Pure & App!. Chem., Vol. 48, pp. 163—178. Pergamon Press, 1976. Printed in Great Britain.

HETEROGENEOUS REACTIONS INNON-ISOTHERMAL LOW PRESSURE PLASMAS:PREPARATIVE ASPECTS AND APPLICATIONS

S. VEPfEKInstitute of Inorganic Chemistry, University of ZUrich, Switzerland

Abstract—Non-equilibrium properties of low pressure plasmas open new ways to preparative solid state chemistrysince they enable strongly endothermic reactions, as well as reactions requiring high activation energies, to take placeat low temperatures. For example, the plasma techniques are now replacing many conventional methods insemiconductor technology such as thin film deposition, removal of photoresists and etching, and significantprogress is expected in the application of low pressure plasmas to chemical vapour deposition (CVD). Chemicaltransport appears to be a simple method for investigating the behaviour of heterogeneous systems, solid/gas, underplasma conditions, as well as a promising technique for crystal growth. The unique features of plasma chemistry aredemonstrated by two examples along with a description of adequate diagnostic investigations. Several significantaspects of plasma production and characterization are discussed from a plasma chemist's point of view. A number ofheterogeneous systems are described to illustrate the variety of plasma syntheses and of the interactions between lowpressure plasmas and solid surfaces.

INTRODUCTION

Ten years ago, heterogeneous reactions in low pressureplasmas were an underdeveloped field of plasma chemis-try.'2 On the other hand, the major part of the recent bookby Hollahan and Bell is devoted to the applications of theplasma to heterogeneous systems—solid/gas.3 There areseveral reasons for this fast development: (a) The highinternal energy of the plasma as well as a high activationenergy, which is available atlow temperatures, offer uniquepossibilities forchemical synthesis. (b)Alargevarietyof thedifferent types of discharges allows one to choose themostsuitable experimentalarrangementforaparticularproblem,etc. (c) The most important reason is, however, the highreaction yields obtained by this technique.

The low yields arising from the low pressure usedmakes the application of these plasmas to homogeneoussystems in industrial processes unfavorable when aconventional method is available. However, this consider-ation does not apply to heterogeneous processes such ascrystal growth and deposition of thin films. In suchprocesses the growth rate is limited by the rate of build-upof crystal lattice. The rate of lattice growth is a function oftemperature and the processes which control the nuclea-tion. In many systems the activation energy of nucleationcan be supplied by the plasma. Moreover, linear growthrates of the order of lO_3_l0_2 cm hr', which are usuallyobtained by the conventional technique below 1000°C,4can also be reached in the plasma.

Let us consider a simple example: a depositiontemperature of 100—1000°C, a concentration of theparticular monomer '3M 5 X 1014_1016 cm3 (i.e. partialpressure PM < 1 torr) and a sticking probability a 10-'—iO. The Hertz—Knudsen equation4'5 leads to depositionrates between 6 x iO' and 2.5 x i0' species cm2 sec';i.e. a linear deposition rate of the order of l0—l cm hr'.

Table 1 shows the growth rates of several solidmaterials experimentally obtained in low pressure plas-mas. It can be seen that the theoretically expected ratesare reached in the experiments. In some cases, the use ofplasma allows a faster rate of solid growth than obtainedby the conventional technique. This is especially valid forsystems such as the transport of red phosphorus (seeSection 5) and the deposition of cadmium sulfide, which

Table 1. Linear deposition rates of solid materials obtained in lowpressure plasmas

SolidTemperature

(°C)Linear deposition

rate (cm hr') Ref.

A1N 5x103crystals 1000

TiN 2x1(r2 6CdS 100-300 3 x 10 7P(red) 150 5 x 10 8P3N5 265 lO—10 9SiO2Al203 100—500 10—10 10, 11

Si3N4Si02/Ge02 1000 0.6 12, 13

involve some kinetically hindered steps in either theevaporation or in the deposition. If, for example,cadmium sulfide were deposited at the rate of 3 xl0- cm hr' by the thermal evaporation, the layers wouldshow very low resistivity due to the sulfur deficiency.7

The electrical and optical properties of conventionallydeposited cadmium sulfide ifims are usually improved bylong term annealing. Kassing and Deppe have shown thatannealing in a H25-plasma for about 2 mm leads to thesame effect.'4 Further examples of enhanced depositionrate by discharge applications can be found elsewhere.'5"6

The transport of phosphorus demonstrates a purifica-tion effect. If one can reach the desirable high degree ofpurity, an apparatus with a 2 kW high frequencygenerator could supply several hundred kilograms of purephosphorus per year. Today's annual world production isa few tons.

