Wafer Fabrication

The present invention relates to an apparatus and a method for providing a reactive radical species at a substrate surface. The apparatus and method can serve for cleaning the surface, etching, with or without patterning, the surface, and deposition of layers on the surface of any desired thickness. The apparatus and method are particularly useful for the processing of semiconductor wafers but is also useful for coating objects for protective purposes and can further be used to grow crystalline objects such as diamonds.

The semiconductor industry is continually attempting to develop smaller and finer delineations on the semiconductor devices and to, at the same time, increase the throughput of acceptable devices. There are a number of factors involved which effect the transferring onto wafers of contaminants. These include the use of high temperatures in wafer processing, the use of high electric fields near the surface, the use of liquids in contact with wafers during processing, the use of relatively high gas pressures (greater than one Torr), for example, during wafer processing and the handling of wafers by processing personnel.

At higher temperatures the mobility of contaminants in and on the wafer is greatly increased. Such elevated temperatures are used during epitaxy, diffusion and post-ion implantation annealing.

The use of high electrical fields near the wafer results in at least two types of problems. Charged particles on the surface have their mobilities greatly enhanced by the high fields. Furthermore, the high fields used in ion implantation and reactive ion etching impart large momenta to charged particles in the vicinity of the wafer. These large-momentum charged particles can drive surface contaminants into the wafer and can cause crystal damage to the wafer (as in the case of ion implantation). A crystal defect can be considered to be a contaminant or a set of contaminants relative to several semiconductor device functions.

Liquids are well known to exhibit a `bathtub ring` effect. This effect is the removing of a contaminant from one surface (or area on a surface) to another surface (or another area on the surface). Furthermore, dissolved materials in the liquid can deposit out on the surfaces in contact with the liquid.

When vapor phase chemistries are carried out on the surface of the wafer the by-products must be able to escape from the surface, or they will interfere with the desired chemistries, and they become trapped on the surface (or in the material being deposited when deposition chemistries are being used).

Finally, it is well-known that the `dirtiest things` in a clean room are the people in it. Some handling of wafers by people at this stage of development of wafer fabrication is unavoidable. It is, however, essential that this handling be minimized.

The present invention is directed to overcoming one or more of the problems as set forth above.

In accordance with an embodiment of the invention an apparatus is set forth for providing a reactive radical species at a substrate surface. The apparatus comprises a wall structure which defines an interior chamber. Mounting means serves for mounting a substrate in the chamber at a target position therein with a surface region of the substrate oriented in a selected direction. First vaporous chemical introducing means serves for introducing a first vaporous chemical to the chamber with the first vaporous chemical flowing toward the surface region of the substrate. First radical species generating means intermediate the first vaporous chemical introducing means and the substrate surface serves for generating a first radical species from the first vaporous chemical at a pressure of no more than about 0.1 Torr without producing significant amounts of ionized species with the first radical species flowing toward the surface region of the substrate. Pumping means serves for evacuating the chamber and for maintaining a pressure within the chamber of .ltoreq.10.sup.-7 Torr during the introduction of the first vaporous chemical.

In accordance with another embodiment of the invention a method is set forth for providing a reactive radical species at a substrate surface. The method comprises mounting a substrate in a sealed chamber at a target position therein with a surface region of the substrate oriented in a selected direction. A first vaporous chemical is introduced to the chamber with the first vaporous chemical flowing toward the surface region of the substrate. A first radical species is generated from the first vaporous chemical at a position in the chamber intermediate the place of introduction of the vaporous chemical and the surface region of the substrate without producing significant amounts of ionized species with the first radical species flowing toward the surface region of the substrate. The chamber is evacuated at a rate sufficient for maintaining a pressure within the chamber of 10.sup.-7 Torr or below during the introduction of the first vaporous chemical.

In accordance with another embodiment yet of the invention a method is set forth of cleaning a substrate to remove one or more solids or molecules from a surface thereof. The method comprises generating a neutral activated chemical species without generating a substantial quantity of ionized activated species at a pressure of no more than about 0.1 Torr within a sealed evacuatable chamber. The neutral activated chemical species is flowed against the surface, the neutral activated chemical species being selected to react with and convert at least one of the solids or molecules to a vaporous species. The chamber is evacuated at an evacuation rate sufficient to maintain the pressure therein at .ltoreq.10.sup.-7 Torr during the flowing until a desired quantity of the solid or molecules has been removed by vaporization.

Another embodiment still of the invention is a method of etching into a surface of a substrate. The method comprises generating a neutral activated chemical species without generating a substantial quantity of ionized activated species at a pressure of no more than about 0.1 Torr within a sealed evacuatable chamber. The neutral activated chemical species is contacted with the surface, the neutral activated chemical species being selected to react with and convert the surface to a vaporous species until a desired quantity of the surface has been removed by vaporization. The chamber is evacuated at an evacuation rate sufficient to maintain the pressure therein at .ltoreq.10.sup.-7 Torr during the flowing until a desired quantity of the surface has been removed by vaporization.

An embodiment of the invention provides a method of depositing a chemical species on the a surface region of a substrate. The method comprises mounting a substrate in a sealed chamber at a target position therein with the surface region of the substrate oriented in a selected direction. A first vaporous chemical is introduced to the chamber with the first vaporous chemical flowing toward the surface region of the substrate. A first radical species is generated from the first vaporous chemical at a position in the chamber intermediate the place of introduction of the vaporous chemical and the surface region of the substrate without producing significant amounts of ionized species with the first radical species flowing toward the surface region of the substrate. The chamber is evacuated at a rate sufficient for maintaining a pressure within the chamber of 10.sup.-7 Torr or below during the introduction of the first vaporous chemical. A second vaporous chemical is introduced to the chamber intermediate the position of generation of the first radical species and the surface region of the substrate. The second vaporous chemical flows toward the surface region of the substrate. The second vaporous chemical is selected to react with the first radical species such that substantially all of the second vaporous chemical reacts with the first radical species to produce a second radical species flowing toward the surface region of the substrate. The second radical species, on contacting the surface region of the substrate, deposits the chemical species.

The present invention makes use of very far from thermodynamic equilibrium (VFFTE) chemistries thereby minimizing or avoiding entirely the above-mentioned problems of the prior art. With VFFTE chemistries heating of the wafers are not required; no charged particles are used in the vicinity of the wafers; no liquids are used; only total gas pressures below one Torr are used; and handling of wafers by people is minimized.

The invention will be better understood by reference to the Figures of the drawings wherein:

FIG. 1 illustrates, schematically, an apparatus in accordance with an embodiment of the invention which is suitable for carrying out methods in accordance with embodiments of the invention; and

FIG. 2 illustrates, graphically, equilibrium and very far from equilibrium (VFFE) energy/entropy relationships;

Illustration 3 (which is a part of the text rather than a separate drawing) illustrates, schematically, energy levels and energy interactions.

Microwave generators are useful in certain embodiments of the invention. Microwave generators useful in the accordance with the present invention will generally be operated at less than the avalanche mode of operation. The frequency is not critical. Typically, commercial 2.45 Gigahertz microwave generators can be utilized. They generally will have a power above one kilowatt and at times as much as or even more than 20 kilowatts. The current will be generally low, for example, no more than 2 amperes. The electron energy is kept low enough so that substantially no ionic species are formed. Instead, only radical species are formed. When electron guns are utilized in accordance with various embodiments of the present invention they are generally utilized at a maximum of about 20 volts since at higher voltages ionization would occur. Typically the electron gun would operate at relatively high amperage, normally from 3 to 1,000 amps.

When tunable lasers are utilized, they are generally subject to the same parameters as are the electron guns. Typically they will be operated at a wavelength which falls within a range from about 600 to about 3,000 Angstroms.

When either electron guns or lasers are utilized, they are preferably utilized in a symmetrically placed mode preferably with an odd number of electron guns or tunable lasers, the electron guns being positioned so as to provide a relatively even sweep of energy across the chemical species being introduced whereby the activated species impinging on the wafer surface are relatively evenly distributed over the entire surface of the wafer.

Chemicals introduced can be utilized as diluents, as reactants with surface species or as sources of secondary radicals formed by the reaction of the primary radicals formed in the neutral activated species generating portion of the vacuum chamber. In the latter instance substantially all of the species impinging on the surface of the substrate will be in the nature of secondary radicals produced by reaction of the initial activated species with the vaporous chemical being introduced nearer the target, possibly with some excess of the first radical species also being present. If the chemicals introduced serve as reactants with surface species they will arrive at the surface along with the primary radicals. Diluents are useful in removing photoresists and the like so as to moderate the reaction with the primary radicals.

In operation a substrate is mounted in a sealed chamber at a target position therein with a surface region of the substrate oriented in a selected direction. A first vaporous chemical is introduced to the chamber with the first vaporous chemical flowing toward the surface region of the substrate. In the particular embodiment illustrated this would occur at an upper chemical introduction circle. A first radical species is generated from the first vaporous chemical at a position in the chamber intermediate the place of introduction the vaporous chemical and the surface region of the substrate without producing significant amounts of ionized species with the first radical species flowing toward the surface region of the substrate. The chamber is evacuated at a rate sufficient for maintaining a pressure within the chamber of 10.sup.-7 Torr or below during the introduction of the first vaporous chemical.

As mentioned above, a second vaporous chemical can be introduced to the chamber intermediate the position of generation of the first radical species and the surface region of the substrate, with the second vaporous chemical flowing toward the surface region of the substrate, the second vaporous chemical being selected to react with the first radical species to produce a second radical species flowing towards the surface region of the substrate. The geometry of introduction of the first vaporous chemical and the geometry of the first species generating means is preferably such as to provide substantial uniform distribution of the second radical species over the surface region of the substrate.

The specific spacing and positioning of the upper circle, a lower circle at which the second vaporous species can be introduced, the wafer and even the walls of the vacuum chamber can be important to the practice invention. When a microwave cavity is utilized to produce the first activated species the microwave cavity will generally be in the region between the upper and the lower circles or rings, a distance of 3 to 6 inches. The distance between the lower ring to the wafer surface will generally be from about 6 inches to about 12 inches. Generally the width of the chamber should be about three times of the diameter of the wafer being treated. This, along with the high pumping rate provided, serves to assure that the surface of the wafer will be relatively uniformly impinged upon by the activated or radical species.

FIG. 1 illustrates an apparatus 10 in accordance with an embodiment of the invention. The apparatus 10 includes a wall structure 12 which defines an interior chamber 14 accessible via an evacuatable entry port 13. Mounting means 16, in the embodiment illustrated a platform 18, serves for mounting a substrate 20, suitably a semiconductor wafer, in the chamber 14 at a target position 22 therein with a surface region 24 of the substrate 20 oriented in a selected direction (upwardly in the particular environment illustrated but the orientation is arbitrary and the direction which gravity acts is not important to the invention).

First vaporous chemical introducing means 26, in the embodiment illustrated an upper ring 28 having a plurality of holes 30, is located within the chamber 14 and is oriented to direct the first vaporous species toward the surface region 24 of the substrate 20. In the particular embodiment illustrated the holes 30 are evenly spaced about the ring 28 to provide as even a flow and as uniform a distribution as possible of vaporous species over the surface region 24 of the substrate 20.

First radical species generating means 32 is provided intermediate the first vaporous species introducing means 26 and the substrate surface region 24. The first radical species generating means 32 serves for generating a first radical species from the first vaporous chemical at a pressure of no more than about 0.1 Torr without producing significant amounts, and preferably zero, of ionized species with the first radical species flowing toward the surface region 24 of the substrate 20.

There are several alternatives for providing generation of the first radical species. For example, the first radical species generating means may be, or may include, a microwave generator 34 along with a microwave cavity 35 defined by walls 37. The microwave generator 34 may be used in combination with a defocussed electron gun and power supply 36. One or more additional defocussed electron guns 38 may be present. Also, and in what is a preferred embodiment, a plurality of defocussed electron guns 36,38 may be utilized without the microwave generator 34 (and without the microwave cavity walls 37). Instead of, or in addition to the electron guns, one can provide a .beta.-emitter, for example, a nickel .beta.-emitter, positioned so as to provide electrons in the cavity 35. The .beta.-emitter can be present in elemental or combined form. The result will be a less expensive but generally less efficient apparatus. As still another embodiment useful alone or in combination with the microwave generator 34 and the defocussed electron gun 36 or guns 36,38, a defocussed laser and power supply 40, generally a tunable laser, may be used to generate the first radical species. The location of the first radical generating means 32 and the location and orientation of the first vaporous species introducing means 26 are such as to provide a substantially uniform distribution of the first radical species over the surface region 24 of the substrate 20.

In accordance with a preferred embodiment of the invention the chamber 14, where the substrate 20 is mounted, is at least about three times the width of the largest cross-dimension of the surface region 24.

In accordance with an embodiment and certain uses and methods of the invention it may be desirable or necessary to introduce a second vaporous species between the first radical species generating means 32 and the surface region 24 of the substrate 20. In such instances second vaporous chemical introducing means 42 is provided for introducing the second vaporous species intermediate the first radical species generating means 32 and the surface region 24 of the substrate 20. The second vaporous species is flowed towards the surface region 24 of the substrate 20. It is selected to react with the first radical species to produce a second radical species flowing toward the surface region 24 of the substrate 20. In the particular embodiment illustrated the second vaporous species introducing means 42 is similar in construction to the first vaporous species introducing means 26. More particularly, it includes a ring 44 with a plurality of oriented holes 46. When two vaporous species introducing means are present in the desired orientation it becomes possible to select and use a first vaporous species which can readily form the first radical species and, through reaction of the first radical species with the second vaporous species, to produce a second radical species which could not be readily directly produced by radical species generating means alone, i.e., by defocussed electron guns, defocussed lasers and/or microwave generators.

Means 48 is provided for evacuating the chamber 14 at a rate sufficient for maintaining a pressure within the chamber of 10.sup.-7 Torr or below during the introduction of the first vaporous species. In the particular environment illustrated the evacuating means 48 is in the nature of a high speed pump 50. Relatively large conduits 52 lead from the chamber 14 to the pump 50. Preferably, the conduits 52 open onto openings 54 which are located about and downstream from the substrate 20 whereby vaporous species generated at and/or flowing past the substrate 20 are efficiently carried through the large openings 54 and the large conduits 52 to the high speed pump 50.

The invention provides a number of capabilities. For example, a radical species can be produced which can carry out a chemical reaction on the surface region 24 of the substrate 20. Deposition of a chemical can be carried out in this manner. If desired, the radical species can be used to clean (remove one or more solids from) the surface 24 of the substrate 20. The surface region 24 of the substrate 20 can be etched using this technique. This can be done in connection with etching techniques such as masking to provide a desired etching pattern. An advantage of the apparatus and method of the invention is that a single apparatus can be used in sequential steps to carry out different chemistries. For example, a selected radical species, e.g., atomic hydrogen, may be generated for a time sufficient to clean the surface 24 of the substrate 20. Thereafter, through introducing one or more different vaporous chemicals thereby changing the first and/or second radical species, deposition can proceed in the same apparatus without changing the substrate or its orientation.

