than about one half micrometer, holographic lithography is the preferred method of exposing resists.
If one overlaps two beams from a single coherent source, such as a laser, and if environmental vibrations are sufficiently low that the beams do not move relative to one another by more than about p/2, it is nearly guaranteed that one will be able record in a resist film a recognizable diffraction grating. However, to achieve reliable and repeatable results in holographic
lithography, and to do so over large area, has proven to be an extremely difficult task, and for many
investigators an elusive goal.
An important object of this invention is to provide a novel means of conducting holographic
lithography in which the stability in space and time of the intensity distribution at the recording place is ensured by means of a feedback loop. This invention greatly enhances the contrast in the exposing intensity distribution, as well as the reproducibility of
holographic lithography. It also enables one to increase the area of exposure.
This invention results from the recognition that the quality of the "image" (i.e., the pattern of fringes and the depth of modulation) in holographic lithography is critically dependent on the phase relationship between the two (or more) beams that form the image. In this
invention, there is means for directly detecting and stabilizing this phase relationship.
According to the invention, a beam-splitting optical element is above and behind a plate used to hold substrates during holographic lithography. This beam splitter intercepts portions of two beams used in
holographic lithography, and produces at least two interference patterns whose variations reflect variations in the relative phase difference between the two
overlapping beams. The at least two interference
patterns are monitored by two photodetectors, and the difference in the output signals from the two detectors is used in a feedback loop to adjust the potential on a pockel-cell phase shifter so as to minimize this
difference.
Numerous other features and advantages of the invention will become apparent from the following
description when read in connection with the accompanying drawings in which:
FIG. 1 is a pictorial representation of a holographic lithography system;
FIG. 2 is a diagrammatic representation of an exposure station;
FIG. 3 is a top view of an assembly according to the invention including elements in FIG. 2;
FIG. 4 is a pictorial representation illustrating the function of the beam-splitting optical flat;
FIG. 5A is a pictorial representation of
overlapping beams illustrating the principle of
holographic lithography;
FIGS. 5B and 5C illustrate the intensity as a function of distance of high contrast and low-contrast images, respectively, produced by the interference pattern;
FIG. 6 is a diagrammatic representation of structure for strongly attenuating reflected beams;
FIGS. 7A and 7B show relief structures for high and low contrast images, respectively;
FIG. 8 is an electronmicrograph of a grating pattern in resist obtained by holographic lithography without the feedback loop in operation;
FIG. 9 is an electron micrograph of a grating pattern in resist obtained by holographic lithography with the feedback loop activated according to the
invention.
FIG. 10 illustrates a process for obtaining a relief grating of higher aspect ratio; and
FIGS. 11A and 11B are partial pictorial
perspective and front views, respectively, of structure in FIGS. 2-4 helpful in understanding the relationship of the system elements.
With reference now to the drawing and more particularly FIG. 1 thereof, there is shown a pictorial representation of a holographic lithography system.
A laser 11, such as an argon ion laser operating at the UV wavelength 351 nm, emits a beam which strikes shutter 12. Shutter 12 can be opened electronically to commence the operation of holographic lithography by directing the beam to the remainder of the apparatus, comprising an interferometer. It is typically fixed to a table of low vibration amplitude.
An advantage of this invention, because it involves active stabilization of the phase difference between interfering beams, is that it relaxes the
constraints on the stability of the table to be
achievable in typical lab or shop environments with a low cost table.
