JPH1087316A - Low volatile solvent-base precursor for nanoporous aerogel - Google Patents

Abstract

PROBLEM TO BE SOLVED: To deposit thin film-shaped nanoporous aerogel having controlled porosity without controlling the atmosphere and to age and dry the aerogel. SOLUTION: A sol of a precursor of aerogel contg. a reactive body of a precursor of metal-base aerogel and a polyol-contg. 1st solvent is prepd. The molar ratio of molecules of the 1st solvent to metallic atoms in the reactive body is at least l:16. The 1st solvent is preferably glycerol and the reactive body is preferably selected from among a metal alkoxide, a partially hydrolyzed metal alkoxide, a granular metal oxide and a combination of them. The molar ratio of molecules of the 1st solvent to metallic atoms in the reactive body is <=2:1, preferably 1:2 to 12:1.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates generally to precursors for producing nanoporous (microporous, nanometer-sized pores) aerogels, and more particularly to the bulk and thin film aerogels. It relates to precursors suitable for critical and supercritical drying.

[0002]

2. Description of the Related Art Aerogel is a porous silica material,
It can be used for a variety of applications, such as films (eg, electrical insulators or optical coatings provided on semiconductor devices) or bulk (eg, thermal insulators). For simplicity, the embodiments herein are primarily used as electrical insulators provided on semiconductor devices.

Semiconductors are widely used in integrated circuits for electronic devices such as computers and televisions. Semiconductor and electronics manufacturers, as well as end users, desire integrated circuits that consume less power but can operate in shorter packages and in less time. However, many of these demands conflict with each other.
For example, the size of a predetermined circuit is 0.5 micron to 0.2 micron.
Simply reducing to 5 microns can increase energy consumption and heat generation by 30%. Also, reducing the dimensions generally results in increased capacitive coupling, or crosstalk, between conductors that send signals between chips. Such effects not only limit the achievable speed, but also degrade the noise margin used to ensure proper operation of the device. One way to reduce energy usage / heat generation and crosstalk effects is to increase the dielectric constant of the insulator separating the conductors, the dielectric.
US Patent No. 5,470,802 issued outside of Runard
The article provides background art on some of these solutions.

Certain materials, namely nanoporous dielectrics, include some of the most promising new materials for semiconductor manufacturing. These dielectric materials comprise a solid structure, for example silica, which is traversed by an intertwined network of pores, typically having a diameter of a few nanometers. Very high porosity can be formed in these materials, with a corresponding dielectric constant generally less than half that of dense silica. However, it has been found that nanoporous dielectrics having high strength despite high porosity and excellent in compatibility with most current semiconductor manufacturing processes can be manufactured. Thus, nanoporous dielectrics are potential low-k replacements for common semiconductor dielectrics, such as dense silica.

[0005] A preferred method for forming a nanoporous dielectric is using sol-gel technology. Here, the term "sol-gel" is not a term that describes a product, but a term that refers to a reaction mechanism in which a sol, which is a colloidal suspension of solid particles in a liquid, changes into a gel by the growth and interconnection of the solid particles. It is. One theory is that the continuation of the reaction in the sol allows one or more molecules in the sol to eventually reach a macroscopic size and form a solid network extending almost throughout the sol. In this regard (called the gel point), the substance is called a gel. According to such a definition,
A gel is a substance containing a continuous solid skeleton surrounding a continuous liquid phase. If the scaffold is porous, the term "gel" as used herein means a continuous porous solid structure surrounding the fluid in the pores.

One method of forming a sol involves the use of hydrolysis and condensation reactions, which polymerize polyfunctional monomers in solution and can result in relatively large, highly branched particles. Many monomers suitable for such polymerization include metal alkoxy. For example, tetraethoxysilane (TEOS) monomer can be partially hydrolyzed in water by the following reaction.

[0007]

Embedded image Si (OEt) ₄ + H ₂ O → HO-Si (OE
t) ₃ + EtOH

The reaction conditions can be controlled such that the desired number of hydrolysis reactions, such that each monomer partially or completely hydrolyzes it, are carried out on average. The completely hydrolyzed TEOS becomes Si (OH) ₄ . Once the molecules have been at least partially hydrolyzed, the two molecules may be linked together in a condensation reaction as follows.

[0009]

Embedded image (OEt) ₃ Si—OH + HO—Si (OH)
₃ → (OEt) ₃ Si—O—Si (OH) ₃ + H ₂ O or (OEt) ₃ Si—OEt + HO—Si (OEt) ₃ →
_{(OEt) 3 Si-O-} Si (OEt) 3 + EtOH

Thus, the bonding of the two molecules forms an oligomer, releasing water or ethanol molecules. Si—O— in the oligomer formed by these reactions
The Si bond has three sites at each end that are available for further hydrolysis and condensation. Thus, additional monomers or oligomers can be added to this molecule in a somewhat random fashion to form a single, highly branched polymer molecule from literally thousands of monomers. The oligomerized metal alkoxy described herein comprises a molecule formed from at least two alkoxy monomers, but does not include a gel.

The sol-gel reaction is the basis for xerogel and aerogel film deposition. In a typical thin film xerogel process, a sol of an ungelled precursor is applied to a substrate (e.g., by spray coating, dip coating or spinning coating) to form a thin film having a thickness of a few microns or less, It can gel and dry to form a dense film. The precursor sol often contains a stock solution and a solvent, and may also include a gelling catalyst that changes the pH of the precursor sol to speed gelation. During and after coating, the volatile components in the sol film usually evaporate rapidly. Thus, the deposition, gelling and drying steps can occur simultaneously (at least to some extent) as the film rapidly collapses to a dense film. In contrast, the aerogel process differs from the xerogel process primarily by preventing pore collapse during drying of the wet gel. Some methods for preventing pore collapse (US Pat.
There is a method of treating the wet gel with a condensation-inhibiting modifier (as described in No. 70,802) and a method of supercritically extracting the fluid in the pores.

Aerogels and xerogels are preferred as two dry gel materials for nanoporous dielectrics of semiconductor thin films. Typical thin film xerogel processes produce membranes with limited porosity (up to 60% for large pore sizes, but generally less than approximately 50% for effective pore sizes in submicron semiconductor fabrication). Although some of the conventional xerogels have porosity greater than 50%, these conventional xerogels have substantially large pore sizes (typically 100 nm or more). These large pore size gels have rather low mechanical strength. Furthermore, due to such a large hole diameter,
Gels are small on microcircuits (typically smaller than 1 micrometer, potentially smaller than 100 nm)
It becomes unsuitable for filling patterned gaps, limiting the use of those optical films to only longer wavelengths. On the other hand, nanoporous aerogel films can form almost desirable porosity associated with extremely fine pore size. Generally, the nanoporous materials used herein have an average pore size of less than about 25 nm, preferably less than 20 nm (more preferably less than 10 nanometers, and even more preferably less than 5 nanometers). In many formulations using this method, typical nanoporous materials for semiconductors can have an overall average pore size of at least 1 nm, but more often have an average pore size of at least 3 nm. Nanoporous inorganic dielectrics comprise nanoporous metal oxides, especially nanoporous silica.

[0013]

For many nanoporous thin film applications, as optical films, or for aerogels and xerogels used in microelectronics, it is preferable to precisely control the thickness and density of the aerogel. . Several important properties of the membrane, including mechanical strength, pore size and dielectric constant, are related to the density of the aerogel. At present, it has been found that the density and film thickness of aerogel are related to the viscosity of the sol when the sol is applied to a substrate. This raises a previously unrecognized problem.
The problem is that it is extremely difficult to independently and accurately control both the density and thickness of the aerogel using conventional precursor sol and deposition methods.

[0014] Nanoporous dielectric thin films can be deposited on patterned wafers and are often deposited on patterned conductor levels. It is now necessary to complete the sol deposition before gelation can begin to ensure that the gaps between such wires continue to fill correctly and that the gel surface remains substantially planar. I know. For this purpose, it is also preferred that the fluid in the pores does not evaporate significantly after gelling, for example during aging. Unfortunately, it is also desirable to be able to reach the gel point as soon as possible after deposition to simplify the process, and one way to speed up the gelation of the thin film is to leave it to evaporate. The precursor sol suitable for aerogel deposition can control film thickness, aerogel density, gap filling and flatness, is relatively stable before deposition, gels relatively quickly after deposition, and It is recognized that it must be aged without evaporating.

[0015] Methods have now been found that allow controlled deposition of aging films from multi-solvent precursor sols. In this method, the viscosity and the film thickness of the sol can be controlled relatively separately. As a result, the film thickness can be changed in a short time from the first known value to the second known value that can be set by the ratio of the solvent and the spin condition, so that the film thickness is maintained independently of the airgel density. In addition, high-speed gelation becomes possible. However, at the same time, independently of the spin conditions and the film thickness, the ratio of solid to liquid in the film when dried in the precursor sol before deposition (and therefore the density of the aerogel) can be determined accurately.

Even with this new separation of the problem of deposition from the secondary problems of viscosity and density control, our experience shows that thin-film sol-gel techniques for forming xerogels and aerogels generally require drying. Before, for example after gelling and during aging, it has been found that some method, for example atmospheric control, is generally necessary to limit evaporation. In principle, this evaporation rate control can be achieved by controlling the solvent vapor density on the wafer. However, our experience has shown that the evaporation rate of the solvent is very sensitive to small changes in vapor concentration and temperature. To better understand this process, we modeled the isothermal evaporation of some solvents from the wafer as a function of% concentration. The evaporation rates at ambient temperature for some of these solvents are shown in FIG. In order not to make this evaporation a problem during processing, the product of the evaporation rate and the processing time (preferably several minutes) must be significantly smaller than the film thickness. This suggests that for solvents such as ethanol, the atmosphere on the wafer must be maintained at over 90% saturation. However, there can be problems associated with allowing the atmosphere to reach saturation or supersaturation. Some of this problem is related to the condensation of atmospheric components on the film. It has been found that condensation on a gelled or ungelled film causes defects in the film that have not been sufficiently aged. Therefore, it is generally preferable to control the atmosphere so as not to saturate the components.

[0017]

SUMMARY OF THE INVENTION We have found that using low volatility solvents without the use of high volatility solvents and with less control over the atmosphere of the solvent without precise control of the solvent atmosphere. I found it to be a good solution. After examining this prerequisite, it was discovered that glycerol was an excellent solvent.

The use of glycerol allows the necessary control of the atmosphere during deposition, gelation and / or aging to be less intensive (compared to conventional solvents). The reason for this is that while it is still preferable to avoid saturation, the concentration of atmospheric solvent can be reduced without excessive evaporation. FIG. 2 shows how the evaporation rate of glycerol changes with temperature and concentration of atmospheric solvent. Our experience has shown that the use of glycerol allows the formation of acceptable gels by depositing, gelling and aging with little or no control over the atmosphere.

When producing a nanoporous dielectric, it is preferred to subject the wet gel film to a process known as aging. The hydrolysis and condensation reactions do not stop at the gel point, but continue to reconstitute and age the gel until the reaction has stopped working. It is believed that during aging, preferential dissolution and redeposition of the solid structural portion provides increased strength, more uniform pore size, and advantageous results including increased pore collapse resistance during drying. Unfortunately, conventional aging techniques used for bulk gels are not well suited for aging thin films in semiconductor processing, in part because this aging technique generally requires immersing the substrate in a liquid, We also recognize that it can take days or weeks to complete for some reasons. As one feature of the present invention,
There is a gas phase aging technique that does not immerse a wet gel thin film in a liquid or dries it early, and can aging such a thin film in a matter of minutes.

Aerogels are nanoporous materials that can be used for a variety of applications, including membrane or bulk. However, it should be recognized that in practice the film processing is not similar to the bulk processing since the problems that occur in the film manufacturing process are different from the problems during the bulk processing.

In general, it has been discovered that aging in a saturated atmosphere prevents the problems associated with liquid immersion aging. Further in accordance with this aspect of the invention, there are provided several solutions for aging wet gels at elevated temperatures. This method can be used even when the wet gel originally contains a low boiling liquid in the pores. However, these methods work better when using low volatile solvents.
Finally, according to this aspect of the invention, it is contemplated to add an optional gas phase aging catalyst to the aging atmosphere to speed up aging.

