TWI575571B - Non-periodic pulse processing system and method for partially melted film - Google Patents

Description

Non-periodic pulse processing system and method for partially melted film [Reciprocal Reference Related Applications]

US Application No. 61/264,082, filed on Nov. 24, 2009, entitled "Systems and Methods for Advanced Excimer Laser Annealing"; US Patent Application No. 61/286,643, filed on Dec. 15, 2009, entitled "Systems and Methods for Advanced Excimer Laser Annealing"; December 31, 2009 U.S. Patent Application Serial No. 61/291,488, entitled "Systems and Methods for Advanced Excimer Laser Annealing", filed on November 3, 2009, entitled US Patent of "Method For Obtaining Uniformly Sized Small Grain Polycrystalline Silicon With Low Intragrain Defect-Density Films Through Partial Melt Crystallization" Application No. 61/257,657; application on November 3, 2009, entitled "through US Patent Application Method for Full Melt Crystallization to Obtain Uniformly Sized Small Grain Polycrystalline Silicon with Low Intragrain Defect-Density Films Through Complete Melt Crystallization No. 61/257,650; US Patent Application No. 61/291,663, entitled "Advanced Single-Scan SLS", filed on December 31, 2009; January 1, 2010 application, name U.S. Patent Application Serial No. 61/294,288 to SLS (Sequential Firing SLS); and System and Methods for "Non-Periodic Pulse Continuous Lateral Crystallization" (Applications and Methods for Non-Periodic Pulse Sequential Lateral Solidification), U.S. Patent Application Serial No. 12/776,756; and U.S. Patent No. 4, 2010, entitled "System and Methods for Non-Cycles" PCT International Patent Application No. PCT/US2010/0 of Periodic Pulse Sequential Lateral Solidification) The priority of 33565 is attached to the full text of each case for reference.

All patents, patent applications, patent publications, and publications cited herein are hereby incorporated by reference. In the case where the application of the application is inconsistent with the presentation of the document, the application shall be examined and examined.

The present invention relates to a non-periodic pulse processing system and method for a partially molten film.

In the field of semiconductor processing, there have been some technical descriptions for converting an amorphous germanium film into a polycrystalline film. One technique is excimer laser annealing (ELA). ELA is a pulsed laser crystallization process, which can produce a uniform crystal film on a substrate, such as a heat-resistant substrate (such as glass and plastic), but not limited thereto. The ELA system and process examples are described in commonly-owned U.S. Patent Publication No. 20090309104, entitled "Systems and Methods for Creating Crystallographic-Orientation Controlled Poly-Silicon Films", Ea 2009 Application filed on August 20th; U.S. Patent Publication No. 20100065853, entitled "Process and System for Laser Crystallization Processing of Film Regions" On a Substrate to Minimize Edge Areas, and Structure of Such Film Regions), the application filed on September 9, 2009; and US Patent Publication No. 20070010104, entitled "Using a Linear Beam to Laser Crystallize the Substrate Application for Process and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type Beam, and Structures of Such Film Regions, Application for March 9, 2006 case.

Conventional ELA tools use a single linear beam that continuously scans the surface of the sample at low speeds with large overlaps (eg, 95%) between pulses to create a large number of pulses per unit area by a single scan. Therefore, in ELA, the membrane region is irradiated by excimer laser light, causing the membrane to partially melt and then crystallize. Repeating part of the molten film can form a small crystal grain polycrystalline film. However, this method often has the problem of uneven microstructure, which is caused by energy fluctuation between pulses and/or uneven beam intensity distribution. A large number of pulses not only need to cause an additive effect to obtain a more uniform grain size, but also reduce the short-axis beam edge effect. In the beam edge section of the beam, the energy gradually drops to zero. Depending on the position within the film, position dependent variations occur in the order of the initial pulse energy. This variation is not easily removed by subsequent ELA processes and may result in pixel brightness afterimages (ie, mura). Figure 1A illustrates the random microstructure obtained by ELA. The cerium (Si) film is irradiated multiple times to produce a random polycrystalline film having a uniform grain size. FIG. 1B depicts a conventional ELA single scan showing the short-axis cross-section of the linear beam 101 as the beam 101 scans the film 104. Beam 101 travels in the direction of arrow 102, and as light beam 101 moves across film 104, region 103 of film 104 is illuminated by a plurality of laser pulses.

In addition, crystallization methods and tools for obtaining uniform grain structure (UGS) at very high yields have been reported. For example, such a system is disclosed in Processes and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type, entitled "Processing and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type" U.S. Patent Application Publication No. 20070010104 to Beam, and Structures of Such Film Regions. UGS is a single pulse irradiation process involving complete melt crystallization (CMC) and/or partial melt crystallization (PMC) to be crystallized. An additional feature of the UGS process is the controlled ignition of the position of the laser pulse, which will only partially or completely melt in the area where the pixel rows/columns of the thin film transistor (TFT) reside. If the step distance between the pulses exceeds the linear beam width, then the lines are still unirradiated areas of the film (eg, as-deposited amorphous 矽). The selective area crystallization (SAC) process can have high yields when the average number of pulses per unit area is less than one.

However, none of the prior art tools have specifically optimized ELA for large films, such as for televisions with low pixel density. For such substrates, the conventional ELA is an inefficient process in which time and resources are wasted on the Si substrate between the crystal pixel positions. Although the UGS tool can omit these areas, the material obtained is significantly more defective than typical ELA materials, and the material uniformity is insufficient when using typical illumination conditions.

Non-periodic pulse methods and tools for sequentially triggering lasers using positional control are described. Multiple laser implementation systems can be employed to generate separate non-periodic laser pulses in the crystallization process, i.e., each separate laser pulse causes individual partial melting and curing cycles. Multiple lasers are used to coordinate the pulse sequence and illuminate and crystallize selected membrane regions with a single scan or multiple scans.

In one aspect, the invention is directed to a method of treating a film comprising, when advancing a film in a first selected direction, illuminating a first region of the film with a first laser pulse and a second laser pulse, each laser pulse providing plastic a beam of light having a fluence sufficient to partially melt the film such that the first region resolidifies and crystallizes to form a first crystalline region, and the second region of the film is illuminated with a third laser pulse and a fourth laser pulse, each The pulse provides a shaped beam and has a fluence sufficient to partially melt the film such that the second region resolidifies and crystallizes to form a second crystalline region, wherein the time interval between the first laser pulse and the second laser pulse is less than the first Ray Half of the time interval between the shot pulse and the third laser pulse.

In some embodiments, the time interval between the first laser pulse and the second laser pulse is longer than the time interval of a single melting and solidification cycle of the film. In some embodiments, the first laser pulse and the second laser pulse each have the same energy density, the first laser pulse and the second laser pulse each have a different energy density, the first laser pulse and the second The laser pulses each achieve the same degree of film melting, and/or the first laser pulse and the second laser pulse each achieve a different degree of film melting. In some embodiments, the film is an amorphous ruthenium film that does not contain pre-existing crystallites. In some embodiments, the first laser pulse has an energy density sufficient to melt the amorphous tantalum film and produce a crystal structure with a defective core region. In some embodiments, the second laser pulse has an energy density sufficient to re-melt the defective core region to form a uniform fine grain crystalline film.

In some embodiments, the film is an amorphous ruthenium film. In some embodiments, the film is deposited using one of low pressure chemical vapor deposition, plasma assisted chemical vapor deposition, sputtering, and electron beam evaporation.

In some embodiments, the film is a treated ruthenium film. In some embodiments, the treated ruthenium film is an amorphous ruthenium film that does not contain pre-existing crystallites, which has subsequently been processed in the following manner, including propelling the amorphous ruthenium film toward the second selected direction to extend the laser pulse illumination. The enamel film is shaped to extend the laser pulse to have a fluence sufficient to partially melt the amorphous ruthenium film.

In some embodiments, extending the laser pulse is produced by sequentially overlapping laser pulses from a plurality of laser sources, wherein the delay between pulses is short enough to initiate a single melting and curing cycle. In some embodiments, the amorphous ruthenium film is obtained by plasma assisted chemical vapor deposition. In some embodiments, the pulse length of the extended laser pulse is greater than a full width at half maximum of 300 nanoseconds (ns).

In some embodiments, the treated ruthenium film is a ruthenium film treated in the following manner, comprising: illuminating the ruthenium film with a laser pulse when the ruthenium film is advanced in a second selected direction, the laser pulse having a fluence sufficient to completely melt the ruthenium film . In some embodiments, the laser pulses are generated by overlapping laser pulses from a plurality of laser sources.

In some embodiments, the method includes illuminating a third region of the film with a fifth laser pulse and a sixth laser pulse as the film is advanced in the second selected direction, each laser pulse providing a shaped beam and having a portion of the molten film The fluence is such that the third region resolidifies and crystallizes to form a third crystalline region, and the fourth region of the film is illuminated with a seventh laser pulse and an eighth laser pulse, each pulse providing a shaped beam and having a portion of the molten film a fluence such that the fourth region resolidifies and crystallizes to form a fourth crystalline region, wherein a time interval between the fifth laser pulse and the sixth laser pulse is less than between the fifth laser pulse and the seventh laser pulse Half of the time interval. In some embodiments, the second selected direction is opposite the first selected direction, wherein the third region overlaps the second region and the fourth region overlaps the first region.

In some embodiments, the second selected direction is the same as the first selected direction, the third area overlaps the first area, and the fourth area overlaps the second area. In some embodiments, the method includes moving the film in a direction perpendicular to the first selected direction prior to advancing the film in the second selected direction. In some embodiments, each laser pulse is a linear beam with a uniform energy density at the top. In some embodiments, each laser pulse is a flood illumination pulse.

Another aspect of the invention pertains to a film treated in accordance with the above method. Yet another aspect of the invention is directed to an apparatus having a film treated in accordance with the above method, wherein the apparatus includes a plurality of electronic circuits disposed in a plurality of crystalline regions of the film. In some embodiments, the device is a display device.

In one aspect, the invention relates to a system for processing a film using a non-periodic laser pulse, comprising primary and secondary laser sources for generating a laser pulse; and a working surface for securing the film to the substrate; a platform for moving the film relative to the beam pulse to construct a propagayion direction of the laser beam pulse on the surface of the film; and a computer for processing instructions for synchronizing the platform with the laser pulse to provide loading to the mobile The first region of the film of the platform illuminates the first laser pulse from the primary source, the second region of the film illuminates the second laser pulse from the secondary source, and the third region of the film gives the third source from the primary source Laser pulse illumination, wherein the processing command is provided to move the film in a direction of travel relative to the beam pulse to illuminate the first, second, and third regions, wherein the distance from the center of the first region to the center of the second region is less than the center of the first region Half the distance of the center of the third zone, wherein the first, second and third laser pulses have a fluence sufficient to partially melt the film. In some embodiments, the platform moves at a fixed speed.