The deposition and properties of various oxides andnitrides will be described in the next section together withsome further examples of plasma applications. Thepresent paper is intended as a short introduction into thisfield rather than a detailed review of the previous work(see e.g. Ref s 1, 3, 10, 17—20). However, we shall try toshow, with use of several selected examples, "what canbe done" and "how to do it". The choice of the particularsystems considered reflects the philosophy of the author

163

164 S. VEPEK

rather than an attempt to evaluate their importance in thegeneral concepts.

The topics of the title include three types of reactions:(a) chemical evaporation and deposition of solids inwhich only gaseous reactants appear on the right handside of the eqn (1):

A(s) + B(g)==±C(g) + D(g) +

(b) modification of the solids and/or their surfaces such asoxidation and nitridation (2a), changes in the properties ofthe solid and/or surface (2b) and reduction of oxides,halides, etc. (2c):

A(s) + B(g)—>C(s)A(s) + B(g)—A'(s) + B(g)A(s) + B(g)—C(s) + D(g)

(c) reactions in which the solid is involved as a "thirdbody" e.g. surface recombination of atoms and ions,de-excitation of various excited states, and surfacecatalysis.

The main part of this paper will be devoted to type (a)reactions. Although some reactions of type (b) and (c) willalso be mentioned, the reader should consult the abovementioned general literature as well as some specializedarticles for more details; e.g. plasma anodization21 and ionplating,22 morphology changes,23 plasma treatment ofpolymers,24 surface recombination,25 thermal accom-modation and molecular beam scattering from solidsurfaces.26

In fact, the plasma can influence a heterogeneoussystem described by reaction (1) in two different ways,which will be called "the kinetic"—and "the ther-modynamic effects" (Ref s. 6, 27, 28). In the first case, thenecessary activation energy for a thermodynamicallypossible reaction is supplied by the plasma, without whichthe reaction would not occur due to a kinetic barrier. Aweak discharge, which is used to catalyze such a reaction,changes the energy content of the system only slightly.2'27

On the other hand, the high energy of an intense lowpressure discharge (discharge current 10' to severalamperes, pressure 1O_1 to several torr) changesdrastically the chemistry of such systems as comparedwith the conventional chemistry at the same pressure andtemperature.

Usually, both of these effects take place more or lesssimultaneously. We shall give, later in this paper, twoexamples in which these effects can be quite clearlydistinguished (transport of carbon according to

Boudouard's reaction and with hydrogen, Section 3).The last part of this paper deals with some preparative

applications of the plasma. The chemical transport ofsolids in low pressure plasma which has been used inthese studies appears to be a promising preparativemethod as well as a simple tool for the investigation of thebehaviour of heterogeneous systems under plasma

(1) conditions. We shall try to illustrate this point with severalexamples.

Any theoretical treatment of heterogeneous systemsunder plasma conditions is very difficult because of thelarge number of elementary processes involved and thelack of data on the reaction rates. A simple,phenomenological theoretical approach was developed

(2a) previously by the author in order to obtain some(2b) fundamental ideas on the behaviour of such systems.2(2c) Any substantial progress of the theory requires an

accumulation of more experimental and preparativeexperience, a better and more detailed diagnostic approachand, last but not least, more data on the elementaryprocesses, both in the gas phase as well as on the surface ofthe solid.

1. APPLICATIONS

Considering the high deposition rate indicated in Table1 one may ask about the quality of the deposited films.Table 2 shows the dielectric properties of variousmaterials obtained by the plasma technique. For compari-son, the deposition temperatures necessary for conven-tional and plasma techniques are also given. Of principalimportance for semiconductor technology is the lowtemperature which can be used in plasma as comparedwith the high temperature necessary for the conventionaltechnique. Such high temperatures cause technologicaldifficulties due to an enhanced diffusion of doping materialsin the semiconductors. The dielectric properties of theplasma deposited ifims—which are given on the right handside of the table—are excellent.

A typical method for the preparation of inorganicoxides is the low temperature pyrolysis of organometalliccompounds. Several methods of nitride deposition will bedescribed later.