The composition of the gaseous species being evacuated from the chamber 14 can be monitored by monitoring means 56 which can include a probe 58 which is positioned in the chamber 14 adjacent the openings 54. The monitoring means 56 can utilize any convenient analytical technique, e.g., mass spectroscopy, gas chromatography or visible, laser fluorescent spectroscopy, ultraviolet or infrared spectroscopy. In the Figure a laser 60 powered by a power source (not shown) can provide a laser beam 63, shown in phantom, across and above the surface 24 of the substrate 20 and a detector 62 located at 90.degree. from the direction of the laser beam along with an optical spectrometer 64 for monitoring the chemical reactions. When the composition being monitored changes such can be used as an indication or determinant that the desired layer has been fully or sufficiently removed.

In Classical Thermodynamics the First Law (The Conservation of Energy) is written

where dE is the differential change in the internal energy of the system, q is an infinitesimal amount of work done by the environment on the system. (In the earlier literature the convention used was that of the work done by the system rather than on the system. Thus, [I-1] was formerly written dE=q-w.) Although dE is an exact differential (independent of path), q and w may not be. It is therefore desirable to define the paths in ways such that dE is expressible in terms of exact differentials. There are several ways in which this can be done. q can be replaced with C dT, where C is the heat capacity, and T is the absolute temperature. (It will be convenient to represent extensive variables--those proportional to the amount of matter in the system--in bold print and intensive variables--those not proportional to the amount of matter in the system--in ordinary print). w can be replaced by a.sub.1 dA.sub.1 +a.sub.2 dA.sub.2 +. . . +a.sub.n dA.sub.n where a.sub.1 is an intensive variable such as pressure, a mechanical force, the electromotive force, the magnetic field or a chemical potential; and A.sub.1 is the associated extensive variable such as volume, distance, the electric charge, the magnetization or the number of moles of a chemical species, respectively. Using the Einstein summation convention we can now write [I-1] as ##EQU1## Equation [I-2] has two peculiarities. The first differential on the right side of the equation is that of an intensive variable, whereas all the other n+1 differentials are those of extensive variables, and there is no intensive variable associated with the internal energy.

We shall define a new extensive variable,

called `Entropy`. We can now rewrite [I-2] as ##EQU2## Recognizing that integers are intensive variables, we can rewrite [I-4] as ##EQU3## where d/A.sub.o =dE and a.sub.0 =-1. At the first blush [I-5] looks strange. The right hand side is the sum of products of intensive variables and their associated extensive variables. The left hand side is simply zero, an integer. The answer to this apparent inconsistency is that the a.sub.1 's are the elements of a vector. The dA.sub.1 's are the elements of a second vector which is orthogonal to the first vector. Hence, their scalar product is zero. Although [I-5] has some interesting mathematical consequence, [I-4] is more useful for the present discussion.

We shall consider a closed system which is of constant volume and is surrounded by an `energy controlling` bath. Traditionally a closed system is defined as one for which energy can be transferred between the system and the environment, but mass cannot be transferred. By Relativity, however, energy and mass are two different facets of the same entity. Thus, the traditional definition of a closed system is `leaky`. We shall, therefore, define a closed system as one for which only zero rest mass particles can be transferred between the system and the environment.

Traditionally the environment here would be called a `temperature controlling` or `temperature controlled` bath. Since, as we shall show temperature is not always defined, we prefer the appellation `energy controlling`. (If one wishes to be strictly accurate, one should note that an `energy controlling` bath is not the same as a `temperature controlling` bath. With a `temperature controlling` bath a system can undergo a phase transition during which the temperature of the system does not change, but the energy does. With an `energy controlling` bath a phase change cannot occur unless the bath is reset to allow the net transfer of energy between the system and the bath.) At equilibrium

and

(For near-equilibrium conditions we should use the partial derivative with the notation that all of the A.sub.1 's of [I-6] are held constant). In FIG. 2, the curve labelled `EQUILIBRIUM CURVE` represents [I-7] with the acknowledgement of the Third law requirement that entropy start at the value of zero, and with the acknowledgement that Relativity and Quantum Mechanics require that the internal energy start at a value greater than zero. We shall return to FIG. 2 when we discuss Non-Equilibrium Thermodynamics. Returning now to [I-4], we can rewrite the equation as

We now define a new extensive variable, A.sub.H, such that

A.sub.H is the `Helmholtz Free Energy`. It is standardly represented by `A`. We have appended the H subscript in order to distinguish it from the other A.sub.1 's. Substituting [I-9] into [I-8] we obtain

By virtue of [I-6]

One of the terms on the right hand side of [I-10] is PdV.sub.e, where P is the pressure of the system and V.sub.e is the volume of the environment. Since

where V.sub.s is the volume of the system, we can rewrite [I-10] as

It is traditional to define A.sub.H for constant volume conditions. Thus,

Since so much chemistry is carried out under constant pressure rather than constant volume conditions, it is convenient to introduce two more extensive variables, H-`Enthalpy`, and G-`Free Enthalpy`, more usually called `Gibbs Free Energy`, by ##EQU4## From [I-9], [I-12] and [I-15] we obtain

For constant pressure conditions the requirement for equilibrium is

Thus,

Standardly, it is .DELTA.H, .DELTA.G and S (constant pressure) which are measured and tabulated, rather than .DELTA.E, .DELTA.A.sub.H and S (constant volume). If we restrict our considerations to systems at low pressures, the ideal gas law,

or

is applicable. For vapor phase reactions for which the sum of the number of reactant molecules is equal to the sum of the number of product molecules, .DELTA.n=0, and hence PAY=0. Thus,

We can therefore, use the data of the standard tabulations of .DELTA.H.sub.f.sup.0 '(standard Enthalpies of formation), to calculate the .DELTA.E.sub.r.sup.0 's (standard Internal Energies of reaction) of reaction for which An=0.

From Equilibrium Statistical Mechanics, using the Principle of the Equipartition of Energy, we can obtain at equilibrium

where N(j) is the number of items under discussion in states for which the enthalpy is .DELTA.H(J), g(j) is the number of such states (the `multiplicity`), k is the Boltzmann constant, T is the absolute temperature and .DELTA.R(ji) is the change in enthalpy of the item in going from one of the i-th states to one of the j-th states. For the purposes of [I-22] it is assumed that

Since the exponential term in [I-22] is positive definite by virtue of [I-23], it follows that

for all values of T. In order that

T would have to be negative. This cannot, however, be the case since T is an absolute temperature. In VFFTE systems such as those which we shall be discussing later, [I-25] is the case. Thus, we must conclude that temperature is not defined for VFFTE systems.

Equation [I-22] can be rearranged to a more familiar form.

or

Let

and

Using the definition of the Gibbs Free Energy, [I-15], we have

(The reader may be surprised to find an extensive variable `sitting by itself` in [I-22], [I-26] and I-29]. This is because the Boltzmann constant's units are J/K/molecule. It, thus, is an `extensive constant`. Equation [I-26] is the equation developed by Max Karl Ernst Ludwig Planck in his work on the `Black Body Problem` for which he invented the constant `k`, which is known as `Boltzmann's Constant` rather than `Planck's Constant`. The appellation `Planck's constant`, of course, is reserved for `h`--the quantum of action. Planck invented h in his work on the `Black Body Problem`.)

Although equilibrium thermodynamics can be, and is, applied to near-equilibrium systems, it cannot be applied to VFFTE systems. The discussion of [I-22] showed that T is not defined for VFFTE systems. If T is not defined, then by virtue of [I-19] P is not defined. If it were, then by [I-19] we could define T. Similarly, A.sub.H, H and G--by virtue of [I-9], [I-14] and [I-15], respectively--cannot be defined. We are, thus, left with only E, S, V and the N(j)'s for VFFTE systems.

Some very `strange` things happen in VFFTE systems. Let us consider, as a simple example, a system with three energy levels and three items which can be in any of the three energy states (cf. ILLUSTRATION 3]. As we did relative to ILLUSTRATION 2, we shall assume that the system has a constant volume and is in an energy-controlled bath. The energy levels have the energy values E.sub.1 =0, E.sub.2 =1 and E.sub.3 =2. We start with all three items in the E.sub.3 state. E=6 and g=1. By [I-26], if g=1, S=0. We now stepwise remove one unit of energy. We must consider two Cases--the items are distinquishable and the items are indistinguishable.

In either case as the energy of the system is reduced the entropy increases initially reaches a maximum value, and then decreases. This is not the behavior which we noted for equilibrium systems (FIG. 2. EQUILIBRIUM CURVE). There we observed a decrease in the entropy with any decrease in the internal energy.

Consider the point A in FIG. 2. The system is near equilibrium. Holding the energy of the system constant, we find that the entropy of the system increases until it reaches the equilibrium value. Now consider the point B. Here we have a VFFTE system like that in ILLUSTRATION 3. Holding the energy of the system constant, we find that nothing happens. The entropy does not change! There is no mechanism available by which it can. As we lower the energy of the system the entropy follows the `VFFTE CURVE`. When the energy has lowered sufficiently, the curve joints the `EQUILIBRIUM CURVE`. Of course, not all systems are like that of ILLUSTRATION 3. The `trick` is to choose those systems and conditions so that the system will follow the `VFFTE CURVE`. Before we consider how to make such choices, we must discuss first some of the factors involved in the kinetics of chemical reactions.

Absolute Reaction Rate Theory is also known in the literature as `Activated Complex Theory` and Transition State Theory`. In order to clarify the use of these different terms let us consider a simple model reaction.

It is assumed that the reactants combine for a short period of time to form the `activated complex` AB.sup. which then decomposes to form the products. Thus, [I - 30 ] is written

It is further assumed that the reactants are in equilibrium with the activated complex, or--more accurately--that the reactions

and

are fast compared with

Thus, [I-30a] becomes

If we consider traditional chemical kinetics, we can write ##EQU5## where the square brackets here indicate concentrations (and not equation numbers), t is time and k.sub.r is the rate constant. The rate with which AB.sup. decomposes is equal to [AB.sup. ] times a frequency term, kT/h, where k is Boltzmann's constant, T is the absolute temperature and h is Planck's constant. We shall not prove here that kT/h is the correct frequency term because of the complexities of the proof which involves concepts from statistical quantum mechanics. Thus, we can also write ##EQU6## where K.sup. is the `equilibrium constant`, we can combine [I-31] and [I-32] to obtain ##EQU7##

From the viewpoint of dimensional analysis [I-34] appears to be in error. The dimensionality of k.sub.r appears to be time.sup.-1, but in [I-31] its dimensionality is concentration.sup.-1 time.sup.-1. Going back to the derivation of [I-29] we see that [I-34] really should have been written ##EQU8## where .DELTA.H.sup. is the enthalpy of activation. Since the g's are proportional to concentrations, the dimensionality of g (AB)/9(A)g(B) is concentration.sup.-1. This fact is hidden by [I-26] in which g is treated as though it were dimensionless. By virtue of entropy being an extensive variable, g must be proportional to the amount of material. Thus, [I-34] is valid whether the reaction is unimolecular, bimolecular, trimolecular, etc.

At the thermodynamic equilibrium the principle of the equipartition of energy applies. Each translational and rotational degree of freedom `receives` kT/2 as its energy `allotment`. Each vibrational degree of freedom `receives` a `double allotment`, kT. This is because a vibrational degree of freedom is actually two--a kinetic energy degree of freedom and a potential energy degree of freedom. Let W be the total amount of energy available in the system, and n be the number of degrees of freedom, then at equilibrium

and

Under VFFTE conditions W is not equipartitioned among all degrees of freedom. It is restricted to just a few degrees of freedom. Let us assume that there are p such degrees of freedom where

and define

then ##EQU9## and

Since T and .DELTA.H are not defined under VFFTE conditions, we must rewrite [I-34] as ##EQU10## By virtue of [I-40] the first factor on the right is much greater than the corresponding factor in [I-35]. The second factors in the two equations are essentially equal. There is a small difference between .DELTA.S under constant pressure conditions and .DELTA.S under constant volume conditions. The third factor in [I-41] is much greater than that in [I-35] even when the difference between .DELTA.H and .DELTA.E is taken into account. Thus, a reaction will be much faster under VFFTE conditions than under Near Thermodynamic Equilibrium (NTE) conditions.

Since under NTE conditions W is partitioned among many degrees of freedom, it is possible for several different reaction paths to have sufficient activation energy (or enthalpy) for reactions to occur. As a result several other products besides the desired one may be produced. The desired product is, therefore, contaminated by these other products. If the purity of the desired product is crucial for its intended use, the presence of these contaminants can be a very serious problem.

If the system is properly prepared for the VFFTE reaction, all or most of W will be in the degree (or degrees) of freedom which will lead to the desired product. Thus, the product will be produced much purer than that produced under NTE conditions as well as being produced much faster. (How to prepare the system properly for the VFFTE reaction may not be obvious and may require a considerable amount of ingenuity.)

Let us consider a specific example of these principles. Suppose that we wish to generate methyl radicals from methane. We could use dioxygen or monoxygen as the oxidant.

For a proposed chemical reaction to be useful it must be thermodynamically permitted and kinetically favorable. The greater the exo-ergonicity (and/or the exo-thermicity) of a chemical reaction is, the more likely it is that the reaction will be fast. This is, of course, not a guarantee. It is, however, guaranteed that a reaction which is highly endo-ergonic (and/or endothermic) will be very slow. As a test of this Principle let us look at the rate constants for these four reactions. The data are taken from <I-1>.

and

and

Thus we find that the Principle holds for this case.

By raising the temperature of the system for [I-42] the reaction rate of [I-43] can be attained. What is the required temperature?

While specialized equipment is required to operate at so high a temperature, it can be done. At so high a temperature, however, numerous side reactions can be expected to occur. For example, for the reaction

Thus, at equilibrium only about 0.1% of the carbon is in the form of methane. The vast majority of the carbon has been deposited on the surfaces of the reactor. Since .DELTA.G.sub.f.sup.0 (C, g, 4600K)=+0.5kJ mol.sup.-1, there will also be a negligible quantity of carbon vapor at 1729K.

Since .DELTA.H.sub.f.sup.0 (O, g, 298K)=+249 kJ mol.sup.-1, a significant amount of energy, we should also consider how much energy is required to heat the methane and dioxygen from 298K to 1729K. The proper way in which to calculate these quantities is to integrate C.sub.p with respect to T from 298K to 1729K. A simple way to approximate these quantities is to multiply the average value of C.sub.p over the temperature range by the temperature range. Thus, ##EQU11## The factor of 10.sup.-3 is present because heat capacities are usually given in J mol.sup.-1 K.sup.-1. Similarly, ##EQU12## To these two energy requirements we must add the energy required to heat the reactor from 298K to 1729K. Although iron melts at 1809K, let us assume that the reactor is made of 100 gram atoms of iron (around twelve pounds). ##EQU13## where 0.9 kJ/gram atom and 0.8 kJ/gram atom are the heats of transition from the alpha phase to the gamma phase and from the gamma phase to the delta phase, respectively. Thus, although it requires more energy to generate the monoxygen than is required to heat the methane and dioxygen from 298K to 1729K, the monoxygen requirement is quite small when the total 1729K energy requirement is considered.

