Plasma Chemistry and Plasma Processing, VoI. 9, No. 1, 1989 (Supplement)

Applications and Trends of Nonequilibrium Plasma Chemistry with Organic and Organometallic Compounds H. Suhr1 Received January 15, 1988; revised February 20, 1988

The possibilities and limitations of available apparatus for nonequilibrium discharges are discussed. Especially for synthetic work there is a lack of suitable equipment. The unique possibilities which plasmas offer to chemistry are demonstrated by examples from the homogeneous gas phase, and from plasma liquid and plasma solid interactions. Various applications and major trends are being described. The most rapidly increasing field of plasma chemistry is presently the formation of thin films of metals, oxides, carbides, or nitrides by plasma enhanced CVD. The latest results, and especially the use of organometallic compounds and starting material, are being discussed.

KEY WORDS: Nonequilibrium; organic compounds; organometallic. 1. I N T R O D U C T I O N In recent years, nonequilibrium plasma techniques have become a key technology providing essential steps in a variety of fields like chemistry, physics, materials science, and electronics. The following paragraphs will focus on the importance of plasmas in organic chemistry and related areas. Chemistry is an example in which plasma technologies are already very helpful, and their importance is liable to experience a strong increase in the future. An appropriate way to obtain information on the present state, trends, and future possibilities of a field would be to consult with experts. The present organizers, possibly in order to save on compensations for such services, preferred to invite a number of specialists to a "workshop." However, although the participants of this workshop are outstanding in their own fields, they are not familiar with organic plasma chemistry and were thus not able to contribute to an evaluation of its potential. Therefore, 1Department of Organic Chemistry, Universityof Tfibingen, Tfibingen, F.R.G. 7S 0272-4324/89/0300-007S$06.00/0

(~) 1989 Plenum P-ablishing Corporation

8S

Suhr

the following paragraphs do not present the opinion of the workshop as a whole, but only a single man's view of the topic. The following review will not try to cover all the work which has been done in this field; it will rather demonstrate the various possibilities of using plasma techniques in chemistry by selected examples. 2. E Q U I P M E N T FOR P L A S M A CHEMISTRY

Before discussing applications and trends in plasma chemistry, it is necessary to describe the equipment available for these techniques. The apparatus strongly dictates what presently can be done. Frequently, successful laboratory experiments cannot be transferred to industrial processes for lack of suitable equipment. The main types of apparatus for nonequilibrium plasmas, their present state of development, and the future possibilities shall be discussed briefly. Figure 1 shows a parallel plate type reactor that is widely used in microelectronics for etching and coating processes. It consists of two circular plates of 10-50 cm diameter with a spacing of a few centimeters. In deposition processes, both electrodes are of the same size; in reactive ion etching, the lower electrode 'is smaller to generate a self bias. Under conditions of low pressure (10-100 Pa) and low power density (0.1-1 W/cm2), a uniform discharge is generated between the two electrodes. A serious problem in etching, and even more in deposition, is the generation of a gas flow which is constant over the entire surface area. To achieve this uniformity, special arrangements concerning the gas inlet and exhaust system have to be used. Parallel plate reactors have been used for about two decades in surface treatment and have replaced most other types of reactors. Their application is, however, limited to fiat and small samples. By using parallel plate type reactors, anisotropic etching has become possible. This has been the strongest stimulus for plasma technologies in recent years. Because of their stability and flexibility, parallel plate type reactors have become the standard equipment in microelectronics and are by now commercially available from a number of manufacturers. Their only disadvantage is the intensive ion bombardment, which might cause radiation

i"-"Z.i:iZ:.::::! ::::". Fig. 1. Parallelplate reactor for DC and AC-< 100 MHz.

Organic and OrganometallicCompounds

9S

:!i : i :