Thin films of organic polymers are obtained bypolymerization of a suitable monomer gas in electricdischarge at low pressure.333 The possible fields ofapplication of the polymer films are in semiconductortechnology,"32 integrated optics,34 semipermeable mem-branes35 and protective coatings.

Removal of photoresists and etching are among theseveral important steps in the fabrication of semiconduc-

Table 2. Dielectric properties of plasma deposited thin films

Deposition tem(°C)

peratureProperties of plasma deposited films

'

Dielectricconstant tan 6

Dielectricstrength(V cm1)

Surface chargedensity

(cm2 eV) Ref.Compound Conventional Plasma

Si3N45i02A1203

700—900

900—1200

700—1000

250—500

200-300100—500

6—10

4.5—5.5

7.5—8.5

10s10"10_2

0.5—1 100.5—1 . 1O0.7 l0

101010_loll>2 1010

10, 11

10, 11, 1410, 11, 29, 30

Organicpolymers

P3N5100

200—300

2.5—6.2

4.410_2s10

0.1—1 100.1—0.9 l0

1012

?10, 31, 32, 33

9

Non-isothermal low pressure plasmas 165

tor devices. The photoresists are photosensitive organicpolymers which are used by the photolithographictechnique for maskingthe device duringfabrication. At theend of the particular process the photoresist has to beremoved. The conventional technique involves several wetstepsandrequirestheuseofhighestgradechemicals. Ontheother hand, the photoresist can easily be removed byoxidation in a low pressure discharge."36

In a similar way silicon, silicon oxide and nitride can beetched using CF4-plasma37 which is fast, simpler, cleanerand cheaper than the conventional technique. The etchingrate can be significantly enhanced if a combinedr.f. sputtering/etching technique is used.38

A comparison of the two techniques for the manufac-ture of semiconductor devices has recently been given byKirk."

The success of plasma technology in this field is wellknown. A more recent example will now be described, i.e.the application of the plasma for fabrication of low lossoptical fibres.

Optical fibres will be an important part of futurecommunication systems. The fibre consists of a core witha high refractive index and a sleeving with a low one,providing total reflexion of light at the boundary. Adefinite radial index profile and low attenuation—i.e.extremely high purity of the glass—is desirable for theapplication of fibres to communication purposes. Al-though the main problems in the preparation of low lossoptical fibres have already been solved some problems,such as the achieving the definite radial index profileremain to be overcome. One way to achieve the definiteprofile is to approximate the desired profile by repeatedlayer deposition followed by diffusion equalization. Thismeans that layers of thickness less than 1 m are required.Such layers are deposited in a microwave discharge asshown in Fig. 1 (for more details see Ref s. 12, 13).

Gaseous starting substances, e.g. SiCl4/GeCl4/02 areintroduced into a quartz tube where they pass a microwavecavity. In the discharge region, the transformation ofchlorides to oxides takes place. Oxide layers of a uniformthickness are obtained by moving the microwave cavityalong the enclosing tube. The typical experimentalconditions are as follows: total pressure 10—30 torr,frequency 2.450 GHz, inner diameter of the surroundingtube 6mm; a deposition rate of about 0.3—0.6cm hf' isobtained at a temperature of about 1000°C.

Optical fibres of an outer diameter of 100 jm are drawnfrom such preforms. Optical losses below 10 dB km' inthe spectral region 740—920 nm with a minimum of3.6 dB km' at about 1040 nm have been obtained so far.These losses are quite comparable with those of fibresproduced by the conventional technique, but the use of aplasma allows much better control of the radial refractiveindex profile. Geittner et al. expect, that after elimination

of some impurities (e.g. Fe2 from stainless steelcomponents inthe experimentalequipment)the totallosseswill become even less.'2

The last example we shall mention in this introductionis the continuous wave HCN-laser."°'4' Due to thelongwavelength, 337 j.mand311 j.m,thislaseris suitableforavarietyof applications including plasma diagnostics, direct mixingof the far i.r. light with electromagnetic radiation producedby klystrons, study of nonlinear phenomena, etc.

The strongest radiation at 337 m corresponds to thevibrational-rotational transition (1 1'O) J = 10—* (04°0) J =9of the HCN molecule. This transition is allowed becauseof a strong Coriolis coupling between these levels and thepopulation inversion is obtained due to the high de-excitationprobabiityofthelowerlasinglevel(04°0). Fortheoperation, HCN molecules must be formed and/or excitedinto the (11'O) state.

