6
Ion Processes, Neutral Chemistry, And Thermochemical Data

Introduction

In this chapter, the panel assesses the needs and status of cross sections and rate coefficients for ion processes and neutral chemistry in plasma processing reactors. It also assesses the availability of thermochemical data. Requirements for ion transport and cross section data are particularly stringent, because low-pressure plasma etching reactors typically operate at high plasma densities, producing ionization fractions of 10-4 to 10-2. Therefore ion collision processes (ion-ion neutralization and scattering, ion-molecule reactions, electron-ion recombination) are proportionally more important compared with neutral chemistry. The range of ion energies for which these cross sections and rate coefficients are required is large. Conventional reactive ion etching (RIE) reactors use capacitively coupled power for ion generation and acceleration and have applied potentials of hundreds of volts. This results in ions having directed energies of tens to hundreds of eV in the sheaths, while their random thermal temperatures in the bulk plasma tend to be small (<< 0.05 eV). Inductively coupled plasma (ICP) and microwave-excited electron cyclotron resonance (ECR) reactors differ from conventional RIE reactors in that they typically operate at lower pressures (<< 10 mTorr), higher power deposition, and larger ionization fraction. These conditions allow bulk ion temperatures to climb significantly above the gas temperature (0.1-0.5 eV).

The Knudsen number (Kn = mean free path / characteristic dimension) of ICP and ECR reactors may exceed 0.01 to 0.1. These long mean free paths complicate modeling in that noncontinuum algorithms must be employed as transport approaches the molecular flow regime. Large Knudsen numbers also imply that hot atom transport is more prevalent. Hot atoms have translational energies that significantly exceed their random thermal temperature. (The term "hot atom transport" is used here to refer to all hot neutral species: atoms, molecules, and radicals.) Hot atoms are generated by reactions of energetic ions with neutrals (charge exchange) or other ions (ion-ion neutralization); by energetic ions reflecting from surfaces and returning to the plasma as neutrals; and by dissociative electron collisions of molecules. The latter category includes electron impact dissociation of neutral molecules and dissociative recombination of molecular ions. Hot atoms are important for at least two reasons. First, they may impact on the wafer, thereby modifying the etching or deposition. Second, by virtue of their large translational energy, they may participate in reactions having activation energies that are otherwise energetically disallowed.

The availability of neutral chemistry cross sections and rate coefficients varies greatly depending on the specific reaction chemistry being used. Depending on the specific chemistry, rate coefficients may be available from the literature on combustion, upper atmospheric chemistry, or chemical vapor deposition (CVD). An important difference between the database needs in neutral chemistry for plasma etching reactors, as opposed to CVD, is the contributions of excited states. Significant fractions of the neutral species in a plasma processing reactor will have stored internal energy (vibrational or electronic) resulting from collisions with electrons or ions. Reactions that have activation energy barriers for reactants in their ground states may be energetically allowed (or accelerated) if the reactants are either vibrationally or electronically excited, or they may have different branching ratios.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 47
--> 6 Ion Processes, Neutral Chemistry, And Thermochemical Data Introduction In this chapter, the panel assesses the needs and status of cross sections and rate coefficients for ion processes and neutral chemistry in plasma processing reactors. It also assesses the availability of thermochemical data. Requirements for ion transport and cross section data are particularly stringent, because low-pressure plasma etching reactors typically operate at high plasma densities, producing ionization fractions of 10-4 to 10-2. Therefore ion collision processes (ion-ion neutralization and scattering, ion-molecule reactions, electron-ion recombination) are proportionally more important compared with neutral chemistry. The range of ion energies for which these cross sections and rate coefficients are required is large. Conventional reactive ion etching (RIE) reactors use capacitively coupled power for ion generation and acceleration and have applied potentials of hundreds of volts. This results in ions having directed energies of tens to hundreds of eV in the sheaths, while their random thermal temperatures in the bulk plasma tend to be small (<< 0.05 eV). Inductively coupled plasma (ICP) and microwave-excited electron cyclotron resonance (ECR) reactors differ from conventional RIE reactors in that they typically operate at lower pressures (<< 10 mTorr), higher power deposition, and larger ionization fraction. These conditions allow bulk ion temperatures to climb significantly above the gas temperature (0.1-0.5 eV). The Knudsen number (Kn = mean free path / characteristic dimension) of ICP and ECR reactors may exceed 0.01 to 0.1. These long mean free paths complicate modeling in that noncontinuum algorithms must be employed as transport approaches the molecular flow regime. Large Knudsen numbers also imply that hot atom transport is more prevalent. Hot atoms have translational energies that significantly exceed their random thermal temperature. (The term "hot atom transport" is used here to refer to all hot neutral species: atoms, molecules, and radicals.) Hot atoms are generated by reactions of energetic ions with neutrals (charge exchange) or other ions (ion-ion neutralization); by energetic ions reflecting from surfaces and returning to the plasma as neutrals; and by dissociative electron collisions of molecules. The latter category includes electron impact dissociation of neutral molecules and dissociative recombination of molecular ions. Hot atoms are important for at least two reasons. First, they may impact on the wafer, thereby modifying the etching or deposition. Second, by virtue of their large translational energy, they may participate in reactions having activation energies that are otherwise energetically disallowed. The availability of neutral chemistry cross sections and rate coefficients varies greatly depending on the specific reaction chemistry being used. Depending on the specific chemistry, rate coefficients may be available from the literature on combustion, upper atmospheric chemistry, or chemical vapor deposition (CVD). An important difference between the database needs in neutral chemistry for plasma etching reactors, as opposed to CVD, is the contributions of excited states. Significant fractions of the neutral species in a plasma processing reactor will have stored internal energy (vibrational or electronic) resulting from collisions with electrons or ions. Reactions that have activation energy barriers for reactants in their ground states may be energetically allowed (or accelerated) if the reactants are either vibrationally or electronically excited, or they may have different branching ratios.

