1. Field of the Invention
The present invention relates to laser thermal processing, and in particular relates to apparatus and methods for performing laser thermal processing with laser diode radiation.
2. Description of the Prior Art
Laser thermal processing (“LTP”) (also referred to as “laser thermal annealing”) is a technique used to anneal and/or activate dopants of source, drain or gate regions of integrated devices or circuits, to form silicide regions in integrated devices or circuits, to lower contact resistances of metal wiring coupled thereto, or to trigger a chemical reaction to either deposit or remove substances from a substrate.
Various devices for performing LTP of a semiconductor substrate are known and used in the integrated circuit (IC) fabrication industry. LTP junction annealing is preferably done in a single cycle that brings the temperature of the material being annealed up to the annealing temperature and back down in a single cycle. If a pulsed laser is used, this requires enough energy per pulse to bring the surface of the entire chip or circuit up to the annealing temperature. Because the required field size can exceed four (4) centimeters-squared (cm2) and the required dose can exceed one (1.0) Joules/cm2, a relatively large, expensive laser is required. It is also difficult to achieve good dose uniformity over a relatively large area in a single pulse because the narrow spectral range of most lasers produces a speckled pattern due to interference effects.
Laser diode bars are well-suited to serve as a source of radiation for performing LTP because their wavelengths of 780 nm or 810 nm are readily absorbed in the top layer (i.e., ˜21 microns) of silicon. Diode bars are also efficient converters of electricity to radiation (˜45%) and emit a variety of wavelengths that may be scrambled to provide uniform energy coverage over an extended field size.
U.S. Pat. No. 6,531,681 (the '681 patent) describes how a linear laser diode array, or several linear diode arrays, can be used to form a uniform, narrow line image that can be scanned across a substrate to thermally anneal integrated circuits thereon. The '681 patent also describes how the line image can be placed on a mask and imaged through a projection system to process selected areas of a substrate scanned in synchronism with the mask. However, performing laser thermal processing with a linear array of laser diode bars as described in the '681 patent is problematic. Applications involving silicon substrates have system requirements (i.e., line image width and dwell time) that require relatively high energy densities (e.g. in the range of 1300 W/mm2 for a 200 μs dwell time).
U.S. patent application Ser. No. 10/287,864 describes the use of a P-polarized CO2 laser beam incident at near Brewster's angle to perform LTP of a silicon substrate with integrated circuits formed thereon. As described therein, the use of incident angles at or near Brewster's angle produces very uniform heating of substrates that are otherwise spectrally non-uniform at normal incidence. For example, at normal incidence and at 10.6 microns bare silicon has a reflectivity greater than 30% and silicon oxide has a reflectivity of less than 14%. One benefit of using a CO2 laser when performing LTP is its ability to deliver a well-collimated beam having relatively high energy density. Another benefit is that the 10.6 μm wavelength emitted by the CO2 laser is large compared to the various film thicknesses likely to be found on a wafer ready for the annealing step. Small variations in film thickness therefore do not result in large variations in reflectivity as would be the case for a shorter annealing wavelength.
However, the CO2 laser wavelength of 10.6 μm is best suited for annealing heavily doped silicon substrates, which can absorb sufficient radiation in the top 50 to 100 μm of material. However, for annealing lightly doped substrates or substrates that are doped only in a shallow layer near the top surface, the CO2 laser radiation passes right through with very little of the incident energy resulting in useful heating.
Laser diodes, on the other hand, emit radiation at wavelengths of 780 nm or 810 nm. These wavelengths are readily absorbed in the top 10 to 20 μm of a silicon wafer independent of the doping level. With laser diodes operating at the short time scales (i.e., 100 μs to 20 ms) associated with LTP, the heating depth is determined by thermal diffusion rather than by the radiation absorption depth (length).
It would therefore be useful to have systems and methods for performing laser thermal annealing at or near the Brewster's angle with polarized laser diode radiation delivered at relatively high energy densities.
A first aspect of the invention is a system for performing laser thermal processing (LTP) of a substrate having a Brewster's angle for a select wavelength of radiation. The system includes a two-dimensional array of laser diodes adapted to emit polarized radiation at the select wavelength. The system also includes an LTP optical system having an image plane and arranged to receive the emitted radiation and form an original (first) image at the substrate. The radiation beam is P-polarized and is incident the substrate at an incident angle that is at or near the Brewster's angle. The system further includes at least one recycling optical systems arranged to receive radiation reflected from the substrate and direct the reflected radiation back to the substrate as corresponding at least one recycled radiation beams.
