This invention relates to electronic devices and associated fabrication processes. More specifically, the invention relates to highly tensile dielectric films on heat sensitive substrates formed by UV curing, for example implemented in strained transistor architecture for NMOS devices in which a highly tensile silicon nitride capping layer is provided on the source and drain regions to induce tensile strain in the NMOS channel region.
As transistors are scaled to smaller dimensions there is a need for higher switching speeds. One solution to increase transistor speed is to strain the silicon in the channel. Adding strain to the silicon lattice structure is believed to promote higher electron and hole mobilities, which increase transistor drain current and device performance.
One explanation for an increase in strain to improve device performance is a change in electron symmetry. When a silicon lattice is under tensile strain, its physical symmetry is broken, and with it the electronic symmetry. The lowest energy level of the conduction band is split, with two of the six original states dropping to a lower energy level and four rising to a higher energy level. This renders it more difficult for the electrons to be ‘scattered’ between the lowest energy states by a phonon, because there are only two states to occupy. Whenever electrons scatter, their motion is randomized. Reducing scatter increases the average distance an electron can travel before it is knocked off course, increasing its average velocity in the conduction direction. Also, distorting the lattice through tensile strain can distort the electron-lattice interaction in a way that reduces the electron's effective mass, a measure of how much it will accelerate in a given field. As a result, electron transport properties like mobility and velocity are improved and channel drive current for a given device design is increased in a strained silicon channel, leading to improved transistor performance.
Transistor strain has been generated in NMOS devices by using a highly tensile post-salicide silicon nitride capping layer on the source and drain regions. The stress from this capping layer is uniaxially transferred to the NMOS channel through the source-drain regions to create tensile strain in the NMOS channel. For example, a 1000 Å silicon nitride capping layer with a stress of 1 GPa has been shown to provide a 10% NMOS IDSAT gain (saturation drain current) from tensile channel strain (Ghani, et al., A 90 nm High Volume Manufacturing Logic Technology Featuring Novel 45 nm Gate Length Strained Silicon CMOS Transistors, IEEE (2003), incorporated by reference herein in its entirety for all purposes). However, a tensile stress in excess of 1E10 dynes/cm2 is necessary for optimal results.
A highly tensile silicon nitride capping layer may be deposited using a thermal CVD process, e.g., LPCVD. However, these processes generally require temperatures of greater than 500° C. to remove hydrogen from the capping layer to induce tensile stress in the capping layers, and at these higher temperatures the underlying NiSi (silicide/salicide) substrate for the capping layer undergoes a phase transformation that increases its resistivity. NiSi structures require temperature of less than about 400° C. to avoid the resistivity increase. NiPtSi gate electrodes has better thermal stability then NiSi, but even NiPtSi structures still require temperatures of less than about 480° C. Hence, thermal budget constraints for future transistor architectures require the films to be deposited at temperatures below than about 480° C., preferably below than about 400° C. A lower temperature thermal anneal may be used, for example, one in which the anneal temperature does not exceed 480° C. However, the duration of a thermal anneal process at that temperature that is necessary to obtain the benefit (e.g., about 2 hours) is not economically viable, and neither is the stress achieved sufficiently high.
Accordingly, fabrication processes for generating greater NMOS transistor channel strain are desirable.
The present invention addresses this need by providing a processing technique to generate a highly tensile dielectric layer on a heat sensitive substrate while not exceeding thermal budget constraints. Cascaded ultraviolet (UV) irradiation is used to produce highly tensile films to be used, for example, in strained NMOS transistor architectures. Successive UV radiation of equal or shorter wavelengths with variable intensity and duration selectively breaks bonds in the SiN film and minimizes shrinkage and film relaxation. Higher tensile stress than a non-cascaded approach may be obtained.
Ultraviolet (UV) irradiation is used to produce highly tensile films to be used, for example, in strained NMOS transistor architectures. UV curing of as-deposited PECVD silicon nitride films, for example, has been shown to bring about significant hydrogen removal and shrinkage, resulting in tremendous stress changes. Films with stresses of at least 1.65 GPa have been successfully produced using this technique at a temperature of 480° C. Even higher stresses are in demand, since device performance continues to increase with stress. The methods of the present invention create greater stresses by varying UV wavelengths, intensity, and duration to target specific bonds in the SiN film
In one aspect, the present invention pertains to a method of fabricating a highly tensile dielectric layer on a heat sensitive substrate. The method includes providing the heat sensitive substrate, depositing a dielectric layer on the substrate, exposing the dielectric layer UV radiation of a first wavelength or wavelength range at a temperature less than about 480° C., and repeating the exposure one or more times with shorter wavelengths or wavelength ranges in successive exposures. Thus, each exposure operation may apply UV radiation of a particular wavelength or wavelength range, and a mixture of specific wavelength or wavelength ranges may be used in a sequence. Each wavelength or wavelength range is at most equal to or shorter than the previous wavelength or range. The cascaded approach results in a higher tensile stress in the dielectric film than the non-cascaded approach. The dielectric film may be a silicon nitride, a silicon carbide, oxygen-doped silicon carbide, nitrogen-doped silicon carbide, silicon boron nitride, silicon boron carbide, silicon oxide, and combinations thereof. In certain embodiments, the dielectric film is silicon nitride or a doped silicon nitride capping layer on the source and drain regions of an NMOS device.
In certain embodiments, the first wavelength or wavelength range may include only wavelengths longer than about 300 nm, preferably about 300-800 nm. After the first UV exposure, subsequent exposures having shorter wavelengths are conducted. In one example, the UV exposure operation may be repeated three times. A first repeat exposure may have a wavelength within or range of about 295-800 nm; a second range may have a wavelength within or range of about 280-800 nm; and a third range may have a wavelength within or range of about 225-800 nm. In certain embodiments, the substrate may be exposed only to UV radiation having a wavelength longer than about 200 nm, or 225 nm, or preferably longer than about 250 nm. In some cases, depending on the film being cured and rate of cure, a final stage cure wherein the film is exposed to all the available wavelengths, even if the wavelength is shorter than the cut-off in the previous stage from the broadband source (or specific wavelngths like 222 nm, 172 nm etc. if using an excimer lamp) may be used.