Still another aspect of the present invention is directed to a method of converting an amorphous ruthenium film containing no pre-existing crystallites into a small-grain film, the method comprising: propelling the amorphous ruthenium film toward the first selected direction to extend laser irradiation without The ruthenium film is shaped to extend the laser pulse to have a fluence sufficient to partially melt the amorphous ruthenium film, wherein the average grain size of the grain of the small grain film is less than the film thickness. In some embodiments, the pulse length of the extended laser pulse is greater than half full width by 300 ns and is a flood illumination pulse. In some embodiments, extending the laser pulse is produced by delaying overlapping laser pulses from a plurality of laser sources, wherein the delay between pulses is short enough to initiate a single melting and curing cycle. In some embodiments, the amorphous ruthenium film is obtained by plasma assisted chemical vapor deposition.

Another aspect of the invention is directed to a method of processing a film comprising providing a semiconductor film onto a substrate having a bottom interface adjacent the bottom surface of the substrate and a top surface facing the bottom surface; and, wherein the energy density is greater than the film integrity A laser beam of 1.3 times the melting threshold illuminates the film, and the energy density is selected to completely melt the film; and when the curing begins, the cap layer forms a surface interface on the top surface of the semiconductor film; wherein after the irradiation and complete melting of the film, the top interface Heterogeneous nucleation occurs at the bottom interface, and once cooled, heterogeneous nucleation will form less defective germanium grains on the bottom surface of the film. In some embodiments, the pulse duration of the laser beam is greater than 80 ns, 200 ns, or 400 ns. In some embodiments, the semiconductor thin film comprises a tantalum film having a thickness of from about 100 nanometers (nm) to about 300 nm. In some embodiments, the substrate is glass or quartz. In some embodiments, the grains are small equiaxed grains. In some embodiments, the energy density of the laser beam is 1.4 times the local full melting threshold. In some embodiments, the cover layer is formed by depositing a thin layer to the top surface of the film prior to illumination. In some embodiments, the cover layer is an oxide layer having a thickness of less than 50 nm. In some embodiments, the cover layer is formed by illuminating the film in an oxygenated environment. In some embodiments, the oxygenated environment is air. In some embodiments, the oxygenated environment is only oxygen. In some embodiments, the substrate is a patterned metal film overlying an insulating film, wherein the energy density is greater than 1.3 times the full melting threshold of the film. In one aspect, the invention is directed to a bottom gate thin film transistor (TFT) fabricated in accordance with the above method, wherein the patterned metal film is a bottom gate and the insulating film is a gate dielectric.

Non-periodic systems and methods enable high yield ELA and selective zone crystallization. This process is suitable for active matrix organic light emitting diode (AMOLED) television (TV) and ultra high definition liquid crystal display (UD-LCD). For both products, amorphous 矽 lacks performance and stability, and current low performance low temperature polysilicon (LTPS) technology is not yet cost competitive in terms of required panel size (eg eighth generation, up to 2.2 x 2.5 square) meter).

This invention relates to systems and methods for forming uniform polycrystalline films using a non-periodic pulsed laser technique in combination with partial melt crystallization and complete melt crystallization techniques. In some embodiments, the non-periodic pulse ELA is used to fabricate a uniform fine-grained crystalline film from an as-deposited amorphous germanium (Si) film that does not contain pre-existing crystallites, such as by low pressure chemical vapor deposition. (LPCVD), plasma assisted chemical vapor deposition (PECVD), sputtering or electron beam evaporation. In some embodiments, the floodlighting method is used to fabricate a uniform fine grain crystalline film, or to fabricate a precursor film for a non-periodic pulsed illumination method. The flooding method can be a two-shot partial melting process in which an amorphous ruthenium film (such as a PECVD film) without any pre-existing crystallites is converted into a uniform fine-grained crystal film through two steps, and the crystal The average lateral dimension of the granules exceeds the film thickness, i.e., the small crystal column. The flooding method can also be a part of the melting process of extending a single shot, wherein an amorphous ruthenium film (such as a PECVD film) which does not contain any pre-existing crystallites is converted into a uniform fine-grained crystal film, and the crystal The average lateral dimension of the granules is less than the film thickness. The flooding method can also be a complete melting process in which any type of amorphous ruthenium film having an oxidized interface at the top and bottom of the film is converted into a small, equiaxed grain Si film with less defects.

Non-periodic pulse ELA methods and tools for continuously triggering lasers using position control are described. The system may employ multiple lasers to produce separate non-periodic laser pulses in the crystallization process, such as each separate laser pulse causing individual partial melting and curing cycles with different aperiodic intervals between pulses. Multiple lasers are used to coordinate the pulse sequence and illuminate and crystallize selected membrane regions with a single scan or multiple scans. Multiple scans are performed to achieve a large number of melting and curing cycles in a predetermined area to benefit from the multiple irradiation accumulation effect of the ELA, thereby forming a uniform polycrystalline film having a tighter grain size distribution, for example.

Non-periodic pulse

An example laser pulse sequence is shown in Figure 2A-2C. The y axis represents energy density and the x axis represents time. Figure 2A depicts a periodic laser pulse rate that can be used in conventional ELA processes. The periodic laser repetition rate results in a laser pulse pattern that is equally spaced in the time domain. FIG. 2B shows the example of the non-periodic pulse in which the second pulse 105 ignites in a time approaching the first pulse 106. Next, the third pulse 107 ignites at a time interval different from the interval between the first pulse 106 and the second pulse 105. Figure 2C illustrates an embodiment of a laser pulse with different pulse rates and laser power (energy density). The film thus irradiated will experience a non-periodic pulse rate and variable illumination energy. Since the time interval between the first pulse 106 and the second pulse 105 is shorter, the areas illuminated by the first pulse 106 and the second pulse 105 are more overlapped.

The time delay between the first pulse 106 and the second pulse 105 is less than half the time interval between the first pulse 106 and the third pulse 107. In some embodiments, the time interval of the first pulse 106 and the second pulse 105 is less than 1/10, or less than 1/20, or less than 1/100 of the time interval of the first pulse 106 and the third pulse 107. The time delay between the first pulse 106 and the second pulse 105 can be from about 3 microseconds to about 1 millisecond, from about 5 microseconds to about 500 microseconds, and from about 10 microseconds to about 100 microseconds.

Figures 2B and 2C illustrate non-periodic pulse patterns using two closely spaced laser pulses or two "train" laser pulses; however, more closely spaced pulses may be used, such as 3-5 , which corresponds to 3-5 or more laser or laser cavities. In this embodiment, if more closely spaced pulses from different lasers are used, such as laser beams from two different laser energy sources or two different laser carriers of the same laser energy source, the target area corresponds accordingly. It is irradiated more times. For example, n pulses from n laser sources are closely spaced to form a column of n pulses, such that a single region will be illuminated n times after a single scan. The width of the beam is similar to the conventional ELA process.

The two consecutive pulses of the pulse train do not need to have the same energy density. For example, if the film still has heat due to the first pulse, the energy density of the second pulse can be lower than the first pulse. Similarly, a higher energy density can be used to compensate for changes in optical properties caused by the first pulse (amorphous 矽 absorbs ultraviolet (UV) light slightly better than crystallization). Considering the above two effects and other possible factors, the energy density of the second pulse is appropriately selected to cause the film to undergo the same degree of melting. Here, the degree of melting is reported to be a melting measurement independent of the melting details, which may be related to the precursor phase (amorphous or crystalline), heterogeneity (eg, a defect defect or a defect core surrounded by large and uniform grains) And the surface topography (smooth or rough, such as periodic like light wavelength) changes dramatically. The same degree of melting can be achieved when the melting range during the second pulse is equal to the melting range of the first pulse, for example about 80% of the film. In a multi-scan process dedicated to benefiting the additive effect to obtain a more uniform polycrystalline film, it is desirable that the pulses mostly produce the same degree of melting, so that the process is most efficient.

As shown in Figure 2C, the first laser pulse and the second laser pulse have different energy densities. In particular, Figure 2C illustrates that the first laser pulse has a lower energy density than the second pulse. In some embodiments, however, the second laser pulse has a lower energy density than the first laser pulse. In addition, in the multiple scanning process, the energy density offset between the first pulse and the second pulse at different scans is different or absent. For example, the energy density shift between the first and second pulses at the first scan is selected to compensate for optical property changes, while the offset at the second scan is selected to compensate for temperature. In some embodiments, even if the two pulses have different energy densities, and because the heat caused by the first pulse remains in the film, the second low energy pulse can also cause the same amount of film melt as the first high energy pulse.

In one embodiment, the system utilizes coordinated trigger pulses from a plurality of laser sources (a single laser source having multiple laser cavities (eg, tubes)) to generate non-periodic laser pulses, thereby generating a series of Pulses that are closely spaced in the time domain. Multiple laser sources can be combined into a single laser system. The laser system is a computer control system that uses computer control technology to illuminate the substrate in a predetermined manner, such as a computer to control laser ignition and platform movement, and one or more laser chambers to generate one or more laser beams. Each laser beam corresponds to a laser source. Each of the laser beams can be produced by a separate laser, or a portion of a plurality of laser chambers contained in a laser system.

Tools having multiple laser cavities (e.g., tubes) have been previously disclosed, (1) increasing pulse energy by simultaneously triggering and subsequently combining multiple pulses, and (2) triggering different tubes by delay and subsequent bonding The pulse width is increased as described in "Systems and Methods for Processing Thin Films", U.S. Patent No. 7,364,952, issued Apr. 29, 2008. In other words, the pulses can be combined to provide a modified single melt and cure cycle. The difference in non-periodic pulse ELA is that it is a pulse that utilizes different lasers in individual melt/cure cycles. However, since the pulse time domain is close enough, the platform still effectively overlaps when traveling at high speed.

In addition, non-periodic pulsed ELA methods and tools can also be used to perform selective regional crystallization of the film whereby only the film regions of the electronic device to be formed are crystallized. Non-periodic pulsed ELA methods and tools provide selective crystallization of the crystal, causing the first region of the film to grow, followed by a repetition rate of the laser for a period of time, and then allowing the second pulse of two or more lasers to substantially overlap, The second region of the film crystal grows. The laser pulse timing results in a non-periodic laser pulse sequence and substantially overlaps in the illuminated area, which will be detailed later. This method and system can be applied to ELA processes at high throughput.