A second aspect of the invention is a method of performing laser thermal processing (LTP) of a substrate. The method includes emitting radiation of the select wavelength from a two-dimensional array of laser diodes. The method also includes receiving the emitted radiation with an LTP optical system and forming therefrom a linearly P-polarized radiation beam that forms an image (e.g., a line image) at the substrate. The method also includes irradiating the substrate with the radiation beam at a first incident angle corresponding to a minimum substrate reflectively for the select wavelength, while scanning the image over at least a portion of the substrate. The method further includes directing radiation reflected from the substrate back to the substrate as a recycled radiation beam during scanning, while preserving the P-polarization of the radiation.
The various elements depicted in the drawings are merely representational and are not necessarily drawn to scale. Certain proportions thereof may be exaggerated, while others may be minimized. The drawings are intended to illustrate various implementations of the invention, which can be understood and appropriately carried out by those of ordinary skill in the art.
The apparatus of the present invention is first described, followed by its methods of operation. The power density requirements and system throughput capabilities are then set forth.
Apparatus
Laser Diode Array
Laser diode array 12 includes a plurality of laser diodes 18 positioned at regularly spaced intervals along a two-dimensional emission face 20 of the array. In an example embodiment, laser diode array 12 is formed by combining (e.g., “stacking”) linear diode arrays that make up rows or columns of the array.
A typical commercially available laser diode array bar (i.e., linear diode array) is a stack of one (1) centimeter linear arrays each containing 60 emitters and spaced about 160 μm apart along the length of the array. Each emitter is about 1 μm wide and about 150 μm long. The orientation of the emitter is such that the largest dimension of the emitter is aligned with the length of the array. The laser diodes 18 typically emit radiation 14 that diverges 10° in a plane defined herein as the Y-Z plane and containing the axis of the individual linear arrays. Further, radiation beam 14 diverges by an amount (e.g., 30°) in a plane orthogonal to the axis of the individual linear diode arrays (defined herein as the XZ plane).
Suitable laser diode array bars are commercially-available from numerous suppliers, including SDL, 80 Rose Orchard Way, San lose, Calif. 95134-1365 (e.g., the SDL 3400 series includes linear arrays 1 cm long and capable of 40 Watts (W) output power), Star Technologies, Inc. of Pleasanton, Calif., Spire, Inc. of One Patriots Park, Bedford, Mass. 01730-2396, Siemens Microelectronics, Inc., Optoelectronics Division, of Cupertino, Calif. (Model SPL BG81), Spectra Diode Labs, Thompson CFS of 7 Rue du Bois Chaland, CE2901 Lisses, 91029 Evry Cedex, France, and IMC, 20 Point West Boulevard, St. Charles, Mo. 63301.
Because the heat generated in the operation of the laser diodes 18 can be substantial and limits the maximum available output power, the laser diode array bars are typically water-cooled to prevent overheating during use.
In a specific example embodiment, laser diode array 12 is made up of 25 rows of laser diodes 18, with each row separated by 1.9 mm and containing 49 laser diodes 18 each measuring 100 μm along the Y-axis and 1 μm along the X-axis (i.e., along the cross-row direction). Each laser diode row is 10mm long and the laser diode array 12 is 24×1.9 mm=45.6 mm wide. The radiation emitted from each laser diode in the Y-Z plane diverges 10° full-width half-max (FWHM) and by 35° FWHM in the X-Z plane. A suitable two-dimensional laser diode array 12 is available from Coherent, Inc, in the line of LightStone™ products (e.g., the diode array sold under the tradename LIGHTSTACK).
In example embodiments, laser diode array 12 generates radiation 14 at a wavelength in the range from about 350 nanometers (nm) to 950 nm, and in a particular example embodiment at 780 nm or at 810 nm. Such wavelengths are particularly effective for processing a silicon substrate having integrated devices or circuit features on the order of one micron or less with source/drain regions of a few tens of nanometers (nm) in thickness.