The interferometer comprises the beam splitter 14 which creates two beams, a variable attenuator 15 for
adjusting the irradiance in one of the two beams until both match, and mirrors 16a, 16b and 16c for redirecting the beams onto the exposure station 17. In some cases, mirror 16c may be omitted. Spatial filter 18 includes a lens which focuses the laser beam onto a plane that has a single pinhole. The pinhole is chosen to be equal to the diffraction-limited focal spot of the lens. In this way, essentially pure spherical waves emanate from spatial filter 18. A collimating lens 20 (shown in dotted outline) may be placed downstream from spatial filter 18 to yield a parallel beam. However, the distance between spatial filter 18 and exposure station 17 may be
sufficiently great so that lens 20 is unnecessary. The pinhole of spatial filter 18 is adjusted in position by means of a commercially available piezoelectric displacer attached to a micrometer drive (not shown) to allow for remote adjustment. Phase shifter 19 operates on the principle of a Pockell cell: application of a voltage across the cell causes a change in index of refraction and hence shifts the phase of the optical beam that is traversing the cell.
Exposure station 17 is shown in greater detail in FIG. 2. FIG. 2 shows an arrangement of photocells 22 built into the vertical plate 21 of the exposure station. Photocells 22 are used for centering the two interfering beams on vertical plate 21. Rotary table 23 can be swung so that photocells 22 face either the right or the left beam. After such centering is completed, the substrate to be exposed is held in front of the array of photocells 22 by means of a low pressure vacuum holder or "chuck" (not shown).
FIG. 3 is a top view of the vertical plate 21 showing substrate 24 held onto the plate 21, a cover 25 , which prevents exposure of substrate 24 during adjustment of the photodiodes 26 and the feedback loop. A beam-
splitting optical flat 27 is positioned on top of
vertical plate 21. In place of the optical flat one could instead use a beam splitting cube as item 27.
The function of the beam-splitting optical flat 27 is illustrated in FIG. 4. The beam 31 coming in from the left is partially reflected from the flat 27 and
partially transmitted (shown as dotted lines.) The beam 32 from the right is also partially reflected and
partially transmitted. The interference patterns due to overlap of transmitted and reflected beams are sketched below the left and right photodiodes 26. These sketches show left and right "bulls-eye" patterns 33 and 34, respectively, with a dark center (left) and a bright center (right), respectively. The character of the bulls-eye pattern depends on the phase difference between left and right beams 31 and 32 and can be adjusted by applying a voltage to phase shifter 19. The photodiodes 26 are typically centered on the two bulls-eye patterns. The bulls-eye pattern is characteristic of the
interference of two spherical waves of large but slightly different radii. If the left and right beams are not spherical waves, but, for example, plane waves, the interference pattern would consist of parallel fringes and can be processed as well accordingly to the
invention. The relationship of system elements is helpfully illustrated in FIGS. 11A and 11B showing pictorial perspective and front views respectively of certain elements described above.
In the absence of feedback, the bulls-eye interference patterns 33 and 34 oscillate rapidly due to relative phase shifts between left and right beams.
These oscillations produce brightening and darkening of the central spot and the successive rings of the bulls-eye pattern. For example, a relative phase shift of radians between left and right beams causes the central
spot to change from maximum darkness to maximum
brightness. A relative phase shift of 2 radians will further shift the central spot of the bulls-eye pattern back to maximum darkness. The relative phase shift between left and right beams can arise from any of several environmental factors: vibration of the
interferometer table, the mirror or other components; changes in temperature; changes in the density or
pressure of the air through which the beams propagate. These environmental variations are typically radom and uncontrolled. Thus, changes in the interference patterns at the locations of the two photodiodes 26 reflect changes in the relative phases of the left and right interfering beams at the location of the substrate.
FIG. 5 depicts structure helpful in understanding a basic principle of holographic lithography. Beam 31' coming from the left overlaps with beam 32' coming from the right. In the region of overlap 35, there is an optical standing wave whose spatial period is given by equation (1) above. The vertical hatching depicts this standing wave, which consists of sinusoidally alternating dark and light fringes (i.e., regions of high and low irradiance), as plotted in the graphs of FIGS. 5B and 5C showing high-contrast and low-contrast interference patterns or holographic images, respectively. The interference pattern is recorded in the photosensitive film or resist 36.
