Method of forming a silicide region in a Si substrate and a device having same

[0002] This application is a Continuation-In-Part of an application entitled “Method of Forming a Silicide Region in a Si Substrate and a Device Having Same” filed on Apr. 12, 2000, and having Ser. No. 09/547,836 now U.S. Pat. No. 6,274,488.

[0003] 1.Field of the Invention

[0004] The present invention pertains to a method of forming a silicide region on a silicon substrate in the manufacturing of semiconductor integrated devices, and such devices having a silicide region formed by the method of the present invention.

[0005] 2.Description of Related Art

[0006] Improvements in semiconductor technology and semiconductor manufacturing are the main drivers to the reduction of cost and the increase in speed of computers. There have been many advances for increasing the speed of semiconductor devices, ranging from packaging of integrated circuits (“chips”) to the wiring of the devices on the chip, to the devices themselves.

[0007] Improvements in chip manufacturing techniques include new processes (or improving an existing process) for making the chip. Such improvements may include new, faster and more efficient ways of creating the necessary device structures. For example, many techniques have been developed for forming silicide regions for integrated device contacts, particularly for metal-oxide-semiconductor (MOS) devices formed on silicon substrates. Most of these techniques involve the formation of a metal layer over a gate, drain or source region upon which the silicide is desired to be formed. These techniques then use a thermal treatment to react the metal with the silicon composing the gate, drain and/or source regions, to form the low-resistivity silicide regions. The substrate is further processed by removing the unreacted metal layer, then forming an electrically-insulating layer over the substrate, and finally forming conductive metal lines on the insulating layer. These metal-lines can contact the silicide regions over the gate, source and drain regions to form the desired circuit connections for an integrated MOS device.

[0008] Techniques for forming suicides are subject to several stringent process constraints that must be met in order for such techniques to be effective. These constraints include: (1) the metal used to form the silicide must be carefully selected to be a species that diffuses into the silicon, to avoid the formation of leakage paths between the source, drain and gate of an integrated device; (2) for self-aligned silicidation techniques, the metal layer must not react with the insulative material composing the self-aligning side walls of the gate; (3) the dopants must not segregate into the suicide regions, which makes it difficult to achieve low contact resistance; (4) the technique should have a process window that allows the silicide region to be formed on both crystalline-silicon and poly-silicon; (5) the silicide formation should be insensitive to dopants present in the silicon; and (6) the metal atoms should not diffuse beyond the silicide regions, that is, into the silicon. If metal atoms diffuse into the silicon, junction leakage will likely increase. The simultaneous fulfillment of all of the above-stated criteria is at best difficult for most conventional silicidation techniques, especially those that use relatively extensive thermal treatments. Most often, a failure to perform the conventional technique within its relatively narrow process margins manifests itself as defects due to thermal drift of the metal atoms beyond desired boundaries during the relatively prolonged thermal treatment periods. If the silicide region extends beyond its design dimensions, it can result in-leakage paths between the gate and the source and drain or between the source and/or drain and the substrate. Therefore a great need exists for a technique that enhances silicidation process margins beyond those conventionally available.

[0009] U.S. Pat. No. 5,888,888 (the '888 patent) discloses a method of producing a suicide region on a Si substrate. The '888 patent is incorporated herein by reference. The method disclosed in the '888 patent is useful for a variety of purposes, including the reduction of the electrical contact resistance to the Si substrate or an integrated electronic device formed thereon. The method includes the step of irradiating a metal layer formed atop the source and drain regions with front side irradiation i.e., irradiating the metal layer from the front side of the silicon substrate to initiate diffusion of the metal into the silicon substrate. While this method is very useful, it would still be preferable to be able to heat the metal layer by other means that result in quicker silicidation.

[0010] The present invention pertains to a method of forming a suicide region on a silicon substrate in the manufacturing of semiconductor integrated devices, and such devices having a silicide region formed by the method of the present invention.

[0011] Accordingly, a first aspect of the invention is a method of forming a silicide region in a crystalline Si substrate having an upper surface and a lower surface. The method comprises the steps of first, forming an amorphous doped region in the Si substrate at or near the upper surface to a predetermined depth. This results in an interface being formed between the amorphous doped region and the crystalline Si substrate. This interface is referred to herein as the amorphous-crystalline interface. The next step is forming a metal layer on the upper surface of the Si substrate, in contact with the amorphous doped region. The next step is then performing backside irradiation with a first radiation beam to heat the interface to initiate explosive recrystallization (XRC) of the region. This provides heat to the metal layer sufficient to cause diffusion of metal from the metal layer into the amorphous doped region, as described in greater detail below. The diffusion of the metal, in turn, results in the silicide region being formed in the amorphous doped region at or near the upper surface of the Si substrate.