In certain embodiments, a long pass filter is used to narrow the wavelength ranges of a broadband UV source, which may emit radiation from about 170 nm to the infrared region (about 1-10 micron). A long pass filter may be used to attenuate, or reduce, intensity of UV wavelengths passing through to filter those that are shorter than a certain cut-off wavelength. A long pass filter may be a number of dielectric layers with optical properties and thicknesses tuned to certain UV wavelengths. A long pass filter may be used with a short pass filter to create a bandpass filter that only allows UV radiation of certain wavelength range to pass. A bandpass filter may pass radiation of a broad range, such as 300 nm to 800 nm, or a narrow range, such as 250 nm to 260 nm. The wavelengths not transmitted are either reflected or absorbed by the filter material.
In addition, or instead of, using a broadband UV source, a singular wavelength source such as a laser may be used. An excimer laser can emit UV radiation in specific wavelengths depending on the molecule used for the electrical stimulation. For example, using Xe2*, KrCl*, KrF, XeBr and XeCl, UV radiation having a wavelength of 172 nm, 222 nm, 248 nm, 282 nm and 308 nm, respectively can be generated. In some cases, the radiation emitted by the excimer laser is scanned across the face of the substrate to completely expose the entire substrate. In other cases, multiple lasers may be used to form a pattern, which may be also scanned. Also, excimer lamps may also be used, wherein the emitted UV wavelength is quasi-monochromatic. A quasi-monochromatic wavelength is a spectrum having a defined peak and a defined range for full width at half maximum (FWHM). For example, a quasi-monochromatic spectrum may have a peak at 222 nm with a FWHM of 10 nm.
In certain embodiments, the methods of the present invention may employ a multi-station chamber. The stations in the chamber include apparatus for exposing the substrate to one or more UV radiation wavelengths or a wavelength range. The substrate may be transferred sequentially from station to station and be exposed to differing UV wavelengths, intensities, and for different durations. An example chamber may have four stations, each with one or more UV radiation sources and, optionally, one or more filters.
By utilizing this cascaded approach, very high tensile stresses may be achieved. The tensile stress may be higher than 1.5E10 dynes/cm2 or 1.5 Gigapascal (GPa), preferably higher than 1.65E10 dynes/cm2 or 1.65 GPa, or even more preferably higher than 1.8 GPa.
In certain embodiments, the present invention pertains to a method of fabricating a highly tensile dielectric layer by providing a heat sensitive substrate, depositing a dielectric layer on the substrate, and exposing the dielectric layer to ultraviolet radiation having a wavelength or wavelength range longer than about 250 nm. In these embodiments, the tensile stress induced may be higher than the stresses achieved with ultraviolet radiations that include wavelengths shorter than about 250 nm. The exposure may be repeated one or more times with UV radiation of shorter wavelengths with successive exposures, but at all times longer than 250 nm.
These and other aspects and advantages of the invention are described further below and with reference to the drawings.
Reference will now be made in detail to specific embodiments of the invention. Examples of the specific embodiments are illustrated in the accompanying drawings. While the invention will be described in conjunction with these specific embodiments, it will be understood that it is not intended to limit the invention to such specific embodiments. On the contrary, it is intended to cover alternatives, modifications, and equivalents as may be included within the spirit and scope of the invention as defined by the appended claims. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The present invention may be practiced without some or all of these specific details. In other instances, well known process operations have not been described in detail in order not to unnecessarily obscure the present invention.
Introduction
The present invention provides a processing technique to generate a highly tensile dielectric layer on a heat sensitive substrate while not exceeding thermal budget constraints. Ultraviolet (UV) radiation is used to produce highly tensile films to be used, for example, in strained NMOS transistor architectures. UV curing of as-deposited PECVD silicon nitride films using broadband UV sources, for example, has been shown to produce films with stresses of at least 1.65 E10 dynes/cm2, or 1.65 GPa. Other dielectric capping layer film materials show similar results. In transistor implementations, the stress from a capping layer composed of such a film is uniaxially transferred to the NMOS channel through the source-drain regions to create tensile strain in NMOS channel. The architecture has been developed for 90 nm logic technology on 300 mm substrates, although it is not so limited in application. As the technology nodes advances to 65 nm and smaller, higher stresses are required to maintain the same device performance as the films becomes thinner. Additionally, device performance continues to improve with increased stresses. Thus, greater stresses continue to be desired.
In the conventional process, broadband UV radiation is used, having wavelengths range from less than 200 nm all the way to infrared. Absorption of the radiation in the UV region, followed by bond-breaking and H2 removal, and subsequent stress generation has been the mechanism. The broadband UV radiation breaks the Si—H, N—H, and Si—N bonds in the film. After the Si—H and N—H bonds are broken and the hydrogen diffuses out of the film, the dangling silicon and nitrogen bonds form a SiN network. After Si—N bonds are broken, the bonds reform in a preferred energy pathway.
Significant tensile stress is generated within the first few minutes of cure (<10 min) accompanied by high shrinkage. With increasing cure time, the rate of stress generation drops off significantly, and continued curing brings about a stress reversal (stress starts decreasing). Although not intending to be bound by this theory, the reason for the stress saturation and reversal may be two-fold: first, decreasing hydrogen concentration in the film and increased compaction (due to shrinkage) makes it difficult to continue evolving hydrogen from the film. Hydrogen diffusion through the film depends on many factors, including time, film temperature, micro-structure of the film, and concentration of hydrogen in the film. As the film shrinks, the hydrogen evolution decreases to a point where further bond breaking does not increase the SiN network because the hydrogen is unable to diffuse out. While a decrease in hydrogen diffusion accounts for stress saturation, the stress reduction is likely due to a relaxation phenomenon.