As another example, let us consider the generation of formyl from formaldehyde. Again the kinetic data are from <I-1>.

and

and

and

Again we find that the reactions which are more exo-ergonic are faster. The temperature of 3679K is much too high for an iron reactor. Because formaldehyde is more stabile relative to the elements than methane is, [I-64] is much smaller than [I-51] even though the temperature is much higher.

B. Collision Theory

Although Absolute Reaction Rate Theory, being based on Quantum Mechanics, is more elegant and more definitive than Collision Theory, the latter nonetheless remains quite valuable for the insights which it provides. Unlike Absolute Reaction Rate Theory all reactions, being based on the Kinetic Theory of Gases, Collision Theory is applicable only to gas phase reactions and heterogeneous reactions in which one of the phases is the gas phase.

We begin with the assumption that the gas phase consists of two kinds of molecules, which we designate by A and B, and that the molecules can be approximated by spheres. The diameters of these spheres are d.sub.A and d.sub.B, respectively. If the centers of an A molecule and a B molecule are within a distance

collision will occur. The collision frequency will be given by

where v.sub.AB is the mean relative speed of the two molecules, N.sub.A and N.sub.B are the number of molecules of A and B, respectively, and V is the volume of the gas phase.

From the Kinetic Molecular Theory of Gases ##EQU14## m.sub.A and m.sub.B being the masses of molecule A and molecule B, respectively. It should be noted that, although the units of [I-67] appear to be ms.sup.-1 (or cm s.sup.-1), the units which must be used are actually ms.sup.-1 molecule .sup.-1 because v.sub.A, for example is the sum of the speeds of all of the A molecules divided by the number of A molecules. Most texts ignore this `nicety`. We will find in what follows that our results will have the wrong units if we ignore this `nicety`.

The substitution of [I-67] into [I-66] yields ##EQU15## If A and B are the same kind of molecule, ##EQU16## where the initial factor of 1/2 is to correct for counting each molecule twice--once as a collider and once as a collidee, R is the molar gas constant (8.31 J mol.sup.-1 K.sup.-1) and M.sub.A is the molar mass in kg.

The number of collisions experienced by one molecule in unit time is ##EQU17## and the mean speed is ##EQU18## Thus, the average distance traveled between collisions, the `mean free path`, is ##EQU19##

The triple collision frequency can be estimated by means of the assumption ##EQU20## on the idea that the probability of a third molecule colliding with two molecules which have collided, before they `bounce apart`, is proportional to the ratio of its collision diameter and its mean free path. Thus, ##EQU21##

It will be instructive to evaluate [I-73], [I-70] and [I-75] for some specific cases. At 273K and one atmosphere pressure (STP) 6.02.times.10.sup.23 molecules occupy ##EQU22##

From <I-2> we obtain the following collision diameters ##EQU23## The average of these three values is 0.294 nm. Thus, we can consider dioxygen to be a `typical` molecule. ##EQU24## It should be noted that J=kgm.sup.2 s.sup.-2.

If every collision leads to a chemical reaction, the rate constant will be ##EQU25## where the subscript on the k indicates a bimolecular reaction. From <I-3> we obtain the following

and

Thus, we see that each collision of a monohydrogen and a trifluoromethyl undoubtedly leads to a reaction. It should be noted that, if we had used ms.sup.-1, rather than ms.sup..sup.-1 molecule.sup.-1, as the units for [I-67], we would not have obtained the correct units for [I-79]. ##EQU26## Thus, for every trimolecular collision at STP about 350 bimolecular collisions occur.

At 0.1 Torr N/V is reduced by a factor of 1/7600 (1 atmosphere=760 Torr). Thus, ##EQU27## Thus, for every trimolecular collision at 273K and 0.1 Torr about 2.6 million bimolecular collisions occur. At 0.1 Torr, therefore, there are effectively no trimolecular collisions, and the mean free path is almost one millimeter!

No discussion of Chemical Reaction Kinetics would be complete without at least mentioning the Arrhenius Equation.

an empirical equation which is useful in describing the change in k.sub.r with temperature E* is the `activation energy`. A is known as the `frequency factor`, the fraction of the collisions which have the proper orientation of the molecules. A more sophisticated form of [I-87] takes into account that A is actually a function of temperature.

The comparison of [I-88] with [I-35] yields ##EQU28##

If in [I-34] we assume that .DELTA.H.sup. =0 (no enthalpy of activation), ##EQU29## where the factor of V/N is introduced because of the form of the ratio of multiplicities in [I-35]--two molecules combine to form one activated complex,

For the formation of a chemical bond the entropy decreases by about 1.7.times.10.sup.-22 J bond.sup.-1 K.sup.-1. In the activated complex about half a bond is formed. Thus, if the molecules do not need a specific orientation to react, ##EQU30## It should be noted that this value is 5.6 times as large as that calculated using `rigid sphere` Collision theory, [I-79].

Referring back to [I-80]-[I-82], we see that the experimental values of the rate constants are all somewhat smaller than that calculated in [I-90]. In the calculation of [I-90] we assumed that `the molecules do not need a specific orientation to react`. Since not all of the surface of CF.sub.3 on collision with H will yield HF, an orientation factor is clearly involved. This is even more so the case with CF.sub.2 and a fortiori the case with CF. Since ##EQU31## this would be the fraction of the surface of the CF molecule which will yield HF on collision with H if CF were a sphere. The area of a portion of a sphere is given by

If we equate [I-91] and [I-92] divided by 4.pi.r.sup.2 (the total surface area of a sphere), we obtain ##EQU32## and

If we use a circle as the model of CF, then [Io91] is a measure of the fraction of the circumference which is reactive. ##EQU33##

Both of these models are, of course, very poor approximations to the CF molecule, but the fact that I-93] and [I-94] are not that much different is indicative that the value is in the `right ball park`.

Before we begin our discussion of the details of the VFFTE chemistries for wafer fabrication, it will be useful to describe the VFFTE apparatus (Apparatus). At this time we will limit our discussion to those aspects of the Apparatus related to wafer-wide chemistries. Those aspects of the apparatus related to patterning will be discussed later.

The Apparatus comprises a vacuum system with a large chamber and a high speed pumping system which will keep most of the chamber at a maximum of 100 nanoTorr. The diameter of the chamber should be at least three times the diameter of the wafer. The chamber is connected to two load locks which can be pumped down independently of the main chamber and each other. The load locks--main chamber complex is equipped with a robotics system which will transfer wafers between the load locks and the main chamber, and which will hold the wafer in place in the main chamber with a minimum of linear and rotational motions during the carrying out of the chemistries.

Above the main chamber and connected to it is the `activation system` which consists of an upper vapor inlet circle, the activation chamber and the lower vapor inlet circle. The activation chamber can be a microwave cavity, a chamber which is irradiated by an electron gun (or guns) providing a defocused beam(s) of high current of low energy (1-25 eV) electrons or a chamber which is irradiated by a tunable laser (or lasers) with a defocused beam(s) with a high flux of photons in the desired energy range.

Since the charge on an electron is 1,602.times.10.sup.-19 C,

Furthermore,

therefore ##EQU34## where .nu. is in wave numbers (cm.sup.-1). Thus, ##EQU35##

Into the upper circle is introduced the vapor species which is to be activated. The proper valving and controls are connected to the circle so that several different species are available for activation at controlled flow rates. (Only one species is being activated at a time.) Into the lower circle is introduced the vapor species (or more than one species) which is (are) to react with and/or dilute the activated species. (In some cases no species is introduced into the lower circle. All the chemistries are carried out by the activated species reacting with the materials on the surface of the wafer including, in some cases, the surface of the wafer itself.) As with the upper circle the proper valving and controls are connected to the lower circle so that several different species are available at controlled flow rates.

Since many of the possible species used are toxic, corrosive and/or flammable, the source containers must be enclosed in safety chambers with the appropriate monitor/warning systems in case of leaks. Furthermore, since the effluents may also be toxic, corrosive and/or flammable, the Apparatus must have means for safely disposing of the effluents.

Finally, the apparatus must have a control computer which will control the valves, controls and pumps in the proper sequences as well as the movements of the wafers. The instructions to the control computer can be on tapes or discs or can be inputs from design computers. V. UPPER CIRCLE CHEMISTRIES

A. Introduction

As was indicated in SECTION I. high electric fields near the wafer are to be avoided. This includes the presence of charged particles near or on the surface of the wafer. Thus, activation involving ionization is to be avoided as much as possible. This means that the conditions in the activation chamber should be adjusted to maximize electronic excitation (including dissociation) and minimize ionization. Because the cross section versus energy curves for electronic excitation and ionization frequently peak at widely different energies, these adjustments can be made for the vapor species which we have chosen for discussion here.

Since the excitation-dissociation energies are 3 to 15 eV above the thermodynamic dissociation energies, the particles leaving the activation chamber can have energies 145 to 725 kJ mol.sup.-1 above the ground states. The tabulated thermodynamic data are generally for ground states. Thus, reactions may be much more exothermic than the ground state calculations indicate. Furthermore, the excited state species reaction energies may be so great as to cause thermal damage to the wafer. In these cases we dilute and partially de-energize the excited state species by introducing the ground state diatomic species by means of the lower circle. B. Carbon Monoxide

Carbon monoxide is a valuable reagent for the etching of metals which form volatile carbonyls such as Cr(CO).sub.6 Fe(CO).sub.6, Mo(CO).sub.6, Ni(CO).sub.4 and W(CO).sub.6. It should be noted that all of these metals have unpaired d-electrons. The ground state of CO is X.sup.1 .SIGMA..sup.+. Since this state has no unpaired electrons, its rate of reaction with metals at `room temperature` is not very great. The first electronic excited state is a.sup.3 .pi. which has two unpaired electrons.

The ionization potential for CO is 14.01 eV <I-4>. The dissociation energy for the X .sup.1 .SIGMA..sup.+ state is 11.09 eV <I-5>. The only triplet excited state with a reasonable lifetime is the a.sup.3 .pi. state with an excitation energy of 6.01 eV and a lifetime of 7.5-9.5 ms (depending on the vibrational state) <I-6>. By [I-72] ##EQU36## In 9 ms the CO molecule will travel about 1.3 m. Thus, the excited CO molecules will have sufficient time to traverse the distance from the bottom of the activation chamber to the surface of the wafer and react with the metal there.

From cross section curves <I-7> we estimate that the cross section for excitation to the a.sup.3 .pi. state at .about.7 eV is .about.1.times.10.sup.-16 cm.sup.2. The dissociation and ionization cross sections are, of course, zero at .about.7eV.

Monofluorine is an extremely powerful oxidizing agent. In addition there are numerous volatile fluorides. The ionization potential of difluorine is 15.7 eV <I-8>, and the dissociation energy is 1.56 eV <I-9>.

Difluorine has a continuous absorption region with a maximum at 2845 .ANG. (4.36 ev) <I-10>. From potential energy curves we estimate the Franck-Condon region excitation energy from the .sup.1 .SIGMA..sub.g.sup.30 ground state to the .sup.1 .sup.[.sub.u repulsive state to be .about.2.7.times.10.sup.4 cm.sup.-1 (3. 3 eV) to .about.4.2.times.10.sup.4 cm.sup.-1 (5.2 ev) <I-11>.

Just as monofluorine is a powerful oxidizing agent, monohydrogen is a powerful reducing agent. In addition there are numerous volatile hydrides. The ionization potential of dihydrogen is 15.4 eV <I-12>. The dissociation energy is 4.48 eV <I-13 >.

Dihydrogen can dissociate to the b.sup.3 .SIGMA..sub.u.sup.30, H(1s)+H(1s), state or to the B' .sup.1 .SIGMA..sub.u.sup.+, H(1s)+H(2s), by excitation from the X .sup.1 .SIGMA..sub.g.sup.+ ground state <I-14>. For the excitation to the B' .sup.1 .SIGMA..sub.u.sup.+ state

<I-13>. Since H (25) and H(2p) have the same energy in the absence of a magnetic field, both products are given. It should be noted that the lifetime of the H(2p) state is only 1.6.times.10.sup.-19 s, whereas the H(2s) state is metastabile with a lifetime of 1.4.times.10-1 s <I-15>. The peak in the photo dissociation curve is .about.840 .ANG. (14.76V) with a cross section of .about.3.1.times.10.sup.-17 cm.sup.2 <I-15>. The excitation to the b.sup.3 .THETA..sub.u.sup.+ state is reported to have a cross section of .about.9.times.10.sup.-17 cm.sup.2 at .about.15 eV <I-16>.

Since the transition from the H(2s)/H(2p) states to the H (1s) state is 10.2 eV <I-17>, and since the dissociation energy of dihydrogen is 4.48 eV, the sum is 14.68 eV. Thus, it is possible that the experimental data have been misinterpreted. It may be that the only excitation is that of [I-98]. The apparent dissociation from the b.sup.3 .SIGMA..sub.u.sup.+ state could be the result of the rapid decay of the H(2p) state to the H(1s) state. In any case the required excitation energy for the dissociation of dihydrogen is -14.72-14.8 eV. Since this is so close to the ionization energy of 15.4 eV, the excitation energy must be controlled carefully to avoid the formation of a significant amount of H.sub.2.sup.+.

Monoxygen is a powerful oxidizing agent, but it is not as powerful as monofluorine. There are numerous volatile oxides. The ionization potential of dioxygen is 12.06 eV <I-18>, and the dissociation energy is 5.12 eV <I-19>.

The ground state of dioxygen is the X .sup.3 .SIGMA..sub.g.sup.- state. Some of the excited states, their excitation energies <A> and their lifetimes <I-20> are ##EQU37##

Except for the first two excited states all of the excited states are repulsive. The cross section for dissociation is approximately ##EQU38## For ionization plus dissociative ionization Q.sub.ION (gross).apprxeq.1-2.times.10.sup.-17 cm.sup.2 <I-21>. The ratio of Q.sub.DISS to Q.sub.ION (gross) is probably much higher than these numbers would indicate since R. L. McCarthy <I-22> did not find any charged particles outside of the microwave cavity on dissociating dioxygen in a 2.45 GHZ microwave field. Furthermore, he observed a yield of 0.80 gram atoms/kW-hr for monoxygen. ##EQU39## Thus, his system was operating at about 43% efficiency in generating monoxygens if the system was using electrons with an average effective energy of about 20 eV. McCarthy also reported a yield of 1.0 gram atoms of monohydrogen per hW-hr. If the efficiencies are the same for oxygen and hydrogen, the average effective electron energy in the hydrogen case was 16 eV.