In selective area crystallization, the film crystallizes at the location of the electronic device to be fabricated (subsequent process manufacturing, not discussed herein). However, not all electronic devices need to be equally uniform, or even have the same conductive material. For example, small TFTs are far more demanding for crystal uniformity than large TFTs or large capacitors. Also, TFTs for current driving require better uniformity than TFTs for switching. Therefore, in terms of the total area of a particular crystalline region, only a portion of the region that needs to be crystallized with a large number of laser pulses to obtain high crystal uniformity and conductivity, the remainder is treated with fewer pulses, or even a single pulse. The selective region crystallization non-periodic pulse ELA provides an architecture that scans only selected film regions, thereby reducing processing time.

Non-periodic pulse ELA

Aperiodic ELA systems include one or more of the following features: a plurality of lasers or laser tubes, and means for delaying the firing of subsequent pulses with sequential pulses within a short continuous period of time. The system also includes a position control trigger pulse such that the laser beam pulse illuminates a particular location on the substrate. Position control ensures that when the illumination area is properly positioned on the substrate, the closely spaced pulse timing of the two time domains should allow the illuminated portion of the film to cure between pulses to, for example, fabricate a row of pixel TFTs or circuits. The additional laser beam pulse has a high hat beam profile and the beam width is sufficient for a series of pulses to overlap the selected area.

The number of laser sources depends on various considerations such as throughput, laser power, panel size, display size, system design, and tool maintenance. A large number of lasers generally have a high crystallization rate, but must also have a large number of optical components, which will lead to complex and expensive system designs. Also, a large number of lasers may increase tool downtime due to more frequent maintenance (such as changing tubes). Example lasers have a value of two to four or more lasers each having a power of about 600 watts (W) or more to process a glass panel greater than 2 square meters and possibly up to 5 or 7.5 square meters. A display with a size of 30, 40 or 50 inches or more is manufactured.

Compared to conventional ELA and/or UGS tools, the non-periodic pulse ELA tool offers the following advantages:

(1) Effective delivery of power to the preselected area: Due to position control, the area between the pixel TFT/circuit does not need to be crystallized, so the crystallization rate is more efficient.

(2) Eliminate the residual image of the edge of the beam: the edge of the beam does not impinge on the pixel TFT/circuit area, so the crystalline regions within it undergo exactly the same pulse sequence.

(3) Optimized pulse sequence: During multiple scans, the region is illuminated by a series of pulses from multiple laser sources, which optimizes the sequence (eg pulse energy, pulse width, pulse warm-up).

(4) Use the vertical displacement between scans to mitigate beam non-uniformity on the long axis. (Using the effective parallel displacement within or between scans, ie, moving the lateral position of the beam relative to a predetermined area, the beam inhomogeneity on the short axis can also be mitigated.)

Non-periodic pulsed ELA typically requires multiple scans to achieve a desired material uniformity. The yield caused by the SAC operation of the non-periodic pulse ELA is generally higher than the conventional ELA. In addition, the non-periodic pulse ELA requires less pulses than is required for conventional ELA to obtain an acceptable uniform crystal structure. In the conventional ELA, the edge of the beam overlaps with the predetermined area, resulting in a change in the crystal structure of the irradiated area along the scanning direction. Crystal structure changes are described, for example, in Im and Kim on October 4, 1993, in Appl . Phys . Lett . 63, (14) . "Phase transformation of excimer laser crystal amorphous decidua (Phase transformation) "dissociation in excimer laser crystallization of amorphous silicon films"", which discusses the case where the grain size of a partially melted low pressure chemical vapor deposition (LPCVD) film varies with energy density; the salt crystal LPCVD amorphous Si film contains small crystallites, It initiates crystallization, causing the grain size of the film to increase with energy density. In plasma-assisted chemical vapor deposition (PECVD) films, the melting and solidification process is further complicated by the lack of crystallites. Therefore, the crystallization progresses first by forming a crystal through a nucleation process. If the nucleation density is low, a disc-shaped crystal structure is produced, for example, as shown in Fig. 2D, visible at the edge of the PECVD amorphous Si film which is very close to a single shot (i.e., a laser pulse). Figure 2D shows the edge region 120 of a single-shot PECVD amorphous Si film. The edge region 120 has both the amorphous Si portion 122 and the crystalline Si portion 124. However, the transition region 126 between the amorphous Si and the crystalline Si is not a sharp edge, but a heterogeneous region containing a mixture of crystalline and amorphous materials. The unevenness of the film after the first irradiation is affected by the change in grain size or the presence of a disc-shaped crystal structure. The unevenness is not easily removed by subsequent irradiation. In the conventional ELA, the influence of the energy density gradient of the edge of the first pulse beam can be seen even after 10 or more pulses. Therefore, a large number of pulses are required to eliminate the history of the edge of the first pulsed beam.

As described, the SAC using the non-periodic pulse ELA requires less pulses to achieve the same uniform crystalline film. As detailed below, the energy profile throughout the minor axis of the linear beam includes a leading edge and a trailing edge of the energy density gradient, and a central flat zone having a relatively constant energy. Here, the "linear beam" refers to a beam having a width substantially smaller than the length of the beam, that is, the beam has a high aspect ratio. In conventional ELA, the beam edge is the main source of material non-uniformity. In the non-periodic pulse ELA, the edge of the beam is located outside the predetermined area, allowing the predetermined area to be illuminated through the high-hat portion of the first pulse. In addition, the energy density of the beam can be optimized to produce the most uniform starting material for the accumulation process to reduce the number of pulses required to achieve the desired material uniformity.

System for performing non-periodic pulse ELA

Figure 3A shows a non-periodic pulse ELA system. The system includes a plurality of laser pulse sources 110, 110' that operate, for example, at 308 nm (yttrium chloride; XeCl), 248 nm or 351 nm. A series of mirrors 206, 208, 212 direct the laser beam to the sample platform 180, which is capable of scanning in the y-direction. The beam is shaped to form a linear beam of, for example, about 360 nm, or about 470 nm, or about 720 nm, or any length suitable for one, two or more scans of the glass panel. The system also includes a slit 140 (which is used to control the spatial profile of the laser beam) and an energy density meter 216 for reading the reflection of the slit 140. The selective mask 228 is used to block the beam when there is no sample or when it is not desired to be illuminated. Sample 170 is placed on platform 180 for processing. In addition, a homogenizer can be used to provide a more uniform high hat beam profile. Attenuators are still used. Beam energy is controlled by direct control of the laser. Platform 180 can be a linear mobile platform with the ability to move laterally. Depending on the situation, the system can include a pulse extender 213 and a mirror 214 to generate pulses of extended pulse width.

The sample moving platform 180 is preferably controlled by a computing arrangement that causes the sample 170 to move in the y-plane direction and selectively move in the x and z directions. Accordingly, the relative position of the equipment control sample 170 relative to the illumination beam pulse is calculated. The repetition rate and energy density of the illumination beam pulses are also controlled by the computing equipment. Those skilled in the art will appreciate that in addition to the beam sources 110, 110' (e.g., pulsed excimer lasers), the illumination beam pulses may be other semiconductors suitable for at least partially melting (and possibly completely melting the entire thickness) of the sample 170 (e.g., germanium). The known short energy pulse source of the selected region of the film is produced in the following manner. Known sources may be pulsed solid state lasers, continuous chopping lasers, pulsed electron beams, and pulsed ion beams. Typically, the illumination beam pulses produced by beam sources 110, 110' provide a sample-level beam intensity of 400 mJ/cm ² to 1 or 1.5 mJ/cm ² or more, 10 to 300 nanoseconds. Wide (full width at half maximum (FWHM)), and pulse repetition rate from 10 Hz (Hz) to 300 Hz to 600 Hz or 1.2 kHz (kHz) or more.

The example system of Figure 3A is used to perform semiconductor film processing of sample 170 in a manner that is further detailed below. The exemplary system of the present invention can use masks/slits to define the resulting masked beam pulse profile and reduce the non-uniformity of the contiguous and edge regions of the portion as the semiconductor film portion is illuminated by the masked beam pulse and then crystallized.

For example, a linear beam width for a non-periodic pulse ELA process is from about 100 or less to from 300 microns to about 400 to 600 microns or more. The fluence of the ELA beam is chosen not to cause the film to melt completely. Therefore, the fluence of the ELA beam should be less than about 5% to 30% or more of the fluence value that initiates the complete melting of the particular film. The fluence value that initiates complete melting depends on the film thickness and pulse width. Additionally, the ELA beam has a lower repetition rate of from about 300 Hz to about 600 Hz. The disclosed high power laser provides sufficient energy per pulse to provide a suitable energy density throughout the length of the illumination zone to cause the pulse to melt the film in this region.

The ELA linear beam can be generated by a low frequency laser source, for example, from JSW (Japanese Steel Works, Ltd.), Gate City Ohsaki-West Tower, 11-1, Osaki 1-chome, Tokyo, Japan. , Shinagawa-ku) specific system. High-frequency lasers (such as those taken from TCZ) are not suitable for non-periodic pulsed ELA processes, as the required scan speed (which is subject to pulse repetition rate and TFT or circuit pitch) will become very fast.

As shown in FIG. 3B, the semiconductor film 175 of the sample 170 is placed directly on, for example, the glass substrate 172, and one or more intermediate layers 177 may be disposed therebetween. The semiconductor film 175 has a thickness of 100 angstroms ( ) to 10000 (1 micron) as long as at least a portion of the desired area is completely melted at least partially or entirely.

According to an exemplary embodiment of the present invention, the semiconductor film 175 is composed of germanium (e.g., amorphous germanium film), germanium, germanium (SiGe), etc., which preferably contains a small amount of impurities. The semiconductor film 175 can also be made of other elements or semiconductor materials. The intermediate layer 177 immediately below the semiconductor film 175 is composed of a mixture of cerium oxide (SiO ₂ ), cerium nitride (Si ₃ N ₄ ), and/or an oxide, a nitride or other materials.

Figure 4 illustrates an exemplary cross-section of beam pulse 200, which may also be shaped by the optics of the 3A system and/or produced by a mask. In this exemplary embodiment, the energy density of beam pulse 200 is contour 220, where the energy density is less than the full melting threshold, i.e., the beam pulse energy density at which the film is completely molten. In particular, the profile 220 includes a top portion 205, a leading edge portion 210, and a trailing edge portion 215. The top 205 of this embodiment extends a width C in which the energy density is nearly constant. The width C is between 100 microns and 1 millimeter (mm). The leading edge portion 210 extends a distance D1 (e.g., 50 microns to 100 microns) and the trailing edge portion 215 extends a distance D2 (e.g., also from 50 micrometers (μm) to 100 μm). The leading edge portion 210 has a section of length D1P that extends from a nearly constant point of energy density to a low point of crystallization threshold, i.e., the beam pulse energy density at which the film crystallizes. Similarly, trailing edge portion 215 has a section of length D2P that extends from the crystallization threshold point to an almost constant high point of energy density. The top 205 is often referred to as the "high hat" portion of the beam.