It is noted here that the present invention is not limited to a laser diode array 12 generating radiation only within the above-stated wavelength range. Commercially available laser diodes emit radiation at wavelengths extending from 380 nm (e.g., GaN blue diodes) through 931 nm. The wavelengths and types of laser diode arrays commercially-available on the market have rapidly expanded, and this trend will likely continue so that numerous additional arrays both in and out of the above-stated wavelength range are expected to become available from manufacturers in the future. Arrays of such future laser diodes may be useful for implementation in the subject invention, particularly those that emit wavelengths absorbed by silicon. Some commercially-available laser diode array bars are capable of generating radiation 14 at a relatively intense power level of 50 W to 100 W in a 1 cm long bar containing a single row of diodes.
In an example embodiment, laser diode array 12 generates radiation having a power density of 150 W/mm2 or greater as measured at the substrate.
LTP Optical System
With continuing reference to
Radiation beam 23 is P polarized and is incident on substrate 16 at an angle at or near the Brewster's angle θB (in the
In an example embodiment of the present invention, the incident angle of radiation beam 23 is within ±10° of the Brewster's angle for the material of the substrate being processed (e.g., silicon). In another example embodiment, the incident angle is between 60° and 80°.
The use of incidence angles near Brewster's angle produces uniform heating on substrates that are spectrally non-uniform at normal incidence because of the uneven distribution of circuit elements containing different films having different spectral characteristics. For example, a given wafer can have one region that is predominately bare crystalline silicon, and another region that is predominately covered with isolation trenches filled with SiO2 to a depth of 0.5 μm. A third region may have areas containing a 0.1 μm film of poly-silicon on top of an oxide trench in silicon. The reflectivity of each of these regions varies with the angle of incidence as measured relative to surface normal N ( see
Another advantage of operating in this angular range is that the reflectivities of all the films are very low in this region and therefore incident radiation beam 23 is coupled into substrate 16 very efficiently. At normal incidence, about 33% of the incident, 800 nm radiation beam is reflected from bare silicon, and about 3.4% is reflected from the surface of an infinitely thick SiO2 layer. At an incidence angle of 68°, only about 3% of the radiation is reflected from the bare silicon and from the top surface of the SiO2 layer. When interference effects from multiple surfaces are considered the result is more complicated, but the total variation in reflectivity from the various possible films is minimized when the P-polarized incident radiation beam 23 is incident at or near the Brewster's angle for silicon.
With reference first to
In an example embodiment, the focal length of each cylindrical lens element is relatively short, e.g., about 3 mm. The N collimated beams 110 (e.g., N=25) are equivalent to a single collimated output beam 112 of a given width (e.g., 47.5 mm). Theoretically the angular spread of the rays in the (substantially) collimated beam 112 could be very small (e.g., 0.024°) and limited only by the 1 μm size of the emitter or by diffraction. In practice, the diode rows wind up slightly bent resulting in a misalignment with the cylindrical lens elements 102. This limits the minimum divergence angle of output beam 112 (e.g., to about 0.3° FWHM).
An example of a suitable cylindrical lens array 100 is available from Limo Micro-Optics & Laser Systems, Bookenburgweg 4, 44319 Dortmund, Germany. The polarization direction of beams 110 is oriented such that the electric field vector is perpendicular to the row direction, i.e., the polarization is in the X-direction. In this case, and with the optical arrangement shown in
For the sake of description and ease of illustration, laser diode array 12, cylindrical lens array 100 and optional half-wave plate 120 are grouped together and considered herein to constitute an effective laser radiation source 140 that emits output beam 112 (
With reference again to
In this example, cylindrical collimating lens 204 and cylindrical focusing lens 228 form a telecentric, anamorphic relay with a reduction power (ratio) of about 2 which generally can vary between about 1.5 and about 4.5 in the Y-Z plane. Note that a reduction power of 2 corresponds to a magnification magnitude of ½. These cylindrical lenses contribute no power in the X-Z plane (
Normally, it would be desirable to have as large a reduction ratio as possible to concentrate the power in line image 24 formed at substrate 16. However the larger the reduction ratio, the larger the cone angle at the substrate and the larger the angular variation in the range of incidence angles in radiation beam 23 as seen by the substrate. For example, if laser diode array 12 was imaged 1:1 onto substrate 16, then the angular spread of the radiation leaving the laser diode array would be duplicated in the radiation beam at the substrate.
To keep the optical design relatively simple, and to limit the variation of incidence angles at substrate 16, it is desirable to limit the angular spread of radiation beam 23 at the substrate to about 20°, which corresponds to the aforementioned demagnification ratio of about 2 in the Y-Z plane. Thus, by way of example, a 10 mm long row of diodes is imaged into a line image 5 mm long.