OCR for page 47
--> Cross Sections and Rate Coefficients Ion Processes The needs for fundamental data on ion processes can be subdivided by energy range as indicated in Table 6.1. The division in energy range is based on two premises. First, the ion distribution in the bulk plasma for the majority of plasma processing reactors can be well described by a Maxwellian or moderately drifting Maxwellian. The temperatures are usually less than 0.1 eV. Transport coefficients and cross sections for processes in this range of energies are usually measured using a swarm or drift tube technique, and characterized by a "random" temperature. Ion energies greater than 0.1 eV are usually found only in the presheath or sheath regions of the plasma. Cross sections for collisional processes in the higher energy range are measured by beam techniques, and usually characterized by a directed energy. TABLE 6.1 Categorization of Data Needs for Ion Processes Process Thermal Superthermal   (e < 0.1 eV) (0.1 < e < 1 keV) Momentum transfer • • Ion-molecule and charge exchange • • Ion-ion neutralization •   Electron-ion recombination •   Ion-neutral and neutral-neutral excitation   • Momentum Transfer Momentum transfer collisions are elastic and/or inelastic processes resulting in a change in momentum and in which the identity of the ion does not change. Beam-measured cross sections should be resolved in energy and angle. Swarm-measured cross sections or rate coefficients should be temperature dependent. The availability of ion swarm parameters (usually mobilities, diffusion coefficients, and characteristic energies vs. E/N) is good for ions of interest in atmospheric or combustion applications, and poor for ions of interest to plasma processing. These data are typically scattered in the literature. A compendium of ion transport coefficients dating to 1984 is available from Ellis et al.1 Ion-Molecule and Charge Exchange Reactions Collisions resulting in transfer of positive or negative charge from the incident ion are ion-molecule or charge exchange reactions. If the identity of the ion does not change, this process is known as a symmetric charge exchange and is sometimes difficult to distinguish from a momentum transfer collision. Beam-measured cross sections should be resolved in energy and angle. Swarm-measured cross sections or rate coefficients should be temperature dependent. Distribution of products from the target molecule should also be identified. The availability of data for thermal ion-molecule and charge exchange reactions for molecules of atmospheric or combustion interest is relatively good. Many compendia are available, such as those of Sieck and Lias,2 Albritton,3 and Ikezoe et al.4

OCR for page 47
--> There are isolated instances of measurements of cross sections for these reactions in gases of interest to plasma processing. For example, Morris et al.5 have measured reaction rate coefficients for Tsuji et al.6 have made measurements of charge exchange of rare gases with etching gases. For example, Bohme has measured and assembled a database for ion-molecule reactions for silicon-bearing ions.7 Mandich and Reents8 have performed measurements of ion-molecule cross sections in silane systems using Fourier transform mass spectrometry, The latter work has shown that there are bottlenecks for silane ion association reactions that may be breached by the internal energy of the reactants or by the presence of impurities such as water. (See Figure 6.1.) A subset of ion-molecule reactions are ion association reactions The rate coefficients for these processes tend to be system specific even for rare gases. Scaling laws for their temperature dependence, however, have been developed by Johnsen9 Superthermal ion-molecule and charge-exchange collisions usually occur in the sheaths. These are processes that are not allowed at thermal energies in the bulk plasma but nevertheless should be addressed in the database for use in higher pressure systems where sheaths may be collisional. Figure 6.1 Reaction sequence for ion chemistry initiated by SiD+ in SiD4. (Reprinted, by permission, from M.L. Mandich and W.D. Reents, J Chem. Phys. 95:7360 (1991). Copyright © 1991 by the American Institute of Physics.)