The system also includes a plurality of projection lenses whereby multiple sections of the film are scanned simultaneously. A system capable of simultaneously scanning a plurality of sections of a film is disclosed in U.S. Patent No. 7,364,952, entitled "System and Method for Processing Thin Films." Although the method and system use a dual laser source, additional lasers may be employed.

The non-periodic laser pulse pattern is preferably obtained by offsetting a plurality of lasers of the same repetition rate. As described above, the computer system controls the laser to produce the pulse energy profile shown in Figure 2B-2C. As noted above, although the disclosed embodiment uses two laser tubes, the non-periodic pulse ELA can employ more than two laser tubes. For example, three, four, five or more laser tubes each emitting individual laser pulses and providing up to three, four, five or more illuminations of each film portion during each scan.

Film 170 can be an amorphous or polycrystalline semiconductor film, such as a tantalum film. The membrane can be a continuous membrane or a discontinuous membrane. For example, if the film is a discontinuous film, it can be a lithographic patterned film or a selectively deposited film. If the film is a selectively deposited film, it may be a film that is subjected to chemical vapor deposition, sputtering or solution processing, such as inkjet printed ruthenium-based ink.

Non-periodic pulse ELA method

Figure 5A shows a non-periodic pulse ELA process. Figure 5A shows an exemplary film that has been illuminated with two sets of double laser pulses, with the first two laser pulses appearing intensively, experiencing a delay (during which period, the substrate continues to move in the -y direction indicated by arrow 980), followed by the next two Laser pulses also appear intensively in time. The process includes at least four illumination steps, wherein the two illumination steps (steps 1 and 3) correspond to pulses from the primary laser, and the second illumination step (steps 2 and 4) corresponds to pulses from the secondary laser.

Figure 5A illustrates the film 175 of the sample 170 being continuously moved relative to the optics shaped by the optics of the 3A system and/or by the pattern of the mask patterned linear beam 164. Fig. 5B is an exploded view of the area 590 of Fig. 5A. In the illustration of the semiconductor film 175 on the illuminating sample 170, the sample 170 is moved in the negative y direction (arrow 980) with respect to the direction of the linear beam 164. When the sample 170 is moved in such a manner that the linear beam 164 is directed toward the first row 510 of film 175, the beam source 110 is activated by the computing device to illuminate at least a portion of the first linear beam pulse 410 from the primary laser source 110. One or more portions 511-519 of the first row 510 of semiconductor film 175 are fused. The outline and length of the first linear pulse 410 shown in Fig. 5 substantially corresponds to the contour and length of the pulse 200 shown in Fig. 4. The width C of the high hat top 205 of the first pulse 410 is preferably wide enough to illuminate and at least partially sweep the entire cross-section of portions 511-519 within the region 910. These sections can be designed to place specific structures (such as TFTs) so that they can be used to define pixels. The partially melted resolidified portion may have small grain areas, but includes a relatively uniform material therein. The molten portions 511-519 are re-solidified and crystallized to contain uniform grain growth.

Second, the second linear beam pulse 410' from the secondary laser source 110' illuminates the film 175 to initiate partial melting of the film 175. The high hat portion of the second linear beam pulse 410' illuminates the second region 920 of the film 175 to partially fuse the entire cross-section of the portions 511-519. As shown in FIG. 5, the region 910 and the region 920 largely overlap and form a first crystal region 960. In the non-periodic pulse ELA process, the overlap of the first region and the second region is greater than 70%, greater than 85%, greater than 90%, greater than 95%, or greater than 99%.

After illuminating and partially melting the first row 510 with the linear pulses 410, 410', the sample 170 is moved in the negative y direction (by computational equipment control, the beam 164 is directed at the second row 520 of semiconductor film 175 on the sample 170). As with the first row 510, upon reaching the second row 520, the primary laser source 110 is activated by the computing equipment to produce a third line pulse 420 from the primary laser that is illuminated in substantially the same manner as the first row 510 of illumination described above. And at least partially or completely melting one or more sections 521-529 of the second row 520 region 940. Next, a fourth linear beam pulse 420' from the secondary laser source 110' illuminates the film 175 to initiate partial melting of the film 175, including sections 521-529. The high hat portion of the fourth linear beam pulse 420' illuminates the fourth region 950 of the film 175. As shown in FIG. 5, the third region 940 and the fourth region 950 are largely overlapped to form a second crystal region 970. In the non-periodic pulse ELA process, the overlap of the first region and the second region is greater than 70%, greater than 85%, greater than 90%, greater than 95%, or greater than 99%.

The movement of the sample 170 is performed at a distance D (such that the linear beam 164 is directed from the first row 510 of the semiconductor film 175 to the second row 520). The distance D is also referred to as the pixel line periodicity or the pixel pitch, and the sample 170 is also subjected to the movement of the distance D by the sample 170.

The movement of the sample 170 relative to its impact by the beam 164 may continue (if not stopped). The computing equipment can control the lasers 110, 110' to generate corresponding pulses 410, 410', 420, 420' at predetermined frequencies. Thereby, the continuous moving speed V of the sample 170 with respect to the semiconductor film 175 at the linear pulse 410, 410', 420, 420' can be defined, such that the pulse can accurately illuminate the rows 510, 520 of the film 175. For example, the moving speed V of the sample 170 can be defined as follows: V = D x f _laser , where the f _laser is the respective pulse frequency. Therefore, if the distance D is 200 μm and the f _laser is 300 Hz, the velocity V is about 6 cm/sec, which can be a fixed speed.

Although the sample 170 does not necessarily continue to move with respect to its beam 164, the primary laser source 110 and the secondary laser source 110' can be activated in response to position signal control provided by the mobile platform 180. This signal may indicate the position of the sample 170 relative to its location at which the linear beam 164 is being fired. Based on the signal related information, the computing device can control the activation of the laser sources 110, 110' and the movement of the sample 170 to effectively illuminate a particular portion (e.g., row) of the semiconductor film 170. Thus, positional control illumination of at least a portion of the semiconductor film 175 can be achieved using the linear beam 164.

The four shots are partially melted, and the molten region is then rapidly solidified to form a crystalline region. The region of the film 175 where the first region 910 overlaps the second region 920 forms a first crystalline region 960. The region of the film 175 where the third region 940 overlaps the fourth region 950 forms a second crystalline region 970.

The film velocity and the repetition rate (frequency) of the first and second laser pulses determine the location of subsequent crystalline regions on the film. In one or more embodiments, the first and second crystalline regions 960, 970 may also overlap, and if the film is scanned in the y direction, the entire film surface will crystallize.

As shown in Fig. 5A, the first and second crystal regions 960, 970 do not overlap. Thus, the non-periodic pulse sequence can be used to selectively crystallize specific regions, such as pixel TFTs or circuits 511-519 and TFTs or circuits 521-529 of active matrix devices such as displays or sensor arrays. In this SAC embodiment, there is no overlap between the first and second crystalline regions 960, 970. Since there is no overlap, the platform holding the sample can move at high speed to increase the spacing between the first and second crystalline regions 960, 970 to match the periodicity of the matrix type electronic device. Speeding up the platform can dramatically increase overall processing throughput. For example, in a pixel array of a display, the density of the electronic device is very low, for example, a pixel pitch of several hundred micrometers or more, such as more than 1 mm or more, so that by crystallizing only these regions, the yield can be greatly increased. Thus, for a particular laser pulse rate, the platform can move at a faster rate to achieve a selected area on the fully crystalline film. Example yield values for SAC non-periodic pulse ELA systems can be found in the "Examples" section of this application. The increased production of non-periodic pulsed SACs can make large panel production as required for large TV manufacturing more competitive, such as the eighth generation panel (~2.20 x 2.50 ^m2 ).

Figure 6 depicts a scan similar to the scan of Figure 5A except that the energy density of the first and third linear beam pulses 1000, 1010 is lower than the energy density of the second and fourth linear beam pulses 1020, 1030. This figure corresponds to the energy density shown in Figure 7C. The energy density is from about 20% to about 70% of the full melting threshold. Generally, in non-periodic pulsed ELA, the first melting and curing cycle can be optimized to provide the most uniform crystal structure to benefit from the additive process of ELA, thereby forming a material that is very uniform and has a low defect density. For example, the energy density of the first pulse is greater than the full melting threshold. The high energy density can be easily achieved, for example, by simultaneously igniting the first two pulses to cause only a single melting and curing cycle (i.e., no difference). Similarly, the first two pulses can be triggered with a small amount of delay to form a longer pulse width combining pulse, which further benefits the uniformity of the partially molten material, especially the amorphous material (a-Si) whose starting material is PECVD deposition. The condition of the membrane.

Figure 7 illustrates the first non-periodic pulse scan as shown in Figure 5A, and includes a second scan opposite the direction of the film 1100. In the first scan of FIG. 7, when scanning in the first direction 1120, five regions 1110, 1112, 1114, 1116, 1118 are illuminated. As shown in FIG. 5A, each of the five regions 1110, 1112, 1114, 1116, 1118 corresponds to a region illuminated by the first linear beam pulse 1122 and a region illuminated by the second linear beam pulse 1124. Each irradiation causes partial melting of the irradiated area and subsequent crystallization. The overlap region generated between the region illuminated by the first linear beam pulse 1122 and the region illuminated by the second linear beam pulse 1124 corresponds to the first region 1110. After all of the five film areas have been scanned for the first time, the film is moved in the positive x direction and a second scan is performed in the direction of arrow 1130 in the opposite direction to the first scan. The conventional multi-scan ELA technique is described in the patent application WO 2010/056990 entitled "Systems and Methods for Crystallization of Thin Films". In some embodiments, the film does not move in the x direction prior to scanning, or the film moves in the negative x direction between the first and second scans. As shown in Fig. 7, the second scan produces illumination areas 1132, 1134, 1136, and the like. Multiple scans provide a better quality crystalline film. The membrane can be scanned once, twice, three times, four times, five times or more.