[0012] A second aspect of the invention is a method of forming a MISFET device in a Si substrate having upper and lower surfaces and silicide regions. The method comprises the steps of first, forming spaced apart shallow trench isolations in the Si substrate. The next step is forming, atop the upper surface of the Si substrate and between the shallow isolation trenches, a gate having sides. The next step is then performing a first amorphization and doping of first and second portions of the Si substrate adjacent the gate to form extension regions having a first predetermined depth from the upper surface. The next step is forming sidewall spacers adjacent the sides of the gate. The next step is then performing a second amorphization and doping of the substrate adjacent the sidewall spacers to transform portions beneath the extensions to deep drain and deep source regions having a second predetermined depth from the upper surface, and first and second amorphous-crystalline interfaces within the Si substrate. At this point, the deep drain and the extension above it can be considered a single “drain” region. Likewise, the deep source and the extension above it can be considered a single “source” region. The next step is depositing a metal layer atop the upper surface of the Si substrate over the drain and source regions. The next step is performing backside irradiation. This is accomplished using a first radiation beam incident the lower surface of the Si substrate. The energy from this beam is used to initiate XRC at the first and second amorphous-crystalline interfaces such that heat is provided to the deep drain and source regions and propagates toward the upper surface of the Si substrate and the metal layer so as to initiate diffusion of metal in the metal layer into upper portions of the drain and source regions. This results in the formation of the silicide regions in the upper portions of the drain and source regions. The silicide regions so formed have a low sheet resistance, i.e., 5Ω/sq or less. The final step is removing the metal layer that remains after the diffusion of metal is completed.

[0013] FIG. 1A is cross-sectional schematic diagram of a Si substrate with a mask layer, the substrate being irradiated with an implant beam so as to amorphize a region in the substrate;

[0014] FIG. 1B is a cross-sectional schematic diagram of the Si substrate of FIG. 1A being further irradiated with an implant beam so as to implant dopants into the amorphous region;

[0015] FIG. 1C is a cross-sectional schematic diagram of the Si substrate of FIG. 1B having a metal layer formed thereon and being further subject to at least one of front-side and backside irradiation to form a silicide region in the doped amorphous region;

[0016] FIG. 1D is a cross-sectional schematic diagram of the Si substrate of FIG. 1C, showing the resultant suicide region;

[0017] FIG. 1E is a cross-sectional schematic diagram of the Si substrate of FIG. 1C, but with a conformal layer formed atop the metal layer;

[0018] FIG. 2A is a cross-sectional schematic diagram of a Si substrate having formed therein shallow trench isolations and a gate atop the substrate upper surface;

[0019] FIG. 2B is a cross-sectional schematic diagram of the Si substrate of FIG. 2A, further including a doped amorphous region;

[0020] FIG. 2C is a cross-sectional schematic diagram of the Si substrate of FIG. 2B, further including sidewall spacers adjacent the sides of the gate;

[0021] FIG. 2D is a cross-sectional schematic diagram of the Si substrate of FIG. 2B, showing the deep dopant amorphizing implant and deep dopant implant steps;

[0022] FIG. 2E is a cross-sectional schematic diagram of the Si substrate of FIG. 2B, further including a metal layer, an optional conformal layer, and front-side and backside irradiation to initiate XRC;

[0023] FIG. 2F is a cross-sectional schematic diagram of a resulting integrated MISFET device according to the present invention; and

[0024] FIG. 3 is a graph that shows the intrinsic carrier concentration versus temperature and the absorption length for two different radiation wavelengths versus temperature for Silicon.

[0025] The present invention pertains to a method of forming a silicide region on a silicon substrate in the manufacturing of semiconductor integrated devices, and such devices having a silicide region formed by the method of the present invention.

[0026] The method of the present invention produces a silicide region on a Si substrate that is useful for a variety of purposes, including the reduction of the electrical contact resistance to the Si substrate or an integrated electronic device formed thereon. First described is a generalized method of forming the silicide region in a silicon (Si) substrate, as a first embodiment of the present invention. Then, a method of fabricating a metal-insulator-semiconductor field effect transistor (MISFET) is described as a second embodiment of the present invention. The insulator could be silicon dioxide (SiO₂) or a composite layer of SiO₂with other insulating materials and the semiconductor could be silicon or III-V compounds or Si-Ge alloys or III-V compound alloys. The metal could be, for example, doped polysilicon, aluminum, or a refractory metal.