The left and right incident beams 31' and 32' are partially reflected from substrate 37, and the
interference of these reflected beams can lead to an orthogonal standing wave, as discussed in the paper by N. N. Efremow, N. P. Economou, K. Bezjian, S. S. Dana, and H. I. Smith, "A Simple Technique for Modifying the
Profile of Resist Exposed by Holographic Lithography, "J.
Vac. Sci. Tech. 19, 1234 (1981). It is generally
desirable to strongly attenuate these reflected beams.
To accomplish this attenuation reference is made to the configuration shown in FIG. 6. The antireflection coating 38 is a commercial product, a film which contains a dye that strongly attenuates the beams 31' and 32' as they pass through it. In this way, the resist 36 records primarily the interference due to the incoming beams 31' and 32', and only a very small remnant of the beams reflected from the substrate 37. XL, a commercial product of Brewer Scientific has been used as the ARC 38, and Microposit 1400-17, as resist 36.
The proper exposure of the resist with a high contrast holographic image, and the subsequent
development, yields a relief structure such as depicted in FIG. 7A. If the image has low contrast, a relief structure such as depicted in FIG. 7B can result. An important function of the present invention is to
reliably provide a high contrast image. The contrast of the image can be degraded by many factors, including:
(1) imbalance in the power density in the left and right beams; (2) a variations, over the course of the exposure, of the relative phase of the left and right beams. The former problem is easily rectified. An object of this invention is to solve the second problem in a manner that is widely applicable. Stabilizing the relative phase of the two beams greatly improves the image contrast in holographic lithography. This stabilization of the relative phase occurs according to the invention by monitoring the signals from photodiodes 26, and using these signals to control the voltage applied to the phase shifter 19 in FIG.1. This is accomplished by a feedback loop circuit, a method well-known to anyone familiar with control electronics. The essential purpose of this circuit is to apply a voltage to phase shifter 19 such
that the bulls-eye interference pattern remains
unchanged. This, in turn, guarantees that the relative phase difference between beams 31 and 32 is unchanged.
FIG. 8 is an electron micrograph of a grating pattern in resist obtained by holographic lithography without the feedback loop in operation, and FIG. 9 shows the result of activating the feedback loop.
Once the relief grating is obtained in the resist, several processes can be brought to bear to obtain a relief grating of higher aspect ratio. One such process is shown in FIG. 10. The relief structure in resist is coated at an oblique angle (so-called shadowing) with titanium, SiO or other suitable material. The substrate is then placed in a reactive-ion-etching system
(commercially available) supplied with oxygen gas.
Activation of the reactive-ion-etching leads to rapid etching of the polymeric ARC 38, below those areas not protected by the Ti or SiO.
There is attached as Appendix A Decription of Technology helpful in further understanding the
invention.
Other embodiments are within the claims.
APPENDIX A
1 Description of Technology
1.1 Introduction
A simple and effective feedback system has been developed to stabilize the interference pattern used in producing holographic gratings. Gratings have a very large number of scientific and engineering applications (see [1] for example) and can be produced by a variety of different techniques. One of the most effective techniques is to expose a photosensitive material to the interference pattern of two optical fields. The photosensitive material changes in some useful way when exposed to light and then, perhaps, further processed. "Holographic'' gratings are produced by taking advantage of the change in the photosensitive material either directly or as part of a multistep process to make a grating structure.
In order to successfully fabricate holographic gratings the interference pattern must be stable in space to within a small fraction of the grating period during the entire time required to expose the photosensitive material. This requirement becomes more and more difficult to achieve as the grating period becomes smaller, as the exposure time becomes longer, and when the optical system must be in a less than ideal environment such as a clean room or production area where noise and vibration levels are high. If the interference pattern cannot be held steady because of vibration, noise, thermal drift, and other instabilities the contrast recorded in the photosensitive material decreases quickly. This makes the fabrication of useful gratings difficult and unreliable and, when the contrast is small enough, impossible.