Therefore, the non-periodic pulse ELA system can perform multiple scans to achieve a predetermined number of pulses. For example, a four-beam system can be used for five scan processes, resulting in a total of 20 pulses per unit area of the membrane. This technique precisely controls the pulse energy sequence of each membrane segment. For example, in a non-periodic pulse ELA, during the first scan, the first pulse of each pulse train may have a lower fluence than the subsequent scan. In some embodiments, the pulse of the final jet surface has a low energy density to initiate surface melting to reduce the surface roughness of the ELA treated film. In addition, the segments of the pixel TFT or circuit or any of its components may have exactly the same sequence of pulse energy densities since they completely avoid beam edge illumination. Avoiding the beam edge impinging on the predetermined area means that the accumulation process can be more quickly concentrated into a material having a predetermined uniformity, so that the total number of pulses for this material can be reduced compared to the conventional ELA process. Therefore, the benefits of this method will be doubled: the average number of pulses is reduced by selective region crystallization, and after the first pulse, the initial unevenness of the material is reduced by avoiding beam edge illumination, thereby reducing the number of pulses in the predetermined region.

Compared to the aforementioned ELA method, the beam width in the non-periodic pulse selective area crystallized ELA tends to be small; it only needs to be as wide as the width of the area to be crystallized. Therefore, excess energy can be used to increase the beam length. Long beam lengths can be obtained using large diameter projection lenses. Again, the beam can be split into individual optical paths whereby multiple regions of the film are simultaneously crystallized during beam pulse scanning. Increasing the length of the processing area during scanning reduces the total number of scans required to completely crystallize the film.

In addition, the selective region crystalline non-periodic pulse ELA can be used to precisely align the high hat portion of the beam such that the predetermined region is not illuminated by the trailing edge of the beam. Ideally, the first illumination of the predetermined area should utilize the high hat portion of the beam, or at least a portion of the linear beam having a similar energy density greater than the film crystallization threshold. Accordingly, by selectively illuminating the film so that the edge of the beam does not illuminate the predetermined film region, the number of scans required to produce the necessary microstructure and uniformity within the film can be reduced.

In some embodiments, the optics are used to split the beam into two or more line beams, each directed to another column of pixel TFTs or pixel circuits (or at least to the location where the pixel TFTs or circuits are later fabricated). In this manner, the use of a beam split into a two-line beam will double the number of pulses per unit area, so complete crystallization can be achieved with fewer scans. A plurality of parallel line beams can be used to shoot adjacent columns of pixel TFTs/circuits or to shoot non-adjacent columns. A plurality of linear beams can be generated in a known splitting manner and directed to individual optical tracks. The splitting can also be recombined while passing through a partial optical path, for example by a projection lens, or even immediately after splitting. The beam splitting may be parallel to each other and/or at a slightly offset angle to each other. To split the beam and maintain the beam length, a beam of approximately 1/m width will be produced, where m is the number of line beams.

The special parameters of the non-periodic pulse ELA method depend on the beam width, which in turn depends on the width of the crystalline region. For example, the size of the active matrix device means a specific pixel size. The pixel size results in a new pixel layout that utilizes non-periodic ELA processing capabilities. For example, a 55-inch display with a 660 μm pixel pitch requires only a crystalline region with a width of 300 μm. Further reduction in pixel size (eg for ultra high definition displays) and optimized layout design make it more suitable for non-periodic ELA crystallization, reducing the area size to, for example, less than 150 μm. Optimization further includes having two adjacent columns having different pixel layouts: adjacent columns of TFTs/circuits can be placed close together so that they overlap within a single illumination, and then the distance traveled to the next area to be illuminated may be greater.

In addition to the pixel TFTs, the TFTs are also placed around the display to, for example, fabricate row and column drivers. Line drivers require high performance to handle image signals. In some embodiments, the SAC provides sufficient area of crystalline material to integrate the intended driver around the display. In other embodiments, the non-periodic pulse ELA is followed by an individual crystallization step whereby the periphery of the display is more completely crystallized. This can be achieved by the same laser and optical path as the conventional scanning ELA performed in these areas. Alternatively, this can be accomplished by solid state lasers that are shaped to form a narrow line beam for continuous lateral crystallization (SLS) or ELA. Alternatively, use a two-dimensional (2D) projection illumination tool such as a double-shot SLS (ie, two laser pulses per unit area, which can be found on October 31, 2008, entitled "Using a high-frequency laser to uniform U.S. Patent Application Serial No. 12/063,814, or SLS (i.e., SLS using dot pattern masking, in the System and Methods for Uniform Sequential Lateral Solidification of Thin Films Using High Frequency Lasers). See U.S. Patent No. 7,645,337, issued January 12, 2010, entitled "Systems and Methods for Creating Crystallographic-Orientation Controlled Poly-Silicon Films". . These can be integrated into the same tool and benefit from a precision platform. Here, the x-irradiation process means that each target film region is irradiated x times.

As described above, selective region crystallization involves crystallizing only predetermined regions such as matrix type electronic devices or circuits. The position of the crystalline region needs to be aligned with the node location of the matrix type electronic device or circuit. Therefore, for the implementation of SAC, sample alignment technology should be adopted. The sample alignment step can be achieved in accordance with various techniques. In one technique, a crystallization system can be utilized to establish sample alignment that is more capable of locating a sample in a manner that reproducibly positions the sample during other processing steps in the fabrication of the electronic device. A common method is to detect the crystallization process and align it before the crystallization, when the panel is provided with a reference point or an alignment mask. This sample alignment method is commonly used in lithography procedures to fabricate thin film transistors that cover different features of the device for sub-micron accuracy. SAC sample alignment does not need to be as accurate as lithography. For example, each side of the crystalline region may be a few microns or 10 microns or more larger than the predetermined area.

In another technique, the position of the crystalline region is detected prior to fabrication of the electronic device to establish sample alignment. The positioning can be achieved by detecting the area of the electronic device to be set. The area can be detected because the phenomenon of changing from amorphous to crystalline can be observed by a microscope, which is caused by a change in optical properties.

The system for sample alignment can include an automated system for detecting a reference point and aligning the sample relative to the reference point to a known location. For example, the system can include computing equipment to control the moving and responsive optical detector that detects the reference point on the film. The optical detector is, for example, a charge coupled device (CCD) camera.

Uniform partial melt crystallization of PECVD amorphous Si film

As noted above, the partial melt crystallization technique crystallizes the ruthenium film with one or more shots, wherein at least the last pulse does not initiate complete melting of the film. In some embodiments, a partial melt flooding method is used to fabricate a uniform fine grain crystalline film, or to fabricate a precursor film for a non-periodic pulsed illumination method. The partial melting flooding method can be a partial melting process of double exposure, wherein an amorphous ruthenium film (such as a PECVD film) without any pre-existing crystallites is converted into a uniform fine-grained crystal film through two steps, and the average of the crystal grains The lateral dimension exceeds the film thickness. The partial melting flooding method can also be a partial melting process for prolonging the pulse width single-shot, wherein an amorphous ruthenium film (such as a PECVD film) which does not contain any pre-existing crystallites is converted into a uniform fine-grained crystal film, and the average of the crystal grains The lateral dimension is less than the film thickness.

Professor James Im's work shows that a single-illumination process with an energy density close to the full melting threshold may trigger super lateral growth (SLG), resulting in "almost complete melting" (Im et al., APL 63, 1993, p1969), resulting in low crystals. The grain size of the intragranular defect density grows laterally. This material can be used to manufacture TFTs with mobility up to 100 cm ² /Vs or more. However, the uniformity of the TFT of this material is very poor, because the grain size is easily affected by (1) pulse energy density, (2) heterogeneity of the precursor film, and (3) when a completely amorphous film is used, The randomness of the crystal nucleation process. However, multiple exposures in the SLG required regime produce more uniform grains. This is attributed to the formation of a periodic surface roughness commensurate with the wavelength of the illumination light in the film, which in turn results in a self-stabilizing process. This approach has been commercialized as an ELA, which most often uses a linear beam. As described above, the ELA process is an accumulation process in which an initially uneven polycrystalline film is concentrated into a more uniform state by multiple irradiations in an almost complete melting range. However, if the initial polymorphism is uniform, the ELA process will be more efficient.

As described above, with the UGS system or the non-periodic pulsed ELA system, a more uniform polycrystalline film can be obtained in which the predetermined region is not irradiated by the beam edge. However, even the region that is initially irradiated by the high-hat portion of the beam may have a problem of unevenness due to the heterogeneity of the precursor film. If it is a completely amorphous film, there is a random problem of the crystal nucleation process. SUMMARY OF THE INVENTION The present invention is directed to methods and systems for performing partial melt crystallization to produce an initially uniformly crystalline polycrystalline film that facilitates the efficiency of the ELA processes described above (known and non-periodic pulses). In other embodiments, the resulting uniformity-enhanced PMC material itself can be used to fabricate thin film electronic devices without further processing with ELA. This is advantageous for low performance thin film devices (e.g., less than 100 cm ² /Vs or 10 cm ² /Vs) which are sufficient but film uniformity is still critical.

Previously published by Im and Kim on October 4, 1993 in Appl. Phys. Lett. 63, (14) "Phase transformation mechanisms involved in excimer laser The crystallization of amorphous silicon films) refers to the application of partial melt crystallization (i.e., crystallization when the energy density is lower than the almost complete melting threshold) to the LPCVD deposited amorphous Si film. This study indicates that the LPCVD Si film is not completely amorphous, and small crystallites exist in the film as crystal seeds. Due to the high crystallite density, the lateral spacing of the crystallites is extremely small, and the crystal growth mainly occurs in the direction of the vertical film surface. Since the material has a small crystal grain, it is suitable for manufacturing a uniform TFT. The single-crystal crystallization of the LPCVD film is one of the UGS methods, which is performed using a floodlighting tool, which enables the platform to be synchronized with the laser pulse irradiation (see the name of the film area on the laser crystallization processing substrate to reduce the edge area). Process and system for laser crystallization processing of film regions on a substrate to minimize edge areas, and a structure of such film regions, and US Patent Application Publication using a two-dimensional (2D) projection system No. 2006-0030164 A1, and Processes and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type, "Processes and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type""Beam, and Structures of Such Film Regions" and U.S. Patent Application Publication No. 2007-0010104 A1). It may be a method of manufacturing an LTPS device and has an extremely high throughput. This device is currently applied to UD-LCD TV products (eg, approximately 2000×4000 pixels, 480 Hz, and 80”). For this reason, the performance level of amorphous germanium is insufficient (compared to n up to 30 or even 50 cm ² /Vs) Channel UGS TFT, n-channel a-Si TFT only up to about 1 cm ² /Vs).