Since the thin film is aged, it contains a very small amount of pore fluid to be kept very constant for a predetermined period of time, so that it is difficult to age the wet gel during the formation of the thin film. If pore fluid evaporates from the membrane before aging strengthens the network, the membrane tends to be xerogel-like. On the other hand, if excessive pore fluid condenses from the atmosphere onto the film before the network is strengthened, this may locally disrupt the aging process.

Accordingly, certain methods of controlling the rate of evaporation of pore fluid during aging have been found to be beneficial for aging thin film fabrication. Basically, the control of the evaporation rate during aging can be achieved by actively controlling the concentration of the fluid vapor in the hole on the wafer. However 150
The total amount of pore fluid contained in a 1 mm thick 70% porous wet gel, for example, deposited on a 1 mm thick wafer, is only about 0.012 mL, and this amount is equivalent to one 3 mm diameter droplet of fluid. Easy to fit. A typical thin film used for a nanoporous dielectric on a semiconductor wafer is about 10
The thickness is 1 / 00th. Thus, by actively controlling the vapor concentration of the bore fluid (eg, by adding or removing solvent from the atmosphere) such that only 1% or less of the bore fluid can evaporate during aging, Difficult things happen. That is, the surface area of the thin film is large, and the tolerance for the variation of the fluid in the pores becomes extremely small. Evaporation and condensation control is particularly important for high-speed aging, especially at high temperatures, where film-making processes have heretofore been virtually impossible.

We have overcome the problem of evaporation control rate without actively controlling the fluid vapor concentration in the pores on the wafer. Instead of such a vapor concentration, by processing the wafer in a very small volume chamber and naturally evaporating a very small amount of pore fluid contained in the wet gel membrane, the processing atmosphere is substantially reduced within the pore fluid. It becomes saturated state. Without cooling the wafer at any point in a substantially saturated processing atmosphere, this method also solves the condensation problem that must generally be avoided, especially during high temperature processing.

An Aerogel precursor sol is disclosed herein. The aerogel precursor sol comprises a metal-based aerogel precursor reactant and a first solvent containing a polyol, wherein the molar ratio of molecules of the first solvent to metal atoms in the reactant is at least 1:16. Preferably, the first solvent is glycerol, and preferably the aerogel precursor reactant can be selected from the group consisting of metal alkoxides, at least partially hydrolyzed metal alkoxides, particulate metal oxides and combinations thereof. Generally the first to metal atom in the reactant
The molar ratio of the solvent molecules is 12: 1 or less, preferably the molar ratio of the first solvent molecules to metal atoms in the reactants is between 1: 2 and 12: 1. In some embodiments, the molar ratio of the molecules of the first solvent to metal atoms in the reactants is between 2.5: 1 and 12: 1. In this method, it is also preferred that the nanoporous dielectric has a porosity of 60% or more and an average pore size of less than 25 nm. In some embodiments, the aerogel precursor also includes a second solvent. Preferably, this second solvent has a higher boiling point than glycerol.
In some embodiments, the second solvent can be ethanol. In some embodiments, the first solvent also includes a glycol, wherein the glycol is ethylene glycol, 1,4 butylene glycol,
1,5 pentanediol and a combination thereof. In some embodiments, the first
The polyol is 1,2,4-butanetriol;
3-butanetriol; 2-methylpropanetriol; and 2- (hydroxymethyl) -1,3, propanediol; 1-4,1-4, butanediol; and 2-methyl-1,3-propanediol, and Selected from the group consisting of combinations thereof. In some embodiments, the first polyol is selected from the group consisting of ethylene glycol, 1,4 butylene glycol, 1,5 pentanediol, and combinations thereof.

Accordingly, the present invention provides for depositing, gelling, aging and drying thin-film nanoporous aerogels of controlled porosity without controlling the atmosphere. According to another feature, the present invention provides passive atmosphere control, e.g., depositing, gelling, and rapidly aging at high temperatures thin-film nanoporous aerogels with controlled porosity by only limiting the volume of the aging chamber. , So that it can be dried.

The present invention, including its various features and advantages, is best understood by referring to the following drawings.

[0028]

DETAILED DESCRIPTION OF THE INVENTION A typical sol-gel thin film process disintegrates on drying, producing a dense gel and thus forming a xerogel having only a few percent porosity. Under uncontrolled drying conditions of xerogel film formation, it is difficult or impossible to completely separate the deposition, aggregation, gelation and drying steps during thin film formation, as the entire process can be completed within seconds. Met. However, it has been found that such methods are generally not suitable for depositing controllable low density, high porosity thin films. The reason for this is that in an aerogel-type drying process, the thin film remains less dense after drying, and the final density is mainly determined by the ratio of solid to liquid in the film during gelation. Also, it has been found that the following criteria are preferred for aerogel thin film deposition, especially when the patterned wafer requires a thin film to planarize and / or fill gaps.

1) Initial viscosity suitable for spin-on use 2) Stable viscosity at the time of deposition 3) Stable film thickness at gel time 4) Ratio of predetermined solid to liquid at gel time 5) Immediately after deposition Gelation

Conventional precursor sols and methods satisfying these conditions have not yet been found. However, according to the present invention, it has been found that sols prepared with a certain ratio of at least two solvents can be used to meet these conditions. With reference to FIG. 15, one can best understand how to deposit and gel such precursor sol.

As shown in FIG. 15, at time t = 0, the multi-solvent type precursor sol can be spun on the wafer at the initial film thickness D0 and the initial viscosity H0. This spinning is preferably performed in a controlled atmosphere having a partial thickness of the low volatility solvent that significantly suppresses evaporation of the low volatility solvent from the wafer. Therefore, the low volatile solvent is maintained after the spin-on operation, and thus the film thickness is reduced to D1.
The highly volatile solvent is preferentially removed from the wafer during the evaporation time T1 while decreasing. The viscosity also changes to h1 during this time, mainly by removal of the solvent. Ideally, during this time, little crosslinking of the primer clusters in the sol occurs. At the end of T1, almost all of the highly volatile solvent should evaporate, at which time the film thickness should stabilize or continue to decrease at a lower rate, so that the gel time is The ratio of solids should be a given value and the thickness should be a given value.

The main purpose of time T2 is to separate the end point of evaporation time T1 from the gel point occurring during gel time T1. This time T2 is preferably longer than 0. However, certain precursors, particularly those with solvents that promote faster gelling, such as glycerol, will gel by the end of period T1. Further time T
A catalyst in a gas phase, for example, ammonia may be introduced into the controlled atmosphere during 1 or T2. The catalyst is capable of diffusing into the thin film, further activating the sol and promoting rapid crosslinking. Preferably, little or no evaporation occurs during T2, but the viscosity should begin to increase substantially as the crosslinks continue to link the polymer clusters.

Evaporation after the gel point can result in poor gap fill and flatness for patterned wafers. Therefore, it is preferred that the film thickness be kept substantially constant before passing through the gel point by limiting evaporation after the gel time T3. Sometimes the viscosity changes significantly during time T3 as the sol approaches the gel points where the large polymer clusters eventually join and form a continuous spanning cluster across the thin film.

Several advantages of this new method are apparent from FIG. Both sol viscosity and film thickness are observed to change rapidly, but generally not simultaneously. Further, the film thickness also changes from the first known value to a second known value that can be set separately according to the solvent ratio and the spin condition. Using this method, a low-viscosity film is applied, rapidly thinned to a predetermined thickness, and easily gelled at a desired density.

The preceding paragraph envisages a method of changing the sol viscosity of the precursor separately from the dried gel density.
However, this paragraph does not answer the question of which solvent is most suitable. From our experience, it has been found that the solvent evaporation rate for traditional Aerogel solvents can be very sensitive to small changes in vapor concentration and temperature. To better understand this process, we modeled isothermal solvent evaporation from the wafer as a function of% saturation. This modeling is based on mass transfer theory. A good literature on this mass transfer theory includes R.C. B. Bird, W.M. E. FIG. Stewart,
E. FIG. N. There are transport phenomena by light foot (especially Chapters 16 and 17). These calculations have been performed for a range of solvents. FIG. 1 shows the ambient temperature evaporation rates for some of these solvents. In order not to make this evaporation a problem during processing, the product of the evaporation rate and the processing time (preferably several minutes) must be significantly smaller than the film thickness. This suggests that for solvents such as ethanol, the atmosphere on the wafer must be maintained at over 90% saturation. However, there can be problems associated with allowing the atmosphere to reach saturation or supersaturation. Some of this problem is related to the condensation of atmospheric components on the film. It has been found that condensation on a gelled or ungelled film causes defects in the film that have not been sufficiently aged. Therefore, it is generally preferable to control the atmosphere so as not to saturate the components.

We have found that it is a better solution to use low volatility solvents without using high volatility solvents and without precise control of the solvent atmosphere and with less control over the atmosphere. I discovered that. After examining this prerequisite, it was discovered that glycerol was an excellent solvent.

The use of glycerol allows the required control of the atmosphere during the deposition, gelling and / or aging (relative to conventional solvents) to be less severe. The reason for this is that although it is still preferable to avoid saturation, the concentration of atmospheric solvent can be reduced without excessive evaporation. FIG. 2 shows how the evaporation rate of glycerol changes with temperature and concentration of atmospheric solvent. Our experience has shown that the use of glycerol allows the formation of acceptable gels by depositing, gelling and aging with little or no control over the atmosphere. In such most preferred methods, atmosphere control during deposition and gelation (substantially uncontrolled atmosphere) is generally limited to clean room temperature and humidity control, while wafer and / or precursor sols are maintained at separate temperatures. Control can be provided.

An attractive feature of the use of glycerol as a solvent is that the evaporation rate is sufficiently low at ambient temperature, so that several hours at ambient conditions do not result in dramatic shrinkage of the film. In our experience, the use of glycerol can form acceptable gels by depositing, gelling and aging in an atmosphere with little or no control. With glycerol, the evaporation rate at ambient temperature is sufficiently low so that the ambient film does not undergo dramatic shrinkage in a matter of hours. Also, according to our experience, it has been found that the use of ethylene glycol can form an acceptable gel by depositing and gelling in an atmosphere that is completely uncontrolled or substantially uncontrolled. With ethylene glycol, the evaporation rate at ambient temperature is higher than that of glycerol, but low enough that minutes of ambient conditions do not cause dramatic shrinkage of the film. However, ethylene glycol-based sols are much less viscous than the corresponding glycerol-based sols, thus simplifying deposition. Also, the pore fluid in the glycerol-based sol has a much higher surface tension than the corresponding ethylene glycol-based sol, making drying with low shrinkage more difficult.

In addition to acting as a low vapor pressure and miscible solvent, ethylene glycol and glycerol can also participate in sol-gel reactions. The exact response in this process has not been thoroughly studied, but certain reactions can be expected.
When using tetraethoxysilane (TEOS) as a precursor, ethylene glycol can be replaced with an ethoxy group.

[0040]

Embedded image Si (OC ₂ H ₅ ) ₄ + xHOC ₂ H ₄ OH
Si (OC ₂ H ₅ ) _4x (OC ₂ H ₄ OH) _x + xC ₂
H ₅ OH

Similarly, when using tetraethoxysilane (TEOS) as a precursor with a glycerol solvent, glycerol can be replaced with an ethoxy group.

[0042]

Embedded image

Basically, the presence and concentration of these chemical groups alter the reactivity (eg, gel time) of the precursor, alter the microstructure (surface area, pore size distribution, etc.) of the gel, alter the aging characteristics, Almost all other properties can be changed.

Using the new solvent system, a wide range of processing parameters can be varied, including gel time, viscosity, aging conditions and dry shrinkage. Many of these properties, such as gel time, are difficult to measure on thin films. Although the bulk and thin film properties may be different, to better understand how changing the solvent system affects the nanoporous silica process, the bulk sample (eg, 30
It is often beneficial to perform a series of experiments (mm length x about 5 mm diameter).