With reference to
In the above example, the magnifications in the X-Z and Y-Z planes were determined by setting an upper limit of 20° to the cone angle in radiation beam 23 as seen by the substrate. However, there is no fundamental limit for the range of incidence angles, although a small range of angles can yield less variation in the energy absorbed across the wafer. If the beam collimation produced by the diode and cylindrical lens arrays had been tighter, then a higher magnification in the X-Z plane could have been used to obtain a narrower line image. Similarly, there is no fundamental reason why the numerical aperture of the laser beam on the substrate has to be identical in both planes. Thus, the reduction power in the Y-Z plane could have been, say between about 1.5× and about 4.5×, and the reduction power in the X-Z plane could have been, say between about 50× and about 150×. The reduction power in the X-Z direction depends on the angular spread in the radiation beams 112 after collimation by cylindrical lens array 100.
A close-up view of cylindrical focusing lens 228 and cylindrical relay lens group 230 forming line image 24 at substrate 16 as viewed in the Y-Z plane and the X-Z plane is shown in
The optical design data for an example embodiment of LTP optical system 22 as described above is set forth in Table 1, below. In the Table, the first column is the surface number, the second column is the surface radius, the third column is the distance to the next surface (thickness or spacing) and the fourth column identifies the lens material. The letter “S” stands for “surface number” S1, S2, etc., and TH stands for “thickness.” All the thickness and radius values are in millimeters (mm). An asterisk (*) indicates an aspheric surface for surfaces S3 and S10, and the aspheric surface data is provided separately below.
Wherein k is a toroidal aspheric constant defined by the equation:
z=cy2/(1+(1−(1+k)c2y2)0.5)
In an example embodiment, each row of diodes is capable of generating about 80 W of optical power with water cooling. Assuming an overall efficiency of 70%, the image power density (i.e., the intensity in image 24) is about:
Power=25(80 W)(0.7)/(1.62 mm)(5 mm)=173 W/mm2
This amount of power is significantly less than the 1300 W/mm2 (associated with a 200 μs dwell time) needed in the prior art LTP system of the '681 patent.
In an example embodiment, the intensity (power density) in line image 24 is 100 W/mm2 or greater.
Control system
With reference again to
In an example embodiment, control system 25 includes a reflected radiation monitor 39A and a temperature monitor 39B. Reflected radiation monitor 39A is arranged to receive radiation 23 reflected from substrate surface 16S. Reflected radiation is denoted by 23′. Temperature monitor 39B is arranged to measure the temperature of substrate surface 16S, and in an example embodiment is shown arranged along the surface normal N so as to view the substrate at normal incidence at or near where line image 24 is formed. However temperature monitor 39B could also be arranged to view the substrate at the Brewster's angle corresponding to the wavelength band used to measure temperature. Monitors 39A and 39B are coupled to controller 26 to provide for feedback control based on measurements of the amount of reflected radiation 23′ and/or the measured temperature of substrate surface 16S, as described in greater detail below.
In an example embodiment, controller 26 is a microprocessor coupled to a memory, or a microcontroller, programmable logic array (PLA), field-programmable logic array (FPLA), programmed array logic (PAL) or other control device (not shown). The controller 26 can operate in two modes of operation: open-loop, wherein it maintains a constant power on the substrate and a constant scan rate; and closed-loop, wherein it maintains a constant maximum temperature on the substrate surface or a constant power absorbed in the substrate. Since the maximum temperature varies directly as the applied power and inversely as the square root of the scan velocity, in an example embodiment a closed loop control is used to maintain a constant ratio of incident power divided by the square root of the scan velocity. I.e., if P23 is the amount of power in radiation beam 23 and V is the scan velocity, then the ratio P23/V1/2 is kept constant.
For closed loop operation, controller 26 receives at least one parameter via a signal (e.g., an electrical signal), such as the maximum substrate temperature (e.g., via signal 232 from temperature monitor 39B), the power P23 in radiation beam 23 (e.g., via signal 42 from detector 38), the reflected power in reflected radiation beam 23′ (e.g., via signal 230 from reflected radiation monitor 39A. Further, controller 26 is adapted to calculate parameters based on the received signals, such as the amount of power absorbed by wafer 16 as determined, for example, from the information in signals 230, 232 and/or 42.