The stress relaxation phenomenon may be caused by the reformation of Si—N bonds in a lower energy state. As the film cures, the number of Si—H and N—H bonds decreases, leaving more Si—N in the film. Since the film is exposed to a broadband source, radiation in the shorter wavelengths (higher energy) continues to break the Si—N backbone. While these bonds immediately reform, they reform in a less-strained configuration. Hence it is not possible to generate higher stresses by curing longer as more configurations are reformed in a less-strained manner.
Film stress may be increased if breaking of the Si—N backbone is minimized while continuing to break hydrogen bonds and evolve the hydrogen. The present invention reduces breaking of the Si—N backbone by exploiting the difference in absorption spectrum of the Si—H, N—H, and Si—N bonds.
In one aspect, the present invention involves generating tensile stress in dielectric films by treating the films with UV radiation of shorter wavelengths or wavelength ranges in successive exposures, while maintaining all UV radiation at a wavelength of a cut-off value and longer. In one embodiment, the present invention involves generating tensile stress by treating a silicon nitride film with UV radiation at a wavelength of 225 nm and longer. The cut-off limits on the wavelengths are guides to determine the appropriate wavelength to use. The silicon nitride film has a cut-off wavelength of 225 nm. Other variants of the same SiN film, or other films such as SiC, SiCN, SiBN, etc. will each have their own cut-off limits.
Post-Deposition UV Treatment of Dielectric Layer
The post-deposition UV treatment, also referred to as UV curing, technique of the invention generates a highly tensile dielectric layer on a heat sensitive substrate without thermally degrading the substrate. In a particular implementation, source/drain salicide capping layer stress in NMOS transistors is increased without thermally degrading the salicide. The technique involves post-deposition UV treatment of a capping layer film on a heat sensitive substrate, e.g., a silicide (or salicide) such as NiSi. In a specific embodiment, the capping layer is composed of silicon nitride (SiN or Si3N4) deposited by PECVD and the invention is primarily described herein with respect to this embodiment. In alternative embodiments, however, the capping layer may be deposited by other thermal CVD process, e.g., low pressure CVD (LPCVD), atmospheric pressure CVD (APCVD) or other suitable techniques such as spin-coating, print-coating, and dip-coating, and be composed of other dielectric materials including silicon carbide (SiC), oxygen-doped silicon carbide (SiCO), nitrogen-doped silicon carbide (SiCN), silicon boron nitride (SiBN), silicon boron carbide (SiBC), and silicon oxide (SiO or SiO2), and the invention should be understood to apply to these capping layer materials as well. Further the present invention also applies to other situations where greater stresses are desirable and may be induced by tailoring successive UV exposures at a wavelength known to break particular bonds to form a desired cured structure.
PECVD films in general contain a considerable amount of hydrogen. For example, PECVD silicon nitride films contain generally contain about 15-30% hydrogen in the form of Si—H and N—H bonds. Again, while the invention is not limited by this theory, it is generally accepted that irreversible tensile stress develops in PECVD films from the reduction of the amount of hydrogen in the film, and due to shrinkage of voids. The loss of hydrogen and shrinkage of voids result in a volume reduction in the film. But the constraint of the substrate prevents any lateral shrinkage, thus imposing tensile strains in the film. The change in hydrogen concentration has been shown to be proportional to the irreversible stress change. Thermal annealing at temperatures in excess of 500° C., e.g., 600° C., are also known to remove the hydrogen from such a film (silicon nitride) by providing sufficient energy to attain the right bonding configuration and stoichiometry, in particular the removal of H and formation of extended Si—N bonds. These temperatures, however, exceed the thermal budget constraints, leading to, for example, higher resistivity of NiSi. In some cases, annealing at these temperatures may adversely affect the partially fabricated circuits and films.
It has now been found that UV irradiation is a source of energy that makes compressive films tensile and tensile films even more tensile. It is believed that the photons from the UV source for example, a “H bulb” of a Hg lamp having a nominal wavelength from about 150 nm to 800 nm and an intensity of between 1 μW/cm2 and 10 W/cm2, provide sufficient energy to break most H-bonds in a dielectric film, e.g., for a silicon nitride film, the Si—H and N—H bonds. UV irradiation has a penetration depth sufficient to treat a film of full thickness, for example, between about 50 and 30,000 Å, e.g., about 300-1500, such as 700 Å. A following gas, such as He, Argon or N2 may be used as a purging gas during UV curing to remove evolved hydrogen. Alternatively or in addition, the chamber may be evacuated. Other reactive gases such as O2, CO2, N2O, Hz, H2O vapor, and vapors of alcohols (such as methanol, ethanol, isopropyl alcohol [IPA]) may be used to further control the UV curing process. The process pressure may vary between 10 to 1000 Torr. At a moderate substrate temperature (e.g., between about 25-500° C.) and UV exposure, the H atoms from neighboring broken bonds combine to form H2 that diffuses out of the film. The removal of hydrogen leaves behind micro voids in the film, along with the voids formed during deposition. The photon energy from the UV source, coupled with the thermal energy due to the substrate temperature cause these voids to shrink (in order to minimize surface energy). This shrinkage imposes significant tensile strains in the film.
The substrate temperature during the UV treatment is affected by the device and film on the partially fabricated circuit. The use of nickel monosilicide (NiSi) layers constrains the substrate temperature to less than 400° C., and nickel platinum silicide (NiPtSi), to less than about 480° C.