Glycerol can react with TEOS, resulting in a dry gel having properties that are significantly different from those of the ethanol / TEOS gel. Unexpected property changes in glycerol / TEOS based gels generally include (at least in most formulations):

Low density can be achieved without performing supercritical drying or predrying surface modification. -Significantly simplified aging. -The gelation time is shortened without using a catalyst. -The strength of the bulk sample is increased by an order of magnitude (at a given density) over conventional TEOS gels. Very large surface area (（1,000 m ² / g) High optical clarity of the bulk sample (due to the narrow pore size distribution).

Low density--According to the present invention, it is possible to form a dried gel at an extremely low density without performing predrying surface modification or supercritical drying. These low densities generally 0.3~0.2g / cm ³ (nonporous SiO ₂
It decreased until 0.1 g / cm ³ when) longitudinal or attention has a density of 2.2 g / cm ^3. Porosity (this porosity means% of hollow structure)
The term is about 86% to 91%
(For a porosity of about 95%, the density is 0.1 g / cm ³ ). As shown in FIG. 3, these porosity corresponds to a dielectric constant of about 1.4 for a porosity of 86% and a dielectric constant of 1.2 for a correspondence of 91%. The actual mechanism that allows these high porosity is not completely known. However, the reasons may be that the gel has high mechanical strength and that the gel does not have many surface OH (hydroxyl) groups, combinations thereof or other factors. This method also seems to provide excellent uniformity across the wafer.

If desired, the process can increase the porosity to 90
% To about 50% (by changing the TEOS / solvent ratio). Typical conventional dried gels having small pore sizes required either a supercritical drying or a surface modification step before drying to obtain these low densities. While some conventional xerogels have porosity greater than 50%, these conventional xerogels have significantly larger pore sizes (typically 100 nm).
Above). These large pore size gels have significantly lower mechanical strength. Furthermore, these large pore sizes make the pores unsuitable for filling small (typically less than 1 μm) patterned gaps on microcircuits. If desired, this process can increase the porosity to 5
It can also be adjusted to be lower than 0% (by changing the TEOS / solvent ratio). Care can be taken to prevent premature gelling, resulting in porosity of up to 20%.

Thus, the present invention has enabled a new and simple method for producing nanoporous low density dielectrics. The new glycerol-based method allows both bulk and thin film aerogels to be produced without supercritical drying or a pre-drying surface modification step. Prior airgels required at least one of these steps to prevent substantial pore collapse during drying.

Density Prediction--By varying the ratio of glycerol to silicon (or other metal), the density after drying can be accurately predicted. This accuracy is due to the well controlled evaporation allowed by the low volatility glycerol solvent. Since the process shows excellent shrinkage control during aging and during drying, the process can accurately predict the density (and therefore porosity) of the dried gel. For bulk gels, density prediction has not generally been considered a major problem, but until now it has generally been difficult to predict the final porosity of thin film gels. The ability to provide accurate density predictions for such low porosity dried gels is one reason that this new process is preferred over current xerogel processes for forming low porosity gels.

Simplified Aging--In producing nanoporous dielectrics, it has been discovered that it is preferable to subject the wet gel film to a process known as aging. The hydrolysis and condensation reactions do not stop at the gel point, but continue to reconstitute and age the gel until the reaction has stopped working. During aging, beneficial dissolution and redeposition of the solid structural parts produces beneficial results.
These beneficial results include increased strength, more uniform pore size, and increased collapse resistance of the pores during drying. However, it is difficult to age the wet gel during thin film formation because the membrane contains only a very small amount of pore fluid which must be kept fairly constant for a given period of time to cause aging. If pore fluid evaporates from the membrane before aging strengthens the network, the membrane tends to become xerogel-like. On the other hand, if excess pore fluid condenses from the atmosphere onto the thin film before the network is strengthened, this can locally disrupt the aging process and cause film defects.

Our novel glycerol-based process greatly simplifies the aging of thin film nanoporous dielectrics. Other thin film nanoporous dielectric aging processes allow significant evaporation, fluid condensation,
Or need a controlled aging atmosphere.
These glycerol-based processes behave at least in part during deposition and gelling in a manner similar to the ethylene glycol-based processes described below. However, ethylene glycol-based gels generally require atmospheric control to prevent significant evaporation during aging, even at room temperature. In contrast, glycerol-based gels have significantly reduced evaporation and shrinkage rates during aging. This makes it possible to relax or omit the atmosphere control during aging. We can produce high-quality thin-film glycerol-based nanoporous dielectrics simply by performing passive atmosphere control during room temperature or high temperature aging.

Shorter gel time--The use of glycerol also significantly reduces gel time. Many typical ethanol-based precursors have a gel time of at least 400 seconds when catalyzed (longer for water-in-oil catalysis). However, it has been discovered that even without catalysis, certain glycerol-based precursors gel during spin-on to the wafer. Such rapid gelation is not only faster than ethanol-based gels, but also much faster than ethylene glycol-based gels. FIG. 4 shows the gel times of two different ethylene glycol based compositions depending on the ammonia catalyst used. These gel times are for bulk gels without evaporation of ethanol and / or water, such as for thin films. Evaporation increases the amount of silica and thus shortens the gel time. Thus, these gel times can be an upper limit for a given precursor / catalyst.
The gel time shown in FIG. 4 is substantially shorter than the gel time of the conventional ethanol-based precursor. In general, the gel time also exhibits a first order dependence on the concentration of the ammonia catalyst. This means that the gel time can be easily controlled

These novel glycerol-based gel thin films can be gelled within a few seconds without using a gelling catalyst. We have elucidated several mechanisms that can be used to initiate gelation in thin films without the addition of a catalyst.
One method is to concentrate the precursor sol by evaporating the volatile solvent. Another method is to increase the pH by evaporating the acid in the precursor sol. Such basification by evaporation relies on increasing the pH of the precursor sol to help initiate gelation. However, this basification process generally does not require changing the pH from below 7 to above 7. Such evaporative basification acts similarly to typical basic catalysis processes and thus greatly speeds up gelation. At room temperature and pressure, certain acids, such as nitric acid, have evaporation rates comparable to ethanol.
The concentration of highly volatile solvents and / or stabilizing acids and / or
Alternatively, a simple and very flexible method for adjusting the gel time by changing the type is obtained.

Higher strength--the properties of the glycerol-based sample are significantly different from normal gels, as can be seen both at low dry shrinkage and at the difference in the quantitative handling of wet and dry gels. It is. Thus, upon physical examination, the glycerol-based dry gel appears to have improved mechanical properties as compared to both the conventional gel and the ethylene glycol-based dry gel. FIG. 5 shows the bulk modulus measured during an isothermal compression measurement of one sample prepared using an ethylene glycol-based gel and a conventional ethanol-based dry bulk gel (both having the same initial density). After an initial change contributing to the buckling of the structure, both samples show that the modulus depends on the density according to the power law. This exponential series dependency is what is usually observed in dried gels. However, what is surprising is the strength of the ethylene glycol-based dry gel. At a given density (and therefore dielectric constant), the modulus of this sample of ethylene glycol dried gel is an order of magnitude greater than a conventional dried gel. The previous evaluation indicates that the glycerol-based gel has higher strength than the ethylene glycol-based gel. These assessments include information based on qualitative handling tests and shrinkage during drying. It is not entirely clear why this strength increases. However, according to previous experiments,
It can be seen that the short gelation time and / or narrow pore size distribution is the reason for this high strength.

Large Surface Area--We measured the surface area of some dry bulk gels. These surface areas are 600-
The size was 1000 m ² / g compared to a representative ethanol-based dry gel having a surface area in the range of 800 m ² / g. Thus, a large surface area may mean a small pore size and improved mechanical properties. At this time, it is unknown why these surface areas are large in glycerol-based dry gels.

Pore size distribution--The optical clarity of these dry bulk gels was higher than the ethanol-based dry gels previously produced at this density. Such excellent optical transparency may be due to the extremely narrow pore size distribution. However, the reason why glycerol has such an effect is unknown. Previous experiments have shown one possible explanation that short gelation times are associated with narrow pore size distributions. FIG.
Has a pore size distribution (measured by BJH desorption measurements) of a bulk gel sample with a density of about 0.57 g / cm ³ . The average pore size (desorption law) of this sample is 3.7.
6 nm. Since typical pores are not truly cylindrical, diameter as used herein refers to the diameter of an equivalent cylinder having a surface area to volume ratio that is actually the same as the ratio of total gel surface area to volume.

As noted above, certain properties of glycerol-based gels apply to both bulk gels and thin films. However, some advantages are most apparent when applied to thin films, such as nanoporous dielectric films on semiconductor wafers. The most important advantage is that this novel method allows the processing of high quality nanoporous membranes without atmosphere control during deposition or gelation.

It is important that a nanoporous thin film can be deposited and gelled without controlling the atmosphere.
It is also preferable to age the nanoporous thin film without controlling the atmosphere. This proved to be a bigger challenge than the deposition method. The main reason for this is that deposition and gelling at room temperature can occur in minutes or even seconds, but aging at room temperature generally requires several hours. Thus, an evaporation rate that produces an acceptable shrinkage in a short process can cause an unacceptable shrinkage if the process time is increased by an order of magnitude.

As an example, it has been found that with a certain glycerol-based gel, the magnitude of aging that is satisfactory at room temperature is one day long. However, FIG. 1 shows that with higher temperatures, thin films can be aged in a few minutes of time. These aging times are comparable to the preferred ethylene times of many typical aging-based and ethylene glycol-based gels. Thus, combining these times and temperatures with the evaporation rates of FIGS. 1, 7 and 2, results in a general loss of thickness during aging as shown in Table 2. These expected thickness losses need to be compared to acceptable thickness losses, especially when used for thin films. While no solid guideline exists for acceptable thickness loss, one proposed guideline for certain microcircuits, such as nanoporous dielectrics, should have a thickness loss of less than 2% of film thickness That is. Hypothetical nominal film thickness of 1 μm (actual film thickness can be much smaller than 0.5 μm to several μm
), The acceptable thickness loss is 2
It is shown to be 0 nm. As shown in FIG.
The glycerol group does not control the atmosphere at room temperature,
Such a goal can be achieved. Accordingly, the present invention is to deposit, gel, age and dry thin-film nanoporous aerogels with controlled porosity without controlling the atmosphere. According to another feature, the present invention deposits, gels, and rapidly ages at high temperatures thin-film nanoporous aerogels with controlled porosity, with only passive atmosphere control to limit the volume of the aging chamber. , Which can be dried.

[0061]

[Table 1]

[0062]

[Table 2]

To consider improved yield and reliability, a loss of less than 2%, for example 0.5% or 0.1%
% Lower thickness loss is required. By using passive atmosphere control, the present invention can be extended to these values and even lower evaporation losses. Such passive control is to place the gel in a relatively small sealed container at least during aging. According to such a feature of the invention, the evaporation from the wafer acts to increase the saturation ratio of the atmosphere in the closed vessel. At a given temperature, the vapor continues until the partial pressure of the vapor is high enough to equal the vapor pressure of the liquid. Thus, a combination of solvent and temperature at a lower vapor pressure cannot evaporate the same amount of liquid solvent as is possible with a higher vapor pressure combination. FIG. 8 shows how the vapor pressure changes with temperature for several solvents. FIG. 9 shows the expected value of how thick a layer of solvent can potentially evaporate when a 70% porous gel is placed in a 5 mm high cylindrical container of the same diameter as the wafer. FIG. 10 shows a similar expectation for a container with a 1 mm high air space above the wafer. These figures show that with a 5 mm high air space, the target of 20 nm can be achieved up to 120 ° C. for glycerol based gels and up to 50 ° C. for ethylene glycol based gels. In an air space of 1 mm height, up to 150 ° C for glycerol-based gel, up to 80 ° C for ethylene glycol-based gel
m goals can be achieved. Of course, lower temperature processing can reduce evaporation. According to passive evaporation control using a 1 mm container, it is possible to reduce the thickness loss of the glycerol-based gel to less than 1 nm (0.1% for a 1 μm thick film) even at 100 ° C.

There are various modifications of such a passive control method. One variant allows the size of the container to be increased. Thickness loss increases linearly with container volume. However, a volume of 1000 cm 2 can evaporate glycerol by only 5 nm at 80 ° C. Another variation is
The purpose is to control the porosity of the gel. That is, higher porosity gels generally have higher thickness loss, while lower porosity gels generally have lower thickness loss.