PMC microstructures with very small crystal columns are generally not achievable in this partially molten energy density environment. Studies have shown that some of the melt crystallization currently known cannot be reproduced for the fabrication of uniform small-grain LTPS TFTs. Mariucci et al. ( Thin Solid Films 427 (2003) 91-95 ), for example, exhibit materials that are quite heterogeneous, localized, and have many defects (defective cores that are surrounded by large and uniform grains by lateral growth).

Figure 8A shows the atomic force microscopy (AFM) scan of the membrane surface after one irradiation at the lower end of the necessary environment of the PMC. It shows a disc-shaped structure surrounded by a large protrusion, which represents lateral growth due to the solidification expansion of Si and corresponding lateral mass flow. Figure 8B is a schematic view of the crystal structure of Figure 8A. The crystal structure of Fig. 8B has a defect core 800. This structure is caused by a low nucleation density, which is a laterally crystallized seed crystal and produces a disk-shaped structure. The initial lateral conditions are extremely unbalanced. It is because the crystal has many defects. As the growth front approaches each other, sufficient heat is released to effectively reheat the membrane. Reheating causes a low defect density to grow laterally.

Figure 8C depicts AFM scanning of the film surface after one irradiation at a higher energy density, but still within the necessary environment of the PMC. Fig. 8D is a schematic view showing the crystal structure of Fig. 8C. Here, the higher energy density illuminates the introduced heat to re-melt the defective core region formed in the initial phase of the phase change. The melting threshold of the defect core region is less than the melting threshold of the low defect density outer ring and will therefore melt first. By growing at this energy density, seeding can be carried out from the outer ring and proceeding inward. The seed crystal will form a small protrusion at the center due to the solidification of Si. These protrusions can be seen by the AFM scan of Figure 8C. Remelting the defective core region produces a more uniform film. Fig. 8D illustrates a crystal structure obtained by an energy density sufficient to completely melt the film. Fig. 8E shows an annular region which is formed after the lateral crystallization of the unmelted seed crystal.

The secondary melting of the defective core region is affected by the temporal profile of the laser pulse. For example, excimer lasers taken from Coherent, Inc. (Santa Clara, Calif.) have a temporal profile showing intensity peaks. The first peak causes the film to begin to blast crystallize, and the second peak causes selective remelting of the defective core region formed in the initial stage. The time profile of the laser is known to vary over time, especially as the aging of the laser gas changes. Eventually, as time passes, a third intensity peak will appear. Thus, although the material becomes more uniform after remelting the core, it is not easily reproducible between multiple pulses of the laser tool. Other lasers have only a single intensity peak and the details of remelting within the same pulse may be different.

One way to improve the reproducibility of microstructure is to illuminate the film twice. The first pulse can be optimized to obtain the defective core material, and the second pulse can be optimized to re-melt and thereby clean the core region. This is accomplished by using a secondary scan or a step and illumination procedure in which the regions are illuminated with two pulses before the platform is moved to the next position.

The present invention is directed to a system for providing a secondary illumination partial melt crystallization process in a more efficient manner (i.e., a single scan). A non-periodic pulse ELA system is used to generate a first laser pulse of a secondary process to obtain an intermediate microstructure having large grains but poor film uniformity, and a second pulse is used to clean the intermediate microstructure to produce a final uniformity membrane. The method of the present invention teaches delaying the triggering of a second pulse (and perhaps controlling the fluence of the first or second pulse) to achieve an optimal energy density window to re-melt the core region. Delay triggering has been previously proposed, followed by simulated pulse width extension, and no light loss due to the mirror. The overlap may occur due to the pulse abutment, meaning that when the second pulse arrives, the film is not completely cooled, or may not even fully cure, so that the energy density is utilized more efficiently. Additionally, the energy densities of the first and second pulses may be the same or different. However, before the arrival of the second pulse, the membrane may not be completely cooled, so the membrane may suffer from a second pulse that may be different from the first pulse.

The initial film thickness on SiO ₂ coated glass, quartz or oxidized Si wafers is typically from about 40 nm to 100 nm, or even up to 200 nm. The film is usually preferably thin, as this reduces the deposition time and reduces the energy density required to reach a predetermined degree of melting. The pulse width of the pulse can be about 30 ns or more for the FWHM, such as up to about 300 ns or more for the FWHM. Generally, short pulses can melt the Si film more efficiently, and the heat loss to the underlying substrate is less, resulting in higher yield. The film can be illuminated using the entire partial melting energy density range.

In another embodiment, when a film containing no crystallites (such as those obtained using PECVD) is used, it is preferable to completely avoid the disk-shaped region. By increasing the nucleation density, disc-shaped regions can be avoided. High nucleation density will cause vertical crystallization processes and reduce lateral growth and lateral mass flow. By transferring to a long pulse width, a high nucleation density can be achieved, and the use of a long pulse width will cause the amorphous Si melting front to move more slowly. From Figure 9 interface response function (IRF) visible (solid-liquid interface which depicts the relative velocity of the temperature), which is further subcooled temperature than the melting temperature of the crystalline Si T ^x _m. The IRF of Figure 9 shows the temperature (x-axis) and the velocity of the crystal front (y-axis). The solidified zone is in the positive y zone of the pattern and the molten zone is in the negative y zone of the pattern. The dotted line corresponds to an amorphous crucible, and the solid line corresponds to a crystalline crucible.

For long pulses 900 with slow melting characteristics, nucleation will begin quickly and then reach extreme supercooling conditions as indicated by punctuation 905 on the amorphous Si IRF curve. From the classical nucleation theory, extreme supercooling causes high nucleation rates. Therefore, a large number of crystal nuclei will be formed in a short time before the crystal nucleus begins to grow and fuses and exotherms, so that the film is reheated (referred to as a recalescence phenomenon). High nucleation density virtually eliminates lateral growth in the region, as nucleation growth will proceed in the vertical direction. Substantial lateral growth produces an inhomogeneous structure and an uneven surface. By using a long pulse width pulse, which gives less energy per unit time to the film, a film similar to (some) LPCVD film can be obtained in which a high crystallite density is pre-stored.

On the other hand, when the short pulse 910 is used, the melting front will move quickly and will not be too cold. This condition shows the punctuation 915 corresponding to the IRF. Although the supercooling condition is less than that of the long pulsed film, it is still sufficient for nucleation, albeit at a slower rate. Therefore, fewer crystal nuclei will be formed in a short time interval before a significant re-glow phenomenon occurs, so that the film is further heated to a temperature at which further nucleation is stopped. Due to the low nucleation density, such films will undergo more lateral growth resulting in heterogeneous crystal growth.

Generally, the excimer laser pulse is short enough to generate a short pulse situation, but using an 8x pulse extender (to generate a pulse of about 300 ns for the FWHM), the pulse will become long enough to develop a growing pulse scenario. Alternatively, the extension pulse can be generated by a plurality of laser tubes, each ignited in a short sequence to initiate a single melting and curing cycle.

Therefore, a homogeneous crystal film can be obtained by a single pulse partial melting process using a long pulse having a slow melting property. This film can be used as a precursor film for conventional or non-periodic pulsed ELA processes.

Complete melt crystallization

In another aspect, the illumination in the environment necessary for complete melting is used to produce a uniform fine grain crystalline film, or to fabricate an initially crystalline polycrystalline film that is beneficial to subsequent accumulation processes. Complete melt crystallization (CMC) is a technique in which a Si film is completely melted by single irradiation, and then the film is permeable to nucleation crystallization (see the process and system for filming the film area on a substrate for laser crystallization to provide substantial uniformity and film). USSN 10/525, 288) of the process and system for laser crystallization processing of film regions on a substrate to provide substantial uniformity, and a structure of such film regions. CMC is one of the UGS methods, which is performed using a floodlighting tool that synchronizes the platform with laser pulse illumination (see the process and system and film for the area of the membrane on the laser crystallized substrate to reduce the edge area). Process and system for laser crystallization processing of film regions on a substrate to minimize edge areas, and a structure of such film regions, and USSN 10/525,297 using a two-dimensional projection system, and the name "using a linear beam of light "Processes and Systems for Laser Crystallization Processing of Film Regions on a Substrate Utilizing a Line-Type Beam, and Structures of Such Film Regions" and using a linear beam USSN 11/373, 772) of the ELA system.

The disclosed CMC method focuses on a small equiaxed grain Si film that promotes heterogeneous nucleation of the film to form less defects. The system uses high energy density pulses, for example, greater than 1.3 to 1.4 times the membrane complete melting threshold. This treatment is carried out in ambient air or in any oxygen-containing atmosphere. The process is carried out using a film having an oxidized surface layer or a cover layer having a thickness of less than about 50 nm. The system employs a relatively long pulse width of from about 80 ns to about 500 ns (e.g., 200 ns or 400 ns) combined with a relatively thin Si film (100 nm to 300 nm) on SiO ₂ glass, quartz wafer. By selecting process parameters to initiate a predetermined heterogeneous nucleation scenario, rather than a homogeneous nucleation scenario taught by prior art techniques, nucleation can be performed at the interface between the membrane and the oxidized surface layer and the membrane to the substrate. With the above parameters, a low defect density crystal can be formed.

The CMC method described can be used to fabricate low performance LTPS devices with very high throughput. This device is currently applied to UD-LCD TV products (eg, approximately 2000×4000 pixels, 480 Hz, 80”). For this reason, the performance level of amorphous germanium is insufficient (compared to n up to 30 or even 50 cm ² /Vs) Channel UGS TFT, n-channel a-Si TFT only up to about 1 cm ² /Vs).

It is known that complete melting produces various nucleation-initiating microstructures depending on the conditions of the irradiation and the conditions of the sample construction; the process description can be found in S. Hazair et al ., Mater. Res. Soc. Symp. Proc. Vol. 979 (2007) . The article "Nucleation-Initiated Solidification of Thin Si Films". This microstructure has a high degree of heterogeneity (variable grain size, multiple defect areas) which will result in poor device uniformity. For example, Hazair's thesis focuses on the formation of flower-like grains (flg-Si) in which the defect core region is surrounded by a "valve"-like grain ring of low defect density.

A microstructure is a special case, which is first described in articles published by SR Stiffler, MO Thompson, and PS Peercy in Phys. Rev. Lett. 60, 2519 (1988) . This microstructure consists of small grains uniformly distributed throughout the film thickness and has a very low intra-grain defect density. This microstructure is expected to result in good device uniformity and possibly a reasonable level of device performance. This is true, even if it is applied to the gate TFT, because unlike many other ways of preparing small-grain Si (including deposition techniques), the crystal at the bottom/near has low defect density and large size. However, there are still problems behind the microstructure formation mechanism, so appropriate conditions are required to reproduce it.