The controller 26 is also coupled to receive an external signal 40 from an operator or from a master controller that is part of a larger substrate assembly or processing tool. This parameter is indicative of the predetermined dose of radiation to be supplied to process the substrate or the maximum temperature to be achieved the substrate. The parameter signal(s) can also be indicative of the intensity, scan velocity, scan speed, and/or number of scans to be used to deliver a predetermined dose of radiation to substrate 16.
Based on the parameter signal(s) received by controller 26, the controller can generate a display signal 46 and send it to display unit 30 to visually display information on the display unit so that a user can determine and verify the parameter signal level(s). The controller 26 is also coupled to receive a start signal that initiates processing performed by the apparatus 10. Such start signal can be signal 39 generated by input unit 28 or external signal 40 from an external unit (not shown), such as a master controller.
Method of Operation
The method of operation of LTP apparatus 10 is now described. With continuing reference to
In an example embodiment, controller 26 is preprogrammed to generate a scan control signal 206 based on the parameter signals indicative of the predetermined scan speed and number of scans. The controller 26 generates the scan control signal 206 in coordination with intensity control signal 200 and supplies the scan control signal to stage controller 34. Based on scan control signal 206 and a predetermined scan pattern preprogrammed into the stage controller, the stage controller generates a scan signal 210 to effect movement (e.g., raster, serpentine or boustrophedonic) of stage 36 so that line image 24 is scanned over the substrate 16 or select regions thereof.
In an example embodiment, detector 38 generates a detector signal 42 indicative of the amount of power in radiation beam 23 received at substrate 16, which is a function of the power level of radiation 14 from laser diode array 12 and the transmission of LTP optical system 22. In an example embodiment, controller 26 (or a user directly) determines the intensity control signal 200 and the scan speed. The maximum temperature produced on substrate 16 is approximately proportional to the radiation intensity I23 (i.e., P23/(unit area)) divided by the square root of the scan speed, i.e., I23/V1/2. Hence, in an example embodiment, controller 26 is preprogrammed to achieve a desired maximum temperature by varying either the scan rate, or the laser intensity, or both, to obtain a value of intensity divided by root scan velocity corresponding to the desired maximum temperature. In a further example embodiment, the desired maximum temperature is maintained constant during scanning.
In another example embodiment, an amount of reflected radiation 23′ is measured by reflected radiation monitor 39A, and provides a signal 230 corresponding to the measured power to controller 26. The proportion of radiation beam 23 absorbed by the substrate and the corresponding absorbed power level is then calculated using the incident radiation measurement (e.g., from detector 38) and the reflected radiation measurement. Signal 230 is then used by controller 26 to control the absorbed radiation beam 23 power level provided by laser diode array 12 to substrate 16 to ensure that the correct maximum temperature is maintained in the substrate.
In another example embodiment, substrate temperature monitor 39B measures the temperature of substrate surface 16S and provides a signal 232 to controller 26 that corresponds to the maximum substrate surface temperature. Signal 232 is then used by controller 26 to control the amount of radiation 23 provided by laser diode array 23 to the substrate to ensure that the correct maximum temperature is maintained in the substrate during scanning.
The method also includes scanning line image 24 over at least a portion of the substrate so that each scanned portion sees a pulse of laser diode radiation that takes the surface temperature of the silicon substrate 16 to just under (i.e. to within 400° C. or less) the melting point of silicon (1410° C.) for a period of between 100 μs and 20 ms.
Power Density Requirements for Silicon LTP
The absorbed power density required for annealing silicon substrates (wafers) varies with the “dwell time,” which is the amount of time line image 24 resides over a particular point on substrate surface 16S (
Assuming that a minimum power of 170 W/mm2 is required to perform LTP for silicon-based applications, a laser diode array 12 capable of producing such minimum power can perform LTP with dwell times on the order of 10 ms.