OCR for page 47
--> Figure 6.2 Cross section set for impact of on SiF4. (Reprinted, by permission, from E.R. Fisher and P.B. Armentrout, Chem. Phys. Lett. 179:435 (1991). Copyright © 1991 by Elsevier Science - NL.) The energy dependence and product channels of cross sections for these reactions are required. Measurements of these cross sections for systems of interest to plasma processing are in general sparse; however, definitive experiments for a subset of the gases of interest have been performed by Armentrout et al.10 for energies of 0 to 80 eV. (See Figure 6.2.) These systems include The large variety of ion-molecule systems and possible reaction channels puts a large premium on developing scaling laws for these interactions. In this regard Armentrout et al.11 have proposed scaling laws for cross sections of endothermic ion molecule collisions. Ion-Ion Neutralization Ion-ion neutralization processes are collisions between positive and negative ions that result in the neutralization of both reactants. The rates of these reactions scale with the square of the plasma density, while the magnitude of Coulomb cross sections scales inversely with energy. Therefore ion-ion neutralization cross sections are required for thermal energies, but are not particularly important for superthermal energies. The total rate of ion-ion neutralization reactions will usually be ignorably small in the sheaths where positive ion energies are large and negative ion densities are small. The distribution of products should be identified.

OCR for page 47
--> Cross sections for these long-range Coulomb collisions are not particularly sensitive to the internal structure of the reactants, and therefore they are amenable to scaling laws based on the electron affinity and ionization potential of the reactants. For example, semiempirical cross sections for ion-ion neutralization at low pressure have been proposed by Moseley et al.12 Electron-Ion Recombination Collisions between electrons and positive ions resulting in neutralization of the ion constitute electron-ion recombination. The rates of these reactions scale with the square of the plasma density. Since electron-ion recombination results from a long-range Coulomb force, the cross section often scales inversely with energy. The total rate of electron-ion recombination reactions will therefore usually be ignorably small in the sheaths, and their cross sections (or rate coefficients) are required only in the thermal energy range. At the plasma densities and temperatures of interest for plasma processing, radiative and collisional radiative recombination are not important. Dissociative recombination of molecular ions is the only significant volumetric recombination process. The distribution of neutral products should be identified, as well as their translational energies. Cross sections for these processes are widely scattered in the literature, with few examples for systems of interest to plasma processing. Recent compendia and reviews of cross sections and rate coefficients can be found in Mitchell13 and Adams.14 The temperature dependencies of these processes are also important.15 Ion-Neutral and Neutral-Neutral Excitation Inelastic collisions between ions and neutral species that result in excitation, dissociation, or ionization of the neutral target without transfer of charge are classified as ion-neutral excitation collisions. Hot atom collisions on neutrals can similarly instigate excitation, ionization, or dissociation. Since there is a threshold energy associated with these collisions, they are important only for superthermal ions and usually only in the sheath regions. Similar reactions involving hot atoms may occur throughout the bulk plasma. There is fragmentary data scattered through the literature for energetic ion-neutral and neutral-neutral excitation collisions. Phelps16 has compiled and assessed complete cross section sets (see Figure 6.3) for ion and high-energy neutral impact reaction mechanisms for Neutral Chemistry Many neutral chemistry databases have been developed for combustion and atmospheric chemistry?17 Processes for evaluation of those databases are well established, and so evaluation is not addressed here. Instead, the panel assesses the status of the availability of rate coefficients and proposes methods to address unmet needs. It is important to emphasize that these needs are better stated in terms of mechanisms, as opposed to a collection of rate coefficients.