[0027] First Embodiment

[0028] With reference to FIG. 1A, the first step of the present method is providing a silicon (Si) substrate 10 having an upper surface 12 (also referred herein as the “front-side” of the Si substrate) and a lower surface 14 (also referred to as the “backside” of the Si substrate). Si substrate 10 is subjected to processing to form a silicide region in accordance with the first embodiment of the present invention. Si substrate 10 could be, for example, a silicon wafer, a silicon-on-insulator substrate, a silicon epitaxial layer, or a poly-silicon layer on an insulator.

[0029] The second step is then forming a mask layer 16 on upper surface 12 of Si substrate 10, that is patterned to form a window 20 exposing a portion 24 of upper surface 12. Window 20 defines a portion of Si substrate 10 within which it is desired to form a suicide region. As known to persons of ordinary skill in this technology, mask layer 16 can be any one of a variety of suitable substances (e.g., photoresist or oxide) patterned using a variety of well-known techniques.

[0030] The third step is then amorphizing exposed portion 24 of Si substrate 10, preferably through an ion implantation process 30. The implanted ions destroy the chemical bonds between silicon atoms in Si substrate 10 so that the atomic structure therein is made relatively disordered and random, as compared with the portions of the Si substrate that are not subjected to such ion bombardment. Preferably, the ion species, the implantation energy and the dosage are selected to produce an amorphized region 34 such that it extends to a predetermined depth d. In general, selecting an ion species with a relatively light atomic weight, and/or increasing the ion implantation energy in ion implantation process 30 will each have the effect of increasing the depth d of the amorphous region. Conversely, selecting an ion species with relatively heavy atomic weight, and/or decreasing the ion implantation energy in ion implantation process 30 will each have the effect of decreasing the depth of amorphous region 34. A number of ion species can be used to produce the amorphous region. For example, the ion species can include silicon, argon, arsenic, or germanium. The ion implantation energy can be in a range from 10 to 100 kilo-electronVolts (keV) at a dosage in a range from 10¹³to 10¹⁵ions/cm². In a preferred embodiment, the depth d to which amorphized region 34 extends into Si substrate 10 is predetermined to be less than 1000 angstroms. To produce an amorphized region 34 having a depth d of 300 angstroms, it is preferable that germanium ions implanted at an energy of about 20 keV with a dosage of about 3×10¹⁴atoms/cm²be used. Ion implantation process 30 can be performed with a device such as the 9500 XR ION IMPLANTER™, commercially available from Applied Materials, Inc., of Santa Clara, Calif. An amorphous-crystalline interface I is formed between Si substrate 10 (which is crystalline) and amorphous region 34.

[0031] With reference now to FIG. 1B, the fourth step after the above-described amorphization step is implanting amorphous region 34 with an n-type or p-type dopant, as indicated by dopant implant beam 44, thereby forming an amorphous doped region 40. In practice, any one of a number of conventional dopant implant techniques can be used.

[0032] After performing amorphizing implant 30 and dopant implant 44, the fifth step is removing mask layer 16. This is accomplished, for example, by use of a chemical solvent, by mechanical abrasion, and/or other by techniques well known to persons skilled in the art of semiconductor processing.

[0033] With reference now to FIG. 1C, the sixth step of the present method is forming a metal layer 60 over upper surface 12 of Si substrate 10, including over exposed portion 24, which is the upper surface of amorphous doped region 40. Metal layer 60 supplies the metal atoms needed to form the desired silicide compound for electrical contact with Si substrate 10. A number of metal species can be used to form the silicide compound. For example, metal layer 60 can include titanium, cobalt or nickel used to form the silicides TiSi₂, CoSi₂, or NiSi, with resistivities of 15-20 μΩcm, 17-20 μΩcm, and 12-15 μΩ, respectively. Metal layer 60 is formed preferably by sputtering, but evaporation or chemical vapor deposition can also be used. For example, to form metal layer 60, a suitable sputtering chamber is the ENDURA VHPPVD•, commercially available from Applied Materials, Inc. of Santa Clara, Calif. Preferably, metal layer 60 is formed to a thickness “t” determined by the desired silicide thickness or depth and the proportional amount of silicon and metal consumed to form the silicide. More specifically, the thickness “t” of metal layer 60 should be at least the desired silicide depth divided by the ratio of silicon-to-metal consumed to form the silicide. Thus, to form a TiSi₂silicide region 300 angstroms in depth in which, due to the atomic structure and bonding of this particular silicide, a 2.27 angstroms thickness of silicon is consumed per 1 angstrom thickness of titanium (i.e., a consumption ratio of 2.27), the thickness “t” of metal layer 60 should be at least 300 angstroms/2.27˜130 angstroms. This ensures an appropriate stoichiometric quantity of metal will be available in metal layer 60 to form the desired silicide.