One disadvantage of glycerol is its relatively high viscosity, which can cause problems with gap filling and / or planarization. To reduce the viscosity as described above, a highly viscous solution with low viscosity can be used. FIG. 11A shows the calculated viscosity of a mixture of ethanol and glycerol and a mixture of methanol and glycerol at room temperature. As can be seen, alcohol can significantly reduce the viscosity of these mixtures. FIG. 11B shows the calculated viscosity of a mixture of ethanol and ethylene glycol and a mixture of methanol and ethylene glycol at room temperature. As the figures show, ethylene glycol is less viscous than glycerol, and even small amounts of alcohol significantly reduce the viscosity of these mixtures. Further, if the viscosity of using ethanol in the stock solution is higher than the desired viscosity, further improvement can be realized by using methanol in the precursor solution. The viscosities shown in FIGS. 11A-11B are for pure liquid mixtures only. In practice, depending on the film precursor solution, the precursor solution may include glycerol, alcohol, water, acid and partially reacted metal alkoxide. Of course, the values shown in FIGS. 11A-11B indicate the lower limit because the viscosity can be increased before depositing by catalyzing the condensation reaction.

Such a multi-solvent method may be combined with another method or may be changed to another method.
Another such method is to use high temperatures to reduce the sol viscosity during use. By heating and / or diluting the precursor during deposition (eg, by heating the transfer nozzle and the deposition nozzle of the wafer spin station), the viscosity of the precursor sol can be substantially reduced. Such preheating not only lowers the viscosity of the sol, but also speeds up the gel time and accelerates the evaporation of highly volatile solvents. It is also preferable to preheat the wafer. Pre-heating such wafers requires improved process control, and can improve gap filling, especially with more viscous precursors. However, in many applications, preheating is not required, thus simplifying the process flow. With such a spin-on method without wafer preheating, the spin station does not require a temperature controlled spinner.

The dried gel produced by such a simple thin-film aerogel production process can be used for many purposes. Some of these applications are not as cost effective as using conventional methods. These applications include low dielectric constant thin films (especially on semiconductor substrates), small chemical sensors, thermal insulation structures and thermal insulation layers (including thermal isolation methods for infrared detectors). In general, many low dielectric constant thin films require porosity greater than 60%, and in critical applications require porosity greater than 80 or 90%, so the dielectric constant is significantly reduced. However, structural strength and integrity factors can limit the practical porosity to only 90%. Some applications, including thermal isolation structures and thermal isolation layers, require some porosity to be sacrificed for increased strength and stiffness. These higher stiffness conditions require dielectrics with porosity as low as 30 or 45%. For other high strength / toughness applications, especially sensors where surface area is more important than density, it may be preferable to use a low porosity gel with a porosity between 20 and 40%.

The above description of the thin film focuses on the periphery of the thin film airgel for microelectronic circuits. However, aerogels are also useful for other applications, such as thin films provided on passive substrates. These new high-strength, easy-to-manufacture gels make many of these uses practical. For the purpose of such applications, a passive substrate is defined as a substrate that does not include, or does not include, microelectronic circuitry, or at least has no interaction between the aerogel and the electronic circuitry. C. J. Blinker and G.W.
W. "Sol-Gel Chemistry" by Scherer describes some of these uses in Chapter 14. Some of these passive applications include certain engineering coatings, certain protective coatings, and certain porous coatings.

Anti-reflective (AR) coatings require a wide range of porosity. These coatings are generally 20%
Porosity to 70% porosity, but with appropriate surface protection, higher porosity (90% or more) is effective and provided on high performance coatings or high refractive index substrates. Lower porosity for coating (10%
Up to or less). For some single layer anti-reflective coatings, it is desirable to use a gel having a porosity between 30 and 50%. High performance multi-layer anti-reflective coatings provide a denser layer (e.g., 10-30% porosity) adjacent to the substrate and a less dense layer (e.g., between 45-90%) adjacent the air interface. Porosity). For higher strength / toughness applications, especially where high strength and surface area are primary objectives, 2
It is desirable to use a low porosity gel with a porosity between 0 and 40%. Other thin film coatings may require the lowest density, and thus require porosity greater than 85%, 90% or 95%.

There are also a number of uses for bulk gels that can benefit from these new high strength, easily manufactured aerogels.
Applications for these bulk gels include, but are not limited to, nanoporous (ie, molecular) sieves, thermal insulators, catalyst supports, adsorbents, acoustic insulators, and light separation membranes. Generally, many bulk applications require porosity greater than 60%, while critical applications require 80% or 9% porosity.
Requires a porosity greater than 0%. However,
Structural strength and integrity factors can limit practical porosity to only 95%. Certain applications, which may include sieves, may have to sacrifice porosity for high strength and toughness. These higher toughness conditions may require dielectrics with porosity as low as 30 or 45%. For other high strength / toughness applications where surface area is more important than density (which may include catalyst supports and sensors), use low porosity gels with 20-40% porosity May be desirable.

A typical sol-gel thin film process produces a gel that collapses and densifies on drying, and therefore has a limited porosity (up to 60% for large pore sizes, typically substantially 50% for such pore sizes). %). Under dry conditions where the formation of the xerogel film is uncontrolled, many of the inner pores are permanently destroyed. However, during the formation of the thin film aerogel, it remains almost non-disintegrating, even with a small amount of aging and / or shrinkage during drying, which affects the final density.

Next, refer to FIG. 12A. In this figure, a semiconductor substrate 10 (generally in the form of a wafer) is shown. A typical substrate consists of silicon, germanium and barium arsenide, and the substrate may include active devices, low-level interconnects and insulating layers and many other common structures not shown and known to those skilled in the art. it can. On the substrate 10, several patterned conductive wires 12
(Comprising A1-0.5% Cu composition). The wires 12 extend parallel over at least a portion of their length, separated by gaps 13 of a predetermined width (typically a fraction of a micron). Both the conductors and the gap can have a height-to-width ratio that is greater than that shown, and generally this ratio increases as device size decreases.

According to the first embodiment of the present invention, 61.0
ml of tetraethoxysilane (TEOS) and 61.0 ml.
ml of glycerol, 4.87 ml of water, 0.2 m
Mix with 1 M HNO ₃ and reflux at ６０60 ° C. for 1.5 hours to form a stock solution. Expressed equivalently,
0.27 mol TEOS, 0.84 mol glycerol, 0.27 mol water, 2.04E 4 mol HN
Mix O 2 and reflux at ６０60 ° C. for 1.5 hours. After cooling the stock solution, the solution may be diluted with ethanol to reduce the viscosity. A suitable stock to solvent volume ratio is 1: 8. However, this ratio will vary depending on the desired film thickness, spin rate and substrate. Mix this strongly, generally ~
Store in refrigerator at 7 ° C. and maintain stability until use. This solution is generally warmed to room temperature before film deposition. Distribute 3 to 5 ml of this precursor sol on the substrate 10 at room temperature, and place the substrate 10 on the substrate 10 for about 5 to 10 seconds.
Spin at 00 to 5000 rpm (this speed is determined according to the desired film thickness) to form a sol thin film 14. This deposition can be performed in an atmosphere in which the saturation of the solvent is not specifically controlled (eg, in a clean room with a new type of humidity controller). During and after this deposition and spinning, ethanol, water, and nitric acid evaporate from the membrane 14, but no substantial evaporation of glycerol occurs due to the low volatility of glycerol. Such evaporation temporarily cools the thin film, but the temperature of the film rises within seconds after the evaporation rate has decreased. This cooling is restricted but does not prevent gelling. This evaporation causes the thin film 14 to shrink and concentrate the amount of silica in the sol, forming a thinned film 18. FIG. 12B shows the thinned sol film 18 obtained after substantially all (about 90% or more) of the ethanol has been removed. Concentration, basification by evaporation and / or reheating of such membranes generally results in gelling within seconds.

The membrane 18 has a generally known ratio of silicon to pore fluid at the gel point. This ratio is approximately equal to the ratio of TEOS to glycerol in the sol at the beginning of the deposit (with slight changes due to residual water, continued reaction and concomitant evaporation). This ratio determines the density of the aerogel produced from the sol film, as this method mainly prevents the gel from permanently disintegrating.

After gelling, the thin film wet gel 18 contains porous solids and pore fluids and is preferably aged at one or more controlled temperatures over time. For example, aging is preferably performed at room temperature for about one day. It should be noted that the bore fluid changes somewhat during processing.
Such changes may be due to ongoing reactions, evaporation / condensation, or chemical additives to the film. Aging may include leaving the substrate and gel at about 25 ° C. for about 24 hours, or placing them in a sealed container for about 1 minute.
It is preferably achieved by heating to 30-150 ° C.

[0076] The aged membrane 18 can be dried without substantially increasing its density by one of several methods, including supercritical fluid extraction. However, when using these novel glycerol-based gels, an alternative is to use a solvent exchange method, replacing the aging fluid with a drying fluid and then air-drying the membrane 18 from this drying fluid. This drying method utilizes a solvent exchange method in which the aging fluid is replaced with another fluid. Regardless of whether this fluid is the same as or not the aging fluid, the pore fluid present during drying is often referred to as the drying fluid. When using this dry fluid, the solvent exchange method replaces glycerol and its associated aging fluid, which is dominated by high surface tension, with a lower surface tension dry fluid. This solvent exchange method can be performed as a one-step or two-step process. In the two-step process, about 3 to 8 ml of room temperature (or higher) ethanol is dispensed onto the thin film 18 aged in the first step to replace the aging fluid with an intermediate fluid, and then about 5 to 5 ml. Spin the wafer between about 50-500 rpm for -10 seconds. Often, three to six spin-on sequences are required to replace most of the aging fluid. Preferably, in the second step, the intermediate fluid is replaced by a dry fluid such as heptane. This second step is performed at room temperature (or higher temperature) on the aged thin film 18.
About 3 to 8 ml of heptane, and then spinning the wafer between 50 and 500 rpm for about 5 to 10 seconds. Three to six spin-on sequences are often required to replace most of the intermediate fluid. This solvent exchange method allows us to remove almost all glycerol-containing fluids before drying.
This dry fluid (heptane in this example) is ultimately evaporated from the wet gel 18 to form a dried nanoporous dielectric (dry gel). If the membrane can be dried satisfactorily from a liquid soluble in the aging fluid, no intermediate fluid is required. In many cases, the wet gel can be dried from ethanol or other suitable solvent.

This evaporation can be performed by exposing the wafer surface to an atmosphere that is less saturated with the drying fluid. For example, a wafer can be placed in a substantially uncontrolled atmosphere, and a drying gas can be introduced into the atmosphere. In order to prevent boiling, drying should be started at a temperature slightly lower than the boiling point of the drying fluid, for example, at room temperature. If a higher boiling drying fluid such as glycerol is used (eg, drying without solvent exchange), the onset drying temperature can be increased to a temperature near or equal to the aging temperature. The temperature must be above the boiling point of both the aging fluid temperature and the drying fluid because the film will be overwhelmingly dry (typically within a few seconds). This method prevents catastrophic boiling and allows all fluids to be removed. Glycerol, as well as certain other fluids, decompose at about the same temperature at which they boil, or decompose instead of boiling. When using these fluids, especially fluids such as glycerol that can decompose into harmful substances, care must be taken not to overheat the evaporated fluid or the undried wafer. After drying, the nanoporous dielectric (e.g., 1.
Baking (at 300 ° C. for 5-60 minutes) is preferred. The theoretical permittivity (before surface modification) of this example is 1: 3.

To reduce the dielectric constant, it is preferable to dehydroxylate (anneal) the dried gel. This can be done by placing the wafer in a dry atmosphere with a chemical such as hexamethyldisilazane (HMDS) or hexaphenyldisilazane vapor. This HM
DS replaces hydroxyl groups bound to the surface of most water and / or dried gel pores with methyl groups. Such replacement can be performed at room temperature or higher. Such substitution not only removes water and / or hydroxyl groups but also renders the dried gel hydrophobic (water repellent)
Can be. Hexaphenyldisilazane also removes water and / or hydroxyl groups and makes the dried gel hydrophobic. However, phenyl groups have higher temperature stability than methyl groups at the expense of a slightly higher dielectric constant.