OCR for page 47
--> Status of the Database The database requirements for neutral chemistry are both more lenient and more stringent than those for combustion and upper atmospheric chemistry. The gas temperatures of plasma processing systems are typically low, rarely exceeding a few hundred degrees above ambient temperature. Therefore gas phase chemical reactions that have significant activation energy barriers are not important. Plasma processing systems with high gas temperatures typically operate at low gas pressures, where the rates of gas phase chemical reactions are small compared to wall-activated chemistry. At the same time, there are large densities of molecules and atoms in plasmas that have internal energy (vibrational or electronic) or that are translationally "hot", and that therefore breach activation energy barriers. Figure 6.3 Cross sections for excitation and ionization of H2 resulting from impact of Ar+ on H2, and by impact of H2 on Ar. (Reprinted, by permission, from A. Phelps, J. Phys. Chem. Ref. Data 21:883 (1992). Copyright © 1992 by the American Institute of Physics and the American Chemical Society.) The low gas pressures used in tools with high plasma densities (ICP, ECR) may further restrict the number of reactions that one must address. Association reactions typically operate through a transition state that must be stabilized by colliding with a third body to complete the reaction. For example, the association reaction proceeds as The effective 2-body rate coefficient is If the operating pressure is sufficiently low so that the back reaction of Cl2* to 2Cl is fast compared to the rate of stabilizing collisions, then the effective rate of association is small and the reaction may be ignored. The weakness in the existing databases for neutral combustion and atmospheric chemistry is that they were intended to be used at high pressure, and therefore lack rate coefficients for the low pressure fail-off regime. The important reactions may also depend on the operating conditions of the reactor, such as power deposition and gas residence time. For purposes of discussion, it is useful to define three classes of species: feedstock (F), secondary (S), and primary (P). The feedstock species are those gases that flow into the reactor from the outside. Primary species, usually radicals, axe atoms or molecules that result from direct dissociation of the feedstock gases, by either electron impact or ion-molecule reactions. Secondary species are produced by reactions between primary species or of primary species with feedstock gases. The degree to which F-P, P-P, P-S, or S-S reactions dominate is largely a function of the degree of dissociation of the gas. A convenient measure of the degree of dissociation is

OCR for page 47
--> where P is the specific power deposition, No is the ambient gas density, τ is the gas replenishment time (either by gas flow or surface reactions), and Δε is the energy deposition required for molecular dissociation. If η« 1, then the gas is lightly dissociated and F-P reactions are likely the most important. If η «, then P-P reactions are likely the most important. Finally, if η » 1, then S-S reactions most likely dominate. The availability of reaction rate coefficients for thermal neutral chemistry for plasma deposition systems is in some instances very good. For example, the SiH4/Si2H6/H2 system has been studied extensively in the context of plasma enhanced and thermal chemical vapor deposition, and compendia of rate coefficients have been assembled for use in models.18 The major uncertainties in these databases are in the formation and reaction of higher silanes . These reactions are particularly important with respect to particle formation. The situation is similar for databases for plasma enhanced chemical vapor deposition (PECVD) of SiO2 using SiH4/O2/N2O mixtures. The gas phase chemistry of this system has recently been addressed in the context of modeling remote PECVD of SiO219 and thermal CVD.20 In this system, the mechanisms for the deposition pathways are still uncertain, largely due to the uncertainties in surface reaction rate coefficients. This uncertainty reflects back on the "goodness" of the gas phase database. For example, it has been proposed that Si-O bonds are formed on the growing film21 and that precursors are dominantly fragments of SiH4 and the oxygen donor. If this is the case, the gas phase database is in moderately good shape. A competing school of thought proposes that Si-O bonds are formed in the gas phase and that silanols (SiHnOm) are the deposition precursors. If this is the case, the gas phase database is not adequate since the major uncertainty in the database results from uncertainties in the formation and reactions of silanols. The status of the database for Si3N4 deposition using SiH4/N2/NH3 mixtures is similar to that for SiO2 and is in fairly good shape.22 The deposition precursors for this system, however, are uncertain, and that situation again affects the "goodness" of the database. It has been proposed that the Si-N bonds form on the surface of the growing film for SH4/N2 mixtures, whereas they form in the gas phase for SiH4/NH3 mixtures.23 In the former case, the present status of the database is fairly good. In the latter case, rate coefficients for formation and reactions of the proposed deposition precursors, amino-silanes (SiHn(NH2)m), are largely unknown. Databases for deposition of dielectrics using gases other than SiH4 are currently inadequate. For example, deposition of SiO2 at low temperatures is often performed using TEOS (Si(C2H5O)4); however, the gas phase chemistry is virtually unknown. In all cases, the experimentally derived databases for reactions of vibrationally and electronically excited species range from very poor to nonexistent. Databases for etching chemistries are less complete than those for deposition, largely because of the lack of experimentally derived rate coefficients. The situation is improving based largely on rapid advances in computational chemistry, which have enabled accurate calculation of rate coefficients in ground and vibrationally excited states. A compendium of experimental and computed rate coefficients for the CF4/CHF3/H2/O2 system has recently been assembled for use at high pressure.24 An assessment of neutral chemistry databases for selected chemistries is given in Table 6.2. Excited State Chemistry and Penning Ionization Internal energy (either electronic or vibrational) of reactants increases the amount of energy available and therefore may bridge or reduce activation energy barriers. In plasmas, significant fractions of the atoms and molecules can have internal activation energy. To first order, one might simply reduce the activation energy barrier of the process by the amount of the internal energy. This, however, is a questionable practice, according to Armentrout.25 Electronic energy may not couple directly into the reaction coordinate, and the reaction efficiencies of ground and excited states are often very different even after the effect of total energy has been accounted for.

