1. Field of the Invention
Embodiments of the present invention generally relate to coherent pulsed light sources such as lasers and, more particularly, to pulse stretching and temporally and spatially decorrelating coherent light in an effort to provide intense and uniform illumination.
2. Description of the Related Art
The integrated circuit (IC) market is continually demanding greater memory capacity, faster switching speeds, and smaller feature sizes. One of the major steps the industry has taken to address these demands is to change from batch processing silicon wafers in large furnaces to single wafer processing in a small chamber.
During such single wafer processing the wafer is typically heated to high temperatures so that various chemical and physical reactions can take place in multiple IC devices defined in the wafer. Of particular interest, favorable electrical performance of the IC devices requires implanted regions to be annealed. Annealing recreates a more crystalline structure from regions of the wafer that were previously made amorphous, and activates dopants by incorporating their atoms into the crystalline lattice of the substrate, or wafer. Thermal processes, such as annealing, require providing a relatively large amount of thermal energy to the wafer in a short amount of time, and thereafter rapidly cooling the wafer to terminate the thermal process. Examples of thermal processes currently in use include Rapid Thermal Processing (RTP) and impulse (spike) annealing.
A drawback of RTP processes is that they heat the entire wafer even though the IC devices typically reside only in the top few microns of the silicon wafer. This limits how fast one can heat up and cool down the wafer. Moreover, once the entire wafer is at an elevated temperature, heat can only dissipate into the surrounding space or structures. As a result, today's state of the art RTP systems struggle to achieve a 400° C./s ramp-up rate and a 150° C./s ramp-down rate. While RTP and spike annealing processes are widely used, current technology is not ideal, and tends to ramp the wafer temperature during thermal processing too slowly and thus expose the wafer to elevated temperatures for too long a period of time. These thermal budget type problems become more severe with increasing wafer sizes, increasing switching speeds, and/or decreasing feature sizes.
To resolve some of the problems raised in conventional RTP type processes various scanning laser anneal techniques have been used to anneal the surface(s) of the substrate. In general, these techniques deliver a constant energy flux to a small region on the surface of the substrate while the substrate is translated, or scanned, relative to the energy delivered to the small region. Due to the stringent uniformity requirements and the complexity of minimizing the overlap of scanned regions across the substrate surface these types of processes are not effective for thermal processing contact level devices formed on the surface of the substrate.
Pulsed laser annealing techniques have been used to anneal finite regions on the surface of the substrate to provide well defined annealed and/or re-melted regions on the surface of the substrate. In general, during a pulsed laser anneal process various regions on the surface of the substrate are exposed to a desired amount of energy delivered from the laser to cause the preferential heating of desired regions of the substrate. Pulsed laser annealing techniques have an advantage over conventional processes that sweep the laser energy across the surface of the substrate, since the need to tightly control the overlap between adjacently scanned regions to assure uniform annealing across the desired regions of the substrate is not an issue, since the overlap of the exposed regions of the substrate is typically limited to the unused space between die, or “kerf” lines.
However, light waves produced by a laser often have high temporal and spatial coherence. Coherence is the property of waves that enables them to exhibit interference where at least two waves are combined to add constructively or subtract destructively depending on the relative phase between the waves. Temporal coherence characterizes how well a wave can interfere with itself at a different time and may be defined as the measure of the average correlation between the values of a wave at every pair of times separated by a given delay. Thus, a wave containing only a single frequency (a perfect sine wave or monochromatic light) is perfectly correlated at all times, while a wave whose phase drifts quickly will have a short coherence time. The most monochromatic sources are usually lasers, and higher quality lasers tend to have long correlation lengths (up to hundreds of meters). White light, which comprises a broad range of frequencies, is a wave which varies quickly in both amplitude and phase leading to a short coherence time (approximately 10 periods); thus, white light is usually considered as incoherent. Spatial coherence describes the ability for two points in the extent of a wave to interfere when averaged over time. More precisely, spatial coherence may be defined as the cross-correlation between two points in a wave for all times.