[0034] With continuing reference to FIG. 1C, a first alternative for the seventh step of the present method is irradiating lower surface 14 of Si substrate 10 with a radiation beam 66. This procedure is referred to herein as “backside irradiation”. Preferably, radiation beam 66 comprises pulsed laser light with sufficient irradiance to heat interface I to a temperature sufficient to initiate XRC of doped amorphous region 40.

[0035] XRC occurs when molten Si starts to solidify into crystalline Si from the primary melt at amorphous-crystalline interface I. The latent heat released by this solidification melts a thin layer of the overlying doped amorphous layers. Latent heat is again released during crystallization of this secondary melt. In this manner, a thin liquid Si layer propagates from the original liquid-solid interface (i.e., amorphous-crystalline interface I) to upper surface 12 of substrate 10. This process is quenched at metal layer 60, as the melting temperature of the metal layer is typically higher than that of amorphous Si.

[0036] The temperature sufficient to initiate XRC is in the range between about 1150-1410 degrees centigrade. The upper bound on the temperature range is determined by the lower of the melting temperature of crystalline Si and the melting temperature of metal layer 60. Because amorphous silicon is metastable, it recrystallizes at a lower temperature than crystalline Si, and also melts at a lower temperature than the crystalline portions of Si substrate 10 that are more atomically ordered. As described above, upon XRC, the heat that builds up at interface I propagates toward upper surface 12 of Si substrate 10 (as indicated by arrows 70) as doped amorphous region 40 re-crystallizes. This heat is then absorbed by metal layer 60, which causes metal atoms in the metal layer to diffuse (as represented by arrows 72) into a now-re-crystallized doped region 40′, thereby forming a silicide region 80 within Si substrate 10, as shown in FIG. 1D.

[0037] With reference again to FIG. 1C, radiation beam 66 is preferably applied over several pulses to ensure a silicide reaction of the metal (M) in the silicon (i.e., M+Si−MSi_x). Each radiation pulse in radiation beam 66 preferably has a wavelength between 1-15 microns. In this wavelength range, radiation will travel through crystalline Si but will be absorbed in the doped amorphized region 40′. The temporal pulse width of radiation beam 66 is preferably between 1 nanosecond to 100 microseconds. The repetition rate of the pulses making up radiation beam 66 is preferably between 1 to 1000 Hz. Between pulses of radiation beam 66, Si substrate 10 is allowed to cool to room temperature. To ensure that the backside irradiation step is performed so amorphous region 40 is re-crystallized and that Si substrate 10 and metal layer 60 are maintained in their stable phases, radiation beam 66 preferably has an irradiance (fluence) in a range from about 0.1 to 10.0 J/cm²delivered in a series of 3 to 10 pulses of 10 to 100 nanoseconds, at a repetition rate preferably between 200 Hz and 400 Hz. The exact laser fluence, number of pulses, pulse duration and repetition rate needed to operate within the process margin of the method of the present invention will vary between the different kinds of radiation sources (e.g., lasers) and the suicide thickness of suicide region 80. Accordingly, like most semiconductor processes, these parameters may need to be determined empirically.

[0038] The irradiation of Si substrate 10 is preferably performed in a chamber with an inert atmosphere of nitrogen, argon or helium. A suitable chamber is the P-GILA™ machine commercially available from Ultratech Stepper, Inc., San Jose, Calif. With proper delivery of radiation beam 66 in accordance with the above-described guidelines and principles, the temperature window over which doped amorphous region 40 undergoes XRC while Si substrate 10 and meal layer 60 remain in their solid states, occurs over a relatively large range from about 1150-1410 degrees centigrade. This represents a temperature difference range on the order of 300 degrees centigrade. Thus, the process margin available for successful performance of the method of the present invention relative to conventional silicidation techniques is greatly enhanced.

[0039] Backside radiation beam 66 can be provided by a variety of different radiation sources that have a temporal pulse width preferably between 1 nanosecond to 100 microseconds and a reasonable transmission through the substrate. The choice of laser wavelength depends on how high the dopant concentration is in the region near the top of the substrate that will melt to trigger the salicidation. It is desirable that the backside beam has an absorption length of about a micron or less in the doped region near the upper surface so that it is efficiently absorbed. With a doping concentration of 10¹⁹/cm³this condition is readily met using a 10.6 micron, CO₂laser. The use of other lasers, with wavelengths between 5 and 15 microns, may be possible, however, at this time the CO₂laser appears to be ideal and it is readily available.