OCR for page 47
--> TABLE 6.2 The Status of Neutral Chemistry Databases for Selected Chemistries System Status Comments SiH4 / Si2H6 / H2 (Deposition of a-Si:H, p-Si) Good Database was initially compiled for silane combustion, CVD of p-Si, and PECVD of a-Si:H for photovoltaics. SinHm / O2 / N2O (Deposition of SiO2 oxynitrides) Fair to good Database was initially compiled for CVD of SiO2, silane combustion, and atmospheric chemistry SinHm / NH3 / N2 (Deposition of Si3N4 Fair to good Database was initially compiled for CVD of Si3N4 and atmospheric chemistry CnHm / H2 (Deposition of diamond, diamond-like carbon) Very good Database was initially compiled for combustion. CnFm / H2 / O2 (Etching of SiO2, Si) Fair to good The Plumb and Ryana mechanisms are ''standard'' but applicable to a limited parameter space. A new reaction mechanism and database developed by M. Zachariahb are now available SF6 (Etching of p-Si, W) Fair to poor Interest in modeling circuit breakers and spark gaps at high pressure Database with questionable application to low pressure CCl4 (Etching of p-Si) Fair Need for this database is minimal due to phaseout of the use of CCl4. BCl3 / HBr / C12 / NF3 (Etching of p-Si, metals) Poor These are examples of databases that are currently poor but are amenable to being addressed by calculations a I. Plumb and K. Ryan, Plasma Chern. Plasma Proc. 6:11 (1986); — 6:205 (1986); — 6:231 (1986). b D.R.F. Burgess, M.R. Zachariah, W. Tsang, and P.R. Westmoreland, NIST Technical Note 1412 (U.S. Department of Commerce, Technology Administration, July 1995). A subset of excited state chemistry is electronic quenching and Penning ionization. These reactions are collisions involving excited states of atoms or molecules and resulting in the deactivation of the excited state (quenching) and the transfer of energy to the collision partner. When the collision partner is ionized, the process is termed a "Penning ionization." Quenching reactions are important because they can transfer energy to the collision partner producing dissociation, and they may remove intermediates for multistep ionization. Interest in the development of excimer and metal ion lasers in the 1970s and 1980s resulted in a large database for rare-gas metastable quenching. (See, for example, Velasco et al.26) In the 1980s and 1990s, interest in PECVD of a-Si:H and its alloys for photovoltaics has supplemented that database with reactions involving SiH4, Si2H6, CH4, C2H6, and GeH4.27 In many cases, these rate coefficients and cross sections are for quenching of the excited state on a particular gas, and little information is given on the identity and branching ratios of the products. Summary It is clear that a large resource for cross sections and rate coefficients is currently available in the literature. Unfortunately, this resource was developed largely for use in fields other than plasma processing, and therefore is scattered and difficult to assemble. A first and necessary task is to assemble, evaluate, and disseminate the existing data.