The small equiaxed grain Si (seg-Si) described by Stiffler is due to homogeneous nucleation, ie, the solid nucleates throughout a large amount of liquid compared to the interface only. Stiffler is based on transient reflectance (TR) data and transient conductance (TC) data, which show a simultaneous decrease in anterior reflectance and membrane conductivity. This result is considered to indicate nucleation throughout the membrane block. For twenty years, it has become a model for explaining the presence of grains in the membrane block (ie, not surrounding the surface or bottom interface). Recently, according to the TR study, the Stiffler model is not correct.

The TR study model assumes that seg-Si is produced by heterogeneous nucleation (ie, at the interface), followed by specific volume re-growth, remelting, and resolidification of the defective core structure. The initial stage of this situation is equivalent to the formation of flg-Si, the difference is that the defect core region of the low defect density grain is remelted and resolidified to form seg-Si.

In terms of Stiffler's data, the microstructure features are based on SEM, TEM, and AFM planar topographic images. This is not enough to explain all the characteristics of the TR data. Specifically, the Stiffler model does not account for the decrease in the dorsal TR (BTR) before the fall of the front side TR (FTR), which can be observed from experiments done in a vacuum atmosphere, and the native surface is removed prior to laser exposure. SiO ₂ layer.

At present, based on the planar bottom view and TEM cross-section of the microstructure features, it can be determined that the TR drop will result in smaller grains in the bottom region of the microstructure, which appear to grow upward and become larger at the top of the film. On the other hand, BTR and FTR fall almost simultaneously to the necessary (but not sufficient) conditions for forming the seg-Si microstructure, which was first observed by Stiffler (and is further used to make the best TFT).

Generally, heterogeneous nucleation is reported to occur only at the bottom interface of the membrane. The front side TR drop begins to nucleate at the top interface of the film (ie, at/be near the surface). Then, both sides of the membrane begin to nucleate at the same time (as confirmed by the simultaneous decrease of the TR signal on the front and back sides of the TR), causing about twice as much latent heat to be released into the membrane, thus more effectively/widely remelting/resolidifying the defective core. Area. Nucleation at the surface/near nucleation requires an interface. The interface has, for example, a (native) oxide. The oxide film may be present prior to irradiation or during irradiation, in the presence of oxygen. Depending on the atmosphere, other surface reactions may occur to form an appropriate interface for nucleation. In addition, it was found that in the absence of a top layer (such as removal of native oxide) and the formation of a top layer (such as under vacuum), surface nucleation does not occur and seg-Si as observed by Stiffler does not form. . Finally, in some samples irradiated at lower energy densities, the TR signal was observed to decrease at the same time, but Stiffler seg-Si was not observed. At present, it is believed that this is due to the complete remelting of the solid formed by the nucleation of the top interface. In addition, films thinner than 100 nm can also be seen to simultaneously decrease in TR, although the latent heat within the film volume does not appear to be sufficient to effectively/extensively re-melt/re-cure the defective core regions.

Figures 10A and 10B show the results of the latest TR study. Figure 10A shows FTR and BTR of 150 nm a-Si under vacuum on a glass substrate without a surface oxide layer. The bottom line 1400 in the graph is the case where the film is illuminated. The above reticle is the reflectance value of different CMT values. The x-axis of Figure 10A represents time (nanoseconds) and the y-axis represents the normalized reflectance value. Figure 10B is similar to Figure 10A except that Figure 10B shows the results in air. Figure 10B shows that at the energy density of 1.38 CMT, the BTR signal (the sequence signal at the bottom of the graph and above the laser signal) drops before the FTR falls, and the FTR signal seems to start falling simultaneously with the BTR. Therefore, even in a non-vacuum scenario, high energy is required to obtain the seg-Si microstructure. As shown in Figures 10A and 10B, since the reflectance between the solid and the liquid differs greatly, it is easy to discriminate from the TR data when the solid begins to transform into a liquid, and vice versa. Heterogeneous nucleation can be inferred from the FTR and BTR data and the resulting microstructure (Fig. 11B). Figure 11A shows the normalized reflectance (y-axis) versus time (n-axis) for a 200 nm a-Si film in air at 1.32 CMT and under vacuum at 1.4 CMT. Curves 1500, 1510. Figure 11B is a microstructural image obtained in an air environment. Figure 11C is a microstructural image obtained in a vacuum environment. As can be seen from the second graph, Figure 11B shows that the large crystals are throughout the film thickness of 1520. Fig. 11C shows that the crystal quality in the vicinity of the surface of the film is good, but in the vicinity of the interface with the substrate 1530, it is a small crystal having poor quality. It can be seen that 3D seg-Si is obtained in air rather than under vacuum. When in air, the surface will react to form an oxide layer, so heterogeneous nucleation occurs at the surface and bottom interface, but under vacuum, heterogeneous The core only occurs at the bottom interface.

The method of the present invention is particularly useful in the fabrication of gate TFTs because, unlike many other ways of preparing small grain Si (including deposition techniques), the crystals at the bottom/near have low defect density and large size. Typical bottom gate LTPS TFTs suffer from low mobility and high leakage current. The bottom gate TFT is formed by forming a patterned metal film (gate) under the Si film and separating them by an insulating layer (gate dielectric). During laser exposure, the metal film acts as a heat sink and causes a local full melting threshold (CMT) energy density shift. It was found that the seg-Si formation conditions remained unchanged if local CMT shift was considered. For example, a metal having a thickness of 100 nm and separated from the tantalum film by an oxide film having a thickness of 100 nm, the complete melting threshold shift is usually 15% to 20% or more. Therefore, a seg-Si formation condition is irradiation at an energy density of 1.3 to 1.4 times greater than the local CMT. Care must be taken that the energy density should not be too high to avoid damage to the surrounding film that does not contain the heat sink due to agglomeration and peeling. For example, a film having a thickness of 100 nm and a top of an oxide having a thickness of 100 nm on a 100 nm metal gate is exemplified by a 1.4-fold partial full melting threshold, or about 1.61 to 1.68 times the ambient film complete melting threshold, which is less than the film damage threshold.

The experimental conditions employed by Stiffler are somewhat different from the process conditions of the present invention. Stiffler uses a short laser wavelength (30 ns, about 80 ns for the present invention) and uses a more thermally conductive substrate: SOI (Si film over 250 nm thin SiO ₂ on Si substrate) or overlaid sapphire. In general, homogeneous nucleation requires very rapid quenching. The process conditions described herein including glass substrates and long pulses will cause slower quenching, thereby reducing the likelihood of homogeneous nucleation and increasing the likelihood of heterogeneous nucleation. The oxide used by Stiffler is not thick enough to avoid rapid cooling. The glass substrate can provide a slower cooling rate than the Stiffler configuration. Therefore, the method of the present invention takes useful and feasible conditions, and the Stiffler material can be obtained through understanding of the event.

Samples made in accordance with embodiments of the present invention include forming a 100 to 300 nm Si film on SiO ₂ coated glass, quartz (or oxidized Si wafer). The excimer laser based system (308 nm) was used to illuminate the film with different pulse widths (FWHM of 30-250 ns) and energy density. In situ analysis was performed using front side and back side transient reflectance measurements. The characteristics of the irradiated material were observed by TEM. Also, see Yikang's "Vacuum Experiment Update: Microstructure analysis" (September 2, 2009).

Example

In the case of a large-sized TV, the pixel pitch is 660 μm. With a 600 Hz laser, the scanning speed can reach about 40 cm/s. This condition was achieved using a 0.8 Joule (J) pulse that was shaped to form a 100 μm x 75 cm beam for a pulse of approximately 640 mJ/cm ² and assuming a light efficiency of 60%. Next, using a four-tube laser, this requires five overlapping scans to achieve complete crystallization. For a 2.2 x 2.5 m ² panel, the crystallization time was three parallel scans x (250 cm / 40 cm / s) × five overlapping scans = 93.75 seconds. If the acceleration/deceleration time is 5 seconds, the parallel scan interval is 10 seconds, and the loading and unloading time is 60 seconds, the total process time is about 95 + 5 x 5 + 2 x 10 + 60 = 200 seconds. More conservatively, the process time can be assumed to be 5 minutes. This is equivalent to 60/5×24×30=about 8500 panels/month.

The ELA process of 20 shots (ie, 20 laser pulses per unit area) is required to trigger four laser tubes simultaneously to obtain a beam of 400 μm x 75 cm. For 20 shots, the scan speed was about 1.2 cm/s and the crystallization time was 3 x (250/1.2) = 625 seconds. If the acceleration/deceleration time is ignored, the total process time is 625 + 2 × 10 + 60 = 705 seconds. More conservatively, the process time can be assumed to be 12.5 minutes and the yield is about 3400 panels/month.

Although the present invention has been disclosed in the above embodiments, it is apparent that those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. For example, it should be understood that the manner in which the film is advanced in a selected direction may be an embodiment in which the laser beam is held stationary and the film is moved relative to the laser source, and the film remains stationary and the beam is moved.

101. . . beam

102. . . arrow

103. . . region

104. . . membrane

105, 106, 107. . . pulse

110, 110’. . . Laser source/beam source

120. . . Marginal zone

122. . . Amorphous crotch

124. . . Crystalline crotch

126. . . Transition zone

140. . . Slit

164. . . beam

170. . . sample

172. . . Substrate

175. . . film

177. . . middle layer

180. . . platform

200. . . pulse

205. . . top

206, 208, 212, 214. . . mirror

210. . . Front edge

213. . . Pulse extender

215. . . Trailing edge

216. . . Energy density meter

220. . . section

228. . . Sunshade

410, 410', 420, 420'. . . pulse

510, 520. . . row

511-519. . . Part/circuit

521-529. . . Section/circuit

590. . . region

800. . . Defect core

900, 912. . . pulse

905, 915. . . punctuation

910, 920, 940, 950, 960, 970. . . region

980. . . arrow

1000, 1010, 1020, 1030, 1122, 1124. . . pulse

1100. . . membrane

1110, 1112, 1114, 1116, 1118, 1132, 1134, 1136. . . region

1120. . . direction

1130. . . arrow

1400. . . Bottom line

1500, 1510. . . curve

1520. . . Film thickness

1530. . . Substrate

C. . . width

D, D1, D2. . . distance

D1P, D2P. . . length

The invention will become more apparent and understood by reference to the appended claims.