T

y

Fig. 2. High-frequencydischarge 5-100 MHz. damage to sensitive substrates. Nevertheless, parallel plate reactors will be used in the future without major alterations in size. There may be, however, a development from batch operation to continuous operation by using appropriate slots. Figure 2 shows an electrodeless discharge. It is operated with high frequency, mainly (but not exclusively) with the industrial frequency of 13.56 MHz. In most cases, the reactor has a tubular shape and is surrounded by a coil or by two rings which are connected to a frequency generator. Electrodeless discharges are operated at 10-1000 Pa and powers between 1-1000 W. They have the advantage of simplicity and safety, and are thus ideal setups for exploratory research on laboratory scale. A similar design is realized with certain plasma torches which are, however, operated at a much higher pressure (--1 Bar) and higher power levels (10-50 kW). Figure 3 shows a barrier or ozonizer type discharge. The reactor consists of two AC high-voltage electrodes which are separated by one or two layers of insulating material and a small gap of 1-2 mm. The insulating material limits the total current to a low value. The major advantage of this equipment is its simplicity and the fact that it can be operated at atmospheric pressure. Its design has hardly been changed for more than 100 years. Only recently, the efficiencies have been improved by using higher frequencies. Single units are operated at 1-5 kW. In large ozonizers many small tubular units are operated in parallel. A simplified picture of a corona discharge can be seen in Fig. 4. A typical feature of this kind of discharge is the extreme difference in size between the two electrodes, one being a point or a sharp edge, the other a plate. The currents are low (/xA-mA). Corona discharges can be operated at atmospheric pressure or under vacuum conditions.

!

Fig. 3. Barrier discharge or ozonizer 50 Hz-50 kHz.

10S

Suhr

I

..:: .::... 9 ." 9

9

9

,,,.

.: ""

: ,

- 9149

: ... i

-... "l

F i g . 4. C o r o n a

" "l

discharge,

0 Hz-

MHz.

High-pressure corona discharges are used in the treatment of plastic foils. These pass over a sharp-edge corona discharge in air or oxygen. The corona activates the oxygen or nitrogen molecules which then attack the polymer and change its surface properties. Such treatment is very simple but difficult to control and often suffers from a low reproducibility. A similar design is used in industrial reactors for the polymer coating of car headlights. The reactors have a cylindrical shape and a volume of ~ i 0 m 3 ; they are operated at a low pressure with a discharge burning between a central metal rod and the outer wall. From the electrical data (I ~ 10 A, U = 300 V) it is evident that this is not a corona discharge9 The electrons necessary to sustain the discharge are supplied by thermal emission from the electrically heated rod. Equipment of such kind is especially suited for the coating of three-dimensional objects, provided that they are being turned inside the reactor. Figure 5 shows a microwave discharge with a tubular reactor penetrating through a resonance cavity9 Microwaves are used to a large extent in heating or sputtering but seldom in plasma technology. Difficulties arise from the fact that the wavelengths are of the same dimension as the reaction vessel which causes an uneven distribution of energy and power. There have been several attempts to overcome these difficulties. One promising approach is

!

P 9 ".~..

,~,'.'-.

:' 9

"":r;:.'~. " 4 " Z': .'.'.'""'.' 9 9

,..;2 9149

9 ., . 9

.........

F i g . 5. M i c r o w a v e

i

discharge.

t' 9

)

Organic and OrganometallicCompounds

11S

the use of slow-wave structures, another the combination of microwave plasmas with magnets. Both techniques have already been developed for rather large dimensions, and it is to be expected that such microwave discharges will become an important tool for both volume and surface treatments. The possibility of using electron-beam-sustained discharges in plasma technology has up to now hardly been studied. The apparatus (Fig. 6) consists of an electron gun supplying a beam of fast electrons which can penetrate through a metallic window into a reaction chamber. Here they ionize gas molecules which then are accelerated in a field generated by two sustainer electrodes to effect the chemical reactions. This arrangement, which has been developed for CO2 lasers, seems also very attractive to chemistry because such discharges can be carried out in large volumes and because of their unusual electron energy distribution which permits adjustment to the specific chemical problem. There have been a few attempts to use electron-beam-sustained discharges in chemistry and a flue-gas treatment in pilot plant scale. Summarizing the present state of equipment for nonequilibrium discharges, one can say that the parallel plate reactors have more or less reached their final design and will be used extensively in the future without major changes. A similar situation exists for ozonizer discharges. Electrodeless high-frequency discharges are typical devices for laboratory work and will continue to be restricted to research laboratories. Corona discharges at normal pressure will probably experience a much larger use in the future if a better control of the process parameters becomes possible. The cylindrical reactors with heated electrode might well become the standard techniques for large-scale surface treatments. Presently, there is a great demand for large volume reactors to carry out gas-phase reactions or surface treatments. It is being hoped that microwave discharges of the slow-wave structure or in combination with magnets, or possibly electron-beam-sustained discharges might fill this gap.

,•Sustainer

electrode

win2~

J

!