FIG. 13 shows a flowchart of a general method for obtaining an airgel thin film from a precursor sol according to one embodiment of the present invention. Tables 3 and 4 are a summary of some of the materials used in this method.

[0080]

[Table 3]

[0081]

[Table 4]

According to a second higher density embodiment of the present invention, 150.0 ml of tetraethoxysilane (TEO)
S), 61.0 ml of glycerol and 150.0 m
1 ethanol, 12.2 ml of water and 0.48 ml
Mixing the HNO ₃ of 1M of, refluxed 1.5 hours to 60 ° C., to form a stock solution. Expressed equivalently,
And 0.67 mole of TEOS, 0.84 mole of glycerol, 2.57 mole of ethanol, and 0.67 moles of water, the HNO ₃ of 4 moles of 4.90E mixed, to 60 ° C.
And reflux for 1.5 hours. After cooling the stock solution, the solution may be diluted with ethanol to reduce the viscosity. A suitable stock to solvent volume ratio is 1: 8. This is mixed vigorously and generally stored in a refrigerator at ７7 ° C. to maintain stability until use. This solution is generally warmed to room temperature before film deposition. Distribute 3 to 5 ml of this precursor sol on the substrate 10 at room temperature, and place the substrate 10 for about 5 to 10 seconds.
Spin at 1500-5000 rpm (this speed is determined according to the desired film thickness) to form a sol thin film 14. This deposition can be performed in an atmosphere that does not control the solvent (eg, standard exhaust in a clean room with non-new humidity controls). During and after this deposition and spinning, the membrane 1
From 4, ethanol (viscosifying additives and reaction products from TEOS and water) and water evaporate, but no substantial evaporation of glycerol occurs due to the low volatility of glycerol. Such evaporation shrinks the thin film 14 and concentrates the silica component of the sol, forming a thin film 18. FIG. 12B shows the thinned sol film 1 obtained after substantially all (about 95% or more) of the water has been removed.
8 is shown. Concentration of such membranes generally results in gelling within minutes.

The process described in the first embodiment is generally followed by the following processing. After gelation, wet gel of thin film 1
8 comprises a porous solid and a pore fluid and is preferably aged at one or more controlled temperatures over time. One of several methods involving supercritical fluid extraction
Thus, the aged film 18 can be dried without substantially increasing the density. However, when using lower density formulations of these new glycerol-based gels,
Non-supercritical drying, for example solvent exchange, followed by the first
Preferably, the membrane is dried from a drying fluid as described in the examples. This nanoporous dielectric can then be subjected to a dry baking and / or surface modification treatment as described in the first embodiment. The theoretical dielectric constant (before surface modification) of this example is 1.6.

According to a third higher density embodiment of the present invention, 208.0 ml of tetraethoxysilane (TEO)
S), 61.0 ml of glycerol and 208.0 m
1 ethanol, 16.8 ml of water and 0.67 ml
Mixing the HNO ₃ of 1M of, refluxed 1.5 hours to 60 ° C., to form a stock solution. In an equivalent expression,
And 0.93 mole of TEOS, 0.84 mole of glycerol, 3.56 mole of ethanol, and 0.93 moles of water, the HNO ₃ of 4 moles of 6.80E mixed, to 60 ° C.
And reflux for 1.5 hours. After cooling the stock solution, the solution may be diluted with ethanol to reduce the viscosity. A suitable stock to solvent volume ratio is 1: 8. This is mixed vigorously and generally stored in a refrigerator at ７7 ° C. to maintain stability until use. This solution is generally warmed to room temperature before film deposition. Distribute 3 to 5 ml of this precursor sol on the substrate 10 at room temperature, and place the substrate 10 for about 5 to 10 seconds.
Spin at 1500-5000 rpm (this speed is determined according to the desired film thickness) to form a sol thin film 14. This deposition can be performed in an atmosphere that does not control the solvent (eg, standard exhaust in a clean room with non-new humidity controls). During and after this deposition and spinning, the membrane 1
Although ethanol and water evaporate from 4, no substantial evaporation of glycerol occurs because of the low volatility of glycerol. Such evaporation shrinks the thin film 14 and concentrates the silica component of the sol, forming a thin film 18. FIG.
2B is substantially all (about 95% or more)
Shown is a thinned sol film 18 obtained after the water has been removed. Such concentration generally results in gelling within minutes.

The process described in the first embodiment is generally followed by the following processing. After gelation, wet gel of thin film 1
8 comprises a porous solid and a pore fluid and is preferably aged at one or more controlled temperatures over time. Aging can be achieved by keeping the apparatus at 25 ° C. for about 24 hours. The aged membrane 18 can be dried without substantially increasing its density by one of several methods, including supercritical fluid extraction or solvent exchange and subsequent air drying. However, when using higher density formulations of these novel glycerol-based gels, it is preferable to air dry the membrane 18 from the aging fluid. Such a direct drying method exposes the wafer surface to an atmosphere that is less saturated with the drying fluid. A simple method is to remove the cover from the low volume aging chamber and expose the gel surface to a substantially uncontrolled atmosphere. Another method is to introduce a dry gas into the aging chamber or atmosphere. In such a direct drying method, it is preferable that the starting temperature can be increased to a temperature close to or equal to the aging temperature. Such high temperature drying reduces surface tension and associated shrinkage, speeds up drying and simplifies processing. The temperature must be higher than the boiling point of both the aging fluid and the drying fluid (often the same fluid) because the film is predominantly dry (typically within seconds for high temperature drying). Must. This method prevents catastrophic boiling and ensures that all fluids are removed. Since the drying fluid of this method contains glycerol, which can be broken down into harmful substances, care must be taken not to overheat the evaporation fluid or the undried wafer. Next, as described in the first embodiment, the nanoporous dielectric may be subjected to a baking and / or surface modification step after drying. The theoretical dielectric constant (before surface modification) of this example is 1.70.

According to the fourth embodiment of the present invention, 278.0
ml of tetraethoxysilane (TEOS) and 61.0 ml.
ml glycerol, 278.0 ml ethanol, 22.5 ml water, 0.90 ml 1 M HNO
₃ and reflux at ６０60 ° C. for 1.5 hours to form a stock solution. Expressed equivalently, 1.25 moles of TEOS, 0.84 moles of glycerol, 4.76 moles
Mole of ethanol, mixed with 1.25 moles of water, the HNO ₃ of 4 moles of 9.1E, refluxed 1.5 hours to 60 ° C.. After allowing the stock solution to cool, the solution may be diluted with ethanol to reduce the viscosity. A suitable stock to solvent volume ratio is 1: 8. Mix this strongly, generally ~ 7
Store in refrigerator at 0 ° C and maintain stability until use.
This solution is generally warmed to room temperature before film deposition. Distribute 3 to 5 ml of this precursor sol at room temperature on the substrate 10 and allow the substrate 10 to run for 1500 to 1500 seconds.
Spin at 5,000 rpm (this speed is determined according to the desired film thickness) to form a sol thin film 14. This deposition can be performed in an atmosphere that does not control the solvent (eg, standard exhaust in a clean room with non-new humidity controls). During and after this deposition and spinning, ethanol and water evaporate from the membrane 14, but no substantial evaporation of glycerol occurs due to the low volatility of glycerol. Such evaporation shrinks the thin film 14 and concentrates the silica component of the sol, forming a thin film 18. FIG. 12B shows the thinned sol film 18 obtained after substantially all (about 95% or more) of the water has been removed. Concentration of such membranes generally results in gelling within minutes.

The process described in the third embodiment is generally followed by the following processing. After gelation, wet gel of thin film 1
8 comprises a porous solid and a pore fluid and is preferably aged at one or more controlled temperatures over time. One of several methods involving supercritical fluid extraction
Thus, the aged film 18 can be dried without substantially increasing the density. However, it is preferred that the membrane 18 be air dried from an aging fluid, as described in the third embodiment. This nanoporous dielectric can then be subjected to a dry baking and / or surface modification treatment as described in the first embodiment. The theoretical permittivity (before surface modification) of this example is 1.96.

According to the fifth embodiment of the present invention, 609.0
ml of tetraethoxysilane (TEOS) and 61.0 ml.
ml of glycerol, 609.0 ml of ethanol, 49.2 ml of water and 1.97 ml of 1 M HNO
₃ and reflux at ６０60 ° C. for 1.5 hours to form a stock solution. Expressed equivalently, 2.73 moles of TEOS, 0.84 moles of glycerol, 10.4 moles
Moles of ethanol, 2.73 moles of water, 2.00E
Of HNO ₃ are mixed and refluxed at ６０60 ° C. for 1.5 hours. After allowing the stock solution to cool, the solution may be diluted with ethanol to reduce the viscosity. A suitable stock to solvent volume ratio is 1: 8. Mix this strongly, generally ~
Store in refrigerator at 7 ° C. and maintain stability until use. This solution is generally warmed to room temperature before film deposition. Distribute 3 to 5 ml of this precursor sol on the substrate 10 at room temperature, and place the substrate 10 on the substrate 10 for about 5 to 10 seconds.
Spin at 00 to 5000 rpm (this speed is determined according to the desired film thickness) to form a sol thin film 14. This deposition can be performed in an atmosphere that does not control the solvent (eg, standard exhaust in a clean room with non-new humidity controls). During and after this deposition and spinning, ethanol and water evaporate from the membrane 14, but no substantial evaporation of glycerol occurs due to the low volatility of glycerol. Such evaporation shrinks the thin film 14 and concentrates the silica component of the sol, forming a thin film 18. FIG.
B shows the thinned sol film 18 obtained after substantially all (about 95% or more) of the water has been removed. Concentration of such membranes generally results in gelling within minutes.

The process described in the third embodiment is generally followed by the following processing. After gelation, wet gel of thin film 1
8 comprises a porous solid and a pore fluid and is preferably aged at one or more controlled temperatures over time. One of several methods involving supercritical fluid extraction
Thus, the aged film 18 can be dried without substantially increasing the density. However, it is preferred that the membrane 18 be air dried from an aging fluid, as described in the third embodiment. This nanoporous dielectric can then be subjected to a dry baking and / or surface modification treatment as described in the first embodiment. The theoretical permittivity (before surface modification) of this example is 2.5.

Other solvent to reactant ratios can be used to achieve different values of porosity and dielectric constant. FIG. 14 shows the theoretical relationship between the molar ratio of glycerol molecules to metal atoms and the porosity of the nanoporous dielectric when all the ethanol evaporates from the deposited sol. Generally higher porosity glycerol gels (typically about 0.51 g / c
c) requires solvent exchange or other methods of reducing shrinkage during drying. On the other hand, lower porosity gels require care to prevent premature gelation. Such methods can consist of pH adjustment, temperature control or other methods known to those skilled in the art. In some applications, it may be admitted to allow evaporation of highly volatile solvents after gelling.