The mechanism of the formation of vibrationally excitedHCN molecules remained unresolved for a long time.Recently, Schötzau and the author have shown that theexcited HCN-molecules are formed by both volume andwall processes, the latter contributing apparently more tothe total power emitted by the laser.42

Shortly after first switching on the discharge in anoptimal gas mixture, the inner wall of the discharge tubebecomes covered with a polymer corresponding toapproximately stoichiometric paracyanogen (CN). If apure, stoichiometric paracyanogen is first deposited bydischarge activated polymerization of cyanogen (CN)2,and, afterwards, the discharge is maintained in hydrogenonly, the highest lasing power is obtained from theparticular laser.43

A high degree of dissociation is found in the dischargeeven near the tube wall.42 Hydrogen atoms impinging onthe surface of the polymer are adsorbed there and HCNmolecules are formed:

(CN) +2H—+(CN)-, (HCN)2

(CN)_, (HCN)2 -(CN)_, +2HCN(g) (3)

The desorption can occur either directly, or there may besome intermediate steps involved. In any case, theHCN molecules are most probably vibrationally excitedwhen leaving the surface. For the particular laser we havestudied, the total number of HCN molecules formed persecond on the wall is almost two orders of magnitudelarger than the number of the stimulated transitions whichgive the measured laser gain. The experimentally ob-served gain of the HCN laser, which is constant over thecross section of the resonator tube, is then explained interms of the combined effects of volume processes andreactions on the wall of the discharge tube (for furtherdetails see Ref. 42). Quite recently, Kunstreich andLesieur have been able to distinguish experimentallybetween these two processes by measuring the relaxationtime for the onset of the stimulated emission.'45

It is surely very encouraging to see the application oflow pressure plasmas to a variety of practical problems.However, we must realize, that we still know only verylittle about the basic processes taking place in the plasma.We shall therefore discuss these processes as far as ispossible in the light of our present understanding of theproblem.

Experience has taught us that it is often very difficult oralmost impossible to compare results of various authors

MICROWAVECAVITY

GeC/SiC[4/O2_—

7JNACE

—- PLASMA

\QUARTZ TUBE

Fig. 1. Experimental set up for deposition of optical fibres.'2"3

166 S. VEPkEK

due to insufficient data on the experimental arrangementand, especially, on the discharge conditions. For thisreason, we shall pay attention to these problems in thefollowing section.

2. APPARATUS AND CHARACTERIZATION OFDISCHARGE PARAMETERS

A variety of apparatus for chemical syntheses in lowpressure plasmas has been described in the literature (e.g.Refs. 1, 3, 17). A typical apparatus consists of a vacuumpump, gas sampler, a properly designed discharge tube(reaction vessel), a generator of electric power formaintaining the discharge and necessary instruments formonitoring discharge conditions. Since most of theseitems are standard laboratory equipment we shall selectonly a few problems concerning the production andcharacterization of the plasma which are important forchoosing particular instruments. Let us recall that we areconsidering low pressure plasmas (pressure l0_2 toseveral torr, with an electrical power dissipated in thedischarge 100_104 W).

2.1. Generation of the plasmaWith a few exceptions such as deposition of solid on

thin wires in a corona discharge, plasmas which are moreor less homogeneous over a region of at least severalcentimetres are desirable for application to preparativesolid state chemistry. It turns out, that a radio frequency,elect rodeless discharge is most suitable for these purposes(e.g. Ref. 3, chap. 1 and 10). The discharge is maintained ina quartz or pyrex tube between two external electrodes("capacitive coupling"). At a power of more than severalhundred watts, these electrodes must be water cooled inorder to avoid damage to the tube. It should be pointedout that if the so called "inductive coupling" (see e.g.Ref. 1, p. 39) is used to produce plasma at lower powerand a frequency greater than 10 MHz, the discharge ismaintained due to the r.f. electric field between both endsof the coil, and the coupling is in fact a capacitive one.

In principle, any frequency from 50 Hz up to GHz canbe used to excite gas at low pressure, but for ourpurposes, a frequency of 20—100 MHz is most convenient.