Fig. 6. Electron-beam-sustained discharge 0 Hz-kHz.

12S

Suhr

3. APPLICATION OF PLASMA TECHNIQUES IN ORGANIC CHEMISTRY The term "plasma chemistry" is used for a variety of operations, some of which are mere physical processes, like melting or evaporation. In the following paragraphs, only real chemical transformations shall be discussed. The first question to be asked is what benefits can a chemist get from using plasmas? A rather general answer comes from considering the elementary steps of chemical reactions. Molecules can only react when they have enough energy to overcome activation barriers. In conventional chemistry this energy is transferred to the molecules by collisions with other particles or with the wall. In photochemistry and plasma chemistry, the necessary energy is supplied by photons or by electrons, respectively. Since the different excitation processes involve different amounts of energy, they lead to a variety of excited species (Fig. 7). The conventional excitation is characterized by a stepwise transfer of vibrational energy until the lowest reactive state is reached. In photochemistry, both vibrationally and electronically excited molecules are formed. In collisions with electrons, much larger energies can be transferred and new species like superexcited neutrals, positive, and negative ions are formed. These new species, which cannot be generated by any other method in reasonable quantities, are the main justification for the use of plasmas in chemistry. While these general considerations are valid for both inorganic and organic reactions, there are differences in the experimental requirements. Whereas inorganic molecules are usually simple and have only one or a rather limited number of different bonds, most organic molecules are more complex and have numerous different bonds. Even simple molecules like ethanol (Fig. 8) contain five different bonds which have, however, very similar strengths. Moreover, practically all organic compounds have similar

Energy ranges

I

reactive species

I

M rib

Conventional chemistry

I

I

M rib

M'el

M~i b

M'el

Photochemistry

IPlasma chemistry I

0

)

M**

M- M +

I

5 eV

Fig. 7. Energy ranges and reactive species in conventional chemistry, photochemistry, and plasma chemistry.

Organic and OrganometallicCompounds

13S

Y Y 4.23

3.BO

4.02

3.93

4.50

Fig. 8. Bond strengths in eV in ethanol.

ionization potentials while inorganic compounds exhibit a much larger spread. Accordingly, the plasma treatment of organic starting materials requires far greater selectivities than reactions of inorganic compounds. The main demands on a suitable equipment to be used in plasma syntheses are a tong-time stability and a high-mass throughput. These requirements are difficult to meet with the apparatus presently available. The equilibrium discharges of arc or plasma torches are not suited for organic compounds because of their high temperatures which result in the thermal destruction of the organic matter. Neither does any of the various nonequilibrium discharges discussed previously present a satisfying solution. Ozonizers cannot be used because deposits are formed from organic compounds which change the discharge characteristics and eventually cause their extinction. The various glow discharges are all operated at low pressures and thus do not provide a sufficient mass throughput. In spite of this limitation, numerous syntheses have been developed using mainly electrodeless radio-frequency discharges. Plasma reactions can be carried out either in the glowing zone of a discharge or in the afterglow region. Some are restricted to the luminous zone while others can be accomplished in both regions. Depending on the state of the starting material, reactions can be carried out in the homogeneous gas-phase or as interactions between plasmas and liquids or solids. 4. SYNTHESES IN HOMOGENEOUS GAS PHASE Several hundred organic compounds have been subjected to glow discharges, either in the form of neat organic vapors or diluted with a carrier gas. Frequently, the reactions which took place were unselective and thus not suitable for synthetic work. In a large number of cases, however, the chemical reactions have a high or at least a sufficient selectivity to be useful in preparative work. Since the synthetic possibilities have been reviewed

14S

Suhr

only recently,~1) a few examples of the generation of reactive species, isomerizations, and eliminations may be sufficient. Reactive species which can easily be generated in plasmas are, for example, atomic hydrogen, oxygen, and nitrogen and various radicals. The atoms are easily formed by the dissociation of the molecular gases. While hydrogen and nitrogen atoms are used mainly in inorganic reactions, oxygen atoms are of special interest to organic reactions. Likewise, it is possible to generate radicals by the dissociation of suitable precursors. For example, hexafluoroethane dissociates into C F 3 radicals which can react with metals or metal halides: ~2~ C2F6 ~ 2 CFs+HgCI2 ~

2CF3 Hg(CF3)2+2 C1

Another interesting reactive species is dehydrobenzene which in a plasma can be easily prepared, for example, from phthalic acid anhydride (Fig. 9). Without additives, dehydrobenzene dimerizes to form biphenylene. In the presence of suitable substances, dehydrobenzene and two other intermediates can be scavenged (Fig. 9). Frequently, exposure to a plasma causes the rearrangement of organic molecules. Occasionally, these are related to isomerizations in photochemistry or catalysis. A number of rearrangements are, however, unique

coX,

Z

Fig. 9. Plasma decompositionof phthalic acid anhydride and scavenging of the various intermediatesby ammonia.