[0040] FIG. 3 is a graph that shows the intrinsic carrier (dopant) concentration versus temperature and absorption length for two different radiation wavelengths versus temperature for silicon. From the graph of FIG. 3 it can be seen that it is possible to transmit backside radiation (66) having a wavelength between 1.064 and 10.6 microns through the backside of a 700 micron thick silicon substrate that has an intrinsic carrier concentration of less than 10¹⁶atoms/cm³. In this case the shortest absorption length is for 10.6 micron radiation and this is greater than 103 microns. A much higher doping concentration near the front surface would absorb the radiation transmitted through the wafer from the back-side and result in localized heating.

[0041] For the example illustrated by the graph of FIG. 3 and discussed below, a wafer with an upper surface that is heavily doped with a concentration of dopant atoms greater than 10¹⁹atoms/cm³has been used.

[0042] In the FIG. 3 chart there are three curves, each plotted against temperature in ^OK along the x-axis, each with a horizontal arrow associated with it that points to the y-axis scale that relates to that curve. The y-axis scale on the left is an intrinsic, equivalent carrier (dopant) concentration of the silicon substrate (free carriers generated thermally). The y-axis scale on the right is the absorption length in pure, undoped silicon.

[0043] More specifically, curve 250 having the diamond shaped data points shows the intrinsic effective carrier concentration (left vertical scale) for silicon increasing with temperature. For example, room temperature (273° K) produces a certain number of free electrons which are equivalent to a carrier concentration of about 10¹⁰carriers/cm³. Curve 260 having the triangular data points represents a radiation wavelength (λ) of 10.6 microns, and curve 270 having the square data points represents a radiation wavelength of 1.064 microns. Using curve 250 of FIG. 3, the intrinsic carrier concentration curve, it is seen that at room temperature, 273°K, the intrinsic carrier concentration is between 10⁹/cm³and 10¹⁰/cm³and the absorption path length (curves 260 and 270) is greater than 10¹⁹microns or 10³meters for both wavelengths.

[0044] Viewing curve 250, an equivalent carrier concentration of 10¹⁷/cm³occurs at a temperature of about 800° K. At that temperature, the absorption length of radiation with a 1.064 micron wavelength (curve 270) is about 2.5×10⁴microns (25 mm). Similarly, an equivalent carrier concentration of 1×10¹⁶/cm³occurs at a temperature of about 650° K (curve 250), and for radiation having a 10.6 micron wavelength (curve 260) the absorption length is about 2.5×10³microns (2.5 mm). With these absorption lengths it is possible to drive illumination through the back of a 700-micron thick silicon wafer to the front side without losing much on the way through.

[0045] At the front side of the wafer the implanted impurities raise the carrier concentration to above 1×10¹⁹/cm³and this increases the absorption to both wavelengths dramatically. Since this carrier concentration corresponds to a temperature of about 1400° K, the absorption path length of 10.6 microns wavelength radiation becomes about 2 microns, so most of the radiation incident at the top surface of the wafer would be absorbed in a 2 micron thick implant doped above 1×10¹⁹atoms/cm³. If reflection from the top surface is taken into consideration, then a one micron thick implant layer would effectively absorb most of the incident radiation. The absorption length for 1.064 micron radiation increases as well to about 2.5×10²microns or about 250 microns, which is insufficient to generate much heating in a thin, doped layer. Thus radiation with a wavelength in the 10.6 micron region could be driven through a lightly doped silicon wafer substrate and absorbed in a doped junction region where it is absorbed and heats the junction sufficiently to produce activation or annealing.

[0046] Of course the radiation wavelength doesn't have to be exactly 10.6 microns. This wavelength was used in the above example since radiation at that wavelength is available from commercial CO₂lasers and the properties of silicon at this wavelength are readily available. Clearly there is a range of wavelengths over which this technique will work that extends from about 5 microns to at least 15 microns, if the carrier concentration of the silicon wafer can be raised above the 1×10¹⁹/cm³.

[0047] To obtain 10.6 micron wavelength radiation, a CO₂laser with typical pulse lengths in the 1 to 100 microsecond region can be used.