At lower frequencies the voltage, which has to beapplied to the external electrodes to maintain thedischarge at a given value of the discharge current,increases significantly. Figure 2 shows schematically thedischarge tube and the equivalent high frequency circuit.At a discharge current i, the applied voltage U is given by

U=i (ië+2z+R). (2.1)

Here, w = 2irf, f is the frequency, C denotes the capacityof the condenser consisting of the external electrode,discharge tube wall and conducting layer of the plasmanear the inner wall, Z is the impedance of the electroderegions and R is the ohmic resistance of' the plasmacolumn. The imaginary part of the plasma conductivity iszero in the frequency region considered here.46

The impedance Z increases with decreasing frequencydue to increasing losses of charged particles during oneperiod of the h.f. field. This means, that with decreasingfrequency f, the term (2/ jwC + 2Z) increases. In order tokeep the power dissipated in the plasma column (which isbeing used for the plasma chemistry work) constant, this

U

H c.

\\ 3

b

Fig. 2. (a) Discharge tube; (1) capacitance of the glass wall; (2)electrode regions; (3) plasma column. (b) Equivalent high

frequency circuit.

increase must be compensated by an adequate increase inthe voltage U.

This is illustrated in the following example:47 An h.f.discharge in nitrogen was maintained in a tube of diameter10 cm; the separation of the water cooled electrodes wasabout 25 cm (arrangement as in Fig. 2a) and the pressurewas 1 torr. At a discharge current of between 4—7 A apowerof about 1—2 kW is dissipated in the discharge. The valueof the h.f. voltage applied to the electrodes is between 1and 1.5 kV at 27MHz, but is as much as 9—14 kV at1 MHz. It is evident that the impedance matching is muchmore complicated at 1 MHz than at 27 MHz. Moreover, apower of up to 3 kW can be dissipated in a pyrex tube at27 MHz and the discharge can be maintained for severalthousand hours without damaging the glass. On the otherhand, the same pyrex or quartz tubes only survived a fewhours when the discharge was run at 1 MHz.

Nevertheless, the frequency of 1 MHz can be used ifinternal, water cooled electrodes are applied. Severalarrangements have been developed in our laboratory byE. Wirz.48'49 Figure 3 shows as an example a quartzdischarge tube and the impedance matching for thesynthesis of nitrides. For the preparation of a particularnitride an insert made from the same metal is used toprotect the inner wall of the electrodes and to avoid acontamination of the plasma. Although the 1 MHzgenerator was successfully used for preparative plasmachemistry,48'49 the experimental design, as well as thehandling of the apparatus itself, appears to be much morecomplicated than in our earlier work at a frequency of80MHz.6

The upper frequency limit is determined mostly by thedecreasing efficiency of electron tubes above 100 MHz. Inaddition, sophisticated experimental arrangements mustbe used to produce a homogeneous plasma in largevolumes atvery highfrequencies,forthenthe wavelengthiscomparable to, or shorter than the dimensions of thedischarge tube.

A d.c. discharge is to be preferred when measurementson plasmas have to be done using sensitive electronicinstruments (e.g. mass spectrometry, optical and i.r.spectrometry, etc.). The equivalence between plasma ofthe positive column of a d.c. discharge and that of a h.f.

12 3 2' 1'

Non-isothermal low pressure plasmas 167

HF —generator

1 MHz

4 kW

Fig. . Discharge tube and impedance matching for synthesis of nitrides at a frequency of 1 MHz.

one allows one to correlate the results of measurements ina d.c. discharge with those of capacitive coupleddischarges at frequencies 10—100 MHz.'50'51

2.2. Characterization of the plasmaA comparison of experimental findings of different

authors who have worked with the same systems isfrequently very difficult or impossible due to insufficientdata on plasma parameters. As a minimum the followingmacroscopic parameters of the discharge are needed:diameter of the discharge tube, distance between theelectrodes, frequency, discharge current, gas pressure andflow rate, and, of course, all data on chemical com-pounds used. In addition, the neutral gas temperature andthe axial field strength should be measured. A knowledgeof these parameters allows one to calculate or to estimatefrom literature data the important microscopic parameterssuch as electron concentration and temperature, degree ofdissociation, etc.

Many authors used to give as a discharge parameter thepower dissipated in the whole discharge and/or the h.f.voltage applied to the electrodes. It is evident, thatdue tothefrequency dependence of the usuallyunknownvalues ofthe first two terms in eqn (2.1), (i.e. 1/ftoC and Z) theseparameters do not characterize the plasma of the dischargecolumn.