Organic and Organometallic Compounds

15S

and are only found in plasmas or in mass spectroscopy. Two examples of this sort are shown in Fig. 10. The isomerization of a cyclic compound (pyrrol) to an unsaturated nitrile has been observed for many heterocyclic nitrogen compounds. This reaction might be of interest for plasma polymerizations and shall be discussed later. The second reaction exemplifies the rearrangement of an aryl-alkyl-ether to an alkylphenol. This isomerization is presently studied with regard to its development as a commercial process. Among the multitude of organic plasma reactions, eliminations have been studied most thoroughly. The purpose of this type of reaction is to remove small groups from larger molecules without drastic structural changes. Ring contractions, the generation of multiple bonds, ring formations, and condensation reactions are especially attractive for preparative work (Fig. 11). Figure 12 shows three examples of ring contractions. In all three cases, the decarbonylation is achieved in good yields. The same products can also be synthesized by conventional methods, but not in a single-step reaction. Plasma reactions are especially attractive when they lead from easily accessible starting materials to new products in simple one-step processes. The dissociation of molecules, isomerizations, and eliminations are examples of monomolecular reactions. Due to the low pressure and the relatively high power density these are favored in glow-discharge plasmas. Bimolecular processes are possible but have been studied only for a few cases. Two early research projects aimed at finding new routes to ecaprolactam, a precursor of synthetic polyamid (Fig. 13). In the first, cyclohexane passed jointly with molecular oxygen a discharge and was oxidized to a mixture of cyclohexanol and cyclohexanone. This process has been developed on a pilot plant scale. (3) In the second project, mixtures of cyclohexane and nitrogen oxide are fed into the reactor. The cyclohexyl radicals generated by the plasma interaction are subsequently scavenged

~

CH3-CH=CH-CN

~"~'~O.CH 3 Fig. 10. Examplesof plasma isomerizations.

16S

Suhr

0

C

X--Y

I

\c /

-C--X

i

II

/C\

X--Y

C--Y

,0

X--Y

I -C-X

1

+

I -C-Y

I

I I > -C--C-

-x--Y

I

I

Fig. 11. Examplesof plasma eliminations. by NO to form nitroso cyclohexane, which can easily be isomerized to e-caprolactam. (4) In spite of some promising results, there are nevertheless severe restrictions to gas-phase plasma syntheses. The main problem is the similarity of bond strengths and ionization potentials of all organic compounds which limits the selectivity attainable. Another difficulty arises from the formation of hot products in the course of plasma reactions. Because of the low pressure in the system, the quenching of these hot products is inefficient, and various consecutive reactions result. The addition of CH2 to ethene may serve as an illustration (Fig. 14): This addition is exothermic and the reaction energy exceeds the bond strengths of the resulting cyclopropane. Consequently, in gas-phase reactions the CH2 addition does not lead to cyclopropane but to propene. In the liquid phase, however, the hot primary product is deactivated rapidly, and the cyclopropane synthesis is feasible.

Organic and OrganometallicCompounds

[ • • ~

.

-

17S

( ~ ~

98%

CO

0

~.-> 0

-

CO

-

CO

57%

0 Fig. 12. Plasma decarbonylations as examples for ring contractions.

5. PLASMA INTERACTIONS WITH LIQUIDS The limitations in selectivity encountered in gas-phase plasma reactions stimulated an investigation of the synthetic possibilities of plasma-liquid interactions. In liquids, the hot primary products would be quenched effectively. Furthermore, stirring of the liquid constantly removes product

0

+ 0 2 --.->

+ NO Fig. 13. Potential plasma synthesis of e-caprolactum.