E. C. Zipf <I-23> indicates that the excitation energy in the Franck-Condon region of the X 3.SIGMA..sub.g.sup.1 (.sup.3 P+.sup.3 P) state to the repulsive (.sup.3 P+.sup.5 S) state is around 20 to 27 eV. For 27 eV McCarthy's efficiency is .about.59%. The excitation energy from the monoxygen ground state, .sup.3 P (1s.sup.2 2s.sup.2 2p.sup.4) to the .sup.5 S.sup.o (1s.sup.2 2s.sup.2 2p.sup.3 2s) state is 9.14 eV <I-24>. With a dissociation energy for dioxygen of 5.12 eV and a .sup.5 S excitation energy of 9.14 eV the sum is 14.26 eV. Thus, the monoxygen atoms generated at 20 eV have average kinetic energies of about 2.9 eV. The .sup.5 S.sup.o state has a lifetime of about 600 .mu.s <I-24>. Substituting 16 for 38 in [I-97] we obtain ##EQU40## In 6.times.10.sup.-4 S the .sup.5 S.sup.o atom will have traversed 0.13 m (5.2 inches). Thus, it is likely that very few, if any, .sup.5 S atoms will reach the substrate.

In addition to the .sup.5 S.sup.o excited state there are also the following excited states of monoxygen:

For the dissociation of dioxygen to these excited states the following are reported:

<I-25>. Thus, the average kinetic energies of the generated monoxygen atoms may be as low as 0.6 eV. This is, however, still rather high. If a considerable amount of monoxygen is used as in the removal of photoresists, the wafer can absorb a considerable amount of energy. To prevent this undesirable heating of the wafer the monoxygen is diluted with dioxygen from the lower circle. We shall discuss the details of this chemistry later when we discuss photoresist removal.

As was indicated in SECTION III., by means of the lower circle species are introduced which react with and/or dilute the activated species from the upper circle. The discussions following will be concerned with the thermodynamics and kinetics of vapor phase interactions between the upper circle activated species and the lower circle species in a few cases, but mainly they will be concerned with the thermodynamics and kinetics of the heterogeneous reactions of vapor phase species with the surface of the wafer. The thermodynamic data used in the calculations are for the ground state species rather than excited state species. Thus, in most cases the actual reactions are much more energetic than the calculations indicate.

Since the chemistries used for cleaning and etching are the same, the two subjects are discussed together. The topics to be considered are

SECTION VII. REMOVAL OF PARTICULATES

SECTION VIII. REMOVAL OF ORGANICS

SECTION IX. REMOVAL OF INORGANICS.

The analyses of the inorganic substances are organized by the Periodic Table Groups. If the standard state for the element is a condensed phase, it is discussed first. The compounds considered are the oxides, the chlorides, the nitrates and the fluorides--the oxides because of the exposures to monoxygen and dioxygen (e.g. air), the chlorides because of the ubiquitousness of NaCl etc., the nitrates because of the nitric acid etches, and the fluorides because of the hydrofluoric acid etches and the use of monofluorine. In general only those cases are discussed for which there are adequate thermodynamic data.

Frequently, both the vapor phase and one of the condensed phases exist at 298K. An important factor is the equilibrium vapor pressure of the species. This can be calculated by means of the free energy of sublimation or vaporization. Since the standard state of the vapor species is that at one atmosphere pressure, the equilibrium constant is equal to the equilibrium vapor pressure of the vapor phase expressed in atmospheres. Thus, with p as the equilibrium vapor pressure and K as the equilibrium constant

If we require that the vapor pressure be greater than 0.1 Torr, then ##EQU41## If, however, we require that p be greater than 10.sup.-7 Torr, then

It is obvious that as line widths decrease particulate contaminations become a more severe problem. For 10 .mu.m line widths a 0.5 .mu.m diameter particle is not a major cause for concern. For 0.1 .mu.m line widths a 0.5 .mu.m diameter particle can be a disaster. The minimization of the number of particles in the apparatus and the removal of those that are on the wafer's surface is a must. Particulates can `enter` the Apparatus by airborne contamination, by being carried in on the surface of the wafer and by being generated in the apparatus from the rubbing together of moving parts.

Since the main chamber of the Apparatus is entered only by means of the load locks (except during maintenance), since wafers are loaded into, and unloaded from, the locks in stacks rather than individually, and since the locks are pumped down before being opened to the main chamber, the airborne contamination is minimized. Since the wafer is held in place during the carrying out of all the chemistries, the only moving of parts is the opening of the load locks to the main chamber and the moving of the wafer between the load locks and the center of the main chamber, with one wafer being moved to the `unload` lock while another is being moved to the center of the main chamber. Thus, parts are moving only before and after all of the chemistries are carried out. In addition, since the wafers are never touched from the post-polishing cleaning until they are removed from the `unload` lock, the number of particulates on the surface of the wafer is minimized. It is assumed that the post-polishing cleaning is carried out in a simplified version of the Apparatus.

Because maintenance on the Apparatus is a sine qua non, the room in which the Apparatus is located must be a `relatively` clean room. An `ultra-clean` room, however, is not required. Thus, as an additional feature the enormous costs of `ultra-clean` rooms are not required.

Although the species stream approaching the wafer has a pressure of about 0.1 Torr, the majority of the volume of the main chamber is at a maximum of 100 nanoTorr. This large pressure differential generates a force which will `sweep` the particulates off of the surface of the wafer. For the calculations following we shall assume that the particle is iron and is a sphere. The following facts will be used in the calculations: ##EQU42## and

Let r be the radius of the particle. The cross sectional area of the particle is

and the mass of the particle is ##EQU43## The pressure gradient acting on the particle is ##EQU44## and the force on the particle because of the pressure gradient is ##EQU45## The acceleration of the particle because of F is ##EQU46## For the particle to be moved by the force F, F must be large compared with the gravitational force. Thus, ##EQU47## Thus, we find that the pressure gradient will cause the removal of particulars of the size with which we are concerned provided that the article is not `glued` to the surface by an organic coating or by electrostatic charge. Let us, therefore, now turn to the question of the removal of organics from the surface of the wafer and the surfaces of the particulates. (Since electrostatic charges are found in, or on the surfaces of, insulators, the organic insulators will be removed by the organics removal chemistries. Thus, their electrostatic charges will be removed. Similarly, the inorganic insulators and their electrostatic charges will be removed by the inorganic removal chemistries.)

The study of the removal of organics by monoxygen oxidation is complicated by the fact that the monoxygen may be in an excited state and that the products may be in excited states. For the ground state of monoxygen, .sup.3 p (1s.sup.2 2s.sup.2 2p.sup.4),

By means of the data in SECTION V.E. we can calculate the heats of formation of some of the excited states of monoxygen:

Similarly, the heats of formation of the first two excited states of dioxygen are

Since the two monoxygens attacking a carbon atom on a surface are coming from the same side of the carbon atom, the electronic state of the carbon dioxide formed may be the A .sup.1 B.sub.2 (.DELTA..sub.u) state rather than the X .sup.1 .SIGMA..sub.g.sup.+ ground state. The ground state is, of course, linear whereas the excited state is bent. The excited state is 8.41 eV above the ground state <II-1>. Thus,

The first excited state of hydroxyl is the A .sup.2 .SIGMA..sup.30 state which is 4.02 eV above the X .sup.2 .pi..sub.3/2 ground state <II-2>. Thus,

In SECTION III.A. we discussed the reaction of methane with monoxygen. We shall continue that discussion here.

The .sup.5 S state of monoxygen, of course, is a much more powerful oxidizing agent than even the .sup.1 S state of monoxygen. Because of the short lifetime of the .sup.5 S state, not many of them can be expected to arrive at the wafer surface.

Of all of the hydrocarbons methane is the most difficult to oxidize. For example, ##STR1## The enthalpies of formation of ethane and ethyl are from <B>, and the rate constant is from >II-4>.

The most difficult step in the oxidation of a hydrocarbon is the abstraction of a hydrogen atom. The formation of the carbon-oxygen bond is much faster. ##STR2##

Even dioxygen is an effective oxidizing agent for the formation of a carbon-oxygen bond. In the following reactions M represents a third body which is required for the removal of the energy of the bond(s) formed. k.sub.28 is the second order rate constant under the conditions that the concentration of M is so high that a small change in its concentration has no effect on the rate of the reaction. ##STR3##

The organic molecules discussed above are all small vapor phase species. The organics on the wafer surface, however, are primarily large molecules. Adequate thermodynamic data are not available for most polymers. We shall, therefore, use eicosane, C.sub.20 H.sub.42, as a model compound. The thermodynamic data for eicosane are

For the solid phase we take ##STR4## Clearly, the monoxygen oxidation of eicosane is highly exothermic and highly exo-ergonic even with the use of ground state monoxygen.

The enormous magnitude of .DELTA.H.sub.r.sup.0 is however the source of a serious problem. The generation of so much heat will result in the temperature of the wafer being elevated to undesirably high levels. The goal, therefore, is to find a set of conditions for which the .DELTA.H.sub.r.sup.0 is negative and of sufficient magnitude to ensure a rapid reaction, but not of so large a magnitude as to produce temperature rise problems.

We have noted that the carbon dioxide formed is probably in the .sup.1 .DELTA..sub.u state. The situation for the excited state of hydroxyl is more complicated and requires some discussion. The .sup.3 p ground state of monoxygen has the electron configuration ##EQU48## The .sup.1 D excited state has the electron configuration ##STR5## The .sup.1 S excited state has the electron configuration ##STR6##

The .sup.2 .pi. ground state of hydroxyl has two electrons in the .sigma.-bond formed from an oxygen 2p orbital and the hydrogen 1s orbital. Thus, there are a total of three electrons in the other two 2p orbitals. In the .sup.2 .SIGMA. state the .sigma.-bond has only one electron, and there are a total of four electrons in the other two 2p orbitals of the oxygen. Since the hydrogen is bringing its electron with it to form the .sigma.-bond, the .sup.3 p state of monoxygen can form only the .sup.2 .pi. state of hydroxyl. The .sup.1 D and .sup.1 S states of monoxygen, on the other hand, can form only the .sup.2 .SIGMA. state of hydroxyl.

With these facts in mind let us now consider several variations on [II-16]. ##STR7## Thus, we find that by diluting the monoxygen with dioxygen, a small portion of which have been excited by the monoxygen to the .sup.1 .DELTA. state, we can have the oxidation proceed at a reasonable rate without causing significant heating of the substrate.

If the organic mass is very small, the heat generated is never a problem. When, however, the organic mass is large as is the case with a photoresist layer, the heat generated can be quite large. To analyze the magnitude of the problem let us assume the following:

resist composition--C.sub.20 H.sub.42 (Molecular Mass 282.6)

resist density--1.0 g cm.sup.-3

resist thickness--1 .mu.m (10.sup.-4 cm)

wafer diameter--8 inches (20.3 cm).

Eicosane is actually more difficult to oxidize than a polymer. One out of ten of the carbons is a methyl carbon. In polymers this ratio is much lower. The density of eicosane is about 0.8 g cm.sup.-3, but the density of polymers like the photoresists is about 1.0 g cm.sup.-3.

The number of `moles` of eicosane in the resist layer is ##EQU49## Using the enthalpy of reaction for [II-19] we calculate the total heat generated to be ##EQU50## The heat capacity of silicon is 20.0 JK.sup.1 mol.sup.-1 and the atomic mass is 28.1 g mol.sup.-1 <A>. The density of silicon is 2.33 g cm.sup.-3 <II-10>. Thus, if the wafer is 2.5.times.10.sup.-2 cm thick and the heat is evenly distributed, the temperature rise is ##EQU51## an insignificant temperature rise.

We must now turn our attention to what could be a much more serious problem. We noted in SECTION V.E. that the two monoxygen atoms formed by the dissociation of dioxygen at 20 eV must have average kinetic energies of about 2.9 eV (2.8.times.10.sup.2 kJ mol.sup.-1). By [II-20] there are 1.145.times.10.sup.-4 moles of eicosane, and by [II-19] each mole of eicosane requires 42 gram atoms of monoxygen as well as 20 moles of dioxygen. Thus, the total kinetic energy of the generated monoxygen is ##EQU52## By [II-22] this much energy reaching the wafer, if it were evenly distributed throughout the wafer (which it would not be), would cause a temperature rise of ##EQU53## This much of a temperature rise throughout the wafer would not be serious, but the energy would actually be concentrated at the surface.

The energy required to produce the .sup.1 .DELTA. dioxygen in ]II-19] is ##EQU54## The kinetic energy of [II-23] is, thus, 62.5 times the amount of energy needed for the .sup.1 .DELTA. excitation. We must, therefore, find a way to modify the chemistry of [II-19]. One possibility would be to use less monoxygen and more .sup.1 .DELTA. dioxygen. In the abstraction of a hydrogen atom a dioxygen molecule produces a hydroperoxo (HO.sub.2) rather than the hydroxyl which a monoxygen produces. Thus, if only dioxygens are used, ##STR8## Let us try a `mixture` of [II-19] and [II-26]. ##STR9##

The kinetic energy of the monoxygens in [II-27] is (27/42) times that of [II-23] or 0.87 kJ. The excitation energy necessary to generate the .sup.1 .DELTA. dioxygens is ##EQU55## Thus, the excess energy is 0.49 kJ. By [II-24] the temperature rise is ##EQU56## For [II-27] the ratio of dioxygen entering the lower circle to that entering the upper circle is 35:(27/2)=2.6. Thus, with a ratio of the order of 3:1 the removal of the photoresist can be carried out at a reasonable rate without significant heating of the wafer. The exact value of the ratio has to be determined experimentally since all photoresists are not alike, every monoxygen generated will not reach the wafer (some will diffuse out of the `target` area) and not all collisions between monoxygens and dioxygens will result in the excitation of the dioxygen to the .sup.1 .DELTA. state.

Fluorocarbons play a key role in the deposition of low dielectric constant dielectrics. The vaporization reagent of choice here is monohydrogen. As a demonstration of the efficacy of monohydrogen in the breaking of carbon-fluorine bonds we consider the stepwise reactions of tetrafluoromethane with monohydrogen.

Based on the thermodynamic data the expectation is that [II-28]-[II-31] are fast reactions. This expectation is substantiated by the experimental data. The rate constants for [II-29]-[II-31] are 9.times.10.sup.-11, 3.9.times.10.sup.-11 and 1.9.times.10.sup.-11 cm.sup.3 molecule.sup.-1 s.sup.-1, respectively (cf. [II-80]-[II-82]).

Having demonstrated the efficiency of monohydrogen in breaking carbon-fluorine bonds, we must now consider the efficacy of monohydrogen in breaking carbon-carbon bonds in fluorocarbons.

Finally, we consider the volatilization of polytetrafluoroethylene by monohydrogen.