OCR for page 47
--> As a consequence of past activities in generating databases for combustion, lasers, and atmospheric chemistry, techniques are now available for measuring or calculating many of the missing ion and neutral chemistry rate coefficients and cross sections. It is important to leverage that existing capability to address the database needs discussed here. A task equal in importance to the generation of rate coefficients and cross sections is the construction of "reaction mechanisms." In many cases these mechanisms may consist of reduced reaction sets that address the conditions of interest but may not be applicable to, for example, higher pressures. In constructing them, sensitivity analyses will identify key reaction sets or sequences that are particularly important and for which data may be lacking. In all cases, the choice of the systems of interest must be aligned with the current planning of the semiconductor industry, as indicated in the Semiconductor Industry Association's roadmap.28 Ion Processes The ion-molecule database is fragmentary for most of the chemistries of interest. The majority of ion-molecule reaction rate coefficients that are now available were not produced specifically for the semiconductor manufacturing industry, but rather for upper atmospheric chemistry, high-energy radiation chemistry, and gas-discharge or e-beam excited lasers. The techniques developed for generating those databases may be applied to the systems of interest with little new invention. Data for ion-molecule reactions in which either reactant is a radical are generally missing from existing databases and are of crucial importance to plasma processing. In principle, existing techniques can be applied to producing these data; however, new invention may be required to produce pure radical sources. In all cases, rate coefficients for the distribution of and branching ratios for products, and for reactions with radical species, are the least well known and therefore deserve particular attention. Electron-ion (dissociative) recombination and ion-ion neutralization are also classes of reactions for which experimental techniques exist to measure rate coefficients in the systems of interest. The major gap in the database is the identity of the products. New techniques may be necessary to resolve the identity and branching ratios of the products. Neutral Chemistry The database for neutral chemistry is fairly complete for select deposition systems, and significantly less complete for most etching systems. Incomplete data on rate coefficients for those systems are found predominantly among reactions of electronically, vibrationally, or translationally hot species, and among reactions in the fall-off pressure regime. Given thermodynamic properties, vibrational frequencies, and so on, computational methods are available to generate many of these rate coefficients, particularly for translationally and vibrationally hot species, and for the fall-off pressure regime. There is much greater uncertainty in calculations for reactions of electronically excited species. Applying and improving computational methods to fill the gaps in the neutral chemistry database are to be encouraged. These activities should be coordinated with a less exhaustive experimental program that provides high confidence and well-characterized rate coefficients for validating and benchmarking the computationally derived values. Thermochemical Data Thermochemical data describe the initial and final states of a chemical reaction in equilibrium with a thermal bath. Such data are essential for benchmarking (testing) models of chemical reactions, e.g., testing the validity of theoretical potential energy surfaces used in reaction models. These data are also29 used to reduce the complexity of possible reaction schemes used in chemical kinetics models by screening out energetically unfavorable reactions. In addition, in some cases it is possible to estimate reaction rate coefficients using thermochemical kinetics and transition state theory.

OCR for page 47
--> Thermochemical data of interest for plasma processing include homogeneous reaction energies, entropies, and energy levels for chemical reactions; molecular dissociation, chemi-ionization, and negative ion formation for ground and excited states of neutral and ionic species; heterogeneous reaction energies, such as heats of desorption; and solid state and thin film quantities related to annealing and transport of impurities and defects. In plasma processing discharges, significant deviations from local thermal equilibrium can exist even for neutral species, and this complicates the use of thermochemical information. Thermochemical properties are often of direct use in the modeling of thermal chemical processing systems, where the kinetic and internal energies of the reacting species are more properly described by a local temperature. Most of the currently available thermochemical databases have been developed to serve other fields, such as rocket fuel combustion, upper atmosphere chemistry, and pollution abatement. These databases do not include many of the more reactive gas-phase species of interest for plasma processing. One of the best known sources of thermochemical data, the JANAF Thermochemical Tables,30 provides data on molecular geometries, vibrational frequencies, heat capacities, entropies, enthalpies, and equilibrium constants. The extensive polynomial fits to the temperature-dependent thermochemical data are not usually applicable to models of plasma processing systems. The only relevant compiled sources of thermochemical data found for solids or gas-solid interfaces of interest in plasma processing are represented by the listings of thermochemical properties for solid silicon, carbon, and boron in the JANAF tables. Particularly needed are data on heats of desorption for various adsorbed and absorbed species. Understanding of surface thermochemistry, although not well enough developed to provide an exhaustive compilation, would benefit from data compilations and evaluations to establish the state and availability of the database. Although the first step in this process is to assemble and disseminate the currently available data, the value of evaluated databases must be emphasized. Data and data gathering techniques in original journal articles generally need to be evaluated further by critical comparison with other data and other techniques. Thermochemical data for gas phase etching-related compounds, such as SinClm, CnClm, BnClm, and their derivatives, are generally available. One concern is that there are significant differences (10 kcal/mol or 0.5 eV) among heats of formation of some of the SiClx molecules and ions (P.B. Armentrout, University of Utah, private communication, 1995). Thermochemical data for gas phase compounds related to amorphous hydrogenated silicon, such as SinHm, and their derivatives are also generally available. No reviews or compilations of data are known to the panel, however, regarding the thermochemistry of these species on semiconductor surfaces. Theoretical calculations of thermochemical data have become very useful for certain classes of compounds, such as organic molecules and small inorganic molecules with atoms from the first and second rows of the periodic table.31 In addition, recent work has suggested that the reactions of fluorinated hydrocarbons with oxygen, for example, can be fruitfully analyzed using ab initio techniques coupled with transition state theory.32 One possibility to improve the database is to develop multiple-species discharge systems for reaction measurements, such as the flowing afterglow technique. This apparatus was successfully applied to determining the energetics and kinetics of ion-molecule and radical reactions of interest to upper atmosphere and pollution chemistry. The principal effort here would have to be placed on surface reactions and properties using plasma-generated species. Currently, experiments on surface reactions are based largely on thermal processes, which generally address incident species at energies below about 700 K. The calculation of therrnochemical properties of ground state species using quantum chemistry and empirical approaches has reached the point of considerable utility for estimation of unknown values. In some cases, however, discrepancies with experimental data emphasize the need for additional refinements to the techniques. Although the overlap with thermochemistry data requirements is limited, it is desirable to maintain a collaboration with the proposed data generation efforts for chemical vapor deposition (CVD).33