The purpose of the invention is to counteract the unwanted fringe motion with a simple feedback system so that the resulting interference pattern is stable in space even when large amplitude vibration, thermal, and other instabilities are present. With the stable fringe pattern resulting from active feedback the grating fabrication process is made more reliable and less sensitive to small process variations such as exposure dose and development time. Further, the feedback system makes possible holographic grating exposure in unfavorable environments such as clean rooms where large amounts of noise and vibration exist.
The invention is a. particular implementation of the fringe stabiliza-
tion feedback concept in which the method of detecting the motion of the fringes is different from previous holographic stabilization systems. The new method of detecting the fringe motion uses the fact that a grating is being exposed and not a general hologram. Because of this, a simple and easy to adjust fringe motion detector can be used, resulting in several advantages.
In summary, a feedback system, using a new and simple fringe motion detection method is described which canceles large amplitude vibration, noise, and thermal drift in the optical interference pattern used for the production of holographic gratings. This system results in reliable, large process latitude, holographic grating fabrication even in high noise environments such as clean rooms.
1.2 Technical Description
The feedback system consists of four elements: an electro-optic crystal (Pockels cell), high-voltage amplifier to drive the crystal, a beamsplitter, and a pair of photodiodes. Figure 1 shows the basic configuration. The beamsplitter is placed above and behind the substrate (to be out of the way but still within the overlap of the two beams) forming a Mach-Zehnder interferometer. In the Mach-Zehnder interferometer, one beam is transmitted by the beamsplitter and interferes with the other beam that is reflected by the beamsplitter. For example, in. Fig. 1 the interference intensity measured by diode "1" is proportional to the magnitude squared of the sum of the amplitude of beam "a" reflected by the beamsplitter and beam "b" transmitted by the beamsplitter. A complementary pattern (i.e. at diode "2") is formed from the interference of the two beams with the roles of transmitted and reflected beams reversed. A pair of photodiodes is used to measure the intensity signal of both interference patterns. By power conservation, the sum of the signals measured in each arm is proportional to the laser output power, which is approximately constant. The diodes are connected to a differential amplifier which subtracts one diode signal from the other. With this configuration, a signal proportional to the sine of the phase difference between the two interferometer arms is produced and the common mode signals such as the room lights are effectively canceled. The output of this differential amplifier modulates a high-voltage power
Figure 1: This shows the holographic configuration with feedback stabilization. A beamsplitter is placed in the overlap of the two coherent beams but out of the way of the substrate so that no spurious scattering onto the substrate will occur. Two photodiodes measure the intensity of the large width fringes (not to be confused with the fine period fringes formed at the substrate) formed on both sides of the beamsplitter. The difference of these two intensities is proportional to the sine of the path phase difference between the two arms. This signal is amplified and used to drive the Pockels cell in one arm. The Pockels cell introduces a phase shift (proportional to the applied voltage) that almost completely cancels the noise introduced by mechanical vibration and other disturbances.
supply which in turn drives an ADP-crystal Pockels cell. The Pockels cell modulates the phase of one of the interferometer arms thereby closing the feedback loop. The high gain of the feedback loop keeps the difference between the two photodiode signals very small resulting in a stable fringe pattern. An advantage of using two diodes and taking the difference is that the fringe pattern is largely independent of the ambient light level and the laser output power. A change in the laser output power will only produce a change in the loop gain and the fringe position is insensitive to the loop gain if it is large enough. 1.2.1 Details of Feedback Theory
The difference of the signals from the two photodiodes is proportional to the sine of the phase difference in the path lengths of the two interferometer arms. This phase difference,øTotal, as a function of time consists of two terms: a term representing the phase of the Pockels cell and a phase term due to the noise (vibration, air currents, etc) of the system.
øTotal = øNoise + øPockels (l)
Let Va be the voltage of the amplifier electronics when one diode is dark and the other one is light, the feedback equation becomes:
where V∏ is the ir phase shifting voltage for the Pockels cell. If the feedback system is working properly then the Pockels cell phase, øPockels, will be almost equal and opposite to, the noise phase disturbance, øNoise. Under these conditions the sine function can be expanded in a Taylor series. Keeping only the first order terms we find:
Solving this equation for the total phase difference gives
For a moderate amount of loop gain, πVa/V∏, the total noise term can be substantially reduced.