[0048] With continuing reference to FIG. 1C, a second alternative for the seventh step of the present method is performing front-side irradiation using a radiation beam 90 incident upper surface 12 of substrate 10, in combination with backside irradiation using radiation beam 66. The purpose of this dual irradiation is to facilitate silicidation. In this case, the front side irradiation with radiation beam 90 is used to heat metal layer 60 to further drive the diffusion of the metal (arrows 72) into doped amorphous region 40 and to further drive the subsequent reaction of the metal and the silicon, to form a silicide region 80 (FIG. 1D). Heat from radiation beam 90 may, for example, also be used to assist in melting doped amorphized region 40. In the case of metal layer 60 comprising Ti, the radiation beam 90 preferably has a wavelength of 0.1 to 2 μm, a temporal pulse length of less than 1 ms, and an irradiance of between 0.1 to 2 J/cm². For the backside irradiation with radiation beam 66, the conditions are similar to the first alternative for this step of the present embodiment, as described above. The exact irradiance, number of pulses, pulse duration and repetition rate needed to operate within the process margin of the present method in this second alternative will vary depending on the parameters associated with the front and backside irradiation. Accordingly, these parameters may need to be determined empirically.

[0049] With reference now to FIG. 1E, a third alternative for the seventh step of the present embodiment is a variation of the second alternative set forth above, and includes depositing a conformal absorbing layer 96 atop metal layer 60 to facilitate heat absorption during front-side irradiation. The material making up conformal layer 96 and its thickness are designed to partially absorb radiation beam 90 incident thereon and to diffuse heat from the absorbed radiation to the underlying structure. Accordingly, conformal layer 96 needs to be a natural diffuser of heat and must be capable of withstanding high temperatures, i.e., temperatures in excess of the crystalline Si melting temperature of about 1500 degrees C. Conformal layer 96 must also be easily removable (i.e., strippable) without impacting the layers below. An additional role of conformal layer 96 is to maintain the physical structure of the device during processing. An exemplary material for conformal layer 96 is tantalum nitride (TaN), deposited to a thickness of between 500 and 1000 angstroms via sputtering or by chemical vapor deposition. Other preferred materials for conformal layer 96 include titanium (Ti), titanium nitride (TiN), tantalum (Ta), and tungsten nitride (WN).

[0050] After backside irradiation (or front-side and backside irradiation in combination), the previously doped amorphous region 40 (FIG. 1B) recrystallizes to form re-crystallized doped region 40′. The diffusion of metal atoms (M) from metal layer 60 further results in the formation of silicide region 80 (FIG. 1D) comprising a compound of the form MSi_xhaving essentially the desired composition for the silicide that is to be ultimately formed in Si substrate 10. Region 80 may be in a partially crystallized state in which the atoms of the silicide are relatively ordered, yet not entirely so. For TiSi²silicide, for example, this partially crystallized phase is referred to as its “C49 phase.” In addition, the n-type or p-type dopant added to silicide region 80 becomes activated through dopant trapping upon recrystallization of amorphous doped region 40.

[0051] The eighth step of the process involves optionally irradiating silicide region 80 to re-melt this region to change the silicide stoichiometry. For example, where the metal in metal layer 60 is titanium, additional pulsed front side irradiation of the region using radiation beam 90 can be used to drive the silicide from TiSi₃to TiSi_2.5to TiSi₂, the latter being the preferable form of titanium silicide. This leaves silicide region 80 with the desired compound, which in the case of TiSi₂is the C54 phase. The C49 and C54 phases have the same stoichiometry, but are different phases each containing activated dopants. The eighth step can also be performed by conventional rapid thermal annealing (RTA). However, RTA would prolong the process and likely cause additional unwanted diffusion of the metal and dopants. The above process results in a silicide region having a sheet resistance of 5 Ω/sq or less.

[0052] The ninth and final step in the process involves stripping the remaining unreacted (i.e., undiffused) portion of metal layer 60 (and also conformal layer 96, if used) from Si substrate 10. The stripping of metal layer 60 can be performed with a 4:1 solution of sulfuric acid (H₂SO₄) and hydrogen peroxide (H₂O₂) heated to 120 degrees centigrade. Typically, immersion of Si substrate 10 for about ten minutes is all that is required to remove metal layer 60, although the time necessary to strip the metal layer can vary to a degree depending upon the type of metal and its thickness. Immersion can be performed in a spray etch tool like the EQUINOX™ made by Semitool of Kalispell, Mont.

[0053] Second Embodiment

[0054] The second embodiment of the present invention is a method of forming a semiconductor transistor using the method of the first embodiment of the present invention.

[0055] FIGS. 2A-2F are cross-sectional diagrams outlining the method steps of forming an integrated metal-insulator-semiconductor field-effect transistor (MISFET) device on Si substrate 10. The latter has a low dopant concentration (e.g., 10^15−1016ions/cm³.

[0056] With reference first to FIG. 2A, the first step of the second embodiment is forming spaced apart shallow trench isolations (STIs) 120 to electrically isolate an area of Si substrate 10 in which the MISFET device is to be formed. STIs 120 are formed by first etching spaced apart trenches into upper surface 12 of Si substrate 10, and then filling the trenches with an insulating material (e.g., an oxide such as silicon dioxide). Upper surface 12 of Si substrate 10 is then chemically-mechanically polished, resulting in a planarized upper surface.