Figure 1A illustrates a random microstructure that can be obtained by ELA;

Figure 1B depicts a conventional ELA single scan;

2A-2C are diagrams showing exemplary laser pulse energy profiles in accordance with an embodiment of the present invention;

2D is a graph showing a plasma-assisted chemical vapor deposition (PECVD) amorphous ruthenium film by single irradiation;

3A is a diagram showing a non-periodic pulse ELA system according to an embodiment of the invention;

3B is a diagram of a sample for a non-periodic pulsed ELA system, in accordance with an embodiment of the present invention;

Figure 4 illustrates an example beam pulse profile in accordance with an embodiment of the present invention;

FIG. 5A illustrates a non-periodic pulse ELA process according to an embodiment of the invention;

Figure 5B is an exploded view of a region 590 of Figure 5A, in accordance with an embodiment of the present invention;

6 is a diagram showing a non-periodic pulse ELA process according to an embodiment of the invention;

Figure 7 illustrates a first non-periodic pulse scan as shown in Figure 5A, and further includes a second scan in the reverse direction of the film, in accordance with an embodiment of the present invention;

8A is a view showing a crystal structure of a film after one irradiation according to an embodiment of the present invention;

8B illustrates a crystal structure of FIG. 8A according to an embodiment of the present invention;

8C is a diagram showing atomic force microscopy (AFM) scanning of a surface of a film after one irradiation at a high energy density, but still in the PMC range, according to an embodiment of the present invention;

Figure 8D illustrates a crystal structure of Figure 8C according to an embodiment of the present invention;

Figure 8E shows an annular region formed after lateral crystallization from unmelted seed crystals in accordance with an embodiment of the present invention;

Figure 9 is a diagram showing an interface response function of a film according to an embodiment of the present invention;

FIG. 10A illustrates a front side transient reflectance (FTR) and a back side transient reflectance of a 150 nm amorphous germanium (a-Si) on a glass substrate having a 300 nm oxide layer under vacuum according to an embodiment of the invention. BTR);

Figure 10B is similar to Figure 10A, with the difference that Figure 10B shows the results after vacuum breaking;

11A illustrates a time (nanosecond) of a 200 nm a-Si film having a 300 nm oxide surface layer in air at 1.32 CMT and under vacuum at 1.4 CMT, in accordance with an embodiment of the present invention. a plot of (x-axis) versus normalized reflectance (y-axis);

Figure 11B is a microstructural image obtained in an air environment;

Figure 11C is a microstructural image obtained in a vacuum environment.

110, 110’. . . Laser source

140. . . Slit

170. . . sample

180. . . platform

206, 208, 212, 214. . . mirror

213. . . Pulse extender

216. . . Energy density meter

228. . . Sunshade

Claims (44)

A method of processing a film, comprising: irradiating a first region of the film with a first laser pulse and a second laser pulse while advancing a film toward a first selected direction at a fixed speed, each laser The pulse provides a shaped beam of light and has a fluence sufficient to partially melt the film such that the first region resolidifies and crystallizes to form a first crystalline region; and a third laser pulse and a fourth Ray The pulse illuminates a second region of the film, each pulse providing a shaped beam and having a fluence sufficient to partially melt the film such that the second region resolidifies and crystallizes to form a second crystalline region, wherein the first region The time interval between the laser pulse and the second laser pulse is less than half of a time interval between the first laser pulse and the third laser pulse; wherein the first laser pulse and the second laser The time interval between pulses is longer than a time interval between the single melting, crystallization and curing cycles of the film. The method of claim 1, wherein the first laser pulse and the second laser pulse each have the same energy density. The method of claim 1, wherein the first laser pulse and the second laser pulse each have a different energy density. The method of claim 1, wherein the first laser pulse and the second laser pulse each cause the film to reach the same degree of melting. The method of claim 1, wherein the first laser pulse and the second laser pulse each cause the film to reach a different degree of melting. The method of claim 5, wherein the film comprises an amorphous ruthenium film that does not contain pre-existing crystallites. The method of claim 6, wherein the first laser pulse has an energy density sufficient to melt the amorphous ruthenium film and produce a crystal structure having a defective core region. The method of claim 7, wherein the second laser pulse has an energy density sufficient to re-melt the defective core regions to form a uniform fine-grained crystalline film. The method of claim 1, wherein the film comprises an amorphous ruthenium film. The method of claim 1, wherein low pressure chemical vapor deposition, plasma assisted chemical vapor deposition, sputtering, and electron beam evaporation are utilized. One of them deposits the film. The method of claim 1, wherein the film comprises a treated ruthenium film. The method of claim 11, wherein the treated ruthenium film is an amorphous ruthenium film, the amorphous ruthenium film does not contain pre-existing crystallites, and the amorphous ruthenium film is subsequently processed according to a method comprising: While advancing the amorphous diaphragm in a second selected direction, the amorphous diaphragm is illuminated with an extended laser pulse having a fluence sufficient to partially melt the amorphous diaphragm. The method of claim 12, wherein the extended laser pulse is generated by sequentially overlapping a plurality of laser pulses from a plurality of laser sources, wherein the delay between the pulses is short enough to initiate a single melting and solidification cycle. The method of claim 12, wherein the amorphous ruthenium film is obtained by plasma-assisted chemical vapor deposition. The method of claim 12, wherein the extended laser pulse comprises a pulse length greater than a full width at half maximum of 300 nanoseconds (ns). The method of claim 11, wherein the treated ruthenium film a film processed according to a method, the method comprising: irradiating the film with a laser pulse having a laser pulse sufficient to completely melt the film while advancing the film toward a second selected direction Flux. The method of claim 16, wherein the laser pulse is generated by overlapping a plurality of laser pulses from a plurality of laser sources. The method of claim 1, comprising: irradiating a third region of the film with a fifth laser pulse and a sixth laser pulse while advancing the film in a second selected direction, each of the lasers The pulse provides a shaped beam and has a fluence sufficient to partially melt the film such that the third region resolidifies and crystallizes to form a third crystalline region; and is illuminated by a seventh laser pulse and an eighth laser pulse a fourth region of the film, each pulse providing a shaped beam and having a fluence sufficient to partially melt the film such that the fourth region resolidifies and crystallizes to form a fourth crystalline region, wherein the fifth laser pulse The time interval between the sixth laser pulse and the sixth laser pulse is less than half of the time interval between the fifth laser pulse and the seventh laser pulse. The method of claim 18, wherein the second selected direction is opposite to the first selected direction, the third region is heavier than the second region Stacked, and the fourth area overlaps the first area. The method of claim 18, wherein the second selected direction is the same as the first selected direction, wherein the third region overlaps the first region, and the fourth region overlaps the second region. The method of claim 18, wherein the film is moved in a direction perpendicular to the first selected direction prior to advancing the film in the second selected direction. The method of claim 1, wherein each of the laser pulses comprises a linear beam having a uniform energy density at a top portion of the linear beam. The method of claim 1, wherein each of the laser pulses comprises a floodlighting pulse. A system for processing a film using a non-periodic laser pulse, comprising: a primary laser source and a primary laser source for generating a plurality of laser pulses; and a working surface for fixing a film to a substrate a platform for moving the film relative to the laser pulses to construct a direction of travel of the laser pulses on a surface of the film; A computer having a plurality of processing instructions for synchronizing the platform with the laser pulses to provide a first region of a film loaded to the platform illuminated by a first laser pulse from the primary laser source, A second region of the film is illuminated by a second laser pulse from the secondary laser source, and a third region of the film is illuminated by a third laser pulse from the primary laser source, wherein An equal processing command is provided to move the film in the direction of travel relative to the laser pulses to illuminate the first region, the second region, and the third region, wherein a center of the first region to the second region The distance of the center is less than half the distance from the center of the first region to the center of the third region, and wherein the first laser pulse, the second laser pulse, and the third laser pulse have sufficient to partially melt the film A fluence. A system of claim 24, wherein the platform moves at a fixed speed. A method for converting an amorphous ruthenium film containing no pre-existing crystallite into a small-grain film, the method comprising: elongating the amorphous ruthenium film in a first selected direction while extending A laser pulse illuminates the amorphous ruthenium film, the extended laser pulse having a fluence sufficient to partially melt the amorphous ruthenium film, wherein the small crystal grain film comprises a plurality of crystal grains, and the average of the crystal grains horizontal The dimension is smaller than the thickness of the film. The method of claim 26, wherein the extended laser pulse comprises a pulse length greater than a full width at half maximum of 300 nanoseconds (ns) and is a flooding pulse. The method of claim 26, wherein the extended laser pulse is generated by delaying overlapping of a plurality of laser pulses from a plurality of laser sources, wherein the delay between the pulses is short enough to initiate a single melting and solidification cycle. The method of claim 26, wherein the amorphous ruthenium film is obtained by plasma assisted chemical vapor deposition. A method of processing a film, comprising: providing a semiconductor film on a substrate, the film having a bottom interface adjacent to a bottom surface of the substrate, and a top surface opposite the bottom surface; and a laser A beam of light illuminates the film, the laser beam having an energy density that is 1.3 times greater than a full melting threshold of the film, the energy density being selected to completely melt the film, wherein when curing begins, a coating layer is present for the semiconductor Forming a top interface at the top surface of the film, wherein the top interface and the bottom after illuminating and completely melting the film Heterogeneous nucleation occurs at the interface, and wherein once cooled, the heterogeneous nucleation forms a plurality of less defective germanium grains at the bottom surface of the film. The method of claim 30, wherein the laser beam has a pulse width greater than 80 nanoseconds (ns). The method of claim 30, wherein the laser beam has a pulse width greater than 200 nanoseconds (ns). The method of claim 30, wherein the laser beam has a pulse width greater than 400 nanoseconds (ns). The method of claim 30, wherein the semiconductor film comprises a tantalum film having a thickness of between about 100 nanometers (nm) and about 300 nanometers. The method of claim 30, wherein the substrate comprises glass. The method of claim 30, wherein the substrate comprises quartz. The method of claim 30, wherein the crystal grains comprise Multiple small equiaxed grains. The method of claim 30, wherein the energy density of the laser beam is 1.4 times the partial complete melting threshold. The method of claim 30, wherein the covering layer is formed by depositing a thin layer on the top surface of the film prior to the irradiation. The method of claim 39, wherein the cover layer comprises an oxide layer having a thickness of less than 50 nanometers (nm). The method of claim 30, wherein the cover layer is formed by illuminating the film in an oxygenated environment. The method of claim 41, wherein the oxygenated environment comprises air. The method of claim 41, wherein the oxygenated environment contains only oxygen. The method of claim 30, wherein the substrate comprises a patterned metal film covered by an insulating film, and wherein the energy density is greater than 1.3 times the full melting threshold of the film.