The coherence of laser beams manifests itself as speckle patterns and diffraction fringes, which suggest deviation from the desired uniform illumination in pulsed laser annealing and other applications. A speckle pattern is a random intensity pattern produced by the mutual interference of coherent waves that are subject to phase differences and/or intensity fluctuations. Because the surfaces of most materials are extremely rough on the scale of an optical wavelength (˜500 nm), coherent light from a laser, for example, reflected from such a surface results in many coherent wavelets, each arising from a different microscopic element of the surface. At any moderately distant point from the surface, the distances traveled by these various wavelets may differ by several wavelengths, and the interference of these wavelets of various phases results in the granular pattern of intensity called speckle. In other words, each point in the speckle pattern is a superposition of each point of the rough surface contributing with a random phase due to path length differences. Diffraction fringes are formed when light from a point source, such as a laser, passes by an opaque object of any shape.
Spatial coherence of light sources has been addressed by the use of random phase plates, also known as diffusers. Intended to scatter the light, optical diffusers increase the frequency of modulation due to interference, but they do not eliminate the interference. However, for pulsed laser annealing techniques and other applications, it is not sufficient to simply increase the frequency of modulation with a diffuser; the depth of modulation from coherence effects should be reduced, as well.
Accordingly, what are needed are techniques and apparatus for temporally and spatially decorrelating light from a coherent light source to provide incoherent light.
Embodiments of the present invention generally relate to decorrelating coherent light from a light source, such as a pulsed laser, in both time and space in an effort to provide intense and uniform illumination.
One embodiment of the present invention is a method for decorrelating a coherent light beam. The method generally includes providing a plurality of beam splitters aligned along an optical axis, wherein each of the beam splitters is configured to divide an incident light beam into an on-axis component beam traveling substantially along the optical axis and an off-axis component beam traveling substantially perpendicular to the optical axis; transmitting the coherent light beam to a first beam splitter in the plurality; combining on-axis and off-axis component beams received from a last beam splitter in the plurality in a beam combiner to form incoherent light; and optically steering the off-axis component beams to a subsequent beam splitter in the plurality of beam splitters or to the beam combiner.
Another embodiment of the present invention is a method for decorrelating a coherent light beam. The method generally includes providing N beam splitters, dividing the coherent light beam into 2N component beams using the N beam splitters, and combining the 2N component beams to form an incoherent light beam.
Yet another embodiment of the present invention provides an apparatus for decorrelating coherent light. The apparatus generally includes a plurality of beam splitters aligned along an optical axis, wherein each of the beam splitters is configured to divide an incident light beam into an on-axis component beam traveling substantially along the optical axis and an off-axis component beam traveling substantially perpendicular to the optical axis, a first beam splitter in the plurality being configured to receive the coherent light; a beam combiner configured to combine on-axis and off-axis component beams received from a last beam splitter in the plurality to form incoherent light; and a plurality of optical steering devices configured to direct off-axis component beams to a subsequent beam splitter in the plurality of beam splitters or to the beam combiner.
Yet another embodiment of the present invention provides an apparatus for decorrelating a coherent light beam. The apparatus generally includes N beam splitters configured to divide the coherent light beam into 2N component light beams and a beam combiner adapted to combine the 2N component light beams into a temporally and spatially incoherent light beam.
Yet another embodiment of the present invention provides a laser processing system. The laser processing system generally includes a laser source for providing coherent light, a decorrelator coupled to the laser source, and a target coupled to the decorrelator, wherein the target receives incoherent light. The decorrelator generally includes a plurality of beam splitters aligned along an optical axis, wherein each of the beam splitters is configured to divide an incident light beam into an on-axis component beam traveling substantially along the optical axis and an off-axis component beam traveling substantially perpendicular to the optical axis, a first beam splitter in the plurality being configured to receive the coherent light; a beam combiner configured to combine on-axis and off-axis component beams received from a last beam splitter in the plurality to form incoherent light; and a plurality of optical steering devices configured to direct off-axis component beams to a subsequent beam splitter in the plurality of beam splitters or to the beam combiner.