The datum on polytetrafluoroethylene is from <B>.

In SECTION V.E. we noted that the dioxygen at 20 eV must have average kinetic energies of about 2.9 eV (280 kJ mol.sup.-1). Similarly, in SECTION V.C. we noted that the dissociation of difluorine to two monofluorines is 1.56 eV, but that the Franck-Condon region excitation energy is 3.3 eV to 5.2 eV. If we assume that the excitation energy is 5.0 eV, the average kinetic energy of the monofluorines formed is ##EQU57##

In considering monoxygen or monofluorine as volatilizing reagents these kinetic energies must be added to the enthalpies of formation to obtain the enthalpies of reaction for reactions in which these species are reagents. Not all of the enthalpy of reaction, however, will go into volatilizing the oxides or fluorides formed. Some will go into the kinetic energies of the volatile species, and some will go into the heating of the substrate. In our analyses we shall assume that the enthalpy of reaction must be at least four times that of the enthalpy of sublimation of the oxide or fluoride. The enthalpies of reaction calculated using the ground state enthalpies of all species will be designated in the usual way by .DELTA.H.sub.r.sup.0 (298K). The enthalpies of reaction calculated using the excited state enthalpy for at least one species will be designated .DELTA.H.sub.r.sup.o (298K).

1. THERMODYNAMIC DATA

We note that none of the condensed phase compounds listed in TABLE II-1 are self-volatilizing under the criteria of [II-1] or [II-2].

2. LITHIUM

The enthalpies of sublimation of the oxides and fluoride of lithium are

The numbers in the parentheses are the `times four` values.

Although [II-37] and [II-39] are sufficiently energetic that some lithium peroxide may volatilize, even excess monoxygen reacting with lithium oxide will not generate sufficient energy to volatilize the lithium peroxide formed. Thus, the monoxygen used to remove organics will not by itself also remove all the lithium species.

Of the three reactions with monofluorine only, [II-42] is sufficiently energetic to meet the `times four` criterion. This suggests that it is advisable to add nitrogen dioxide to the dioxygen in the lower circle. Thus,

By this means we can convert all of the lithium compounds to lithium nitrate during the monoxygen reactions. Dioxygen, of course, is used in the lower circle only when there is a large quantity of organics to remove. When this is not the case only nitrogen dioxide need be used in the lower circle. In our analyses of the chemistries of the other alkali metals we shall consider the conversion of all compounds to the nitrates during the monoxygen reactions.

3. SODIUM

The enthalpy of sublimation of sodium fluoride is 285 kJ mol.sup.-1. The `times four` criterion is, therefore, 1140 kJ mol.sup.-1.

The `times four` criterion of 1140 kJ mol.sup.-1 is easily met. Thus, the O/NO.sub.2 +F procedure works well for the removal of all sodium species.

4. POTASSIUM

The enthalpy of sublimation of potassium fluoride is 242 kJ mol.sup.-1. The `times four` criterion is, thus, 968 kJ mol.sup.-1.

The `times four` criterion of 968 kJ mol.sup.-1 is easily met. Thus, the O/NO.sub.2 +F procedure works well for the removal of all potassium species.

5. RUBIDIUM

The enthalpy of sublimation of rubidium fluoride is 227 kJ mol.sup.-1. The `times four ` criterion is, therefore, 908 kJ mol.sup.-1.

The `times four` criterion of 908 kJ mol.sup.-1 is easily met, and the O/NO.sub.2 +F procedure works well for the removal of all rubidium species.

6. CESIUM

The enthalpy of sublimation of cesium fluoride is 199 kJ mol.sup.-1, and the `times four` criterion is 796 kJ mol.sup.-1.

The `times four` criterion is met for [II-66] even without considering the kinetic energy of the monofluorines. The O/NO.sub.2 +F procedure works well for the removal of all cesium species.

1. THERMODYNAMIC DATA

We note that none of the condensed phase compounds listed in TABLE II-2. are self-volatilizing under the criteria of [II-1] or [II-2].

2. BERYLLIUM

The enthalpy of sublimation of beryllium fluoride is 231 kJ mol.sup.-1. The `times four` criterion is 924 kJ [mol.sup.-1. In the absence of any thermodynamic data on beryllium nitrate we are unable to investigate the reactions of nitrogen dioxide with compounds of beryllium.

Thus, [II-70] exceeds the `times four` criterion for beryllium fluoride. The `times four` criterion for BeO is 2976 kJ mol.sup.-1 which neither [II-68] or [II-69] meet. Thus, the O+F procedure works well for the removal of all species of beryllium considered even in the absence of NO.sub.2.

3. MAGNESIUM

The enthalpy of sublimation of magnesium fluoride is 397 kJ mol.sup.-1, and the `times four` criterion is 1588 kJ mol.sup.-1.

The `times four` criterion is met for [II-75] even without considering the kinetic energy of the monofluorines. Thus, the O/NO.sub.2 +F procedure works well for the removal of all magnesium species.

4. CALCIUM

The enthalpy of sublimation of calcium fluoride is 442 kJ mol.sup.-1, and the `times four` criterion is 1768 kJ mol.sup.-1.

The `times four` criterion is met for [II-80] even without considering the kinetic energy of the monofluorines. Thus, the O/NO.sub.2 +F procedure works well for the removal of all calcium species.

5. STRONTIUM

The enthalpy of sublimation of strontium fluoride is 451 kJ mol.sup.-1, and the `times four` criterion is 1804 kJ mol.sup.-1.

The `times four` criterion is met for [II-85] even without considering the kinetic energy of the monofluorines. Thus the O/NO.sub.2 +F procedure works well for the removal of all strontium species.

6. BARIUM

The enthalpy of sublimation of barium fluoride is 405 kJ mol.sup.-1, and the `times four` criterion is 1620 kJ mol.sup.-1.

Once again we find that the `times four` criterion is met even without considering the kinetic energy of the monofluorines. Thus, the O/NO.sub.2 +F procedure works well for the removal of all alkali metal and alkaline earth metal species.

1. INTRODUCTION

Whereas the elements of Groups IA and IIA would not be expected to be present on a wafer as the element or as an `alloy`, this is not the case for the elements of Group IIIA. Aluminium and aluminum-copper alloys are frequently used as electrical conductors. Elementary boron may be present as a by-product of ion implantation. Furthermore, there is a large family of III-V compound semiconductors.

In addition to the O/NO.sub.2 and F reagents we shall introduce a third reagent--methyl. Methyl is generated by introducing an excess of methane into the monofluorine stream by means of the lower circle.

For [II-91] k.sub.r (298K)=8.times.10.sup.-11 cm.sup.3 molecule .sup.-1 s.sup.-1 <II-11>, <II-12>. Thus, reaction occurs with essentially every collision.

In Table II-3 are listed the enthalpies of formation of the methyl compounds of Groups IIB, IIIA, IVA, VA and VIA and their enthalpies of sublimation or vaporization. All of the data are from Ref. <B> except where specifically noted. The symbol `Me` represents the CH.sub.3 radical.

Me.sub.3 B, Me.sub.3 N, MeO and Me.sub.2 O are all gases at 298K. Me.sub.3 requires an additional 39.9 kJ mol.sup.-1 to volatilize by the criteria of [II-1]. All of the other compounds will self-volatilize by the criterion of [11-2]. (These comments, strictly speaking are not correct. The criteria apply to the free energies and not to the enthalpies.) .

The thermodynamic data for both Group III A and Group V A are presented her. This is because the analyses of the chemistries of the III-V compounds obviously require both sets of data.

It should be noted that the nitrates of aluminum, gallium and indium exist only as the hydrates. On dehydration they decompose to the oxides.

3. BORON

The only boron compounds listed in TABLE II-4 which are not gases at 289K are B.sub.2 O.sub.3, BN and BP.

In all of these reactions boric oxide remains as a residue. In [II-95 the phosphorus oxide will volatilize since the `times four` criterion is 106 kJ/mole of BP.

Thus, there are four procedures which are effective in volatilizing all of the boron species present:

4. ALUMINUM

None of the aluminum compounds meet the criteria of either [II-1] or [II-2]. We must, therefore, depend on the `times four` criterion. For aluminum fluoride it is 1204 kJ mol.sup.-1. For trimethyl aluminum it is 248 kJ mol.sup.-1.

The `times four` criterion for arsenic trifluoride is 140 kJ mol.sup.-1. Thus, the `times four` criterion for [II-109] is 1204+140=1444 kJ mol.sup.-1. Monofluorine is therefore, an effective volatilizing reagent for [II-110-[II-116].

Thus, we again find the quartet of effective procedures,

5. GALLIUM

Except for trimethylgallium we do not have sufficient data to determine volatilization criteria. For trimethylgallium the `times four` criterion is 132 kJ mol.sup.-1. Because of the lack of adequate thermodynamic data, we shall not carry out the monofluorine analyses. The reactions of methyl with the phosphorous and arsenic oxides are given above ([II-119] and [II-120]) and will not be repeated here.

Thus, we have two effective procedures. O+Me and Me.

6. INDIUM

The `times four` criterion for trimethylindium is 192 kJ mol.sup.-1. Again because of the lack of adequate thermodynamic data, we shall not carry out the monofluorine analyses.

Again we have two effective procedures

O+Me and Me.

1. INTRODUCTION

At this point we have introduced three reagants O/NO.sub.2, F and Me. In terms of upper circle/lower circle reagents they are

Although this three step process is adequate for the removal of Group IVA elements and compounds, we shall introduce monohydrogen as a useful alternative reagent. The three step process leaves the surface `dangling bonds` attached to fluorine atoms. If a deposition is to be made on the freshly cleaned surface, it is more desirable that the surface atoms be bonded to hydrogens rather than fluorines.

2. THERMODYNAMIC DATA

3. CARBON

Although all of the calculations in this SECTION are for graphite, they are also valid for diamond since the enthalpy and free energy of formation of diamond are so small. Without thermodynamic data on the stepwise reactions, [II-159 must be considered as questionably effective. Only the reaction generating CO of the three possible diatomic products is thermodynamically favorable. None of these reactions, [II-153], [II-155] and [II-160], are kinetically favorable. The oxidation, fluoridation and hydrogenation of both graphite and diamond surfaces are well known phenomena. For the monofluorine and monohydrogen reactions at least two reagent-carbon bonds must be formed before a volatile product is generated. In some cases three or four reagent-carbon bonds are required. For the monoxygen case clearly two-carbon-oxygen single bonds (.sigma.-bonds) will yield a volatile product.

4. SILICON

Reactions [II-164] and [II-165 are expected to form a `surface crust` rather than a volatile compound or a `body` oxide. For kinetic reasons the reaction a monofluorine with silicon is not expected to yield a volatile species unless there are at least two fluorines per silicon. The reaction of monofluorine with silicon dioxide is not expected to yield a volatile silicon species unless there are at least six monofluorines per silicon dioxide.

In the absence of thermodynamic data on (CH.sub.3).sub.x Si for x=1, 2, 3 definitive statements cannot be made with respect to the reactions of methyl with silicon or silicon dioxide. It is clear, however, that between four and eight methyls per silicon dioxide are required to yield a volatile silicon species.

A minimum of two monohydrogens per silicon is required for kinetic reasons as well as thermodynamic reasons for the production of a volatile silicon species. For kinetic reasons a minimum of six monohydrogens are required per silicon dioxide for the production of a volatile silicon species.

5. GERMANIUM

Thus, we find that monofluorine, methyl and monohydrogen are all effective reagents for the removal of all germanium species.

6. TIN

Both SnF.sub.2 and SnF.sub.4 are known compounds. In the absence of thermodynamic data on them, however, we can make no comments regarding the efficacy of monofluorine as a volatilizing reagent for tin species. Both methyl and monohydrogen are effective reagents.

Thus, we find that methyl is effective for the volatilization of all of the lead compounds discussed here.

1. INTRODUCTION

Much of the chemistry of the GROUP VA. elements has been discussed in SECTION IX.D. (GROUP IIIA.). For the convenience of the reader all of the pertinent chemistries except for those of nitrogen will be discussed here.

2. PHOSPHOROUS

The `times four` criterion for P.sub.4 O.sub.10 is 424 kJ mol.sup.-1. Thus, the P.sub.4 O.sub.10 generated by [II-235] will vaporize. By [II-237] monofluorine will remove any of the compound present which is not generated by [II-235].

Thus, methyl as well as monofluorine is an effective reagent for the removal of phosphorous species.

Thus, should it not be desirable to use either monofluorine or methyl, monohydrogen is quite effective in removing phosphorous species.

3. ARSENIC

Except for AsO the oxides of arsenic are not expected to be volatile in the Apparatus. The trifluoride is volatile, and the pentafluoride (for which we have no thermodynamic data) is a gas.

Thus, monofluorine is effective in removing all of the oxides of arsenic as well as arsenic itself.

Thus, both methyl and monohydrogen are effective in removing all of the oxides of arsenic as well as arsenic itself.

4. ANTIMONY

The enthalpy of sublimation of antimony trifluoride is 869 kJ mol.sup.-1, making the `times four` criterion 3476 kJ mol.sup.-1. Thus, monofluorine is not an effective reagent for removing antimony species. (It should be noted that the enthalpy of reaction for [II-266] is for 4 antimony trifluorides.)

Thus, both methyl and monohydrogen are effective reagents for the removal of all of the antimony species discussed here.

5. BISMUTH

Thus, methyl is an effective reagent for the removal of all of the bismuth species discussed here.

1. THERMODYNAMICS

2. SULFUR

By the criterion of [II-1] sulfur trioxide will vaporize spontaneously in the Apparatus. Thus, the only solid species of sulfur which will be considered here is sulfur itself.

Thus, monoxygen, monofluorine, methyl and monohydrogen are all effective volatilizing reagents.

3. SELENIUM

Thus, both monofluorine and monohydrogen are effective volatilizing reagents. Although methyl is probably an effective reagent also, we cannot confirm this in the absence of thermodynamic data.

4. TELLURIUM

Thus, both monofluorine and monohydrogen are effective volatilizing reagents. As is the case with selenium, methyl is probably an effective reagent also, but we cannot confirm this in the absence of thermodynamic data.

As the data in TABLE II-8 show the schedule O/NO.sub.2 .degree.F+CH.sub.3 will remove all main group element species as well as non-fluorocarbon organics. If a large amount of fluorocarbon is present, such as in the case in which a fluorocarbon mask has been used, the schedule should be changed to H+O/NO.sub.2 +F+CH.sub.3. If only a small amount of fluorocarbon is present, or suspected, the schedule O/NO.sub.2 +F+CH.sub.3 +H can be used. This schedule is preferred if silicon expitaxy, for example, is to be carried out immediately after the cleaning schedule. This is because the `hydrogenated` surface usually leads to a better quality expitaxial layer. Since other chemistries must be added to the schedules for the removal of transition metal species and lanthanide species, we shall now turn to the analysis of the chemistries of these species.