The UV treatment may be implemented in a continuous mode or pulsing mode to further optimize the end result on the final film. In a continuous exposure mode, the film is irradiated with a continuous UV source. In a pulsing mode, the film is exposed to pulses of UV radiation, leading to a sequence of curing/quenching/curing/quenching events. By modulating the pulse length, the intensity of the pulses, and the spectra of the radiation, the film stress and other properties may be further optimized.
Table 1, below, provides suitable PECVD deposition conditions for a silicon nitride layer suitable as a capping layer in accordance with the present invention:
The phenomenon of stress saturation and relaxation is shown on
On the other hand, the curves shown for thermal anneal at 500° C. (line 305) and 550° C. (line 307) had different slopes from the broadband UV cure. The stresses increased rapidly at first until about 20 minutes of cure, then increased less rapidly. The stresses continued to increase even after 120 minutes of curing, showing no relaxation phenomenon. The stress obtained at a higher temperature anneal is higher, as evidenced by comparing line 307 to 305. Note however, anneal temperatures of 550° C. and even 500° C. are likely to be above the thermal budget for the underlying NiSi or NiPtSi layers. Line 309 shows the film stresses over time for curing at 480° C. with a broadband UV source with a 320 nm long-pass filter. A long-pass filter allows wavelengths longer than the specified wavelengths to pass. The 320 nm long-pass filter would allow UV radiation of wavelengths of about 320-800 nm to pass and filter out shorter wavelengths for a broadband source having an output of about 170-800 nm. Line 309 has the same slope as that of 307 and 305, both thermal anneals, but a lower substrate temperature than both 307 and 305. The stresses obtained at a lower temperature UV cure (line 309 at 480° C.) were higher than that of higher temperature thermal anneal (line 305 at 500° C.). Although at the highest stress obtained for the filtered UV cure did not reach the peak for the broadband cure, the line had an upward slope and with continued curing would surpass the maximum stress from the broadband UV cure.
Test results depicted in
In order to minimize the shrinkage while driving out the maximum possible hydrogen, a cascading approach may be used. The cascading approach involves sequentially exposing the film to wavelengths above a particular cut-off wavelength that is increasingly shorter. For example,
The next step at that time would be to expose the films only to wavelengths above, e.g., 295 nm. Once the film has sufficiently cured at this wavelength, and the absorption curve has moved below the cut-off wavelength, the film can then be exposed to a lower wavelength radiation, e.g., 280 nm. This progression can continue until finally the film is exposed to radiation at wavelengths near the S—N bond breaking range, e.g., greater than 225 nm, or preferably grater than about 250 nm. The cascaded approach maximizes H2 removal, while minimizing the amount of time that the film is exposed to high energy (lower wavelength) radiation which minimizes the matrix bond-breaking and stress relaxation. Although the example above uses cut-off wavelengths at 305 nm, 295 nm, 280 nm, and 250 nm, the invention is not so limited. Different cut-off wavelengths may be chosen. In some cases, more than one exposure at the same wavelength or range of wavelengths may occur.
Table 2, below, provides an example of typical UV curing conditions for a silicon nitride layer suitable as a capping layer in accordance with the present invention:
In a cascaded approach after the first exposure, a decision is made whether the desired film stress is achieved (408). After the first period of exposure the film may not absorb further at the longer wavelengths, as evidenced by a shifting of absorption curve (see
In an alternative embodiment, a non-cascaded approach with a frequency cutoff may be taken, as shown in
Transistor Architecture
The present invention may be specifically applied to straining an NMOS channel by inducing stress in a silicon nitride capping layer. As shown in
UV curing the capping layer 630 by exposure to UV radiation breaks Si—H and N—H bonds in the film. Hydrogen is removed from the capping layer, and tensile stress in the capping layer is induced. The tensile stress is transferred to the NMOS channel through the source and drain regions resulting in a strained channel. As discussed above, channel strain improves electron mobility and therefore device performance.
Apparatus
The present invention can be implemented in many different types of apparatus. Generally, the apparatus will include one or more chambers (sometimes referred to as process reactors) that house one or more substrates and are suitable for semiconductor processing. At least one chamber will include a UV source. A single chamber may be employed for all operations of the invention or separate chambers may be used. Each chamber may house one or more substrates for processing. The one or more chambers may maintain the substrates in a defined position or positions (with or without motion within that position, e.g., rotation, vibration, or other agitation) during UV treatment operations. For certain operations in which the substrate is to be heated, the apparatus may include a heating platen.
In certain embodiments, the cascading cure process, or generally, multi-operation cure process, is performed using a multi-station cure chamber. As discussed above, in certain embodiments, the multiple operation cure processes of the invention rely on being able to independently modulate the UV intensity, wavelength, spectral distribution and substrate temperature of each operation. Additionally, certain inert or reactive gases may be injected during the cure process at the same or different flowrates at each step. Multi-station cure chambers capable of independent modulation of the above variables are described in U.S. patent application Ser. No. 11/688,695, titled “Multi-Station Sequential Curing of Dielectric Films,” which is hereby incorporated by reference in its entirety for all purposes.
Particularly, the independent control of the substrate temperature and the UV intensity is described in U.S. patent application Ser. No. 11/115,576 and in commonly assigned U.S. patent application Ser. No. 11/184,101, now U.S. Pat. No. 7,327,948 filed Jul. 18, 2005, titled “Cast Pedestal with Heating Element and Coaxial Heat Exchanger,” which are hereby incorporated by reference in their entirety for all purposes. A chamber may decouple substrate temperature and UV intensity by reducing the amount of IR radiation on the substrate, which heats up the substrate, and/or providing independent heat transfer mechanisms to and from the substrate. For example, the chambers may be equipped with cold mirrors or other reflectors to reduce the amount of IR radiation incident on the substrate, the absorption of which results in temperature increase. In addition, each pedestal or other substrate support may have an independent heat transfer mechanism to help maintain a substrate temperature regardless of the UV intensity. Thus, unlike conventional UV cure chambers where substrate temperature is coupled to UV intensity, the substrate temperature and UV intensity may be independently set for a wide range of temperatures and intensities.