Yet another embodiment of the present invention is a method. The method generally includes positioning a plurality of beam splitters along an optical axis and transmitting a coherent pulse of energy through the plurality of beam splitters to form a composite pulse of energy. The composite pulse of energy generally has an amount of energy transmitted through the plurality of beam splitters and an amount of energy reflected at least once in the plurality of beam splitters such that the composite pulse of energy is temporally and spatially incoherent.
So that the manner in which the above recited features of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the appended drawings. It is to be noted, however, that the appended drawings illustrate only typical embodiments of this invention and are therefore not to be considered limiting of its scope, for the invention may admit to other equally effective embodiments.
Embodiments of the present invention decorrelate temporally and spatially coherent light from a light source in an effort to provide intense and uniform illumination. For some embodiments employing a pulsed light source, the output pulse may be stretched relative to the input pulse width. The methods and apparatus described herein may be incorporated into any application where intense, uniform illumination is desired, such as pulsed laser annealing, welding, ablating, and wafer stepper illuminating.
For example, in an exemplary laser processing system 100 of
For a pulsed laser, energy pulse characteristics of the light source 102 may typically include, but are not limited to, the total amount of energy, the peak energy level, the energy flux, the energy density, the pulse profile, the period, and/or the duration of the pulse. In a pulsed laser annealing application if the energy pulse characteristics are not optimized, damage to a substrate may be created by the stress induced from the rapid heating of the melted regions on the surface of the substrate. The rapid heating may generate acoustic shock waves in the substrate that can cause cracks, induce stress, and otherwise damage various regions of the substrate. It should be noted that energy pulse durations that are too long are also undesirable since this may cause dopants in the anneal regions to undesirably diffuse into adjacent regions of the substrate. Therefore, energy pulse characteristics of the light source 102 for a given application should be controlled.
For laser annealing applications, as an example, the dose of energy delivered from the light source 102 may be between about 1 and about 10 Joules over an 8 to 10 nanosecond (ns) pulse duration, which is equivalent to delivering an average total power of between about 100 MW to about 1250 MW in each pulse to the anneal region. It should be noted that the instantaneous power delivered at any time during each pulse may be much higher or lower than the average due to variations in the profile of the energy pulse.
Since the effectiveness of the laser annealing process, for example, is dependent on the transmission, absorption, and reflection of the delivered energy by the material to be annealed, the wavelength (λ) or wavelengths of the energy delivered by the light source 102 may be tuned so that a desired amount of energy is delivered to a desired depth within the substrate. It should be noted that the amount of energy delivered by each photon of light also varies as a function of wavelength (E=hc/λ), and thus, the shorter the wavelength, the greater the energy delivered by each photon of light. However, in some cases the substrate material, such as silicon, has an absorption edge that varies with thickness and wavelength, which limit the wavelengths that are absorbed by the substrate material. Therefore, depending on the thickness and type of material from which the substrate is made, the wavelength(s) of the emitted radiation may be varied to achieve the desired energy transfer to the substrate to minimize damage and promote uniform heating of the exposed region of the substrate. In one embodiment, the light source 102 is adapted to deliver energy at a wavelength less than about 1064 nm to a primarily silicon-containing substrate. In one embodiment, the laser annealing process is performed on a silicon-containing substrate using radiation that is delivered at wavelengths that are less than about 800 nm. In another embodiment, the wavelength of delivered from the light source 102 is about 532 nm to the primarily silicon-containing substrate. In yet another embodiment, the wavelength of the optical energy delivered from the energy source is about 216 nm or about 193 nm to the primarily silicon-containing substrate. For some embodiments, an Nd:YAG (neodymium-doped yttrium aluminum garnet) laser adapted to deliver energy at a wavelength between about 266 nm and about 1064 nm may be used.