We shall pay attention to the measurement of theelectric parameters at the frequencies considered, sincemost chemists are not quite familiar with them.

A modern oscilloscope with high voltage probes for thefrequency range of up to 100 or 200 MHz (at least up to thethird harmonic of the generator frequency) and with acorresponding current and range is the most suitable andversatile instrument for such measurements. The currentprobe has to be placed next to the earthed electrode and thecapacitive current is subtracted from the measured value.The capacitive current is measured either with a dischargetube evacuated below iO torr or filled with an inert gas athigh pressure (no discharge).

The axial electric field strength E can be found bymeasuring the voltage U applied to the discharge tube atconstant current and varying electrode distance (see Fig. 2and Ref. 52, p. 238). The power dissipated in the plasmacolumn of length L is then equal to ELef (ief denotes theeffective value of the h.f. current). One notices again, thatthe plasma conductance of the column (see (3) in Fig. 2)contains only the real term at the frequencies considered.

The relatively high costs of the oscilloscope includingthe probes may prevent many plasma chemists frombuying them. Use of a high frequency voltmeter instead ofthe oscilloscope can significantly reduce the total cost ofthe facilities, but information on the actual time depen-dence of the current and voltage during one period of theh.f. field will be missing. The latter is very important,especially if the discharge tube is placed in a screeningbox ("Faraday cage") where intense standing waves athigh harmonics are induced. These oscillations mayinterfere with the instruments and cause significant errorsin the measurements.

For the "every day" characterization of a discharge inplasma chemistry work, simple, self-made instrumentswhich can be set up in anylaboratory can be used. It may behelpful to describe them here.

The simplest h.f. ammeter consists of two thin, metallicfoils and of a thermometer filled with a non-conductiveliquid (e.g. petroleum, up to 270°C). The chosenthickness of the foil must be much less than the electricskin depth at the given frequency. For example,zirconium foil of thickness 0.02mm or platinum foil ofthickness 0.01 mm can be used up to 80 MHz.53 Themechanical construction is shown schematically in Fig. 4.The thermometer and foils must be rigidly secured (usingteflon), and the whole ammeter has to be thermostatted (e.g.by surroundingitwithacoil of plastic pipe containingwateror oil at constant temperature). Since the resistivity of thefoil does not depend on frequency up to several hundredMHz, the thermometer canbe calibrated using either d.c. ora.c. (50 Hz) current. The current range of the ammeterdepends onthe widthof thefoil. Forexample,anh.f. current

"windings : 15 cm

C: 800 pF

168 S. VEPIEK

of up to 10 A canbe measured when afoil width of 10mm isused. This ammeter can also be used for measurement by a"symmetrical arrangement", i.e. with no electrode earthed.

The h.f. voltmeter consists of a voltage divider (e.g. twocondensers), arectifierand d.c. voltmeter. The arrangementcan be found in any standard textbook, and the instrumentsnecessary for the calibration are available in most physicsdepartments.

A small gas thermometer made of quartz appears to bethe most suitable instrument for measuring the neutral gastemperature.53 It is filled with nitrogen at —250 torr(hydrogen diffuses through quartz at high temperatures)and the pressure differences which are proportional to thetemperature are measured by a "U"-manometer using agallium/indium alloy (liquid at room temperature, negligi-ble pressure up to 1000°C).

The actual temperature measured by the thermometerin a particular discharge depends on the catalyticefficiency and on the emissivity of the surface of thethermometer top. Figure 5 shows the temperaturemeasured by the thermometer when the top was coveredwith various solid materials. The reproducibility of thesemeasurements is illustrated by the curve for platinum.The individual points were obtained with three differentgas thermometers and two different ammeters in twodischarge tubes over a period of six months.

The temperature of the particular solids measured atthe same value of h.f. current depends predominantly onthe catalytic efficiency of these materials for the surfacerecombination of atoms (Ref s. 53,54, pp. 170.-200, 25, 25a).The temperature of quartz poisoned with HPO3 is roughlyequal to the neutral gas temperature. An exact determina-tion of the latter value would include a detailedconsideration of the radial gradient of temperature,emissivity of the surface and plasma, concentration ofatoms, mestastables and ions. However, neglecting theseeffects is quite tolerable within the limits of accuracy ofusual plasma chemical work. It is also evident from Fig. 5,that the temperature measured by a simple thermocouplecan differ significantly from the neutral gas temperature.