As described above, even without solvent exchange,
Higher density glycerol-based gels (typically about 0.64
g / cc) and can be dried with little shrinkage. Unaged wafers can be placed in a small volume furnace or small container, which can be placed on a hot plate. After optional evaporation, seal the container at room temperature. The container raises the temperature according to a ramp function, rapidly aging the membrane, and the container remains sealed when the aging / drying fluid viscosity is reduced. After sufficient aging (possible during elevated temperatures), the gel is ready for drying. At temperatures near the boiling point of glycerol, the viscosity of glycerol can be made sufficiently small (compared to the strength of an aged membrane of a given porosity) to remove glycerol in the furnace atmosphere and dry the membrane directly. Here, in the most demanding low density applications, a somewhat lower surface tension can be obtained by raising the dry air above the boiling point of glycerol. In these cases, the furnace must withstand the pressure (can handle most subcritical drying conditions at temperatures of 1 to 3 MPa). In addition, care must be taken to slowly remove glycerol in the furnace atmosphere, especially at first. Glycerol in the furnace can be removed, for example, by bleeding the pressure with a vacuum pump or by sweeping the glycerol with a gas. The temperature of the furnace may be kept constant while removing the glycerol, or may continue to rise (the furnace may ramp up to the baking temperature while sweeping the glycerol with gas). Some glycerol can be introduced during heating to minimize evaporation from the membrane, but the furnace volume must be small enough so that evaporation does not reduce the film thickness without introducing glycerol during heating. preferable. If the membrane requires supercritical drying, possibly to eliminate even temporary shrinkage, CO ₂ may be used, as is well known to those skilled in the art.
It is preferable to use the solvent exchange by

The same stock solution can be used for bulk aerogel as for thin-film aerogel, but the processing method is substantially different. If different stock solutions are used, the following examples can be applied to give different porosity to the bulk gel. According to an embodiment of the bulk aerogel of the invention, 208.0 ml
Of tetraethoxysilane (TEOS) and 61.0 ml
Glycerol, 3.56 ml of ethanol, 1
And water 6.8 ml, was mixed with HNO ₃ of 1M of 0.67 ml, was refluxed for 1.5 hours to 60 ° C., to form a stock solution. In equivalent terms, 0.93 mole of TE
OS, 0.84 mole glycerol, 3.56 mole ethanol, 0.93 mole water, 6.80E 4
Mixed HNO ₃ molar, refluxed 1.5 hours to 60 ° C.. This is mixed vigorously and generally stored in a refrigerator at ７7 ° C. to maintain stability until use. This stock solution is preferably warmed to room temperature before being placed in the mold. After pouring into the mold, ethanol, water and acid are allowed to evaporate, but no substantial evaporation of glycerol occurs due to the low volatility of glycerol. Such evaporation reduces the volume of the stock precursor sol and concentrates the silica component of the sol. It is observed that at least some of the evaporation occurs before filling the mold. This pre-fill evaporation is particularly effective when the mold is not suitable for substantial evaporation after filling, such as when the exposed surface area of the mold is small, or when the mold shape does not conform to shrinkage. is there. Although such evaporation is not necessary, there are several advantages such as rapid gelation and little shrinkage after gelation without using a catalyst.

After this evaporation, the sol has a generally known value of silicon to pore fluid at the gel point. This ratio is approximately equal to the ratio of TEOS to glycerol in the precursor mixture (with little change due to residual water, continued reaction and accompanying evaporation). This ratio determines the density of the produced aerogel, as this method significantly prevents the gel from permanently disintegrating. If the sol does not gel during evaporation, the sol will gel immediately after almost all of the water, ethanol and acid have evaporated.

Alternatively, the precursor may be catalyzed with 0.5M ammonium nitrate before filling into the mold. With this mixture, the sol generally gels within minutes. Remove the wet gel from the mold and evaporate the ethanol and water.

Generally, the gel shrinks during this evaporation. However, as with other methods, once the evaporation is substantially complete, the sol has a substantially known ratio of silicon to pore fluid at the gel point. This ratio is approximately equal to the ratio of TEOS to glycerol in the precursor mixture (slightly changed due to residual water, continued reaction and incidental evaporation). This ratio determines the density of the produced aerogel, as this method significantly prevents the gel from permanently disintegrating.

After gelation, the wet gel contains a porous solid and pore fluid and is preferably aged at one or more controlled temperatures over time. This aging may include maintaining the substrate and gel at about 25 ° C. for about 24 hours, or
Or 130-150 ° C for about 5 minutes in a closed container
It is preferable to complete the process by heating the mixture. These high temperature aging parameters are valid for 5 mm diameter bulk aerogels. However, the combination of aging time and temperature accelerated at high temperatures due to the low thermal conductivity of the wet gel is highly dependent on the shape of the bulk gel.

After this initial aging, the gel is removed from the mold and dried directly from the mother liquor (ie, the pore fluid remaining at the end of aging when no solvent exchange is performed for aging or drying). Slowly ramp up to about 500 ° C and hold at this temperature,
The gel is dried.

Instead of drying directly from the mother liquor, it is preferred to carry out a solvent exchange method, especially for high porosity gels. This solvent exchange method can be performed in a one-step or two-step process. In the first step, the aging fluid is replaced with an intermediate fluid, and in the second step, the intermediate fluid is replaced with a low surface tension drying fluid,
For example, substitution with heptane is preferred. In this method, it is preferable to remove the gel from the mold, place the gel in a sealed tube containing ethanol, and exchange the bored fluid at 50 ° C. for 8 hours. After the 8 hour period, the gel is rinsed with ethanol and stored in fresh ethanol in a 50 ° C. oven. After 3 to 6 hours, the ethanol is similarly replaced with hexane. With this solvent exchange method, we can remove all of the fluid containing glycerol before drying. The dry fluid (heptane in this case) is finally left to evaporate from the wet gel,
Forms a dried aerogel. No intermediate fluid is needed if the membrane can be dried to a satisfactory state from a liquid soluble in the aging fluid. In many cases, the wet gel can be dried from ethanol or other suitable solvent.

After drying, it is often preferable to bake the aerogel for a short period of time (eg, 15-60 minutes at 300 ° C.) to help remove residual material, eg, organics, on or within the aerogel. In one application, the dried gel is dehydroxylated (annealed)
It is also desirable to This can be done by placing the dried aerogel in a dry atmosphere containing a surface modifier such as trimethylchlorosilane (TMCS), hexamethyldisilazane (HMDS) or hexaphenyldisilazane vapor. HMDS causes a large amount of water and / or hydroxyl groups attached to the surface of the pores of the dried gel to be replaced by methyl groups. This replacement can be performed at room temperature or higher. This substitution not only removes water and / or hydroxyl groups, but also makes the dried gel hydrophobic (water repellent). Hexaphenyldisilazane also removes water and / or hydroxyl groups and renders the dried gel hydrophobic. However, phenyl groups are more temperature stable than methyl groups.

According to the inventive examples based on ethylene glycol, a molar ratio of 1: 2.4: 1.5: 1: 0.0
At 42, tetraethoxysilane (TEOS), ethylene glycol, ethanol, water and acid (1M HNO ₃ )
And reflux at ６０60 ° C. for 1.5 hours. After cooling the mixture, the solution was diluted with ethanol and 70% by volume.
Of the original stock solution and 30% by volume of ethanol. This is mixed vigorously and stored in a refrigerator at ７7 ° C. at once, maintaining stability until use. Warm the solution to room temperature before membrane deposition. The stock solution is combined with a mixture of 0.25M NH ₄ OH catalyst (10: 1 by volume). At room temperature, 3-5 mL of this precursor sol is dispensed onto the substrate 10 and the substrate is allowed to stand for about 5-10 seconds (depending on the desired film thickness).
Spin at 1500 to 5000 rpm to form a sol thin film 14. This deposition can be performed in an uncontrolled atmosphere. However, it is preferred to deposit and gel the sol in a clean room equipped with a standard humidity controller. During this deposition and spinning, and subsequently thereafter, the mixture of ethanol and water evaporates from the membrane 14, but no substantial evaporation of ethylene glycol occurs because of the low volatility of ethylene glycol. This evaporation shrinks the thin film 14 and concentrates the silica component of the sol, forming a thin film 18. FIG.
2B is a reduced thickness sol film 1 obtained after almost all (about 95% or more) of ethanol has been removed.
8 is shown. This concentration in combination with the catalyst generally results in gelling within minutes or seconds.

The membrane 18 has a generally known ratio of silicon to pore fluid at the gel point. This ratio is (residual water,
It is approximately equal to the ratio of TEOS to ethylene glycol in the original sol deposited (slightly changed by continued reaction and accompanying evaporation). This ratio determines the density of the aerogel film made from the sol thin film to such an extent that gel collapse is prevented.

After gelation, the thin film wet gel 18 contains a porous solid and pore fluid and is preferably aged at one or more controlled temperatures over time, eg, at room temperature for about one day. . It should be noted that the bore fluid changes somewhat during processing. These changes may be due to ongoing reactions and / or evaporation / condensation. Aging is preferably performed by placing the device in a low volume aging chamber at about 100 ° C. for about 5 minutes.

The aged membrane 1 can be prepared without any substantial densification by one of a variety of methods, including supercritical fluid extraction or air drying after solution exchange.
8 can be dried. However, it is preferred to air dry the membrane 18 from the aging fluid as described in the third glycerol embodiment. Next, as described in the example using the first glycerol,
The nanoporous dielectric can be subjected to a post-drying baking and / or surface modification step.

According to another ethylene glycol based embodiment of the present invention, tetraethoxysilane (TEOS), ethylene glycol, water and acid (1M HNO ₃ ) in a molar ratio of 1: 4: 1: 0.042. And mixed at ~ 60 ° C
And reflux for 1.5 hours. This is generally stored in a refrigerator at ７7 ° C. to maintain stability until use. This solution is preferably warmed to room temperature before film deposition. At room temperature, 3-5 mL of this precursor sol is dispensed (without using a catalyst) onto the substrate 10 and the substrate is spun at 1500-5000 rpm (depending on the desired film thickness) for about 5-10 seconds, A sol thin film 14 is formed. This deposition can be performed in an uncontrolled atmosphere. However, it is preferred to deposit and gel the sol in a clean room equipped with a standard humidity controller.
During this deposition and spinning, ethanol and water evaporate from the membrane 14 and thereafter, but because of the low volatility of ethylene glycol, no substantial evaporation of ethylene glycol occurs. This evaporation causes the thin film 14 to shrink,
The silica component of the sol is concentrated to form a thin film 18. FIG. 12B shows almost all of ethanol (about 95%).
% Or more) after removal of the sol film 18 having a reduced thickness. This concentration generally results in gelling within minutes.

After gelling, the thin film wet gel 18 contains a porous solid and pore fluid and is preferably aged at one or more controlled temperatures over time, eg, at room temperature for about one day. . It should be noted that the bore fluid changes somewhat during processing. Aging is preferably performed by placing the device in a low volume aging chamber at about 100 ° C. for about 5 minutes.

The aged membrane 1 can be prepared without any substantial densification by one of a variety of methods, including supercritical fluid extraction or air exchange after solution exchange.
8 can be dried. However, it is preferred to air dry the membrane 18 from the aging fluid as described in the third glycerol embodiment. Next, as described in the example using the first glycerol,
The nanoporous dielectric can be subjected to a post-drying baking and / or surface modification step.

The foregoing description illustrates some of the advantages of aging in closed containers. Since there is no suitable aging chamber, we will describe the chamber we have invented to achieve this process. 16A, 16B and 16C show one embodiment of the aging container. In this embodiment, the processing apparatus comprises a body 20 having a substantially planar plate 22, on which an elastic seal 24 is mounted. In operation, plate 22 may be as flat as necessary to provide gaps in the membrane, and may be made of any material compatible with the underlying process (eg, semiconductor manufacturing methods), but with high thermal conductivity. Materials such as stainless steel, glass or aluminum are preferred. The resilient seal 24 is preferably designed to withstand wet gel processing temperatures and pore fluids, and those skilled in the art are well aware of many suitable materials, including Teflon-based and neoprene-based materials. It is preferable to determine whether the seal 24 is substantially substantially insulated or thermally conductive depending on the nature of the temperature control used in the device.

In operation, body 20 need only rest on substrate 26 as shown in FIG. 16C. This substrate may be an optical substrate, for example glass or plastic, or a semiconductor substrate such as a Si wafer. In this embodiment, the seal 24 functions as an atmospheric pressure seal and also as both a substrate surface 28, a chamber surface 30, and a spacer that sets the volume of the chamber 32 formed by the seal 24. For example, the seal 24 can be designed to be compressed to a thickness of about 1 mm by the weight of the plate 22 and thus form a 1 mm high chamber 32 when the body 20 is placed on the substrate 26. For many thin films, the chamber 32 need only be substantially sealed, since a small degree of vapor leakage when processing the substrate 26 does not significantly affect the final film properties.

The body 20 can be used at many points during the Aerogel thin film process. This body 20 can be used as an aging chamber for wet gel films, as a storage or transport chamber for such films, or as a drying chamber to limit evaporation before the sol film gels. It is recognized that both sol and gel films in both of these applications contain extremely small amounts of liquid, so that a limited volume chamber may be used to prevent substantial evaporation from the film. Is done.