In operation, a substrate enters the chamber at station 703 where the first UV cure operation is performed. Pedestal temperature at station 703 is set to a first temperature, e.g. 400° C., with the UV lamps above station 703 set to a first intensity, e.g., 100% maximum intensity, and first wavelength range, e.g., about 305-800 nm. After curing in station 703 for a sufficient time such that absorption at the wavelength range is reduced, the substrate is transferred to station 705 for further curing at the same wavelength range or shorter wavelength range, e.g., 295-800 nm. Pedestal temperature at station 705 is set to a second temperature, which may or may not be the same as the first station and UV intensity is set to a second intensity, e.g. 90% intensity. Stations 707 and 709 may also be used for UV curing. For example, at station 707, the UV radiation may be at a wavelength range of 280-800 nm, and at station 709, 225-800 nm. Each substrate is sequentially exposed to each UV light source.
In order to irradiate the substrate at different wavelengths or wavelengths ranges while using a broadband UV source, which generates radiation in a broad spectrum, optical components may be used in the radiation source to modulate the part of the broad spectrum that reaches the substrate. For example, reflectors, filters, or combination of both reflectors and filters may be used to subtract a part of the spectrum from the radiation. On reaching the filter, light may be reflected, absorbed into the filter material, or transmitted through.
Long pass filters are interference filters, which provide a sharp cut-off below a particular wavelength. They are useful for isolating specific regions of the spectrum. Long pass filters are used to pass, or transmit, a range of wavelengths and to block, or reflect, other wavelengths on the shorter wavelength side of the passband. Long wavelength radiation is transmitted, while short wavelength radiation is reflected. The region of high transmittance is known as the passband and the region of high reflectance is known as the reject or reflectance band. The roll-off region separates the pass-band and reflect-band. The complexity of long pass filters depends primarily upon the steepness of the transition region and also on the ripple specifications in the passband. In the case of a relatively high angle of incidence, polarization dependent loss may occur. Long pass filters are constructed of hard, durable surface materials covered dielectric coatings. They are designed to withstand normal cleaning and handling.
Another type of filter is UV cut-off filter. These filters do not allow UV transmission below a set value, e.g. 280 nm. These filters work by absorbing wavelengths below the cut-off value. This may be helpful to optimize the desired cure effect.
Yet another optical filter that may be used to select a wavelength range is a bandpass filter. Optical bandpass filters are designed to transmit a specific waveband. They are composed of many thin layers of dielectric materials, which have differing refractive indices to produce constructive and destructive interference in the transmitted light. In this way optical bandpass filters can be designed to transmit a specific waveband only. The range limitations are usually dependant upon the interference filters lens, and the composition of the thin-film filter material. Incident light is passed through two coated reflecting surfaces. The distance between the reflective coatings determines which wavelengths will destructively interfere and which wavelengths will be allowed to pass through the coated surfaces. In situations where the reflected beams are in phase, the light will pass through the two reflective surfaces. However, if the wavelengths are out of phase, destructive interference will block most of the reflections, allowing almost nothing to transmit through. In this way, interference filters are able to attenuate the intensity of transmitted light at wavelengths that are higher or lower than the desired range.
Another filter that can attenuate the wavelengths of the radiation reaching the substrate is the window 743, typically made of quartz. By changing the level of metal impurities and water content, the quartz window can be made to block radiations of undesired wavelengths. High-purity Silica Quartz with very little metal impurity is more transparent deeper into the ultraviolet. As an example, quartz with a thickness of 1 cm will have a transmittance of about 50% at a wavelength of 170 nm, which drops to only a few percent at 160 nm. Increasing levels of impurities in the quartz cause transmission of UV at lower wavelengths to be reduced. Electrically fused quartz has a greater presence of metallic impurities, limiting its UV transmittance wavelength to around 200 nm and longer. Synthetic silica, on the other hand, has much greater purity and will transfer down to 170 nm. For infrared radiation, the transmittance through quartz is determined by the water content. More water in the quartz means that infrared radiation is more likely absorbed. The water content in the quartz may be controlled through the manufacturing process. Thus, the spectrum of radiation transmission through the quartz window may be controlled to cutoff or reduce UV transmission at shorter wavelengths and/or to reduce infrared transmission at longer wavelengths.
In addition to changing the wavelengths by altering the radiation that reaches the substrate, radiation wavelength can also be controlled by modifying the properties of the light generator. Broadband UV source can generate a broad spectrum of radiation, from UV to infrared, but other light generators may be used to emit a smaller spectrum or to increase the intensity of a narrower spectrum. Other light generators may be mercury-vapor lamps, doped mercury-vapor lamps, electrode lamps, excimer lamps, excimer lasers, pulsed Xenon lamps, doped Xenon lamps. Lasers such as excimer lasers can emit radiation of a single wavelength. When dopants are added to mercury-vapor and to Xenon lamps, radiation in a narrow wavelength band may be made more intense. Common dopants are iron, nickel, cobalt, tin, zinc, indium, gallium, thallium, antimony, bismuth, or combinations of these. For example, mercury vapor lamps doped with indium emits strongly in the visible spectrum and around 450 nm; iron, at 360 nm; and gallium, at 320 nm. Radiation wavelengths can also be controlled by changing the fill pressure of the lamps. For example, high-pressure mercury vapor lamps can be made to emit wavelengths of 250 to 440 nm, particularly 310 to 350 nm more intensely. Low-pressure mercury vapor lamps emit at shorter wavelengths.