[0057] After STIs 120 are formed, the second step is forming a gate 136 using techniques well-known in the art of semiconductor technology. This involves forming a gate insulator layer 126 on upper surface 12 of Si substrate 10. Gate insulator layer 126 can be an oxide layer, for example, in which case the resulting device is a MISFET. A poly-silicon or amorphous silicon layer 130 is then deposited by low-pressure chemical vapor deposition, for example, atop gate insulator layer 126. Gate silicon layer 130 and gate oxide layer 126 are then patterned using a resist layer to selectively form or etch the silicon layer and the gate insulator layer to form the silicon gate body (“gate”) 136, which includes a portion of gate insulator layer 126. Gate 136, as represented in the cross-sectional view of FIG. 2A, has first and second sides 138aand 138b.

[0058] With reference now to FIG. 2B, the third step of the present embodiment involves forming extension regions 140 and 144, respectively, by amorphizing Si substrate 10 on either side of gate 136 to a predetermined depth, as described above in connection with the first embodiment of the present invention. Extensions 140 and 144 are then doped with appropriate n- or p-type dopants, also as described above. If Si substrate 10, and hence the MISFET channel, is p-type, extension regions 140 and 144 are doped with n-type dopants. On the other hand, if Si substrate 10 is n-type, then extension regions 140 and 144 are doped with p-type dopants.

[0059] With reference now to FIG. 2C, the fourth step of the present embodiment is forming sidewall spacers 150. This is accomplished by forming an insulator layer (not shown) atop upper surface 12 of Si substrate 10 and gate 136. This insulator layer is then anisotropically etched to form insulator sidewall spacers 150 on sides 138aand 138bof gate 136. Sidewall spacers 150 help to achieve self-alignment of the position of the silicide regions yet to be formed, and also prevent horizontal diffusion into gate 136.

[0060] With reference now to FIG. 2D, the fifth step of the present embodiment involves performing a second deeper amorphizing implant 170 and a second deeper dopant implant 174 beneath extension regions 140 and 144, thereby respectively forming an amorphous deep drain region 180 and an amorphous deep source region 184. This process leaves a portion of extensions 140 and 144 under sidewall spacers 150. A first amorphous-crystalline interface I1 separates deep drain region 180 and crystalline Si substrate 10, and a second amorphous-crystalline interface I2 separates deep source region 184 from the crystalline SI substrate. At this point, deep drain region 180 and extension 140 can be considered a single “drain” region, and deep source region 184 and extension 144 can be considered a single “source” region.

[0061] With reference now to FIG. 2E, the sixth step of the present embodiment involves forming a metal layer 60 atop upper surface 12 of Si substrate 10 covering extension regions 140 and 144. Preferably, metal layer 60 is a species such as titanium, cobalt or nickel formed on the amorphous regions by sputtering, evaporation or chemical vapor deposition, as previously described in connection with the first embodiment of the present invention.

[0062] A first alternative for the eighth step of the present embodiment involves performing backside irradiation of Si substrate 10 with radiation beam 66 having a wavelength, irradiance, number of pulses, pulse duration and repetition rate such as described above in connection with the first embodiment of the present invention. Backside irradiation with radiation beam 66 serves to heat interfaces I1 and I2. As described above in connection with the first embodiment, this heat causes XRC, which propagates heat toward upper surface 12 (as indicated by arrows 70) so as to heat metal layer 60, deep drain region 180 and deep source region 184. This heating causes diffusion of metal atoms from metal layer 60 into Si substrate 10 (as indicated by arrows 72), and more particularly, into extension regions 140 and 144, thereby forming silicide regions 80aand 80b, respectively, in the upper portions 140u and 144u of these regions. Note how metal atoms from metal layer 60 do not diffuse into the sections of extensions 140 and 144 extending beneath sidewall spacers 150.

[0063] XRC also serves to activate deep drain and deep source regions 180 and 184 upon recrystallization via dopant trapping. The heat from XRC may be sufficiently large to cause deep drain and deep source regions 180 and 184 to melt, while not melting Si substrate 10. Accordingly, with reference to FIG. 2F, upper portions 140u and 144u of extension regions 140 and 144, respectively, become a silicide regions 80aand 80b, respectively, as described immediately above, and as also described in connection with he first embodiment (FIG. 1D). After irradiation, regions 80aand 80bcool to a chemical form that is at least partially crystalline.