18S

Suhr

C\H2 CH2 + H2C--CH2 C\H2 H2C-CH 2

;

:, H 2 C - C H 2

..C~H2 H2C-CH 2

-90kcallmol

+ 60 Kcall tool

Fig. 14. Addition of C H 2 to ethene. molecules from the surface and consecutive reactions would thus be easily suppressed. Both effects should result in a significant increase in selectivity. There are hardly any reports on such interactions with the exception of the plasma treatment of lubricants, which was developed at the turn of this century (voltaic oil), and a few experiments in plasma electrolysis using one submerged electrode in a conductive liquid. The development of plasma-liquid processes was simple and quite successful. The only necessary requirement concerning the starting material is a low vapor pressure of the liquid. To realize this condition, most liquids have to be cooled down close to their respective melting points. Figure 15 shows a simple arrangement for the interaction of oxygen plasmas with a liquid surface. Plasma oxidations quite nicely demonstrate the advantages of plasmaliquid interactions. When oxygen passes through a glow discharge, atomic oxygen and singlet oxygen are generated. Oxygen atoms can either add to double bonds or abstract hydrogen (Fig. 16). The addition is an exothermic reaction leading to a hot adduct, which easily isomerizes to a carbonyl

Fig. 15. Plasmatreatment of liquids.

Organicand Organometallie Compounds

19S

~

\.

xc

C/ § - ~H

/

"-"-~

add

2~ "0

\/

3p 0

~

\/-'\/

"c-cH

)c-cH

mi9

-C-H '

+ O ~ abs

~

-C. + . O H

-->

'

-C-OH '

Fig. 16. Reactions of oxygen atoms.

compound. Amongthe three products, namely the epoxide, the carbonyl compound, and the alcohol, only the first is of preparative interest. In the gas phase, however, kaaa is about 10 times higher than kabs. Since alkenes have only one C = C bond but several CH bonds, alcohol formation is the predominant reaction. Furthermore, the addition leads mainly to the carbonyl compound and only to a small extent to the epoxide (Table I). In the liquid phase, the addition is about two orders of magnitude faster than the abstraction reaction, and the addition leads predominantly to the epoxide and only to a minor extent to the carbonyl compound. Thus the plasma-liquid interaction shows a much higher selectivity with regard to the desired reaction product. (5) Although these results demonstrated the advantages of plasma-liquid interactions, there are still severe limitations to the application of this approach since most organic compounds, even close to their melting points, have a vapor pressure which is too high for such a treatment. Because of these difficulties the possibility of using solvents for organic plasma reactions was drawn into consideration. Compounds suitable as Table I. Selectivity in Competing Reactions

kaaa/ k~b~ Gas phase Liquid phase

10 200

kcyr 0.3 1-4

20S

Suhr

solvents should have a low vapor pressure, a low melting point, should be inert against oxygen plasmas, and should sufficiently dissolve the compounds to be treated. There exists, however, not a single substance which fulfills all these requirements. As a compromise, solvents can be used whose reactivity is sufficiently lower than that of the compound to be treated. To test these possibilities a number of binary mixtures have been studied. ~6~In rare cases, the two components reacted independently of each other (Fig. 17a). Normally, there is a preferred attack on the substrate while the "solvent" shows a much lower reactivity (Fig. 17b). If the difference in reactivity is sufficient the mixture may contain 50-70% solvent and still show a preferred attack on the substrate. In some favorable cases, the yield could be enhanced by adding a solvent (Fig. 17c). Occasionally, the increase is ~50% effected by the addition of roughly 50% solvent. Naturally, such cases are of great interest to synthetic work. It has also been observed that in some cases the product distribution can be modified by the addition of a solvent, as for example in the oxidation of toluene (Fig. 18). Tffe field of plasma-liquid interactions is quite new but it seems to be very promising for synthetic work. It permits the treatment of liquids, of solutions, and of solid starting materials. Especially, the yield increase and the change of product distribution on dilution deserve further extensive investigation. The present situation of the preparative organic plasma chemistry can be summarized in the following way. A considerable number of gas-phase reactions presenting useful syntheses have been found. Plasma-liquid interactions are also attractive for preparative work because of their high selectivity. All these reactions have been studied and optimized in laboratoryscale equipment. Presently, there is no industrial application. The reason for this is not a lack of demand for such syntheses but the nonexistence of

yield

yield

ield k

k

,,s,,,,,,," \ 0

% solvent I00 a

.... ,, .... ,s,,,,,,,, " "i 0

% solvent I00 b

0

% solvent 1oo C

Fig. 17. Plasma reactions in binary mixtures of a substrate A and a solvent S.