1. THERMODYNAMICS

In the absence of any thermodynamic data on the nitrates we can make no definitive statement on the efficacy of the O/NO.sub.2 +F procedure. We shall return to this question after the completion of the analysis of the lanthanide species chemistries.

2SCANDIUM

The enthalpy of sublimation of scandium fluoride is 382 kJ mol.sup.-1, and the `times four` criterion is 1528 kJ mol.sup.-1. Since the monofluorines are not in their ground state (cf. SECTION IX. A.), the criterion is met even for the sesquioxide.

Thus, for the volatilization of scandium species we have the choice of three procedures:

O+F*

H+F

F* (in the absence of ScF.sub.3).

3. YTTRIUM

The enthalpy of sublimation of yttrium trifluoride is 430 kJ mol.sup.-1, and the `times four` criterion is 1720 kJ mol.sup.-1.

Thus, for the volatilization of yttrium species we have the choice of three procedures:

O+F*

H+F

F* (in the absence of YF.sub.3).

1. THERMODYNAMICS

It is well known that the lanthanides are very similar in their chemistries. The `outer` electronic structures of scandium, yttrium and lanthanum are 3d 4s.sup.2, 4d 5s.sup.2 and 5d 6s.sup.2, respectively. For lanthanum, however, a new factor enters the picture. For lanthanum and the following fourteen elements the 4.function. levels and the 5d levels are very close in energy. Only four of these fifteen elements (the lanthanides) have a 5d electron in their ground states. These are La, Ce, Gd and Lu. These similarities are reflected in many of the thermodynamic values of these elements.

Referring to TABLE II-10. we find that the enthalpies of formation of the sesquioxides are all relatively close together except for that of Eu. The average value, including Eu, is -1829 kJ mol.sup.-1. The average absolute deviation is 43 kJ mol.sup.-1 (2.4%) and the root mean square deviation is 58 kJ mol.sup.-1 (3.2%). The deviation of the value for Eu from the average value is 166 kJ mol.sup.- (9.1%). The deviation of the value for Sc is 80 kJ mol.sup.-1 (4.4%). For the purpose of the calculations in the SECTIONS which follow we shall assume the following enthalpies of formation.

sesquioxide as reactant .DELTA.H.sub..function..sup.0 (O.sub.1.5)=-1898 kJ mol.sup.-1 as product, .DELTA.H.sub..function..sup.0 (O.sub.1.5 ')=-1663 kJ mol.sup.-1

dioxide as reactant, .DELTA.H.sub..function..sup.0 (O.sub.2)=-1089 kJ mol.sup.-1 as product, .DELTA.H.sub..function..sup.0 (O.sub.2 ')=-972 kJ mol.sup.-1

The rule which is followed here is to use the value which is largest in magnitude for the reactant and the value which is smallest in magnitude for the product. (`Magnitude` here means, of course, `absolute magnitude`. This terminology precaution is necessary since all of the values are negative.) By using these extremum values we can use a single calculation (`worst case` analysis) to cover all fifteen elements.

Referring now to TABLE II-11. we find that a similar situation exists relative to the enthalpies of formation of the trichlorides. The average value is -1006 kJ mol.sup.-1 with an average absolute deviation of 31 kJ mol.sup.-1 (3.1%) and a root mean square deviation of 35 kJ mol-1 (3.5%). The deviation of the value for Sc is 81 kJ mol.sup.-1 (8.1%). We shall assume in the calculations which follow the following enthalpies of formation;

dichloride as reactant, .DELTA.H.sub..function..sup.0 (Cl.sub.2)=-824 kJ mol.sup.-1 as product, .DELTA.H.sub..function..sup.0 (Cl.sub.2)=-682 kJ mol.sup.-1

trichloride as reactant, .DELTA.H.sub..function..sup.0 (Cl.sub.3)=-1071 kJ mol.sup.-1 as product, .DELTA.H.sub..function..sup.0 (Cl.sub.3)=-936 kJ mol.sup.-1.

For the trifluorides we require the enthalpies of formation for both the crystalline and gaseous states. As is indicated in TABLE II-12, we have only limited data for the gaseous species. Although we have extensive data on the crystalline states from <II-23>, the values are consistently lower than those from the more recent references, <B>, <II-19>, <II-20>, <II-21> and <II-22>. From <II-23> the average value of the enthalpy of formation of the crystalline trifluorides is -1575 kJ mol.sup.-1 with an average absolute deviation of 42 kJ mol.sup.-1 (2.7%) and a root mean square deviation of 49 kJ mol.sup.-1 (3.1%). From the newer references the average value is -1736 kJ mol.sup.-1 with an average absolute deviation of 43 kJ mol.sup.-1 (2.5%) and a root mean square deviation of 46 kJ mol.sup.-1 (2.6%). The deviation of the value for Sc is 107 kJ mol.sup.-1 (6.2%).

The average value of the enthalpy of formation of the gaseous trifluorides is -1282 kJ mol.sup.-1 with an absolute deviation of 42 kJ mol.sup.-1 (3.3%) and a root mean square deviation of 52 kJ mol.sup.-1 (4.1%). The deviation of the value for scandium is 35 kJ mol.sup.-1 (2.7%). For those elements for which we can calculate the enthalpies of sublimation the largest value is for Er, 481 kJ mol.sup.-1. For this value the `times four` criterion is 1924 kJ mol.sup.-1. In the calculations which follow wse shall use the following extremum values:

monofluoride as product, .DELTA.H.sub..function..sup.0 (F)=-159 kJ mol.sup.-1

difluoride as product, .DELTA.H.sub..function..sup.0 (F.sub.2)=-682 kJ mol.sup.-1

trifluoride as reactant, .DELTA.H.sub..function..sup.0 (F.sub.3)=-1782 kJ mol.sup.-1 as product, .DELTA.H.sub..function..sup.0 (F.sub.3)=-1469 kJ mol.sup.-1

`times four` criterion, 1924 kJ mol.sup.-1

For the dihydrides the average value of the enthalpies of formation from <II-25> (the most recent reference as well as the most complete) is -210 kJ mol.sup.-1 with an average absolute deviation of 9 kJ mol.sup.-1 (4.3%) and a root mean square deviation of 13 kJ mol.sup.-1 (6.2%). The extremum values which will be used are

dihydride as reactant .DELTA.H.sub..function..sup.0 (H.sub.2)=-226 kJ mol.sup.-1

as product, .DELTA.H.sub..function..sup.0 (H.sub.2)=-180 kJ mol.sup.-1.

For the enthalpies of formation of the nitrates we have data for only five elements:

La -1254 kJ mol.sup.-1

Ce -1226

Pr -1229

Nd -1231

Sm -1212.

The average value is -1230 kJ mol.sup.-1 with an average absolute deviation of 10 kJ mol.sup.-1 (0.81%) and a root mean square deviation of 14 kJ mol.sup.-1 (1.1%). The spread of these data is quite small. These elements, however, are those with the lowest number of 4.function. electrons. We shall find that, even when large deviations from this average value are assumed, the reactions involving the nitrates are quite exothermic.

2. MONOXYGEN CHEMISTRIES

3. MONOXYGEN PLUS NITROGEN DIOXIDE CHEMISTRIES

For reactions [II-358], [II-361], [II-362], [II-363] and [II-364] the enthalpy of formation of the nitrate could be zero, and the reactions would still be exothermic. For reaction [II-359] the enthalpy of formation of the nitrate could be -500 kJ mol.sup.-1, and the reaction would still be exothermic. For the worst case, reaction [II-360] the enthalpy of formation of the nitrate could be -800 kJ mol.sup.-1, and the reaction would still be exothermic. Since all of these deviations from -1230 kJ mol.sup.-1 are so large as to be most unlikely, we can reasonably assume that all of the O/NO.sub.2 reactions will go as written.

4. MONOFLUORINE CHEMISTRIES

Thus, we find that even under the extremum conditions monofluorine in its excited state will volatilize all of the lanthanide species analyzed here. Furthermore, even ground state monofluorine will volatilize the dihydrides and the nitrates will volatilize the dihydrides and the nitrates (unless the enthalpy of formation is more negative than -2000 kJ mol.sup.-1 which is most unlikely.)

5. MONOHYDROGEN CHEMISTRIES

6. SUMMARY OF GROUP IIIB. AND LANTHANIDE PROCEDURES

Thus, we find that for the volatilization of the GROUP III B and lanthanide species we have the choice of four procedures.

O+F*

O/NO.sub.2 +F

H+F

F* (in the absence of MF.sub.3).

(In addition it should be noted that the chemistries discussed here can be used to synthesize anhydrous lanthanide compounds. For example, reactions [II-358] and/or [II-359] can be used to synthesize anhydrous nitrates.)

1. THERMODYNAMICS

Of the various chlorides only titanium tetrachloride will vaporize spontaneously in the Apparatus. Zirconium tetrachloride, however, is marginal by the criterion of [II-2]. The enthalpies of sublimation of the tetrafluorides of titanium, zirconium and hafnium are, 98, 237 and 260 kJ mol.sup.-1, respectively. Thus, the `times four` criteria are 392, 948 and 1040 kJ mol.sup.-1 , respectively.

2. TITANIUM

The enthalpy of melting of titanium monoxide is 49 kJ mol.sup.-1. Thus, by the `times four` criterion [II-387] will yield the liquid monoxide. By [II-388[ the reaction will continue with the formation of the sesquioxide. The enthalpy of melting of the sesquioxide is 103 kJ mol.sup.-1. Thus, by the `times four` criterion (412<792+342) the liquid sesquioxide will be formed. By [II-389] the reaction will continue to the dioxide. The enthalpy of melting of the dioxide is 51 kJ mol.sup.-1.

If the sesquioxide is formed from the crystalline monoxide, the liquid sesquioxide will not form. If, however, sufficient monoxygen is present to take the reaction to the dioxide, the liquid dioxide will form, and all of the monoxide and sesquioxide will be converted to the dioxide. Thus, in the subsequent reactions we need consider only the dioxide as the product of monoxygen reactions.

The enthalpy of sublimation of titanium trifluoride is 243 kJ mol.sup.-1, and the `times four` criterion is 972 kJ mol-1. Thus, the trifluoride formed by [II-397] as well as the tetrafluoride formed by [II-398] will volatilize. Thus, we must consider both the trifluoride as well as the tetrafluoride as possible volatilization products.

Thus, even without considering the tetrafluoride as the produce or excited state monofluorine as the reactant monofluorine is an effective volatilizing reagent for titanium species.

Thus, we have a choice of three procedures for the volatilization of titanium species:

O+F

H+F

F (in the absence of TiF.sub.4).

3. ZIRCONIUM

For the three crystalline fluorides of zirconium listed in TABLE II-14. the enthalpies of sublimation are 404 kJ mol.sup.-1, 297 kJ mol.sup.-1 and 237 kJ mol.sup.-1 for the difluoride, the trifluoride and the tetrafluoride, respectively. The `times four` criteria, therefore, are 1616 kJ mol.sup.-1, 1188 kJ mol.sup.-1 and 948 kJ mol.sup.-1.

The enthalpy of melting of zirconium dioxide is 74 kJ mol.sup.-1, and the `times four` criterion is 296 kJ mol.sup.-1. Thus, reaction [II-412] will produce liquid zirconium dioxide. By [II-413] the reaction will continue to the trioxide. Since the .DELTA.H.sub.r.sup.0 for [II-413] is rather small, and since very little is known about the trioxide, we shall assume that the end product of the monoxygen reactions is the dioxide.

Since both [II-424] and [II-424] meet their respective `times four` criteria, we shall consider both the trifluoride and the tetrafluoride as potential volatilized products.

Thus, excited state monofluorine will remove the oxides by way of the trifluoride, and ground state monofluorine will remove the oxides by way of the tetrafluoride.

Thus, we have the choice of three procedures for the volatilization of zirconium species:

0 +F

H+F

F+(in the absence of ZrF.sub.4).

4. HAFNIUM

The paucity of thermodynamic data on hafnium species precludes the extensive analyses which are possible for titanium and zirconium. It is, however, well known that zirconium and hafnium are quite similar in their chemistries. This can be seen by comparing the available thermodynamic data for hafnium with those of zirconium. Consider a given datum for hafnium from which has been subtracted the corresponding datum for zirconium. The difference varies from +4 kJ mol.sup.-1 (gaseous MF.sub.4) to -57 kJ mol.sup.-1 (crystalline MO.sub.2) for the enthalpies of formation. The average for the absolute values for the difference is 19 kJ mol.sup.-1. If we assumed that this difference applies to the dihydride, then

We shall use the value -188 kJ mol.sup.-1 when HfH.sub.2 is a reactant and -150 kJ mol.sup.-1 when it is a product.

Since the `times four` criterion for the volatilization of hafnium tetrafluoride is 1040 kJ mol.sup.-1, monofluorine clearly is an effective volatilizing reagent for these hafnium species.

As was the case for titanium and zirconium we have the choice of three procedures for the volatilization of hafnium species:

O+F

H+F

F (in the absence of HfF.sub.4).

By the criterion of [II-1] VF.sub.5 (l) will volatilize spontaneously in the Apparatus. By the criterion of [II-2] VCl.sub.4 (l), NbF.sub.5 (cf), TaCl.sub.5 (cf) and (CH.sub.3).sub.5 Ta(l) will volatilize spontaneously in the Apparatus. Furthermore, the enthalpy of sublimation of niobium pentafluoride is 74 kJ mol.sup.-1 with a `times four` criterion of 296 kJ mol.sup.1.

Since we do not know the enthapy of formation of tantalum tetrafluoride, we cannot calculate the `times four` criterion for it. We note, however,

Since the data for the niobium and tantalum pentafluorides, are rather close, we can safely assume that the `times four` criterion is less than 600 kJ mol.sup.-1.

2. VANADIUM

The enthalpy of melting of the monoxide is 61 kJ mol.sup.-1 with a `times four` criterion of 244 kJ mol.sup.-1. Thus, the product of [II-457] will be the liquid. The enthalpy of melting of divanadium trioxide is 126 kJ mol.sup.-1 with a `times four` criterion of 504 kJ mol.sup.-1. For one-half mole, therefore, the criterion is 252 kJ mol.sup.-1, and the product of [II-458] is the liquid.

For the dioxide the enthalpy of melting is 48 kJ mol.sup.-1 with a `times four` criterion of 192 kJ mol.sup.-1, and the product of [II-459] is the liquid. The enthalpy of melting of divanadium pentoxide is 60 kJ mol.sup.-1 with a `times four` criterion of 240 kJ mol.sup.-1. For one-half mole, therefore, the criterion is 120 kJ mol.sup.-1, and the product of [II-460] is the liquid.

The reaction of monoxygen with a solid may yield only a surface `skin` of oxide. With a liquid the reaction will go to completion. Thus, for any sufficiently exothermic reaction with monoxygen the product will be divanadium pentoxide.