The temperature difference found for platinum andquartz/HPO3 surfaces can be employed for an approx-imate determination of the degree of dissociation.55 TheWrede—Harteck gauge is suitable for more exactmeasurements (e.g. Ref. 1). A particular construction of aW-H. gauge for measurements in intense moleculardischarges at higher temperatures can be found in our

2 4 6 8 10

Ihf(Fig. 5. Dependence of the temperature of various solids innitrogen discharge on hf. current. Pressure 1.35 torr, frequency

80 MHz, discharge tube diameter 10cm.

previous paper.42 The gradient of the degree of dissocia-tion in the vicinity of the quartz gauge is negligible andproduces no significant errors (for the calculations seeRef. 56).

3. CHEMICAL TRANSPORT OF SOLIDS IN

PLASMA (GENERAL ASPECTS)

Theoretical as well as preparative aspects of chemicaltransport of solids in low pressure plasmas have beendiscussed in our previous papers.2'6'27'28'57° In this sectionwe shall only consider transport of carbon byBoudouard's reaction in a weak CO/C02-discharge and inan intense hydrogen discharge. As pointed out in theintroduction, the first case displays a kinetic effect andthe second one illustrates a thermodynamic effect ofthe plasma. In order to compare conventional and plasmachemistry we shall pay attention to the mechanism oftransport in the light of some recent diagnostic investiga-tions into these systems. It will be seen, that reactions ofneutral atoms and radicals predominate in intensedischarges, whereas ion-molecule reactions are moredominant in weak plasmas at low temperatures.

3.1. The Boudouard's reaction (3.1)This was presented by Schafer61 as an example of a

system where transport (T2 —T1), which should bepossible thermodynamically, does not take place due tothe high activation energy of the reverse reaction.

C(s) + C02(g)==±2C0(g). (3.1)

We have shown earlier, that this activation energy can besupplied to the system by a weak discharge (for theexperiments see Ref. 27). The mechanism of the reverse

THERMOMETERI (°K)

1200

1000

800

600

Fig. 4. Schematic design of hf. ammeter.

Non-isothermal low pressure plasmas 169

reaction 2CO(g)- C02(g) + C(s) remained unresolved atthat time.

This problem was recently clarified in a mass spec-trometric study of the system62'63 using a double focussinginstrument (CEC 21-110) which was modified for acollision-less extraction of a molecular beam of neutralparticles from the discharge tube. The experimentalarrangement is shown in Fig. 6. More details on theapparatus and on the construction of the orifices into thewall of the discharge tube will be published elsewhere.63Three different discharge tubes with conical orifices ofminimal diameter 60,100 and 200 m, and of wall thicknessless than 100 and 200 pm respectively, were used in thisstudy. Carbon monoxide of greater than 99.995% puritywas introduced into the discharge tube. At a total pressurechosen between 0.5 and 6 ton, the steady stateconcentration of CO and CO2 in the discharge, under gasflow conditions, was monitored by the mass spectrometer.The mean residence time Tres of the gaseous species in thedischarge plasma between 0.1 and 200 sec could beestablished by varying the flow rate. Tres is equal to thetime perio. which is available to the system to reach theparticular steady state concentration ratio of C02/CO atthe position of the orifice.

The reverse, plasma catalyzed reaction (3.1) involvesseveral elementary processes. The overall rate of thisreaction, which is predominantly affected by the slowestelementary process, can be expressed in terms of anoverall relaxation time, Trel. At a given pressure anddischarge current, the ratio C02/CO does not depend onTres (i.e. on the flow rate) as long as Tres $ Trel, but it should

MOLECULARBEAM

decrease for Tres Trel. Thus, the relaxation time Trel can bedetermined experimentally by measuring the dependenceof the CO2/CO-ratio on Tres, provided, of course, that therelaxation time of electrons Tel Tres.