In another embodiment, body 20 can include more elements, as shown in FIGS. 17A and 17B. In this embodiment, the main body 20 further includes a substrate holder 36.
And a substrate temperature control means 34. This embodiment illustrates another feature of the seal 24 located outside the substrate (or, in some cases, the seal 24 may be omitted) such that a thin film can be formed over the entire substrate surface 24. . When the chamber 32 is closed, the flat plate 22
And the wafer holder 32 are thermally coupled, so that the thermal control means 3
4 using body 20, substrate 26 and chamber 32
Temperature can be controlled at the same time.

In another embodiment, shown in FIGS. 18A and 18B, seal 24 provides some thermal insulation between planar plate 22 and wafer holder 36. This allows the temperature control means 34 to control the temperature of the substrate, while the separate temperature control means 38 controls the temperature of the planar plate.
Is used. Such means has the advantage that the wet gel film can be dried because the temperature of the planar plate 22 can be selectively lowered to promote condensation on the chamber surface 30.

FIGS. 19A, 19B and 19C illustrate another feature of these aging chambers. For example, FIG. 19A shows a state where the substrate 26 is processed at an inverted position. In this embodiment, accidental or intentional condensate on the chamber surface 30 can be collected so that it does not fall onto the substrate surface 26. In FIG. 19B, the substrate 26 has not only been processed in an inverted state, but also has a first solvent layer 42 (at least one
(Preferably of the same composition as the two bore fluids) is dispensed from the first solvent supply tube 40 to the chamber surface 30 before closing the chamber. In this embodiment, layer 42 can be used to help saturate the processing atmosphere, resulting in less evaporation of pore fluid from substrate 26.

In FIG. 19C, at least one port 46 is provided.
An embodiment is shown in which some atmosphere control means 44 is connected to the chamber 32 via (this port can be closed). The atmosphere adjusting means 44 generates a vacuum, appropriately sets the chamber 32 to an excessive pressure,
It can be used to exchange the atmosphere in the chamber 32 or to supply the vapor of the fluid in the hole to the chamber 32. This embodiment can be used, for example, to age a thin film at a temperature above the boiling point of the pore fluid by operating the chamber 32 at a pressure above atmospheric pressure. This embodiment can also be used to remove at least some of the pore fluid vapors from chamber 32 after aging, thus allowing the film to dry.

Although the present invention has been described above in terms of several embodiments, many of these steps can be varied within the scope of the present invention, and other steps can be added to enhance the overall process. For example, instead of spin coating, the initial thin film may be deposited by other common methods, such as dip coating, flow coating or spray coating. Similarly, in the solvent exchange method, instead of spin coating, dip coating, spray coating or a method of dipping in a liquid or vapor solvent can be used. If a vapor solvent is used, the wafer may be cooled to a lower temperature than the atmosphere, thus promoting condensation on the wafer. Without the above method, water would be considered a solvent in such a process, but for the purposes of this application, water would not be considered a solvent.

While both glycerol and ethylene glycol each have unique advantages, there are other low volatility solvents that can be utilized to produce low shrinkage nanoporous dielectrics. It is preferred to analyze the solvent to determine the expected evaporation rate, but the choice of low volatility estimate can be a priority. Almost all solvents that have a low evaporation rate at room temperature have a boiling point above 140 ° C. Certain solvents having a boiling point below 140 ° C. are available, but boiling points above 160 ° C., more preferably 19 ° C.
It will be appreciated that evaporation rates for solvents having a boiling point above 0 ° C. are generally preferred. Solvents having a boiling point above 203 ° C. also have sufficiently low evaporation rates at 40-80 ° C. for short periods of time with little atmospheric control and suitable for deposition and / or aging. For treatment at 100 to 150 ° C. with almost no atmosphere control, it is preferable to use a solvent having a boiling point higher than 270 ° C. Thereby, the lower limit of the preferred boiling point is roughly prioritized. There is also a general preference for the upper limit of the preferred boiling point. Most solvents with boiling points above 500 ° C. become viscous, so they require further attention during processing. Generally, more effective solvents will have a boiling point below 360 ° C, preferably below 300 ° C. If it is not desirable to dilute or heat the sol during deposition, it may be preferable to use less volatile than having a boiling point below 200 ° C. If no solvent provides all of the desired properties, two or more solvents may be mixed to improve performance. Thus, our earlier prior priority criteria for the selection of low volatility solvents are 170-25.
A boiling point in the range of 0 ° C., and (TEOS
The solvent was (for the gel) miscible with both water and ethanol. Based on these prior criteria, 1,4-butylene glycol and 1,5-pentanediol were nominated as suitable low-volatile solvents other than glycerol and ethylene glycol.

This offers yet another possibility if it is preferred to carry out the deposition and aging at a temperature above room temperature. One improvement is to use a solvent that is not liquid but solid at room temperature. This would potentially allow more material to be used.
Many of these higher melting materials have lower volatility than solvents that are liquid at room temperature (liquid solvents) have at high vapor deposition and aging temperatures. The upper limit of the melting point is not necessary, but in order to simplify the process, these solid solvents at room temperature (solid solvents) must have a melting point lower than 60 ° C, preferably lower than 40 ° C. No. A desirable characteristic for a potential solid solvent is that the solvent readily solidifies to an amorphous phase. Such amorphous solidification reduces the chance of damaging the gel during accidental cooling. Furthermore, the solvent can be removed by freeze-drying due to this property. Another way to keep the temperature of the precursor above the melting point of the solid solvent is to dissolve the solid solvent in the carrier liquid. The carrier liquid can be water, alcohol or any other liquid commonly used for treating thin film aerogels / xerogels. The carrier liquid may be a compatible liquid introduced only as a carrier.

The surprisingly good properties of glycerol and ethylene glycol provide clues to other preferred solvents. We have identified some solvents that give properties that are slightly different from ethylene glycol or glycerol, yet retain many of their benefits. Other most promising solvents are 1,2,4-butanetriol; 1.2,3-butanetriol; and 2-methylpropanetriol; and 2- (hydroxymethyl)
-1,3-propanediol; 1-4,1-4, butanediol; and 2-methyl-1,3-propanediol. Other potential solvents include polyols alone or a combination of polyols with ethylene glycol or other solvents.

The use of low volatility solvents in this way allows the necessary control of the atmosphere during deposition, gelling and / or aging to be less rigorous. The reason for this is that saturation should preferably be avoided, but the solvent concentration in the atmosphere can be reduced without excessive evaporation. Such a wide range of concentrations can be used to allow for wider temperature fluctuations throughout the deposition chamber, especially near the wafer and evaporative cooling effects. An initial goal is to allow a temperature variation of at least 1 ° C.
Therefore, the vapor concentration of the less volatile solvent in the atmosphere should be such that the condensation temperature of the solvent vapor (similar to dew) is at least 1 ° C. below the temperature of the substrate. What really matters is the surface of the deposited sol and / or gel sol. However, the thin film nature of the sol keeps the temperature difference between the sol and the substrate small. In this patent, these two temperatures are used interchangeably as it is easier to measure the substrate temperature. Under certain conditions, a temperature uniformity of 1 ° C. can be obtained, but the volume generation is probably at least 3 ° C., preferably 10 ° C.
Requires a tolerance of ° C. However, the ultimate goal is to deposit, gel, and age in an uncontrolled or substantially uncontrolled atmosphere.
In this most preferred method (substantially uncontrolled atmosphere), the atmosphere control during deposition, gelling and aging is limited to standard clean room temperature and humidity control, while the wafer and / or precursor sol are independent. Temperature control device. If such a substantially uncontrolled atmosphere allows for excessive evaporation, it may be necessary to have active atmosphere control, preferably passive or marginal. In such applications, automatic control is limited to placing wafers in relatively small containers. The container may be partially or completely sealed, and may or may not include a liquid reservoir of the solvent. However, this container is a wafer,
There is no external environmental control for the atmosphere and / or reservoir of the container.

Another example of a modification of the basic method is that prior to drying (and more generally, but not necessarily, after aging), the thin film wet gel 18 has a pore surface modified with a surface modifier. Can be. This surface modification step replaces a significant number of molecules on the pore walls with molecules of another species.
When applying a surface modifier, it is generally preferred to remove water from the wet gel 18 before adding the surface modifier. The wafer can be rinsed in pure ethanol and the water removed by low speed spin coating, preferably as described in the solvent exchange method in the first embodiment.
Since water reacts with a surface modifier, for example HMDS, it is advantageous to remove water in this manner. However, this is not necessary. In our new method using glycerol, surface modification does not need to be performed to help prevent pore collapse, but this surface modification can be used to impart other desirable properties to the dried gel. Examples of potentially favorable properties include hydrophobicity, low dielectric constant, high availability for certain chemicals, and improved temperature stability. Potential surface modifiers that can impart desirable properties include hexamethyldisilazane (HMDS), alkylchlorosilanes (trimethylchlorosilane (TMCS), dimethylchlorosilane, etc.), alkylalkoxysilanes (trimethylmethoxysilane, dimethyldimethoxysilane, etc.) ), Phenyl compounds and fluorocarbon compounds. One effective phenyl compound is hexaphenyldisilazane. According other valid phenyl compounds are generally basic formula _{Ph x A y SiB (4-} x-y), where Ph is phenyl, A is Cl or O
A reactive group such as CH, and B is a ligand, where there are two ligands, which may be the same or different groups. Examples of these phenyl surface modifiers include compounds having one phenol group, such as phenyltrichlorosilane, phenyltrifluorosilane, phenyltrimethoxysilane, phenyltriethoxysilane, phenylmethylchlorosilane, phenylethyldichlorosilane,
Phenyldimethylethoxysilane, phenyldimethylchlorosilane, phenyldichlorosilane, phenyl (3-chloropropyl) dichlorosilane, phenylmethylvinylchlorosilane, phenylethyldimethylchlorosilane,
Examples include phenyltrichlorosilane, phenyltrimethoxysilane, phenyltri (trimethylsiloxy) silane, and phenylallyldichlorosilane. Other examples of these phenyl surface modifiers include compounds having two phenol groups, such as diphenyldichlorosilane, diphenylchlorosilane, diphenylfluorosilane, diphenylmethylchlorosilane, diphenylethylchlorosilane, diphenyldimethoxysilane, diphenylmethoxysilane, There are diphenylethoxysilane, diphenylmethylmethoxysilane, diphenylmethylethoxysilane and diphenyldiethoxysilane. These phenyl surface modifiers also include compounds having three phenolic groups, such as triphenylchlorosilane, triphenylfluorosilane and triphenylethoxysilane. Another important phenyl compound, namely 1,3-diphenyltetramethyldisilazane, is an exception to this basic formula. These lists are not exhaustive, but convey the basic nature of the group. Effective fluorocarbon-based surface modifiers include (3,3,3-trifluoropropyl) trimethoxysilane, (tridecafluoro-1,1,2,2-tetrahydrooctyl)-
1-dimethylchlorosilane and reactive group which forms a covalent bond with hydroxyl groups, include other fluorocarbon groups having, for example, Cl or OCH _3.

The above paragraph lists some of the typical useful properties for conventional applications. However, there are other potential uses for nanoporous dielectrics and aerogels, which may have different desirable properties. Other potentially desirable examples of this include hydrophobicity, high electrical conductivity, high breakdown voltage, high or low reactivity with certain chemicals, and low volatility. It is high. This list is not exhaustive. However, many different types of properties have been found to be desirable depending on the application. Thus, it is clear that many other materials that form covalent bonds with hydroxyl groups are potential surface modifiers that can provide other potentially favorable properties.

The present invention also involves the use of a gelling catalyst, such as ammonium hydroxide. The present invention further allows for the use of other gelling catalysts in place of ammonium hydroxide and / or the addition of a gelling catalyst after deposition. Generally, these other catalysts change the pH of the sol. It is preferred to use a catalyst to raise the pH, but an acid catalyst can be used. Generally, the treatment time is longer with an acid catalyst and the density of the dielectric is higher than in a process catalyzed by a base. Examples of other preferred gelling catalysts include ammonia, volatile amine species (low molecular weight amines) and volatile fluorine species. When adding the catalyst after deposition, it is preferred that the catalyst be added as steam, mist or other steam.