System Throughput
It is important to the commercial viability of an LTP system that it be able to process a sufficient number of substrates per unit time, or in the language of the industry, have a sufficient “throughput.” To estimate the throughput for LTP apparatus 10, consider a 300 mm silicon wafer and a line image 5 mm long and 1.62 mm wide. The number of scans over the wafer is given by 300 mm/5 mm=60. Further, for a dwell time of 10 ms, the scan speed is 162 mm/s. The time for one scan is given by (300 mm)/(162 mm/s)=1.85 s. For a stage acceleration rate of 1 g, the acceleration/deceleration time of the stage is (162 mm/s)/(9800 mm/s)=0.017 s. Thus, the time to process one substrate is 60(1.85 s+(2)(0.017 s))=113 s. If the time to input and output a substrate to and from the apparatus is 15 seconds total, then the throughput is given by (3600 s/hr)/(15 s+113 s)=28 substrates/hour, which is a commercially viable throughput value.
Recycling Reflected Radiation
While it is preferable to irradiate substrate 16 with annealing radiation beam (“radiation”) 23 at an incident angle θ that minimizes reflection of this radiation beam, this is not always convenient or possible. This is because the reflectivity of substrate 16 depends on the nature of surface 16S, which can have an uneven distribution of a variety of thin films and other structures residing thereon.
These structures range from bare silicon in the junction areas, to field oxide, to polysilicon on field oxide. It has been estimated a typical integrated circuit comprises 30% to 50% field oxide, about 15% to 20% bare silicon or polysilicon on silicon, and the rest is polysilicon on field oxide. However these proportions vary from circuit to circuit and even across a circuit.
With reference briefly to
In an example embodiment, if a 20° range of incident angles is considered in the plane of incidence, then the plot of
In practice it is difficult to eliminate the reflection of radiation 23 from substrate surface 16S. Thus, an example embodiment of the present invention involves capturing reflected radiation 23R and redirecting it back toward the substrate as “recycled radiation 23RD, where it can be absorbed by the substrate in order to contribute to the annealing process by further heating the substrate.
There are two main reasons for recycling the reflected energy. One is simply that it improves the efficiency with which energy is coupled into the substrate thus reducing the maximum laser power required and therefore the cost. The second, and even more important reason, is that variations in reflectivity from point-to-point on the wafer lead to variations in the absorbed power and therefore result in an undesirable variation in temperature. Thus, if the resolution of the recycling system can be made high enough, an appreciable improvement in temperature uniformity can be expected.
The required resolution is smaller than the thermal diffusion length (δ) given by:
δ=(DT)0.5 (1)
where D is the thermal diffusivity (0.9 cm2/sec for silicon) and T is the dwell time of the line image over a point on the substrate.
A typical dwell time of one millisecond would yield a thermal diffusion length of about 300 microns, so that a recycling system with 100 microns resolution would provide a substantial improvement in temperature uniformity.
The required numerical aperture (NA) of the recycling system has to match, as a minimum, the numerical aperture of the directly incident beam. Since the patterns on the wafer have a finite contrast, even under illumination conditions designed to minimize this contrast, it is desirable that the recycling system NA be somewhat larger.
Accordingly, with reference now to
Ideally, the recycling optical system 300 needs to reimage the line image 24 back on itself at the same scale and with the same orientation as the original (first) line image. There are a number of simple arrangements that will accomplish this. Two such examples are a lens separated from the object by its focal length followed by a corner-cube reflector, and a relay system that images the object on a plane mirror.
In operation, the polarizer 302 is aligned to the linear polarization direction of the output beam 14 from the laser diode array 12, and the half wave plate 304 is oriented to produce a polarization 45° from that desired on the substrate. The isolation element 306 provides the additional rotation to produce P-polarized radiation on the substrate. This polarization direction is preserved in the recycled radiation beam, however passage of the recycled radiation through the isolation element a second time produces an additional 45° rotation. Thus, the polarization direction of the recycled radiation in the space between the isolation element and the half wave plate is orthogonal to the polarization direction of the radiation 14 coming directly from the laser diode array 12. After making a second pass through the half wave plate 304, the recycled radiation has a polarization direction normal to that passed by the polarizer 302, resulting in severe attenuation of the recycled radiation beam.