Organic and OrganometallicCompounds

21S

OH

<~)'-'CH3+ 0 ~

~CH3 HO @ell HOOCH

ortho 3 meta

3 para

Fig. 18. Changeof product distributionby solvents.The ratio of ortho- meta-parasubstituted products depends on the solventand the degree of dilution. reactors with a sufficient mass throughput. For gas-phase reactions, electronbeam-sustained discharges might be a solution to this problem. The scale-up for plasma-liquid interactions will probably be much easier. In this field, future developments will aim at increasing the surface area of the liquids by using small droplets or sprays. 6. INTERACTION OF PLASMAS WITH SOLIDS The interaction of plasmas with solids is restricted to thin surface layers. Therefore, these processes are not suitable for synthetic work, but they are extremely valuable in a variety of surface treatments. Depending on the physical state of the starting material and the products, etching, surface modifications, or the deposition of coatings can be attained. If the plasma-solid surface interaction leads to a gaseous product, the solid is being etched. This etching can be anisotropic or isotropic depending on the discharge system. Isotropic etching is being used for photomask stripping and for "cold ashing," where the solid organic material is converted into CO2 and H20 by oxygen treatment. Anisotropic etching is possible only in parallel plate reactors and is applied exclusively to inorganic materials (see article by S. Vep~ek, p. 29S). Although etching has been studied extensively and is being widely applied in microelectronics, it is still possible and necessary (e.g., for new materials) to develop new etching processes as, for example, the etching of metals and certain III/V semiconductors by radicals: (7's~ GaAs+CH4 --~ Ga(CH3)3+AsH3 S b + C H 3 - C O - C H 3 --* Sb(CH3)3+CO If the interaction of the plasma gas with the surface leads to a solidreaction product, the upper layers of the surface are being modified. There are several examples for the plasma modification of metals (see S. Vep~ek,

22S

Suhr

p. 29S) and organic polymers. When the latter are subjected to oxygen discharges, oxygen-containing groups are formed in the surface layer. Such a treatment often changes the surface properties significantly; it can, for example, give hydrophobic polymers a hydrophilic surface. When the components of the plasma gas react to form a solid all surfaces exposed to the plasma are coated by a powder or a solid film. "Plasma-enhanced chemical vapor deposition" (PECVD) is at present the area with the highest research activity in the field of plasmas, and it will probably become the most important application in the future. While the conventional techniques of painting, spraying, or dip-coating are satisfying for thick films, thin films prepared by these methods are not coherent enough and show insufficient adhesion to the substrates. In contrast to this vaporphase coatings often have excellent adhesion qualities, even as very thin films. Furthermore, PECVD allows one to prepare films with composition gradients. By suitable choice of parameters, it is possible to deposit films which are pinhole-free, have a good step-coverage, and give no shadows. Thus PECVD processes are useful for coating complicated surface morphologies. PECVD experiments are almost exclusively carried out in parallel plate reactors. Volatile compounds are introduced into the reactor as starting materials where they fragmentate into a solid and gaseous part, of which the latter is being pumped off (Fig. 19). The oldest PECVD process is plasma polymerization. All compounds which can undergo conventional polymerization (alkenes) are also suited for plasma polymerization. While the conventional polymerization leads

~ ,n

t

~\\\\\\\\\\"~ I

+"i -+

M

Fig. 19. Standard reactor for PECVD.