The light source 102 may be optically coupled to a decorrelator 104 for producing incoherent light in an effort to uniformly illuminate a target 106, such as a substrate undergoing semiconductor processing (e.g., pulsed laser annealing) or two components being welded together. Optical coupling between the light source 102 and the decorrelator 104 and between the decorrelator 104 and the target 106 may occur via simple linear alignment of the devices, optical steering devices (e.g., mirrors, lenses, and beam splitters), optical fibers, and/or other optical waveguides depending on the application. Although not shown in the system 100 of
The decorrelator 104 may eliminate certain undesirable effects from the coherent light, such as speckle or fringe formation. For some embodiments, the decorrelator 104 may also function as a pulse stretcher 112 as illustrated in
One exemplary embodiment of a decorrelator 104 is illustrated in
A beam splitter, as used herein, may be generally defined as an optical device that splits a beam of light into two component light beams, a transmitted beam and a reflected beam. Beam splitters may be characterized according to the ratio of reflected intensity to transmitted intensity (R:T). Thus, a beam splitter with a 30:70 ratio reflects about 30% of the energy in the incident light beam and transmits approximately 70% of the energy. There are two main types of beam splitters: plate beam splitters and cube beam splitters.
Plate beam splitters, or plate dividers, consist of a thin plate of optical glass, quartz, or single-axis crystals (e.g., CaF2) with a different type of coating deposited on each side. The first side may be coated with a metallic coating or a dielectric film having partial reflection properties in the optical spectrum. Metallic coatings tend to have considerable absorption, thereby lowering the intensity of reflected and transmitted component light beams after splitting. Dielectric coatings usually are characterized by having no absorption qualities, so such coatings may be used for beam splitters in applications with high-power laser systems, such as a pulsed laser annealing system. The second side may have an anti-reflection coating optimized for 450 (the angle most frequently used in applications employing plate beam splitters) with minimum reflectivity in an effort to avoid unwanted additional reflections. This anti-reflection coating may have a reflectivity of only 0.5% at an angle of incidence of 450.
A cube beam splitter may be formed from two matched right-angle triangular glass prisms that are glued together at their hypotenuses using optical cement, such as Canada balsam, a transparent resin obtained from the balsam fir. The thickness of the optical cement layer may be designed such that a desired portion of the light incident through one face of the cube is reflected and the remaining portion is transmitted for a given wavelength. Prior to cementing, a partial reflection film (e.g., a metallic or dielectric coating) is deposited onto the hypotenuses of the right-angle prisms. The other four faces of both prisms may be antireflection-coated in an effort to minimize ghost images.
Plate beam splitters have a number of advantages over cube beam splitters. First of all, plate beam splitters are devoid of optical cements, which may absorb light energy. Thus, plate beam splitters can withstand significantly higher levels of laser power without damage, an important consideration when using moderate- or high-power lasers. As described above, the light source used in pulsed laser annealing applications may deliver an average total power between about 100 MW and 1250 MW. Plate beam splitters may also be significantly smaller and lighter than cube beam splitters. However, plate beam splitters introduce a shift, or a deviation, into the light beam due to their thickness.
Conversely, cube beam splitters are rugged, easy to mount, and ideal for beam superposition applications. Cube beam splitters deform much less when subjected to mechanical stress when compared to plate beam splitters and do not introduce a shift in the light beam. Most of the unwanted reflections from a cube beam splitter are in the retrodirection (i.e., the opposite direction from the incident direction and along the same optical axis) and thus, do not contribute to ghost images. Furthermore, because the metallic or dielectric coating is sealed within the body of the cube, the coating is very resistant to degradation with time.