In addition to changing light generator properties and the use of filters, reflectors that preferentially deliver one or more segments of the lamps spectral output may be used. A common reflector is a cold mirror that allows infrared radiation to pass but reflects other light. Other reflectors that preferentially reflect light of a spectral band may be used. Therefore a substrate may be exposed to radiation of different wavelengths at different stations. Of course, the radiation wavelengths may be the same in some stations.
In
Pedestals 723 and 725 are electrically heated and maintained at a desired process temperature. Pedestals 723 and 725 may also be equipped with cooling lines to enable precise substrate temperature control. In an alternate embodiment, a large heater block may be used to support the substrates instead of individual pedestals. A thermally conductive gas, such as helium, is used to effect good thermal coupling between the pedestal and the substrate. In some embodiments, cast pedestals with coaxial heat exchangers may be used. These are described in above-referenced U.S. patent application Ser. No. 11/184,101.
In certain embodiments, the substrates are exposed to UV radiation from focused, rather than, flood lamps. Unlike the broadband source embodiments wherein the substrates are stationary during exposure (as in
Other apparatuses designed for other methods involved in previous and/or subsequent processes may be used. For example, methods of the invention may be used with a standard PECVD chamber used to deposit the precursor layer if the chamber is equipped with a UV radiation source. Some supercritical fluid chamber systems may also be configured to include a UV radiation source.
While the invention has been primarily described and exemplified with respect to UV exposure of a deposited film, but also applies to simultaneous UV and thermal and other treatments, or other engineered multi-step processes. For example, while it is preferable both from the perspective of process efficiency and effectiveness to deposit the dielectric to be UV cured in a single step, it is also possible to deposit and cure the dielectric in multiple repeating steps to build up a laminate dielectric with increased tensile stress and without the adverse impacts associated with thermal processing. The film stress will change with respect to the length of the treatment time, UV intensity, UV spectrum, UV operation mode such as pulse and continuous, curing environment, film thickness, and substrate curing temperature.
Although the foregoing invention has been described in some detail for purposes of clarity of understanding, it will be apparent that certain changes and modifications may be practiced within the scope of the appended claims. It should be noted that there are many alternative ways of implementing both the process and compositions of the present invention. Accordingly, the present embodiments are to be considered as illustrative and not restrictive, and the invention is not to be limited to the details given herein.
All references cited herein are incorporated by reference for all purposes.
Number | Name | Date | Kind |
---|---|---|---|
3983385 | Troue | Sep 1976 | A |
4357451 | McDaniel | Nov 1982 | A |
4391663 | Hutter, III | Jul 1983 | A |
4563589 | Scheffer | Jan 1986 | A |
4885262 | Ting et al. | Dec 1989 | A |
5178682 | Tsukamoto et al. | Jan 1993 | A |
5268320 | Holler et al. | Dec 1993 | A |
5282121 | Bornhorst et al. | Jan 1994 | A |
5504042 | Cho et al. | Apr 1996 | A |
5582880 | Mochizuki et al. | Dec 1996 | A |
5686054 | Barthel et al. | Nov 1997 | A |
5700844 | Hedrick et al. | Dec 1997 | A |
5789027 | Watkins et al. | Aug 1998 | A |
5812403 | Fong et al. | Sep 1998 | A |
5840600 | Yamazaki et al. | Nov 1998 | A |
5851715 | Barthel et al. | Dec 1998 | A |
5858457 | Brinker et al. | Jan 1999 | A |
5876798 | Vassiliev | Mar 1999 | A |
5877095 | Tamura et al. | Mar 1999 | A |
6015503 | Butterbaugh et al. | Jan 2000 | A |
6098637 | Parke | Aug 2000 | A |
6132814 | Livesay et al. | Oct 2000 | A |
6136680 | Lai et al. | Oct 2000 | A |
6140252 | Cho et al. | Oct 2000 | A |
6150272 | Liu et al. | Nov 2000 | A |
6228563 | Starov et al. | May 2001 | B1 |
6232248 | Shinriki et al. | May 2001 | B1 |
6254689 | Meder | Jul 2001 | B1 |
6268288 | Hautala et al. | Jul 2001 | B1 |
6270846 | Brinker et al. | Aug 2001 | B1 |