1.2.2 Experimental Configuration
Figure 1 shows the experimental configuration. A conventional holographic lithography configuration is modified by placing an ADP Pockels cell phase modulator in one arm. A beamsplitter is placed above and behind the substrate to be out of the way and thus not scatter light onto the substrate but still be in the overlap of the two beams. The beamsplitter forms a MachZehnder interferometer whose output intensity is monitored by two photo-diodes. The photodiodes are connected to a differential current amplifier to produce a signal proportional to the sine of the phase difference between the two arms. This signal is fed into a high-voltage audio-frequency-bandwidth amplifier which drives the Pockels. cell, closing the feedback loop.
Figure 2 shows the oscilloscope traces when the fringes are locked and not locked for an interferometer configuration that uses the entire 8 foot length of the optical table set up to produce 200nm period gratings. The noise in the fringes without feedback is effectively larger than the grating period resulting in contrast much too small for useful grating exposure. With the fringes locked, good gratings can be generated. Figures 3 and 4 show the difference between exposed grating profiles with and without feed¬back, respectively. 1.3 Advantages and Improvements
1.3.1 Existing methods
The use of feedback to stabilize holographic fringes, of which a grating is a special case, goes back to Naumann and Rose in 1967 [2]. In this work they describe a feedback control system that uses a piezoelectric crystal translator, attached to a mirror, as the phase modulator and a method to detect the motion of the fringes consisting of a lens to magnify the fringes onto a slit connected to a photomultiplier tube (PMT). The PMT translates the optical intensity into an electrical signal which is then used to drive the piezoelectric translator. The next advance in holographic feedback stabilization was to use a previously recorded almost-identical hologram to detect the fringe motion. This scheme was anticipated by MacQuigg in 1974 [3] and proposed by Johansson et al. [4] in 1976. This reference hologram is placed near the hologram to be exposed and produces a moire
Figure 2: The top oscilloscope trace, shows a signal proportional to the sine of the phase noise, without the feedback loop, for a interferometer configuration that uses the entire 4 by 8 foot area of an optics table in a clean room environment. The noise is much larger than the period of the grating to be exposed and therefore the contrast is well below a usable level. The bottom trace shows the signal with a loop gain of about 15. The residual noise is essentially zero, resulting in high contrast fringes that can be used to expose a large area grating and allow a large process latitude.
Figure 3: An SEM micrograph of exposed photoresist profiles using feedback. The feedback greatly increases the fringe contrast.
Figure 4: An SEM micrograph of exposed photoresist without using feedback. The contrast is so poor that there is no hope of producing a good grating from this exposure.
like interference pattern. This moire pattern results from the interference between an incoming wave and its holographic reconstruction. A shift in this pattern corresponds to a proportional shift in the interference pattern. Therefore a shift from light to dark to light again represents a full 2∏ phase shift in one of the arms. This innovation greatly increases the signal level and therefore the signal to noise ratio since not just one fringe, as used by Rose et al., but approximately 103 to 104 fringes contribute to the detected signal. In 1977 MacQuigg contributed another innovation by introducing a small "dither" signal which allows the use of lock-in amplifier technology to measure the error signal [5]. A lock-in amplifier can recover a small signal with a definite frequency, fo, (i.e. the dither frequency) even in the presence of a large noise component. Recent papers on holographic fringe stabilization for applications such as Fourier synthesis of blazed gratings include [6] and [7]. 1.3.2 Advantages of the new system
A serious shortcoming of the previously developed schemes is that a good quality, almost-identical hologram must first be produced. But this is precisely the problem that we are trying to solve with feedback control. Under conditions where the grating period is very small and there is considerable environmental noise disturbing the system it is not possible to make a good quality hologram (i.e. grating). The new phase detector scheme that we propose eliminates this problem for the special case of a grating. The new scheme may not work for a general hologram but works very well for a grating.