OCR for page 47
--> Some currently available sources of relevant thermochemical data are listed in the references.34 Findings The database for ion-molecule and neutral-neutral chemistry varies considerably. For some species and reactions, the data are good. This is especially true for cases in which there is overlap with processes occurring in the upper atmosphere or in some cases in chemical vapor deposition processes. In other cases, however, most notably for etching processes, few data are available. Thermochemical data are sketchy for many species of interest in plasma processing. These data are important in helping to establish boundaries for reaction pathways and in estimating reaction rate coefficients. Techniques, both experimental and computational, are generally available to obtain these quantities, but few efforts are under way at present to meet these needs. References 1. H.W. Ellis et al., Atomic Data and Nuclear Data Tables 7:177 (1976); — 22:179 (1978); — 31:113 (1984). 2. L.W. Sieck and S.G. Lias, "Rate Coefficients for Ion-Molecule Reactions. I. Ions Containing C and H," J. Phys. Chem. Ref. Data 5:1123 (1976). 3. D.L. Albritton, "Ion-Neutral Reaction-Rate Constants Measured in Flow Reactors Through 1977," Atomic Data and Nuclear Data Tables 22:1 (1978). 4. Y. Ikezoe, S. Matsuoka, M. Takebe, and A. Viggiano, Gas Phase Ion-Molecule Reactions Rate Constants Through 1986 (Ion Reaction Research Group of the Mass Spectroscopy Society of Japan). 5. R. Morris, A.J. Viggiano, and J.F. Paulson, J. Phys. Chem. 97:6208 (1993); R. Morris, A.J. Viggiano, J.M. Van Doren, and J.F. Paulson, J. Phys. Chem. 96:3051 (1992). 6. M. Tsuji, T. Fumatsu, H. Kouno, and Y. Nishimura, Chem. Phys. Lett. 192:362 (1992); H. Obase, M. Tsuji, and Y. Nishimura, J. Chem. Phys. 99:111 (1985). 7. D.K. Bohme, "Chemistry Initiated by Atomic Silicon Ions in the Gas Phase: Formation of Silicon Bearing Ions and Molecules," Int. or. Mass Spectrom. Ion Processes 100:719 (1990). 8. M. Mandich and R. Reents, J. Chem. Phys. 95:7360 (1991). 9. R. Johnsen, J. Chem. Phys. 85:3869 (1986). 10. M.E. Weber and P. Armentrout, J. Phys. Chem. 93:1596 (1989); E.R. Fisher and P.B. Armentrout, Int. J. Mass Spectrom. Ion Processes 101:R1 (1990); E.R. Fisher, B.L. Kickel, and P.B. Armentrout, J. Phys. Chem. 97:10204 (1993). 11. B.H. Boo, J.L. Elkind, and P.R. Armentrout, J. Am. Chem. Soc. 112:2803 (1990). 12. J. Moseley, R.E. Olson, and J.R. Peterson, Case Stud. At. Phys . 5:1 (1975). 13. J.B.A. Mitchell, Phys. Rep. 186:215 (1990). 14. J.G. Adams, Int. or. Mass Spectrom. Ion Processes 132:1 (1994). 15. J.S. Chang, R.M. Hobson, Y. Ichikawa, T. Kaneda, N. Maruyama, and S. Teii, J. Phys. B 22:L665 (1989). 16. A. Phelps, J. Phys. Chem. Ref. Data 20:557 (1991); — 21:883 (1992). 17. See, for example, W. Tsang and R.F. Hampson, J. Phys. Chem. Ref. Data 15:1087 (1986); R. Atkinson, D.L. Baulch, R.A. Cox, R.F. Hampson, J.A. Kerr, and J. Troe, J. Phys. Chem. Ref. Data 18:881 (1989); F. Westley, J.T. Herron, R.J. Cventanovic, R.F. Hampson, and W.G. Mallard, NIST Chemical Kinetics Database (1994). 18. M.J. Kushner, J. Appl. Phys. 63:2532 (1988); M.E. Coltrin, R.J. Kee, and G.H. Evans, J. Electrochem. Soc. 136:819 (1989). 19. M.J. Kushner, J. Appl. Phys. 74:6538 (1993). 20. C.J. Guinta, J.D. Chapple-Sokol, and R.G. Gordon, J. Electrochem. Soc. 137:3237 (1990). 21. G. Lucovksy, D. Tsu, and R. Markunas, ch. 16 in Handbook of Plasma Processing Technology, eds. S.M. Rossnagel, J.J. Cuomo, and W.D. Westwood (Noyes Publications, Park Ridge, N.J., 1990). 22. M.J. Kushner, J. Appl. Phys. 71:4173 (1992). 23. D.L. Smith, A.S. Alimonda, and F.J. von Pressig, J. Vac. Sci. Technol. B 8:551 (1990). 24. D.R.F. Burgess, M.R. Zachariah, W. Tsang, and P.R. Westmoreland, NIST Technical Note 1412 (U.S. Department of Commerce, Technology Administration, July 1995). 25. p. Armentrout, Science 251:175 (1991). 26. J.E. Velasco, J.H. Kolts, and D.W. Setset, J. Chem. Phys. 69:4357 (1978).