Though it is not essential to the design of systems 300 of
In the operation of optical system 300 of
One difficulty with the configurations illustrated in
One way of avoiding sending the recycled radiation back to laser diode array 12 is shown in
As is shown in
In the particular example embodiment illustrated in
In the example embodiment of recycling optical system 300 of
In an example embodiment, it is preferred that the reflected radiation be returned by recycling optical system 300 to the same point (e.g., point 321 or the points on line image 24) on the substrate from where it was reflected, to within a fraction of the thermal diffusion length. Otherwise, the reflected radiation can exacerbate the non-uniform heating problems associated with LTP. The embodiment examples of recycling radiation system 300 shown in
One shortcoming of the example embodiments shown in
The period of the grating 462 is chosen to diffract the incident beam back on itself and the grating blaze is optimized for this geometry. Thus, the optimum grating period P is given by P=λ/2 sin θG where λ is the wavelength of the radiation and θG is the angle of incidence onto the grating relative to the grating surface normal NG. The purpose of the grating is to compensate for the tilted focal plane at the substrate, which would otherwise result in the return image being defocused by an amount depending on its distance from the image point 321. Point 321 coincides with the intersection of the substrate 16 with the optical axes A1 and A2 of LTP optical system 22 and recycling optical system 300, respectively. Note that in the geometry shown in
In the operation of recycling optical system 300 of
A shortcoming with the embodiment of
In one example embodiment, LTP optical systems 22 and 22′ are adapted to form images 24 and 24′ that at least butt and may overlap with one another at the substrate. In another example embodiment, images 24 and 24′ are line images. In another example embodiment, at least one of annealing radiation beams 23 and 23′ is incident substrate surface 16S at an incident angle θ23 or θ′23 that is at or near the associated Brewster's angle, which for silicon is ˜75° at 800 nm.
Such an arrangement reduces the demands on the radiation intensity required from the individual laser diode radiation sources 12 and 12′ since their outputs can be effectively combined. The example embodiment of the LTP system of
One of the problems inherent in the simple arrangement shown in
Current commercially available isolators have an aperture limit of 10 mm and a power limitation of 500 W/cm2. This precludes the use of current generation isolators for a silicon annealing application, however isolators might be used for applications requiring appreciably lower power levels. Also, it is anticipated that future generation isolators will have larger apertures and higher power limitations, making them suitable for the silicon annealing applications.
Not only is it possible to use multiple laser diode arrays 12 and 12′ to achieve a desired intensity, in an example embodiment of the present invention, multiple laser diode arrays (radiation sources) are used in combination with an arbitrary number of recycling optical systems while preserving the desired incidence angles. This example embodiment is illustrated in
In practice, an incidence angle θ23 between 60° and 80° would likely be used for annealing silicon. In the example embodiment illustrated in
In most cases, each recycling optical system 300 is employed off-axis so that the input and output beams do not overlap. In the embodiment of
Radiation reflected from the substrate a second time is again collected by corresponding recycling optical systems 300C and 300C′ and imaged back on the substrate as recycled radiation beams 23CRD and 23C′RD. Radiation reflected from the substrate a third time is again collected by corresponding recycling optical systems 300D and 300D′ and imaged back on the substrate as recycled radiation beams 23DRD and 23D′RD. This time, the reflected radiation beams from beams 23DRD and 23D′ RD are returned back to recycling optical systems 300C and 300C′ from where they progress to systems 300B and 300B′ and eventually return to laser diode arrays 12 and 12′.
In the present example embodiment, each of the two input beams 23A and 23A′ reflect from substrate surface 16S seven times before returning to the laser diode array 12 or 12′. Even if a single reflection absorbed only half of the incident radiation, after seven reflections less than 1% of the original radiation would be returned to the corresponding laser diode array. This would be further attenuated by the optical efficiency of the extended optical trains of the recycling optical systems.
The example embodiments discussed above in connection with
Furthermore, in an example embodiment of the present invention, an arbitrary number of recycling radiation systems 300 are arranged similar to the arrangement shown in
In the foregoing Detailed Description, various features are grouped together in various example embodiments for ease of understanding. The many features and advantages of the present invention are apparent from the detailed specification, and, thus, it is intended by the appended claims to cover all such features and advantages of the described apparatus that follow the true spirit and scope of the invention. Furthermore, since numerous modifications and changes will readily occur to those of skill in the art, it is not desired to limit the invention to the exact construction and operation described herein. Accordingly, other embodiments are within the scope of the appended claims.
This application is a continuation-in-part of U.S. patent application Ser. No. 10/653,625, filed on Sep. 2, 2003 and assigned to Ultratech, Inc. This application is also related to U.S. patent application Ser. No. 10/287,864, filed on Nov. 6, 2002, and assigned to Ultratech, Inc.
Number | Date | Country | |
---|---|---|---|
Parent | 10653625 | Sep 2003 | US |
Child | 10838076 | May 2004 | US |