Thus, monoxygen will convert all of the halides to the molten divanadium pentoxide which will solidify at the end of the reaction.

Thus, any reaction with monofluorine which produces the tetrafluoride will proceed further to produce the liquid pentafluoride. As noted above the pentafluoride will spontaneously vaporize in the Apparatus.

Thus, both O+F and F are effective procedures for the volatilization of vanadium species.

Since vanadium hexacarbonyl is a highly volatile compound, it is worth investigating whether carbon monoxide is an effective volatilizing reagent for vanadium species.

Thus, we have the choice of four procedures for the volatilization of vanadium species:

O+F

O+CO

CO

3. NIOBIUM

The enthalpy of melting of the monoxide is 83 kJ mol.sup.-1 with a `times four ` criterion of 332 kJ mol.sup.-1. Thus, the product of [II-480] will be the liquid. For the dioxide the enthalpy of melting is 84 kJ mol.sup.-1 with a `times four` criterion of 336 kJ mol.sup.-1, and the product of [II-481] will be the liquid. The enthalpy of melting of diniobium pentoxide is 69 kJ mol.sup.-1 with a `times four` criterion of 276 kJ mol.sup.-1. For one-half mole, therefore, the criterion is 138 kJ mol.sup.-1, and the product of [II-482] is the liquid.

Thus, monoxygen will convert all of the halides to the molten diniobium pentoxide which will solidify at the end of the reaction.

Thus, we have the choice of two procedures for the volatilization of niobium species:

O+F

4. TANTALUM

The enthalpy of melting of ditantalum pentoxide is 89 kJ mol.sup.-1 with a `times four` criterion of 356 kJ mol.sup.-1.

Thus, nonoxygen will convert all of the halides to ditantalum pentoxide.

Thus, we see that with the monofluorine reactions there is no difficulty in meeting the assumed `times four` criterion for the volatilization of tantalum pentafluoride.

The enthalpy of vaporization of pentamethyltantalum is 42 kJ mol.sup.-1 with a `times four` criterion of 168 kJ mol.sup.-1. Thus, methyl is an effective volatilizing reagent for tantalum species. We, therefore, have the choice of four procedures for the volatilization of tantalum species:

O+F

F (in the absence of Ta F.sub.5)

O+CH.sub.3

CH.sub.3.

1. THERMODYNAMICS

By the criterion of [II-1] MoF.sub.6 (l), WF.sub.6 (l) and MO(CO).sub.6 (cr) will volatilize spontaneously in the Apparatus. By Ref. <II-23> the enthalpy of sublimation of CrF.sub.4 is 82 kJ mol.sup.-1 with a `times four` criterion of 328 kJ mol.sup.-1. The enthalpy of sublimation of Cr(CO).sub.6 is 71 kJ mol.sup.-1 with a `times four` criterion of 284 kJ mol.sup.-1.

The enthalpy of sublimation of WF.sub.4 is 178 kJ mol.sup.-1 with a `times four` criterion of 712 kJ mol.sup.-1. The enthalpy of sublimation of WF.sub.5 is 101 kJ mol.sup.-1 with a `times four` criterion of 40 kJ mol.sup.-1. For (W(CO).sub.6 the enthalpy of sublimation is 76 kJ mol.sup.-1 with a `times four` criterion of 304 kJ mol.sup.-1, and for (CH.sub.3).sub.6 W the enthalpy of vaporization is 33 kJ mol.sup.-1 with a `times four ` criterion of 132 kJ mol.sup.-1.

2. CHROMIUM

The enthalpy of melting of the sesquioxide is 117 kJ mol.sup.-1 with a `times four` criterion of 468 kJ mol.sup.-1 (234 kJ/half mole). Thus, the product of [II-513] is actually the liquid. The enthalpy of sublimation of the trioxide is 297 kJ mol.sup.-1 with a `times four` criterion of 1188 kJ mol.sup.-1. Since

monoxygen will volatilize chromium, but not the sesquioxide nor the dioxide.

Thus, monoxygen is an effective volatilizing reagent for the halides. Since it is not an effective volatilizing reagent for the sesquioxide or the dioxide, monoxygen cannot be used by itself.

Since the `times four` criterion for the tetrafluoride is 328 kJ mol.sup.-1, monofluorine is an effective exchange for chromium.

Since the `times four` criterion for chromium carbonyl is 284 kJ mol.sup.-1, we have four choices for the procedure for volatilization of the chromium species:

O+F

F (in the absence of CrF.sub.4)

O+CO

CO*.

3. MOLYBDENUM

The free energies of sublimation of the tetrachloride, the pentachloride and the hexachloride are 48, 32 and 35 kJ mol.sup.-1, respectively. Since these values lie between the criteria of [II-1] and [II-2], these chlorides will sublime spontaneously in the Apparatus. Thus, in our analyses we need consider only the di- and tri-chlorides.

The monoxide will not form. The dioxide will, but will not volatilize. If the trioxide is formed, it will be as the liquid which will then solidify.

Thus, we find for all molybdenum species the trifluoride and higher fluorides will form. Except for the reactions with the oxides the difluoride will also form.

Thus, we have the choice of four procedures for the volatilization of molybdenum species:

O+F

O+CO

CO*.

4. TUNGSTEN

The trioxide probably will form as the liquid and then solidify.

The tetra-, penta- and hexachlorides all meet the criterion of [II-2] and will volatilize spontaneously in the Apparatus.

The enthalpy of sublimation of the trioxide is 549 kJ mol.sup.-1 with a `times four` factor of 2196 kJ mol.sup.-1. Thus, the trioxide formed in [II-565] will volatilize. Since the monoxygen is at least in part in an excited state, the same comment applies to [II-574].

The `times four` criterion for the volatilization of the di- and tri-fluorides are 1680 and 1260 kJ mol.sup.-1, respectively.

Since 766+394+364=1524 and since the `times four` criterion for the tetrafluoride is 712 kJ mol.sup.-1, the tetrafluoride will volatilize. Higher fluorides may also be formed. They a fortiori will volatilize.

Thus, we have the choice of five procedures for the volatilization of tungsten species:

O+F

O+CO*

O+CH.sub.3

CH.sub.3.

1. [II-THERMODYNAMICS

The `times four` criterion for the species of interest at

2. MANGANESE

Thus, ground state monofluorine is an effective volatilizing reagent for all the manganese species except for the fluorides. Unless one is certain that there are no fluoride species present, the safe procedure is O+F.

Thus, we have the choice of three procedures for the volatilization of manganese species:

O+F

O+CO

CO*.

3. RHENIUM

Since the `times four` criterion for Re.sub.2 O.sub.7 is 552 kJ mol.sup.-1, even excited state monoxygen cannot be expected to be effective in the volatilization of the trioxide.

Thus, we find that even ground state monoxygen is effective in the volatilization of the halides.

Thus, we have the choice of four procedures for the volatilization of rhenium species:

O+F

(questionable if Re.sub.F.sub.6 is present)

O+CO

CO*.

1. THERMODYNAMICS

Fe(CO).sub.5, Ni(CO).sub.4, RuO.sub.4, OSO.sub.4 and IrF.sub.6 will volatilize spontaneously in the apparatus by [II-1]. The `times four` criterion for several species are:

2. BETA-DIKETONATES

For the main group elements we used as reagents O, O/NO.sub.2, F, CH.sub.3 and H. For GROUP VB we introduced CO. For the GROUP VIII B elements we find it useful to introduce the .beta.-diketonyls. The .beta.diketones exist in two forms--the keto-form and the enol-form,

We shall consider here only three of the .beta.-diketones:

R.sub.1 =R.sub.2 =CH.sub.3, acetylacetone, Hacac, B.P.=140.degree. C.

R.sub.1 =CF.sub.3, R.sub.2 =CH.sub.3, trifluoroacetylacetone, Htfac, B.P.=107.degree. C.

R.sub.1 =R.sub.2 =CF.sub.3, hexafluoroacetylacetone, Hhfac, B.P.=70.degree. C.

(All of the data in this SECTION are from Ref. <II-33> unless specifically noted.)

For the reaction

.beta.dik. being the .beta.-diketonyl radical, the .DELTA.H.sub.r is estimated to be 418 kJ mol.sup.-1. If we add to [II-682],

we obtain

For .DELTA.H.sub.t (684)<0, we require that .DELTA.H.sub.t (683)<-418.

Thus, we find that all three reagents are effective, but that monofluorine is the preferred reagent.

The enthalpies of formation of the three gaseous .beta.-diketones are

By [II-682]

and

We now consider the reaction

By the `times four` criterion we require that

For M(acac).sub.n we have all the necessary data for three cases:

Thus, we find that acac(g) is an effective volatilizing reagent for both Cr and Al and is borderline for Cu. (It should be noted that Ref. <II-33> gives two values for .DELTA.H.sub.f.sup.0 {Cu(acac).sub.2 (cr)}, -809.9 and -782.4 kJ mol.sup.-1. The -809.9 kJ mol.sup.-1 is the more recent value.)

For M(tfac).sub.n we have all the necessary data for two cases:

Again we find that .beta.-dik(g) is an effective volatilizing reagent.

For M(Hfac).sub.n we have data only for Cr and Cu, but the data are not complete. We have the enthalpies of sublimation, but not the enthalpies of formation for either the crystalline or gaseous states. We can, however, roughly estimate these values.

where D(M-O) is the average mean metal oxygen dissociation enthalpy. For Cr (hfac).sub.3, D(M-O)=238 kJ mol.sup.-1, and for Cu (hfac).sub.2, D(M-O)=163 kJ mol.sup.-1. (It should be noted that both of these values are given as .+-.11 kJ mol.sup.-1. )

For Cr,

Applying these numbers to [II-686] we have

For Cu,

Applying these numbers to [II-686] we have

If we had taken the extreme value for D(Cu-O)=174 kJ mol.sup.-1, the inequality would have read

Thus, the copper case is borderline. Under the conditions of the Apparatus the procedure is effective.

3. IRON

Since the `times four` criterion for the melting of the ferrous oxide is 88 kJ mol.sup.-1, [II-688] will yield the liquid product. Whether the reaction will stop with [II-689] or continue to [II-690] is uncertain. For our present purposes, however, it does not make any difference.

Thus, O+F is an effective procedure. By virtue of [II-697] even excited state monofluorine alone is not an effective procedure.

Thus, we have the choice of three procedures for the volatilization of iron species:

O+F

O+CO

CO*

4. COBALT

The oxides CO.sub.2 O.sub.3 and CoO.sub.2 are also known, but we have no thermodynamic data on them.

Since the `times four` criterion for the trifluoride is 368 kJ mol.sup.-1, [II-718] is marginally effective in the volatilization of the trifluoride. It is, therefore, not advisable to depend on the F* procedure.

Thus, O+F* is an effective procedure as is O+F is sufficient monofluorine is present to ensure that all of the fluoride formed is the trifluoride

Thus, O+CO is an effective procedure.

Thus, the choices of procedures for the volatilization of cobalt species are

O+F (?)

)+F*

O+CO

CO*.

5. NICKEL

Thus, we have the choice of four procedures for the volatilization of nickel species:

O+F*

F* (in the absence of NiF.sub.2)

O+CO

CO*.

6. RUTHENIUM

The dioxide crystallizes in the rutile structure. Each ruthenium atom is bonded to six oxygens octahedrally, and each oxygen is bonded to three ruthenium atoms. In the tetroxide, however, each ruthenium atom is bonded to four oxygen atoms tetrahedrally, and each oxygen atom is bonded to only one ruthenium atom <II-34>. Reaction [II-754] is, therefore, expected to be slow unless the dioxide is amorphous. Thus, it cannot be relied as a means of volatilizing the dioxide.

Thus, with the caveat that the rate of removal of the dioxide may be quite slow monoxygen can be an effective volatilizing reagent for ruthenium species.

Thus, the choices of procedures for the volatilization of ruthenium species are

O (?)

O+F

F (in the absence of RuF.sub.5)

O+CO

CO*.

7. RHODIUM

In the absence of data on the rhodium fluorides we can make no judgment relative to the efficacy of monofluorine.

Thus, the sesquioxide is the product of the reaction of monoxygen with rhodium.

and monoxygen will also convert the trichloride to the sesquioxide.

Since reaction [II-771] is expected to be slow with ground state carbon monoxide, but fast with the excited state carbon monoxide, only the excited state carbon monoxide is recommended. (The reason that the slow rate is expected is that in the carbonyl compound the rhodium atoms form a tetrahedron, whereas in the sesquioxide each rhodium is surrounded octahedrally by oxygens. The energentics of the excited state carbon monoxide is so great that the rhodium atoms will be quite mobile when the oxygens separating them are removed in the formation of carbon dioxide.) The recommended procedures, thus, are:

O+CO*

CO*.

8. PALLADIUM

The paucity of thermodynamic data on the palladium species precludes an extensive analysis at this time.

The vapor pressures of Pd(acac).sub.2 and Pt(acac).sup.2 are 1 mm at 74.degree.-94.degree. C. and 80.degree.-102.degree. C., respectively <II-35>. Furthermore, Pt (hfa).sub.2 sublimes at 65.degree. C., <II-36>. Thus, hfa is expected to be an effective volatilizing reagent for palladium. The question, however, must be asked whether hfa is an effective volatilizing reagent for palladium species other than the metal itself? Will, for example,

Thus, we see that the C-F, bond strength is of the order of 540-550 kJ mol.sup.-1, whereas the C-H bond strength is only of the order of 340-350 kJ mol.sup.-1. The compound F(hfa) is, therefore, a stabile compound.

There would appear to be a contradiction here since we assumed a .DELTA.H.sub.r of 418 kJ mol.sup.-1 for [II-682]. There actually is no contradiction since H-.beta.dik(g) is the enol. The hydrogen is bonded to oxygen!

(The datum for methanol is from Ref. <B>, and the datum for methoxy is from Ref. <II-13>.) The value for H-.beta.dik(g) is somewhat lower because of the reasonance stabilization of .beta.dik.

9. OSMIUM

Thus, there are two choices of procedures for the volatilization of osmium species:

CO*.

The monoxygen procedure is usually the preferred procedure.

10. IRIDIUM

If O* were used in [II-797] rather than O, the enthalpy of reaction would be -247 kJ mol.sup.-1. Since we know very little about the trioxide at the present time, we cannot recommend O* as a volatilizing procedure.

The possible procedures for the volatilization of iridium species are:

O* (?)

O+F

O+CO

CO.

11. PLATINUM

The possible procedures for the volatilization of platinum species are:

O+F

F (in the absence of PtF.sub.6).

1. THERMODYNAMICS

Of the various species listed in TABLE II-21. only three have significantly low `times four` criteria:

In addition we should note the discussion of Cu(hfa).sub.2 in SECTION IX.O.2. have no volatile silver species to consider.

Cuprous oxide will be formed in the liquid state. It will react further with monoxygen to form cupric oxide. The cupric oxide will not volatilize.