The present invention makes it possible to produce a nanoporous dielectric at room temperature and atmospheric pressure without performing a separate surface modification step. While this new method is not required to prevent substantial densification, it does not preclude the use of a supercritical drying or surface modification step prior to drying. This method is also compatible with the freeze-dry method to the extent that the freeze speed is fast enough to prevent large (eg, 50 nm) finals.
In general, the new method is compatible with most conventional airgel technologies.

Another example reforming method is related to the reaction atmosphere and / or temperature. It is not necessary to perform the coating and gelling in the same chamber or the same atmosphere. For example, the temperature of the substrate can be reduced to slow down the gelation or elevated to increase the rate of surface modification and / or gelation. Further, the overall pressure and / or temperature may be varied to control the evaporation rate and / or gel time. Generally, a high temperature treatment of 40 ° C. or more is performed, but 50 ° C. is preferable, and 70 ° C.
Is more preferred. When operating at high temperatures, care must be taken to prevent boiling of the solvent (eg, the partial pressure in the reaction atmosphere must be sufficiently high).

As a typical example of the reactant, TEO
Although S has been used, other metal alkoxides can be used alone or in combination with TEOS or with each other to form a silica network. These metal alkoxides include tetramethoxysilane (TMO)
S), methyltriethoxysilane (MTEOS), 1,
2-bis (trimethoxylyl) ethane (BTMSE),
These combinations and other silicon-based metal alkoxides known to those skilled in the art. The sol can also form sols from other metal alkoxides known to those skilled in the art, such as aluminum and titanium alkoxides. Other precursor sols known to those skilled in the art include particulate metal oxides and organic precursors. Two representative particulate metal oxides include exothermic (smoke) silica and colloidal silica. Representative organic precursors include melamine, phenolfurfural and resorcinol. Alternative solvents can be used in addition to the alternative reactants. Examples of preferred alternatives to ethanol include methanol and other higher alcohols. Other acids may be used as precursor sol stabilizers instead of praise.

Another variation is to allow and / or facilitate the formation of moderately sized (15-150 monomers per molecule) oligomers in the precursor sol. These larger oligomers can speed up the gelation process in the deposited sol. Sols containing large oligomers can be more viscous than sols containing small oligomers. However, as long as the viscosity is stable, such a high viscosity can be assured by methods well known to those skilled in the art, for example, by adjusting the solvent ratio and spin conditions. The oligomerization needs to be slowed or substantially stopped prior to deposition to help achieve such a desired stable viscosity.
Potential methods of promoting oligomerization include heating the precursor sol, evaporating the solvent, or adding a small amount of a gelling catalyst, such as ammonium hydroxide. Methods that may inhibit oligomerization include cooling the precursor sol, diluting the sol with a solvent, and restoring the precursor sol to a pH that minimizes condensation and gelation (as exemplified above). Nitric acid can be used in combination with ammonium hydroxide).

While the invention has been described with reference to several embodiments, various modifications and changes will occur to those skilled in the art. The present invention includes such modifications and variations that fall within the scope of the appended claims.

The US provisional patent application on which the priority claim of the present application was based is as follows.

With respect to the above description, the following items are further disclosed. (1) A metal-based nanoporous aerogel precursor comprising an aerogel precursor reactant and a first solvent comprising a polyol, wherein the molar ratio of the molecules of the first solvent to metal atoms in the reactant is at least 1:16. Material sol. (2) an airgel precursor reactant selected from the group consisting of metal alkoxides, metal alkoxides at least partially hydrolyzed, granular metal oxides and combinations thereof, and a first solvent containing a polyol, A metal-based nanoporous aerogel precursor sol wherein the molar ratio of the molecules of the first solvent to metal atoms in the reactant is at least 1:16.

(3) The airgel precursor sol according to the above item 2, wherein the polyol is glycerol. (4) The aerogel precursor sol according to (2), wherein a molar ratio of a molecule of the first solvent to a metal atom in the reaction body is 12: 1 or less. (5) The second solvent, wherein a molar ratio of the molecule of the first solvent to a metal atom in the reaction body is between 1: 2 and 12: 1.
The sol of the aerogel precursor according to the above item. (6) The aerogel precursor sol according to (2), wherein the molar ratio of the molecules of the first solvent to the metal atoms in the reaction body is between 1: 4 and 4: 1. (7) The aerogel precursor sol according to (2) above, wherein the molar ratio of the molecules of the first solvent to the metal atoms in the reactant is between 2.5: 1 and 12: 1.

(8) The reactant is a metal alkoxide selected from the group consisting of tetraethoxysilane, tetramethoxysilane, methyltriethoxysilane, 1,1-bis (trimethoxylyl) ethane and a combination thereof. 3. The airgel precursor sol according to claim 2. (9) The aerogel precursor sol according to (2), wherein the reactant is tetraethoxysilane. (10) The aerogel precursor sol according to (9), wherein the tetraethoxysilane is at least partially hydrolyzed. (11) The aerogel precursor sol according to (2), further comprising water. (12) The airgel precursor sol according to (2), wherein the first solvent is ethylene glycol. (13) The airgel precursor sol according to (2), further comprising a second solvent. (14) The thirteenth aspect, wherein the second solvent is an alcohol.
The sol of the aerogel precursor according to the above item. (15) The thirteenth aspect, wherein the second solvent is ethanol.
The sol of the aerogel precursor according to the above item. (16) The aerogel precursor sol according to the above item 2, further comprising a pH modifier. (17) The airgel precursor sol according to the above item 2, further comprising an acid. (18) The aerogel precursor sol according to (17), wherein the acid is nitric acid. (19) The second sol wherein the pH of the sol is between 3 and 5.
The sol of the aerogel precursor according to the above item.

(20) The above-mentioned second aspect, further comprising a gelling catalyst.
The sol of the aerogel precursor according to the above item. (21) the gelling catalyst is ammonium hydroxide;
21. An aerogel precursor sol according to claim 20. (22) The second sol, wherein the sol has a pH of 7 to 9.
Item 10. The airgel precursor sol according to item 0. (23) The aerogel precursor sol according to (2), wherein the sol has a viscosity of 1 to 12 centipoise. (24) the viscosity of the sol is 1 to 5 centipoise,
The precursor sol according to claim 2. (25) The precursor sol according to (2), wherein the reactant is selected from the group consisting of exothermic silica, colloidal silica, and combinations thereof.

(26) Aerogel precursor sol containing at least partially hydrolyzed tetraethoxysilane and ethylene glycol. (27) The aerogel precursor sol according to (26), wherein the molar ratio of ethylene glycol to tetraethoxysilane is between 1:16 and 12: 1.

(28) A silicon-based aerogel precursor comprising a silicon-based aerogel precursor and a first solvent containing a polyol, wherein the molar ratio of the molecule of the first solvent to silicon atoms in the reactant is at least 1:16. Material sol. (29) The first to silicon atoms in the reactant.
The twenty-eighth embodiment, wherein the molar ratio of the molecules of the solvent is 12: 1 or less.
The sol of the aerogel precursor according to the above item.

(30) An aerogel precursor sol containing at least partially hydrolyzed tetraethoxysilane and glycerol. (31) The aerogel precursor sol according to the above (30), wherein the molar ratio of glycerol to tetraethoxysilane is between 1:16 and 12: 1. (32) The aerogel precursor sol of (30), wherein the molar ratio of glycerol to tetraethoxysilane is between 1: 4 and 4: 1.

(33) This specification discloses an aerogel precursor sol. The aerogel precursor sol comprises a metal-based aerogel precursor reactant and a first solvent containing a polyol, wherein the molar ratio of molecules of the first solvent to metal atoms in the reactant is at least 1:16. Preferably, the first solvent is glycerol, and preferably the aerogel precursor reactant can be selected from the group consisting of metal alkoxides, at least partially hydrolyzed metal alkoxides, particulate metal oxides and combinations thereof. Generally, the molar ratio of the molecule of the first solvent to the metal atom in the reactant is 12: 1 or less;
Preferably, the molar ratio of the molecules of the first solvent to the metal atoms in the reactants is between 1: 2 and 12: 1. In some embodiments, the molar ratio of the molecules of the first solvent to metal atoms in the reactants is between 2.5: 1 and 12: 1. In some embodiments, the first solvent comprises a glycol. In one embodiment, the reactant is tetraethoxysilane, which can be at least partially hydrolyzed. In some embodiments, the first polyol is 1,
2,4-butanetriol; 1,2,3-butanetriol; 2-methylpropanetriol; and 2- (hydroxymethyl) -1,3, propanediol;
4,1-4, butanediol; and 2-methyl-1,
3-propanediol, and combinations thereof. The present invention enables the deposition, gelling, aging and drying of thin-film nanoporous aerogels with controlled porosity without controlling the atmosphere. According to another feature, the present invention deposits and gels a thin-film nanoporous aerogel with controlled porosity by only passively controlling the atmosphere, e.g. by limiting the volume of the aging chamber, Aging is performed at a high temperature for a short period of time to enable drying.

[Brief description of the drawings]

FIG. 1 is a graph showing changes in evaporation rate with respect to saturation ratio and type of solvent.

FIG. 2 is a graph showing the evaporation rate of glycerol depending on the temperature and the saturation ratio of the atmosphere.

FIG. 3 is a graph showing the theoretical relationship between porosity, refractive index, and dielectric constant for a nanoporous silica dielectric.

FIG. 4 is a graph showing the change in gel time (without solvent evaporation) of a bulk ethylene glycol-based gel according to a basic catalyst.

FIG. 5 is a graph showing a change in elastic modulus with respect to density in the case of a non-glycol-based gel and an ethylene glycol-based gel.

FIG. 6 is a graph showing a pore size distribution of a bulk glycerol-based nanoporous dielectric according to the present invention.

FIG. 7 is a graph showing an evaporation rate of ethylene glycol according to a temperature and an atmosphere saturation ratio.

FIG. 8 is a graph showing a change in vapor pressure with respect to temperature.

FIG. 9 is a graph showing shrinkage of a thin film when dried in a container having a thickness of 5 mm.

FIG. 10 is a graph showing shrinkage of a thin film when dried in a container having a thickness of 1 mm.

FIG. 11A is a graph showing the change in viscosity of ethanol and a mixture of methanol and glycerol with respect to the volume of alcohol. B is a graph showing the change in viscosity with respect to the volume of alcohol of a mixture of methanol and ethanol with ethylene glycol.

FIG. 12 is a cross-sectional view of a semiconductor substrate at several points during the deposition of a thin film according to the present invention.

FIG. 13 is a flowchart of a deposition process of a nanoporous dielectric according to the present invention.

FIG. 14 is a graph showing the porosity of the nanoporous dielectric according to the present invention and the theoretical molar ratio of glycerol molecules to metal atoms.

FIG. 15 is a graph showing relative film thickness and relative film viscosity over time for one embodiment of the present invention.

FIG. 16A is a cross-sectional view of a sol-gel thin film processing apparatus according to the present invention. B is a plan view of the sol-gel thin film processing apparatus according to the present invention. C is a cross-sectional view of the same device contacting the substrate.

FIG. 17A is a cross-sectional view of another device according to the present invention, emptied. B is a cross-sectional view of another device according to the invention for accommodating a substrate.

FIG. 18A is a cross-sectional view of another device according to the invention, emptied. B is a cross-sectional view of another device according to the invention for accommodating a substrate.

FIG. 19 is a cross-sectional view of another device configuration illustrating another feature of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Substrate 12 Conductor 13 Gap 14 Thin film

 ──────────────────────────────────────────────────続 き Continued on front page (72) William C. Inventor. Ackerman, Albuquerque, New Mexico, Nd. Gunnade Cross Creek, Dallas, Texas, USA 12219 (72) Inventor Richard A. Storz 3321 Swanson Drive, Plano, Texas, USA Inventor Aloc Mascara Albuquerque, New Mexico, USA, Silver Ave 1800 (72) Inventor Teresa Ramos, Albuquerque, New Mexico, NU, El Morro, United States of America

Claims (1)

[Claims] 1. A metal-based nanoporous material comprising an airgel precursor reactant and a first solvent comprising a polyol, wherein the molar ratio of the molecules of the first solvent to metal atoms in the reactant is at least 1:16. Aerogel precursor sol.