6271273 | You et al. | Aug 2001 | B1 |
6290589 | Tolles | Sep 2001 | B1 |
6329017 | Liu et al. | Dec 2001 | B1 |
6329062 | Gaynor | Dec 2001 | B1 |
6340628 | Van Cleemput et al. | Jan 2002 | B1 |
6365266 | MacDougall et al. | Apr 2002 | B1 |
6383466 | Domansky et al. | May 2002 | B1 |
6383955 | Matsuki et al. | May 2002 | B1 |
6386466 | Ozawa et al. | May 2002 | B1 |
6387453 | Brinker et al. | May 2002 | B1 |
6391932 | Gore et al. | May 2002 | B1 |
6392017 | Chandrashekar | May 2002 | B1 |
6394797 | Sugaya et al. | May 2002 | B1 |
6399212 | Sakai et al. | Jun 2002 | B1 |
6420441 | Allen et al. | Jul 2002 | B1 |
6444715 | Mukherjee et al. | Sep 2002 | B1 |
6467491 | Sugiura et al. | Oct 2002 | B1 |
6479374 | Ioka et al. | Nov 2002 | B1 |
6479409 | Shioya et al. | Nov 2002 | B2 |
6485599 | Glownia et al. | Nov 2002 | B1 |
6518130 | Ohno | Feb 2003 | B1 |
6531193 | Fonash et al. | Mar 2003 | B2 |
6534395 | Werkhoven et al. | Mar 2003 | B2 |
6548113 | Birnbaum et al. | Apr 2003 | B1 |
6558755 | Berry et al. | May 2003 | B2 |
6563092 | Shrinivasan et al. | May 2003 | B1 |
6566278 | Harvey et al. | May 2003 | B1 |
6576300 | Berry et al. | Jun 2003 | B1 |
6596654 | Bayman et al. | Jul 2003 | B1 |
6635575 | Xia et al. | Oct 2003 | B1 |
6644786 | Leben | Nov 2003 | B1 |
6677251 | Lu et al. | Jan 2004 | B1 |
6740602 | Hendriks et al. | May 2004 | B1 |
6740605 | Shiraiwa et al. | May 2004 | B1 |
6756085 | Waldfried et al. | Jun 2004 | B2 |
6759098 | Han et al. | Jul 2004 | B2 |
6770866 | Retschke et al. | Aug 2004 | B2 |
6797643 | Rocha-Alvarez et al. | Sep 2004 | B2 |
6805801 | Humayun et al. | Oct 2004 | B1 |
6812043 | Bao et al. | Nov 2004 | B2 |
6831284 | Demos et al. | Dec 2004 | B2 |
6835417 | Saenger et al. | Dec 2004 | B2 |
6848458 | Shrinivasan et al. | Feb 2005 | B1 |
6856712 | Fauver et al. | Feb 2005 | B2 |
6884738 | Asai et al. | Apr 2005 | B2 |
6902440 | Dougan et al. | Jun 2005 | B2 |
6921727 | Chiang et al. | Jul 2005 | B2 |
6958301 | Kim et al. | Oct 2005 | B2 |
6967160 | Paton et al. | Nov 2005 | B1 |
7005390 | Ramachandrarao et al. | Feb 2006 | B2 |
7018918 | Kloster et al. | Mar 2006 | B2 |
7030041 | Li et al. | Apr 2006 | B2 |
7094713 | Niu et al. | Aug 2006 | B1 |
7112541 | Xia et al. | Sep 2006 | B2 |
7132334 | Lin | Nov 2006 | B2 |
7144606 | Huang | Dec 2006 | B2 |
7148155 | Tarafdar et al. | Dec 2006 | B1 |
7166531 | van den Hoek et al. | Jan 2007 | B1 |
7176144 | Wang et al. | Feb 2007 | B1 |
7208389 | Tipton et al. | Apr 2007 | B1 |
7235459 | Sandhu | Jun 2007 | B2 |
7241704 | Wu et al. | Jul 2007 | B1 |
7244672 | Nguyen et al. | Jul 2007 | B2 |
7247582 | Stern et al. | Jul 2007 | B2 |
7253125 | Bandyopadhyay et al. | Aug 2007 | B1 |
7256111 | Lopatin et al. | Aug 2007 | B2 |
7265061 | Cho et al. | Sep 2007 | B1 |
7304302 | Nunan et al. | Dec 2007 | B1 |
7332445 | Lukas et al. | Feb 2008 | B2 |
7381659 | Nguyen et al. | Jun 2008 | B2 |
7390537 | Wu et al. | Jun 2008 | B1 |
7394067 | Soltz et al. | Jul 2008 | B1 |
7402532 | Clevenger et al. | Jul 2008 | B2 |
7481882 | Won et al. | Jan 2009 | B2 |
7504663 | Yamazaki et al. | Mar 2009 | B2 |
7510982 | Draeger et al. | Mar 2009 | B1 |
7611757 | Bandyopadhyay et al. | Nov 2009 | B1 |
7622162 | Schravendijk et al. | Nov 2009 | B1 |
7705431 | Sanganeria et al. | Apr 2010 | B1 |
7790633 | Tarafdar et al. | Sep 2010 | B1 |
7851232 | van Schravendijk et al. | Dec 2010 | B2 |
7858533 | Liu et al. | Dec 2010 | B2 |
7906174 | Wu et al. | Mar 2011 | B1 |
8043667 | Bandyopadhyay et al. | Oct 2011 | B1 |
20010001501 | Lee et al. | May 2001 | A1 |
20010014512 | Lyons et al. | Aug 2001 | A1 |
20020001973 | Wu et al. | Jan 2002 | A1 |
20020015850 | Nakamura et al. | Feb 2002 | A1 |
20020016085 | Huang et al. | Feb 2002 | A1 |
20020034626 | Liu et al. | Mar 2002 | A1 |
20020064341 | Fauver et al. | May 2002 | A1 |
20020106500 | Albano et al. | Aug 2002 | A1 |
20020117109 | Hazelton et al. | Aug 2002 | A1 |
20020123218 | Shioya et al. | Sep 2002 | A1 |
20020123240 | Gallagher et al. | Sep 2002 | A1 |
20020172766 | Laxman et al. | Nov 2002 | A1 |
20020195683 | Kim et al. | Dec 2002 | A1 |
20030013280 | Yamanaka | Jan 2003 | A1 |
20030015764 | Raaijmakers et al. | Jan 2003 | A1 |
20030064604 | Umeda | Apr 2003 | A1 |
20030064607 | Leu et al. | Apr 2003 | A1 |
20030068881 | Xia et al. | Apr 2003 | A1 |
20030134038 | Paranjpe | Jul 2003 | A1 |
20030157248 | Watkins et al. | Aug 2003 | A1 |
20030157267 | Waldfried et al. | Aug 2003 | A1 |