OCR for page 47
--> 27. H. Chatham, D. Robertson, and A. Gallagher, J. Chem. Phys. 79:1301 (1983). 28. Semiconductor Industry Association, The National Technology Roadmap for Semiconductors (SEMATECH, Austin, Tex., 1994). 29. Y.F. Wang and R. Pollard, J. Electrochem. Soc. 142:1712 (1995). 30. M.W. Chase, Jr., C.A. Davies, J.R. Downey, Jr., D.J. Frurip, R.A. McDonald, and A.N. Syverud, JANAF Thermochemical Tables, 3rd edn., J. Phys. Chem. Ref. Data, suppl. 1 (1985). 31. See, for example, P. Ho and C.F. Melius, J. Phys. Chem. 94:5120 (1990). 32. D.R.F. Burgess, M.R. Zachariah, W. Tsang, and P.R. Westmoreland, NIST Technical Note 1412 (U.S. Department of Commerce, Technology Administration, July 1995); M.R. Zachariah, W. Tsang, P.R. Westmoreland, and D.R.F. Burgess, J. Phys. Chem. 99:12512-12519 (1995). 33. See, for example, J.R. Whetstone et al, White Paper for a Chemical Kinetics Database to Support Integrated Circuit (IC) Manufacture, SEMATECH Technology Transfer #94072443A.XFR (September 1994). 34. M.W. Chase, Jr., C.A. Davis, J.R. Downey, Jr., D.J. Frurip, R.A. McDonald, and A.N. Syverud, JANAF Thermochemical Tables, 3rd edn., J. Phys. Chem. Ref Data 14, suppl. 1 (1985) [see NIST/SRD Products Catalogue SP 782 for hard-copy, floppy disk, and on-line versions of this and other databases]; L.V. Gurvich et al., Thermodynamic Properties of Individual Substances, 3rd edn. (English) (Nauka, Moscow, 1978) [similar to JANAF, this reference summarizes rules for estimating thermochemical data from structure]; S.G. Lias, J.E. Bartmess, J.F. Liebman, J.L. Holmes, R.D. Levin, and W.G. Mallard, "Gas-Phase Ion and Neutral Thermochemistry," J. Phys. Chem. Ref. Data 17, suppl. 1 (1988) [see NIST/SRD SP 782 for an updated version on floppy disk]; R.J. Kee, F.M. Rupley, and J.A. Miller, The Chemkin Thermodynamic Data Base, Sandia National Laboratories Report SAND87-8215B (1990) [specific heats, standard state enthalpies, and entropies of species and reactions related to combustion and to CVD of silicon from silane; part of the Chemkin chemical kinetics code; available from the authors; also see Report SAND89-8009B (1993)]; M.E. Jacox, "Vibrational and Electronic Energy Levels of Polyatomic Transient Molecules," J. Phys. Chem. Ref. Data monograph 3 (American Chemical Society, Washington, D.C., 1994); K.P. Huber and G. Herzberg, Constants of Diatomic Molecules (Van Nostrand, New York, 1979).