[0064] A second alternative for the eighth step of the present embodiment, referring again to FIG. 2E, involves performing front side irradiation with radiation beam 90 in combination with backside irradiation with radiation beam 66 to heat metal layer 60. This facilitates the diffusion of metal atoms from metal layer 60 into extension regions 140 and 144, thereby forming regions 80aand 80btherein, as described above in connection with the first embodiment of the present invention.

[0065] A third alternative for the eighth step of the present embodiment is depositing a conformal absorbing layer 96 atop metal layer 60 to facilitate heat transfer to the metal layer and to Si substrate 10, as described above in connection with the first embodiment of the invention.

[0066] The ninth step of the present embodiment involves optionally irradiating silicide regions 80aand 80bto re-melt these regions to improve their stoichiometry, as previously described above in connection with the first embodiment of the invention. This step can be replaced by conventional rapid thermal annealing, but this would prolong the manufacturing process and would cause excess diffusion of dopants, thereby diminishing device performance.

[0067] The tenth step of the present embodiment involves stripping the (remaining) metal layer 60 (and conformal layer 96, if used) from Si substrate 10, also as previously described. This leaves the basic elements of a MISFET device structure, with silicide regions 80aand 80bhaving the desired stoichiometry.

[0068] Further processing steps are then performed to finalize the MISFET device. For example, with reference to FIG. 2F, a first insulator layer 200 is formed atop STIs 120 and silicide regions 80aand 80b, and also adjacent side wall spacers 150. A metallic contact 202 is then formed atop gate 136 using any one of a number of techniques known in the art. Insulator layer 200 is then selectively patterned to expose silicide regions 80aand 80b. Conductive leads 204, 206 and 208 composed of aluminum or other conductive metal, are formed and patterned so as to make physical contact with silicide regions 80aand 80b, and metal contact 202 of gate 136, respectively, of the MISFET device. A second insulator layer 210 comprising silicon oxide, for example, is then formed over conductive leads 204, 206 and 208 to electrically insulate and protect the conductive leads and the MISFET device. At respective ends opposite those in contact with the suicide regions, leads 204, 206 and 208 are coupled to other electronic components, and/or power or signal sources.

[0069] In the present invention, the use of backside irradiation of the Si substrate to initiate XRC to cause the metal layer to diffuse into the amorphous doped region helps reduce heating of the Si substrate so that integrated devices formed on the Si substrate are not subjected to prolonged heating that could otherwise damage such devices, a problem that significantly reduces the effectiveness of conventional silicidation techniques. In the invented method, irradiation of the amorphous doped region(s) can be effectively accomplished within a process margin in which the amorphous doped region recrystallizes or becomes molten, while the Si substrate and the metal layer remain in their solid phases, that spans a temperature range on the order of 300 degrees centigrade. Thus, the process of providing heat to Si substrate by backside or backside and front-side irradiation can deviate significantly from the optimum heating value in the middle of the temperature range without adverse impact to the Si substrate or the silicide region. This feature of the method of the present invention makes possible the formation of silicide contacts without significant increase in the risk of causing a leakage path or other defect in the Si substrate or integrated device formed therein.

[0070] The method of the present invention can be used to perform silicidation at a much faster rate as compared to conventional silicidation techniques, particularly those that require implantation of a stoichiometric quantity of metal ions to form the silicide region. In point of fact, the invented method requires implantation of only 10¹³to 10¹⁵ions cm²whereas conventional silicidation techniques require implantation on the order of 10¹⁶to 10¹⁷ions/cm². Thus, conventional silicidation techniques take ten to ten-thousand times longer to perform in comparison to the invented method. Accordingly, the throughput for the silicidation of Si substrates using the method of the present invention is much greater than is possible with conventional silicidation techniques.

[0071] Also, with ion implantation of the scale needed to implant a stoichiometric quantity of metal ions required in conventional silicidation techniques, the occurrence of ‘knock-on’ and hence the formation of leakage paths, is undesirably increased. In contrast, the relatively low ion implantation dosages required by the invented method minimize the probability of knock-on on a scale that could adversely affect the Si substrate or its integrated devices.

[0072] The many features and advantages of the present invention are apparent from the detailed specification and thus, it is intended by the appended claims to cover all such features and advantages of the described method which follow in the true spirit and scope of the invention. Further, since numerous modifications and changes will readily occur to those of ordinary skill in the art, it is not desired to limit the invention to the exact construction and operation illustrated and described. Moreover, the method of the present invention, like related methods used in the semiconductor arts, are often best practiced by empirically determining the appropriate values of the operating parameters. Accordingly, all suitable modifications and equivalents should be considered as falling within the spirit and scope of the invention.