The new scheme uses an ordinary beamsplitter instead of an almostidentical hologram (i.e. grating) placed near the substrate to be exposed and in the overlap of the two interfering beams (but out of the way so that light is not scattered from the beamsplitter or its holder onto the substrate). Further, the beamsplitter is usually a very efficient optical element so that almost all the light incident on it is converted into a useful signal by the photodiodes. Because of this the signal detected by the photodiodes is sufficiently strong so that a lock-in amplifier is not needed to separate the signal from the noise and, further, wide bandwidth operation is possible. When two photodiodes are used in a differential mode, as in our system,
the "common mode" signals such as the room lights and variations in the laser power are canceled out and do not affect the fringe position. Therefore, the significant advantages of this system over prior systems is that a good quality grating does not need to be fabricated in order for the feedback system to work and that the feedback signal level is very strong and free of common mode interference simplifying the electronics and allowing feedback compensation over wide bandwidths.
1.4 Coπimercial Applications
Gratings have a wide range of scientific, and engineering applications. For many of these applications, with relatively large periods, active feedback of the holographic exposure system is not needed. However, there are important applications, in general where the period is small, where feedback is critical. For example, the permeable base transistor (PBT) is a high speed microwave power transistor with considerable economic appeal. One of the critical steps in the fabrication of this device is producing the fine period grating which is used as the "base." For this step reproducible and high contrast grating exposures are needed. Another application is in the fabrication of distributed feedback lasers. These solid state lasers use a grating structure to provide the feedback. For both of these applications, the final device, (i.e. a permeable base transistor or a distributed feedback laser), sells for several hundred to, thousands of dollars. There is a large market demand for DFB lasers in the telecommunications industry and PBTs in the microwave industry.
In fact, Carl Bozler from Lincoln Lab, whose project is to commercialize the PBT technology, has visited our lab and we have done experiments together. These experiments show that the feedback system on our holographic set up can make a substantial improvement in the quality and reproducibility of his process. Another visitor to our lab, Kiyoshi Fujiϊ, from NEC corporation in Japan mentioned to me that his group had tried to make DFB lasers using holographic techniques and had given up because the process was too unreliable. I believe that if they used the feedback scheme described here that they would have been successful. These two events convinced me that there should be significant economic interest in this technology to warrant patenting and licensing it.
References
[1] T.K. Gaylord and M.G. Moharam "Analysis and Applications of Optical Diffraction by Gratings" IEEE Proc, vol 73, pp. 894-937, May 1985.
[2] D.B. Naumann and H. W. Rose, "Improvement of Recorded Holographic Fringes by Feedback Control," Appl. Opt. vol 6, pp. 1097-1104, June 1967.
[3] D.R. MacQuigg, "The Modulated Grating Hologram," Proc. Soc.
Photo-Opt. Instrum. Eng.-vol 48, pp. 96-100, 1974.
[4] S. Johansson, L.-E. Nilsson, K. Biedermann, and K. Kleveby, "Holographic Diffraction Gratings with Asymmetric Groove Profiles," in Proceedings, ICO Jerusalem 1976 Conference on Holography and Optical Data Processing Pergamon, New York, pp. 521, 1976.
[5] D.R. MacQuigg, "Hologram fringe stabilization method," Applied Optics, vol 16, pp. 291-292, Feb. 1977.
[6] Lars-Erik Nilsson and Hans Ahlén, "Stabilization of the exposing interference pattern in holographic grating production," SPIB vol. 240 Periodic Structures, Gratings, Moiré Patterns and Diffraction Phenomena, pp. 22-26, 1980.
[7] Jaime Frejlich, Lucila Cescato, and Geraldo F. Mendes, "Analysis of an active stabilization system for a holographic setup," Applied Optics, vol. 27, pp. 1967-1976, May 1988.