I

M

oo,

Organic and Organometallic Compounds

23S

almost exclusively to chain molecules, plasma polymers are more or less cross-linked materials. Therefore, films produced via plasma polymerization differ in their physical properties from conventional polymers, especially with respect to solubility, swelling, and hardness. Plasma polymerizations may also be carried out with molecules that cannot be used in conventional polymerization, if these undergo a transformation in the plasma. Pyrrol (Fig. 10) is, for example, isomerized in a plasma into an unsaturated nitrile which easily polymerizes. Similarly, methane fragmentates in plasmas to CH3, CH2, and CH, which are precursors for the formation of polymers. To understand plasma polymerization and other PECVD processes, it is useful to consider the mechanism briefly. The main step in these processes is the adsorption of the starting material on the surface of the solid. Under the influence of the bombardment of ions, electrons, and photons, the adsorbed starting material is converted into intermediates and eventually into the solid film and volatile fragments which are desorbed and pumped off. Reactions and nucleations in the gas phase are only of minor importance. Both the adsorption of the starting material and the desorption of volatile fragments strongly depend on the substrate temperature. If metallic films or films of metal oxides, nitrides, carbides, or silicides are to be prepared by PECVD processes, volatile metal compounds are required as starting materials. For a limited number of elements, volatile halides and hydrides are available which have been used in this sort of processes (see S. Vep~ek, p. 29S). The majority of elements do not form volatile halides or hydrides but many form organometallic compounds which often have a fairly high vapor pressure. For PECVD processes, the organometallic compound should not only be volatile but also thermally stable. This requirement eliminates most organometallic compounds. There are, however, certain metal carbonyls, metal alkyls, chelates, and ~rcomplexes which have a sufficient stability and vapor pressure and can be used in PECVD experiments. A few examples shall be presented. Thus the carbonyls of iron, cobalt, nickel, chromium, molybdenum, and tungsten are especially suited for PECVD because of their volatility and thermal stability. Some of them, e.g., Ni(CO)4, decompose thermally into CO and a very pure metal. However, CO is not an inert molecule in plasmas but decomposes into C02 and carbon which is partly incorporated into the growing films. For this reason, films made by PECVD from metal carbonyls often contain carbon and oxygen. The amount of oxygen depends on the oxygen affinity of the metal, but it can in most cases be drastically reduced when hydrogen is added to the carrier gas. Other candidates of compounds suitable for the deposition of metal films are the metal chelates. In these compounds the metal is bonded to

24S

Suhr

XsC-,~.C/CH ~'~C/"CX5 II 0 -0

I XsC

N

++

I O_ 0

Ii

,tC ~----------~H"/C--... CX5 X = H, F, CH5 Fig. 20. A metal chelate.

the ligand through the oxygen or nitrogen atoms of the ligand (Fig. 20). Depending on the oxygen affinity of the element involved, metallic films or oxide films are being formed. Thus the acetyl acetonate of aluminum gives only the oxide whereas the copper complex decomposes to form pure metal films. Other organometallic compounds which can be used if for PECVD processes are the metal alkyls, like tetramethyltin (Sn(CH3)4), or 7rcomplexes like ferrocene ((CsHs)2Fe) or allyl-cyclopentadienyl palladium (Fig. 21). Since it is also possible to combine different ligands in one molecule, as for example in dimethyl-gold-acetylacetonate (Fig. 22), a great variety of suitable compounds could be synthesized. Among these, only a small number have been tested so far. In all experiments, the rate of film formation and the composition are greatly influenced by the experimental parameters. The substrate temperature seems to be especially important due to its influence on adsorption and desorption processes. With increasing temperature of the substrate the rates of film formation decrease, but at the

Pd

H2C cH CH2 m

Fig. 21. Example of a 7r-complex.

Organic and OrganometallicCompounds

25S

c/'CH%c/CH3 CH3 II /

CH3

Fig. 22. Example of a complex with mixed ligands.

i \

CH3

same time the films become more metallic (Fig. 23). Above a certain temperature, which is characteristic for each set of starting material and experimental conditions, film formation can no longer be observed. In order to obtain pure metallic films, all the experimental parameters have to be optimized carefully. Each individual starting material has its specific optimal PECVD parameters. This makes it difficult to prepare alloys by PECVD techniques. Only recently it has been possible to prepare films of gold-platinum-palladium and of iron-cobalt alloys. Normally, the organometallic compounds are introduced into the reactor by means of an inert carrier gas, sometimes in combination with hydrogen. If mixtures of argon and oxygen, nitrogen or methane are used, it is possible to generate films of oxides, nitrides, or carbides. For example, allyl-cyclopentadienyl-palladium (Fig. 21) with argon as a carrier gas leads to metallic films, whereas with argon/oxygen mixtures, films of palladium

pg/cm2*min

4

Oo~

2

20

40

60

80 100 120 140 t60 Temperature eC

t80

200

Fig. 23. The amount of Pd and organic material in films prepared from allyl-cyclopentadienyI palladium.