Thus, we have the choice of two procedures for the volatilization of copper species:

O+F*

F+hfa.

3. SILVER

Thus, although dioxygen will take silver only to the argentous oxide, monoxygen will take it all the way to the sesquioxide.

In the absence of data on volatile silver species we can make no recommendations at this time.

4. GOLD

Gold trichloride sublimes at 180.degree. at 760 nm <II-37>. At the time that SECTION V. was written we had not considered monochlorine as a potential volatilizing reagent. A discussion of it was therefore, not included. It should be noted that, whereas the datum for crystalline gold trichloride is for the monomer the datum for gaseous gold trichloride is for the dimer.

Thus, we have the choice of four procedures for the volatilization of gold species:

O+F+Cl

O+Cl

F+Cl

1. THERMODYNAMICS

By the criterion of [II-1] the dimethyl compounds of cadmium and mercury will volatilize spontaneously in the Apparatus. The `times four` criterion for some species of interest are

It should be noted that mercurous fluoride is a dimer in the crystalline state and a monomer in the gaseous state.

2. ZINC

It should be noted that although O+F* is an effective procedure, F* is not since it will not volatilize ZnF.sub.2 !

Thus, we have the choice of four procedures for the volatilization of the zinc species:

O+F*

O+CH.sub.3

F+CH.sub.3

CH.sub.3.

3. CADMIUM

Again we have the choice of four procedures:

O+F*

O'CH.sub.3

F+CH.sub.3

CH.sub.3.

4. MERCURY

Thus, although monofluorine will convert mercury to mercurous fluoride and mercurous fluoride to mercuric fluoride, even with excited state monofluorine the reactions are not sufficiently energetic to volatilized the product fluorides. We should, however, consider the reaction between monofluorine and the oxide.

Since monoxygen cannot volatilize the oxide, by itself it is not a satisfactory volatilizing reagent. Similarly, by itself monofluorine is not a satisfactory volatilizing reagent. The procedure O+F, however, is satisfactory.

Thus, methyl by itself is an effective volatilizing reagent.

In SECTION IX.H. we received the chemistries for the Main Group elements. The recommended general procedures are

where

a. monohydrogen is used as the first reagent if a large quantity of fluorocarbons is present, and

b. monohydrogen is used as the last reagent if only a small quantity of fluorocarbons is present.

In the absence of thermodynamic data on the nitrates except for the lanthanides no comments can be made with respect to O/NO.sub.2. We have found CO or CO* to be a useful reagent in many cases and essential for rhodium. In addition, we have found hfa essential for palladium and monochlorine essential for gold.

Summarizing the procedures for the various elements we have:

In order to accommodate Sc, Y and Cu we change F in [II-905] to F*. In order to accommodate Rh we add CO*. For Pd we add hfa, and for Au we add Cl. Thus, the general recommended procedures for the entire Periodic Table are

These procedures are for wafer cleaning. For film etching a specific procedure has to be chosen in accordance with the other materials present.

<A> M. W. Chase, Jr., C. A. Davies, J. R. Downey, Jr., D. J. Frurip, R. A. Mc Donald and A. N. Syverud, `JANAF Thermochemical Tables` 3rd ed., J. Phys. Chem. Ref. Data, 1985, 14, Supplement No. 1.

<B> Donald D. Wagman, William H. Evans, Vivian B. Parker, Richard H. Schumm, Ira Halow, Sylvia M. Bailey, Kenneth L. Churney and Ralph L. Nutell, `The NBS Tables of Chemical Thermodynamic Properties, Selected Values for Inorganic and C.sub.1 and, C.sub.2 Organic Substances in SI Units`, J. Phys. Chem. Ref. Data, 1982, 11, Supplement No. 2.

<C> John A. Dean, ed., Lange's Handbook of Chemistry, 13th ed. McGraw-Hill Book Company, New York (1985).

<I-1> W. Tsang and R. F. Hampson, `Chemical Data Base for Combustion Chemistry. Part 1. Methane and Related Compounds` J. Phys. Chem. Ref. Data, 1986, 15(3), 1087-1279.

<I-2> Walter J. Moore, Physical Chemistry, 4th ed., Prentice-Hall, Inc., Englewood Cliffs, N.J. (1972), p. 157.

<I-3> Cheng-ping Tsai and David L. McFadden, `Gas-Phase Atom-Radical Kinetics of Atomic Hydrogen Reactions with CF.sub.3, CF.sub.2, and CF Radicals`, J. Phys. Chem., 1989, 93, 2472.

<I-4> J. L. Franklin, J. G. Dillard, H. M. Rosenstock, J. T. Herron, K. Draxl and F. H. Field, Ionization Potentials, Appearance Potentials, and Heats of Formation of Gaseous Positive Ions, NSRDS-NBS 26 (June 1969), p. 114.

<I-5> Hideo Okabe, Photochemistry of Small Molecules, John Wiley & Sons, New York (1978), p. 166.

<I-6> Ref. <I-5>, p. 168.

<I-7> S. R. Hunter and L. G. Christophorou, `Electron Motion in Low- and High-Pressure Gases` in L. G. Christophorou, ed., Electron--Molecule Interactions and Their Applications, Vol. 2, Academic Press, Inc., Orlando, Fla. (1984) , p. 200.

<I-8> Ref. <I-4>, p. 145.

<I-9> Ref. <I-5>, p. 184.

<I-10> Robert K. Steunenberg and Richard C. Vogel, `The Absorption Spectrum of Fluorine`, J. Am. Chem. Soc., 1956, 78, 901.

<I-11> A. L. G. Rees, `Electronic Spectrum and Dissociation Energy of Fluorine`, J. Chem. Phys., 1957, 26 (6), 1567-1571.

<I-12> Ref. <I-4>, p. 14.

<I-13> Ref. <I-5>, p. 162.

<I-14> Martin Misakian and Jens C. Zorn, `Dissociation Excitation of Molecular Hydrogen by Electron Impact`, Phys. Rev. (A) , 1972, 6 (6) , 2193.

<I-15> J. E. Mentall and E. P. Gentieu, `Lyman-.alpha. Fluorescence from the Photodissociation of H.sub.2 `, J. Chem. Phys., 1970, 52 (11), 5641-5645.

<I-16> S . Trajmar and D. C. Cartwright, `Excitation of Molecules by Electron Impact`, in L. G. Christophorou, ed., Electron-Molecule Interactions and their Applications, Vol. 1., Academic Press, Inc., Orlando, Fla. (1984), p. 214.

<I-17> M. J. Mumma and E. C. Zipf, `Dissociation Excitation of Vacuum Ultraviolet Emission Features by Electron Impact on Molecular Gases. I. H.sub.2 and O.sub.2, J. Chem. Phys., 1971, 55 (4), 1663.

<I-18> Ref. <I-4>, p. 111.

<I-19> Ref. <I-5>, p. 177.

<I-20> Keith Schofield, `Critically Evaluated Rate Constants for Gaseous Reactions of Severally Electronically Excited Species; J. Phys. Chem. Ref. Data, 1979, 8(3), 775.

<I-21> Y. Itikawa, A. Ichimura, K. Onda, K. Sakimoto, K. Takayanagi, Y. Hatano, M. Hayashi, H. Nishimura and S. Tsurubuchi, `Cross Sections for Collisions of Electrons and Photons with Oxygen Molecules`, J. Phys. Chem. Ref. Data, 1989, 18 (1), 23-42.

<I-22> R. L. Mc Carthy, `Chemical Synthesis from Free Radicals Produced in Microwave Fields`, J. Chem. Phys., 1954, 22 (8), 1366-1365.

<I-23> E. C. Zipf, `Dissociation of Molecules by Electron Impact`, in L. G. Christophorou, ed., Electron-Molecule Interactions and Their Applications, Vol. 1., Academic Press, Inc., Orlando, Fla. (1984), p. 340.

<I-24> Walter L. Borst and E. C. Zipf, `Energy Spectra of Metastable Oxygen Atoms Produced by Electron-Impact Dissociation of O.sub.2 `, Phys. Rev. (A), 1971, 4 (1), 153-161.

<I-25> Ref. <I-17>, pp. 1665-1666.

<II-1> Hideo Okabe, Photochemistry of Small Molecules, John Wiley & Sons, New York (1978), p. 208.

<II-2> Ref. <II-1>, p. 196.

<II-3> R. Atkinson, B. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr and J. Troe, `Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry. Supplement III.`, J. Phys. Chem. Ref. Data, 1989, 18 (2), 891.

<11-4> John T. Herron and Robert E. Huie, `Rate Constants for the Reactions of Atomic Oxygen (O.sup.3 P) with Organic Compounds in the Gas Phase`, J. Phys. Chem. Ref. Data, 1973, 2 (3), 470.

<II-5> John T. Herron, `Evaluated Chemical Kinetic Data for the Reactions of Atomic Oxygen O(.sup.3 P) with Saturated Organic Compounds in the Gas Phase`, J. Phys. Chem. Ref. Data, 1988, 17 (3), 1013-1015.

<II-6> Ref. <II-3>, p. 895.

<II-7> Ref. <C>, p. 9-82.

<II-8> Ref. <C>, p. 9-155.

<II-9 > Eugene S. Domalski, William H. Evans and Elizabeth D. Hearing, `Heat Capacities and Entropies of Organic Compounds in the Condensed Phase`, J. Phys. Chem. Ref. Data, 1984, 13, Supplement No. 1, p. 232.

<II-10> Ref. <C>, p. 4-105.

<II-11> D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, J. Troe and R. T. Watson, `Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry: Supplement II. `, J. Phys. Chem. Ref. Data, 1984, 13 (4), 1354.

<II-12> Ko-ichi Sugawara, Fumiyuki Ito, Taisuke Nakanaga, Harutoshi Takeo and Chi Matsumura, `Vibrational and Rotational Energy Distributions of CH.sub.3 and IF Formed in the Reactions of F Atoms with CH.sub.4 and CH.sub.3 I`, J. Chem. Phys., 1990, 92 (9), 5332.

<II-13> Ref. <II-3>, p. 1094.

<II-14> Ref. Eugene S. Domalski and Elizabeth D. Hearing `Estimation of the Thermodynamic Properties of Hydrocarbons at 298.15K`, J. Phys. Chem. Ref. Data, 1988, 17 (4), 1664-1665.

<II-15> Ref. G. Pilcher and H. A. Skinner, `Thermochemistry of Organometallic Compounds`, in Frank R Hartley and Saul Patai, eds., The Chemistry of the Metal-Cargon Bond, Vol. 1., John Wiley & Sons, Ltd., Chichester (1982 ) , pp. 43-90.

<II-16> Jindrich Leitner, Cestmir Cerny, Petr Vonka and Jan Mikulec, `Theoretical and Experimental Study of Chemical Equilibrium in the Systems Si-Cl and Si-Cl-H. Calculation of Chemical Equilibrium in the Systems Si-Cl and Si-Cl-H`, Collection of Czechoslovak Chemical Communications, 1989, 54 (11), 2899.

<II-17> I. Barin, O. Knacke and O. Kubaschewski Thermochemical Properties of Inorganic Substances, Supplement, Springer-Verlag, Berlin (1977), p. 89.

<II-18> .sub.A I. Barin and O. Knacke, Thermochemical Properties of Inorganic Substances, Springer-Verlag, Berlin (1973), p. 394.

<II-19> Ref. <II-17>, p. 351.

<II-20> Ref. <II-17>, p. 155.

<II-21> Ref. <II-17>, p. 450.

<II-22> Ref. <II-17>, p. 310.

<II-23> C. E. Wicks and F. E. Block, Thermodynamic Properties of 65 Elements--Their Oxides, Halides, Carbides, and Nitrides, Bulletin 605, Bureau of Mines (1963).

<II-24> Ref. <II-17>, pp. 351, 155, 450, 310.

<II-25> G. G. Libowitz and A. J. Maeland, `Hydrides`,in Karl A. Gschneidner Jr., and LeRoy Eyring, Handbook on the Physics and Chemistry of Rare Earths, Vol. 3, North-Holland Publishing Company, Amsterdam (1979), p. 309.

<II-26> Ref. <C>, pp. 4-127, 4-83 and 4-117.

<II-27> G. Dittmer and U. Niemann, `Heterogeneous Reactions and Chemical Transport of Tungsten with Halides and Oxygen Under Steady-State Conditions of Incandescent Lamps`, Philips J. Res., 1981, 36, 89-113.

<II-28> Ref. A. A. Woolf, `An Outline of Rhenium Chemistry`, Quarterly Reviews, 1961, 15 (3), 386.

<II-29> Ref. <II-17>, p. 535.

<II-30,> Keith G. G. Hopkins and Peter G. Nelson, `Comparison of the Thermochemistry of Difluorides and Potassium Hexaflorometallates (iv) of the Elements of the Germanium Triad with that of the Elements of the Nickel Triad: A Study of the Inert-pair Effect`, J. Chem. Soc., Dalton Trans., 1984, 1393.

<II-31> A. A. Bondarenko, M. V. Korobov, O. L. Sharova, A. V. Ryzhov and L. N. Sideroy, `Mass Spectral Study of Equilibrium with Participation of Higher Platinum Fluorides`, Zh. Neorg. Kahim., 1987, 32 (1), 28.

<II-32> M. V. Korobov, V. V. Nikulin, N. S. Chilingarov and L. N. Siderov, `Enthalpy of Formation of Platinum Hexafluoride`, J. Chem. Thermodynamics, 1986, 18 (3), 235.

<II-33> Manuel A. V. Ribeiro da Silva, `Thermochemistry of .beta.-Diketones and Metal-.beta.-Diketones, Metal-Oxygen Bond Enthalpies`, in Manuel A. V. Ribeiro da Silva, ed., Thermo-Chemistry and Its Applications to Chemical and Biochemical Systems, D. Reidel Publishing Company, Dordrecht (1984), pp. 317-338.

<II-34> Ref. A. F. Wells, Structural Inorganic Chemistry, 5th ed., Clarendon Press, Oxford (1984), 540-543.

<II-35> E. W. Berg and F. R. Hartlage, `Fractional Sublimation of Various Acetylacetonates`, Anal. Chim. Acta, 1965, 33 (2), 173-81.

<II-36 > Aesar Catalog, 1989-21990, Johnson Matthey, Seabrook, N.H., p. 181.

<II-37> Susan Budavari, ed., The Merck Index, 11th ed., Merck & Co., Inc., Rahway, N.J. (1989 ) , pp. 710-1.

The present invention provides VFFTE apparatus, methods and chemistries for wafer fabrication and related technologies. It features relatively fast reactions at low temperatures and with minimal human handling of the wafers. Thereby wafer throughput can be greatly increased.

While the invention has been described in connection with specific embodiments thereof, it will be understood that it is capable of further modification, and this application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice in the art to which the invention pertains and as may be applied to the essential features hereinbefore set forth, and as fall within the scope of the invention and the limits of the appended claims.