That being said, the plurality of beam splitters 201-205 may comprise plate beam splitters, cube beam splitters, or a combination of plate and beam splitters as shown. In the decorrelator 104 of
The first beam splitter 201 may receive an incident light beam 210 from the light source 102 in step 304. As described above, the incident light beam 210 may be a series of laser pulses, each pulse having a pulse width of 8 ns, for example. In step 306, the incident light beam 210 may be divided into a transmitted component beam 212 and a reflected component beam 214 by the first beam splitter 201. With the plate beam splitter or the hypotenuse of a cube beam splitter angled at around 45° with respect to the incident light beam 210, the transmitted component beam 212 may remain substantially on the optical axis A-A (i.e., an on-axis component beam), and the reflected component beam 214 may be diverted substantially perpendicular to the optical axis A-A (i.e., an off-axis component beam) as illustrated.
In a similar fashion, each beam splitter 201-205 may receive component light beams from two different directions and transmit component light beams in two different directions. The number of component light beams may be doubled by each beam splitter 201-205 as illustrated in
In a similar manner, if the two off-axis component light beams 414, 416 from the first beam splitter 400 are redirected to a second beam splitter 420 such that the two off-axis component light beams 414, 416 reach one face 422 of the second beam splitter 420 different from the face 424 upon which the on-axis component beams 410, 418 are incident, the second beam splitter may generate four on-axis component light beams 426 and four off-axis component light beams 428. In essence, four component light beams may be produced from two incident component light beams using only a single beam splitter, and the number of component light beams produced may be doubled to eight by using a second beam splitter.
This line of reasoning may be extended to N beam splitters. For N beam splitters where the on-axis component light beams from one beam splitter are incident on the next beam splitter and the off-axis component light beams from the same one beam splitter are redirected to be incident on the same next beam splitter, the number of component beams produced may be expressed as 2N. Thus, for the plurality of beam splitters 201-205 illustrated in
The off-axis component light beams, such as reflected component beam 214 and off-axis component light beams 414, 416, 428 in
For some embodiments using retroreflectors 216, 218 as the optical steering devices, the size of the retroreflectors may dictate the spacing between adjacent beam splitters, or a desired spacing between adjacent beam splitters may lead to the selection of retroreflectors with a corresponding size. Using other suitable optical steering devices, such as an optical fiber or combination of mirrors, may remove such restrictions on the spacing between adjacent beam splitters, allowing more freedom when designing the placement of the beam splitters and the decorrelator size.
For some embodiments as depicted in
Also, such embodiments with the reduced number of optical steering devices may require more lateral space to achieve the same amount of delay. For example, if l is the delay length of one optical path leg in
To achieve the same total delay length d between adjacent beam splitters, embodiments that have a reduced number of optical steering devices (see
With each off-axis component light beam experiencing a delay through an optical path to one or more optical steering devices and back to a subsequent beam splitter, the chart 250 in
For example, if the total delay length of an off-axis component beam is d, then the component light beams generated by delivering the incident light beam 210 to the first splitter 201 shown in
This process may propagate down the remaining beam splitters 203-205 with each component light beam being split into two components that have zero delay added when transmitted as an on-axis component light beam and a 1d delay added when transmitted as an off-axis component light beam. In this example, after the last beam splitter 205 and optical steering devices, the generated component light beams may consist of a component light beam that was not delayed (i.e., with zero delay), five component beams having a 1d delay length, ten component beams having a 2d delay length, ten component beams having a 3d delay length, five component beams having a 4d delay length, and one component beam having a 5d delay length for thirty-two total component light beams as portrayed in the last row of chart 250.
By splitting the incident coherent light beam into component light beams having various time delays, the decorrelator 104 in
In order to be effective, the separation in time due to optical travel through the off-axis optical paths including the optical steering devices (e.g., the retroreflectors 216, 218) should be on the order of a coherence length or more. As used herein, the coherence length may be generally defined as the propagation distance from a coherent light source to a point where the light wave maintains a specified degree of coherence. In optics, the coherence length L may be approximated by the formula:
where λ is the nominal wavelength of the source, n is the refractive index of the medium, and Δλ is the spectral width of the source. Because the spectral width of a source is somewhat ambiguous, however, the coherence length has been defined as the optical path length difference of a self-interfering laser beam which corresponds to a 50% fringe visibility, where the fringe visibility V is defined as
where I is the fringe intensity.