26S

Suhr

oxide are formed. It is also possible to change the carrier gas during an experiment. Thus, hexamethyldisiloxane with argon as carrier gas forms transparent polymer films; with argon/oxygen mixtures, however, a film of SiO2 is obtained. To apply scratch-resistant coatings with a good adhesion to various substrates like lenses or car headlights, the carrier gas composition is changed from argon to argon/oxygen mixtures during the deposition process. As a result the lower layers are more polymer-like and adhere well to the substrate, while the final layers have the hardness of SIO2. For certain applications, metal-polymer combinations with various metal contents are required. This can be achieved by using lower substrate temperatures (see Fig. 23), by choosing starting materials with a higher carbon content, or combining organometallics with metal-free compounds. Tetramethylthin (Sn(CH3)4) is, for example, ideal for preparing films which consists of SnOx, Sn, or of composites with only 50-70% Sn. Starting, however, with tributyl-tinmethacrylate ((C4H9)3Sn--O--CO--C(CH3)=CH2) with a C/Sn ratio of 16, it is easy to prepare films with C/Sn ratios between 2-100. The combination of organometallic compounds and hydrocarbons (e.g., tetramethyltin/propene or of allyl-cyclopentadienyl-palladium/acrylonitrile) have been used to prepare films whose composition can be varied over a wide range. With changing compositions, film properties like the conductivity can be varied over several orders of magnitude. 6

ra

y = 6 , 1 5 2 2 - 0,0491x

R = 0,95

5

E

4

: I

20

40

!

I

60 80 atornio ~ Pd

I

100

Fig. 24. Resistivity of Pd-composites prepared by PECVD.

Organic and OrganometallicCompounds

27S

Metal-polymer composites have recently been prepared by the combination of sputtering or evaporation with plasma polymerization. These techniques lead to films which contain islands of the pure metal in a polymer matrix. (9) When the metal concentration reaches approximately 50 atomic percent, the conductivity increases by several orders of magnitude. By way of contrast, composite films prepared from organometallic compounds and hydrocarbons show an almost linear relationship between metal content and conductivity over the entire range of compositions (Fig. 24). (1~ The field of etching, surface modifications, and coatings is at present extensively studied and will become more and more important in the future. In some cases volatile halides and hydrides will be used as starting materials. Organometallics which have been investigated only recently, probably have a tremendous potential in the future. Many organometallics are known, and far more will be synthesized. The predicted increase in PECVD will not eliminate sputtering or vapor coating, but it will probably substitute them in certain fields, especially in cases where good adhesion to the substrate, edge-coverage, pin-hold-free coating of three-dimensional surfaces, or films with gradients are required.

6. FINAL REMARKS

In the beginning it was asked what chemistry would benefit from the use of plasmas. This question having been answered, we now should go on by asking what chemists can do for the progress in plasma technology. Their part will probably be in the elucidation of reaction mechanisms, in analytics, in the development of new syntheses and new etching and coating processes. Their most important contribution is likely to be in the synthesis and development of suitable starting materials for coating processes. It has been on purpose that the preceding discussion, the selection of examples, and the evaluation of future trends were made by a chemist's point of view. This does not imply that plasma technology will become a part of chemistry. Plasma chemistry is neither a domain of chemistry nor of physics, it is a truly interdisciplinary field. The chemical aspect has been stressed as a reminder of the fact that in difficult cases where the variation of physical parameters alone does not lead to a satisfying solution, the chemical part, mainly the choice of a suitable starting material, presents an additional parameter which might help to solve the problem.

REFERENCES

1. H. Suhr, Plasma Chem. Plasma Process. 3, 1 (1983). 2. R. J. Lagowand J. A. Morrison, Adv. Inorg. Chem. Radiochem. 23, 177 (1980).

28S

3. 4. 5. 6. 7. 8.

Suhr

K. Sugino, E. Inoue, T. Wakabayashi, and T. Matsuda, Denki Kagaku 25, 70 (1957). H. J. Wagenknecht, Ind. Eng. Chem. Proc. Res. Div. 10, 184 (1971). H. Suhr, H. Schmid, and H. Pfreundschuh, Plasma Chem. Plasma Process. 4, 285 (1984). H. Suhr and H. Pfreundschuh, Plasma Chem. Plasma Process. 8, 67 (1988). C. Haag and H. Suhr, Plasma Chem. Plasma Process. 6, 197 (1986). U. Niggebriigge, M. Klug, and G. Garus, Proceedings of the 12th International Symposium on GaAs, Conference Series No. 79, 367, 1986. 9. J. Perrin, B. Despax, V. Hanchett, and E. Kay, J. Vac. Sci. Techn. A 4, 46 (1986). 10. H. Suhr, A. Etspiiler, E. Feurer, and C. Oehr, Plasma Chem. Plasma Process. 8, 9 (1988).