20030199603 | Walker et al. | Oct 2003 | A1 |
20030227087 | Kakamu et al. | Dec 2003 | A1 |
20030228769 | Chen et al. | Dec 2003 | A1 |
20030228770 | Lee et al. | Dec 2003 | A1 |
20040004247 | Forbes et al. | Jan 2004 | A1 |
20040018319 | Waldfried et al. | Jan 2004 | A1 |
20040022960 | Rhee et al. | Feb 2004 | A1 |
20040023513 | Aoyama et al. | Feb 2004 | A1 |
20040029391 | Kirkpatrick et al. | Feb 2004 | A1 |
20040033662 | Lee et al. | Feb 2004 | A1 |
20040058090 | Waldfried et al. | Mar 2004 | A1 |
20040069410 | Moghadam et al. | Apr 2004 | A1 |
20040082163 | Mori et al. | Apr 2004 | A1 |
20040096593 | Lukas et al. | May 2004 | A1 |
20040096672 | Lukas et al. | May 2004 | A1 |
20040099952 | Goodner et al. | May 2004 | A1 |
20040101633 | Zheng et al. | May 2004 | A1 |
20040102031 | Kloster et al. | May 2004 | A1 |
20040152239 | Bao et al. | Aug 2004 | A1 |
20040166240 | Rhee et al. | Aug 2004 | A1 |
20040175501 | Lukas et al. | Sep 2004 | A1 |
20040185679 | Ott et al. | Sep 2004 | A1 |
20040221871 | Fletcher et al. | Nov 2004 | A1 |
20040224496 | Cui et al. | Nov 2004 | A1 |
20050025892 | Satoh et al. | Feb 2005 | A1 |
20050032293 | Clark et al. | Feb 2005 | A1 |
20050064726 | Reid et al. | Mar 2005 | A1 |
20050079717 | Savas et al. | Apr 2005 | A1 |
20050095840 | Bhanap et al. | May 2005 | A1 |
20050101154 | Huang | May 2005 | A1 |
20050112282 | Gordon et al. | May 2005 | A1 |
20050153533 | Hoshino et al. | Jul 2005 | A1 |
20050156285 | Gates et al. | Jul 2005 | A1 |
20050161821 | Lee et al. | Jul 2005 | A1 |
20050164497 | Lopatin et al. | Jul 2005 | A1 |
20050170104 | Jung et al. | Aug 2005 | A1 |
20050191803 | Matsuse et al. | Sep 2005 | A1 |
20050194619 | Edelstein et al. | Sep 2005 | A1 |
20050208758 | Lu et al. | Sep 2005 | A1 |
20050260357 | Olsen et al. | Nov 2005 | A1 |
20050260420 | Collins et al. | Nov 2005 | A1 |
20050272220 | Waldfried et al. | Dec 2005 | A1 |
20060024976 | Waldfried et al. | Feb 2006 | A1 |
20060027929 | Cooney et al. | Feb 2006 | A1 |
20060063662 | Hata et al. | Mar 2006 | A1 |
20060105106 | Balseanu et al. | May 2006 | A1 |
20060118817 | Haisma | Jun 2006 | A1 |
20060121208 | Siegel | Jun 2006 | A1 |
20060142143 | Abrevaya et al. | Jun 2006 | A1 |
20060145304 | Boyanov et al. | Jul 2006 | A1 |
20060216839 | Shenesh et al. | Sep 2006 | A1 |
20060220251 | Kloster | Oct 2006 | A1 |
20060246672 | Chen et al. | Nov 2006 | A1 |
20060260538 | Ye et al. | Nov 2006 | A1 |
20060269693 | Balseanu et al. | Nov 2006 | A1 |
20070015355 | Lin et al. | Jan 2007 | A1 |
20070032024 | Peidous et al. | Feb 2007 | A1 |
20070042581 | Sano et al. | Feb 2007 | A1 |
20070054504 | Chen et al. | Mar 2007 | A1 |
20070105292 | Chen et al. | May 2007 | A1 |
20070132054 | Arghavani et al. | Jun 2007 | A1 |
20070134821 | Thakur et al. | Jun 2007 | A1 |
20070134907 | Ikeda et al. | Jun 2007 | A1 |
20070161230 | Chen et al. | Jul 2007 | A1 |
20070189961 | Iacopi et al. | Aug 2007 | A1 |
20070196011 | Cox et al. | Aug 2007 | A1 |
20070222081 | Chen et al. | Sep 2007 | A1 |
20070224824 | Chen et al. | Sep 2007 | A1 |
20070275569 | Moghadam et al. | Nov 2007 | A1 |
20070281497 | Liu et al. | Dec 2007 | A1 |
20070287240 | Chen et al. | Dec 2007 | A1 |
20080009141 | Dubois et al. | Jan 2008 | A1 |
20080020591 | Balseanu et al. | Jan 2008 | A1 |
20080026579 | Lai et al. | Jan 2008 | A1 |
20080132055 | Nguyen et al. | Jun 2008 | A1 |
20080199977 | Weigel et al. | Aug 2008 | A1 |
20080254643 | Clevenger et al. | Oct 2008 | A1 |
20080286697 | Verhaverbeke et al. | Nov 2008 | A1 |
20080305600 | Liao et al. | Dec 2008 | A1 |
20090017640 | Huh et al. | Jan 2009 | A1 |
20090039475 | Shioya | Feb 2009 | A1 |
20090059406 | Powers et al. | Mar 2009 | A1 |
20100018460 | Singh et al. | Jan 2010 | A1 |
20100261349 | Van Schravendijk et al. | Oct 2010 | A1 |
20100267231 | Van Schravendijk et al. | Oct 2010 | A1 |
20100317198 | Antonelli et al. | Dec 2010 | A1 |
20110045610 | Van Schravendijk et al. | Feb 2011 | A1 |
20110111533 | Varadarajan et al. | May 2011 | A1 |
20110117678 | Varadarajan et al. | May 2011 | A1 |
Number | Date | Country |
---|---|---|
62-229833 | Oct 1987 | JP |
63-307740 | Dec 1988 | JP |
01-107519 | Apr 1989 | JP |
09-306892 | Nov 1997 | JP |
11214364 | Aug 1999 | JP |
2006165573 | Jun 2006 | JP |
2000-0043888 | Jul 2000 | KR |
2006104583 | Oct 2006 | WO |
2006127463 | Nov 2006 | WO |
2007043206 | Apr 2007 | WO |