Spatial decorrelation may be accomplished by an imperfect overlay of the component light beams. In other words, the beam splitters 201-205 and optical steering devices (e.g., the retroreflectors 216, 218) may not be perfectly aligned in an effort to cause displacements, deviations in transmission angles of the component light beams, or both. The separation in space should be sufficient to displace the speckle pattern by at least a speckle dot width at the image plane. The speckle dot width at the image plane may be essentially the minimum resolvable spot for the laser processing system 100 and may be approximated for a laser light source as
where D is the diameter of the laser beam at its narrowest spot, λ0 is the vacuum wavelength of the light, n is the refractive index of the medium, and NA is the numerical aperture. Thus, when air (n=1.0) is used as the transmission medium, the narrowest spot may be approximated by D=0.6λ0/NA.
Once the incident light beam 210 has been decorrelated in both space and time by splitting said beam into 2N component light beams with N beam splitters as described herein, the component light beams from the last beam splitter 205 may be combined into an incoherent light beam 220 in a beam combiner 222 in step 310. The beam combiner 222 may be aligned with the optical axis A-A of the beam splitters 201-205 as shown in
The beam combiner 222 may comprise a polarization rotator or a half-wave plate 230 and a polarizing cube beam splitter 232 aligned with the optical axis A-A of the plurality of beam splitters 201-205. For some embodiments as illustrated in
For some embodiments, the incident light beam 210 (and thus, the generated component light beams in the decorrelator 104) is s-polarized, or perpendicular to the plane of incidence, a plane made by the propagation direction of the light and a vector normal to a reflecting surface. S-polarization is also known as sigma-polarization or sagittal plane polarization. In such embodiments, the s-polarized on-axis component light beams may exit the half-wave plate 230 as p-polarized on-axis component light beams parallel to the plane of incidence. P-polarization is also known as pi-polarization or tangential plane polarization. In a polarizing beam splitter, s-polarized light may be completely reflected, while p-polarized light may be completely transmitted. Therefore, the p-polarized on-axis component light beams from the half-wave plate 230 may be transmitted through the polarizing beam splitter 232 and be combined with the s-polarized off-axis component light beams reflected by the hypotenuse of the polarizing cube beam splitter 232 to form the incoherent light beam 220.
For N beam splitters, the benefit of incoherent summing may increase as the square root of the number of beams being summed (√2N or 2N/2). Thus, for an application, such as a pulsed laser annealing system, with a coherent light source with 40% illumination non-uniformity, for example, the non-uniformity may be improved to 7.1% (=40%/√25) with the decorrelator of
For some embodiments, as shown in
If the incident light beam 210 is a coherent light pulse from a laser, for example, the lateral spacing between the beam splitters 601-610 and the retroreflectors 621-629 may be decreased when compared to embodiments having fewer beam splitters, such as the embodiments of
Although embodiments of the decorrelator 104 are shown in
Furthermore, characteristics of the decorrelator 104—such as the number of beam splitters, the number of optical steering devices, the topology of the layout, and the dimensions of the layout (e.g., the on-axis spacing between adjacent beam splitters, the lateral spacing between a beam splitter and one or more off-axis optical steering devices, and the spacing between the last beam splitter and the beam combiner)—may be selected and altered in an effort to adjust the amount of pulse stretching and to achieve a desired pulse profile when the incident light beam 210 is a coherent light pulse.
While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.
This application is related to the U.S. patent application entitled “Apparatus and Method of Improving Beam Shaping and Beam Homogenization,” by Bruce E. Adams et al. [Docket No. APPM/11251], filed Jul. 31, 2007; and the U.S. patent application entitled “Method and Apparatus for Decorrelation of Spatially and Temporally Coherent Light,” by Dean Jennings et al. [Docket No. APPM/11369], filed Jul. 31, 2007; which are all herein incorporated by reference.