The reverse reaction (3.1) can take place via threedifferent mechanisms (see Ref s. 27, 62, 63):

•O+C—÷O+CO--—*CO2+e (3.2a)

CO+e—* CO*+e_*CO* +CO-+ CO2 (3.2b)

C+O+e—bO+CO-—--b"--—CO2 (3.2c)

Detailed calculations show62'63 that the three possiblereaction mechanisms are characterized by differentoverall relaxation times:

Ta 0.4 sec, Tb 500 sec, 'r 20 sec. (3.3)

Consequently, the reaction path (3.2a) involving theion-molecule process should predominate and the ratioC02/CO should decrease for Tres 1 sec. Figure 7 showsthe measured dependence of the C02/CO ratio on Tres atconstant pressure and discharge current. The agreementof the measured value Trel 0.36 sec with the calculatedvalue Ta 0.4 sec is excellent and we can conclude thatthe ion—molecule reaction mechanism (3.2a) is mostprobably predominant in the weak discharge.

To be sure that the change of the C02/CO ratio was notdue to any effects other than those discussed above, themeasurements were also checked by switching the

ELECTRKSECTOR

Fig. 7. Dependence of the C02/CO-ratio on the mean residence time. Total pressure 3 ton, discharge current 15 mA,inner diameter of the discharge tube 1.2 cm.

ION MAGNETICTO PUMP SECTOR40 I/sec

DETECTOR

Fig.6. Experimental arrangement for mass spectrometric study of the low pressure plasmas.

C02/CO

401

I I

1 2 60

Z'es(sec)

170 S. VEPEK

discharge onandoff andfollowingthe approachof the ratioto the steady state value. However, this measurement waslimited due totheresponsetime of the recording equipmentbeing about 0.8 sec. Nevertheless, it has been proved thatthe relaxation time of the C02/CO ratio is less than thisvalue.

Brown and Bell studied the oxidation of CO and thedissociation of CO2 in an intense h.f. discharge.TM Theyshowed that the contribution of the ion—molecule reactionto the overall reaction rate was small and that thereactions of electrically neutral radicals and atomspredominate in such plasmas. We have found similarresults in previous work on chemical transport.27'28'57'°'65

3.2. The transport of carbon with hydrogenIn an intense low pressure discharge the transport of

carbon with hydrogen displays the thermodynamic effectof the plasma. The experiments and theoretical considera-tions of this system can be found elsewhere.2'27'65 In thepresent paper we shall report some recent work on adiagnostic investigation which has allowed an explana-tion of the transport mechanism. Matrix isolationspectroscopy, used in this work, appears to be a suitabletool for the diagnostics of low pressure plasmas inchemically reacting systems. In addition, some newaspects of the carbon/hydrogen system will be discussed inview of the possible use of carbon in the next generation ofTokamak devices for controlled thermonuclear fusion.56

The transport of carbon in hydrogen plasma has beenattributed to the formation and subsequent decomposi-tion of simple hydrocarbon radicals. A discussion of theelementary processes occurring in the plasma has broughtforth many arguments in support of this mechanism.27'65

However, there has been some doubt regarding thevalidity of this explanation because another equallyplausible argument could be suggested: in the ther-modynamic equilibrium, the gas phase of this systemcontains a number of species, i.e. CH4, H2, H, C2H2, C2H,CH3, CH2, CH, C3, C2, C1.67' At a given pressure, theconcentration of each particular species strongly dependson temperature (see Fig. 13 in Ref. 68). The total amountof carbon in the gas phase, expressed by the ratio C : Hreaches a minimum around 1300°K at a pressure of 1 torr(see Fig. 14 in Ref. 68). Therefore, the chemical transportof carbon in the direction of increasing temperature (e.g.900—p 1300°K27) could also be attributed to the decreasing"solubility" of carbon in the gas phase with increasingtemperature. In such a case, the plasma would merely becatalysing this "thermodynamically possible" reaction.However, the solubility of carbon decreases within thistemperature range due to decreasing concentration ofmethane, i.e. a significant concentration of methane shouldbe found in the plasma.

To determine which of these two possible transportmechanisms is correct, one has to devise a properdiagnostic method for investigation of the gas phasecomposition under the given experimental conditions.Mass and optical spectroscopy are the most frequentlyused tools for such investigations. However, both thesetechniques require the species under consideration to be"activated" by electron impact, which can, however, alsolead to their dissociation. Consequently, determination ofthe true chemical composition of the active dischargeplasma by these methods is difficult. In addition, use ofabsorption spectroscopy (in visible and i.r. region) isfrequently hindered by the plasma's own radiation.

In view of these difficulties, matrix isolation spectros-

copy appeared to be an attractive tool for the plasmadiagnostics in such systems. By this method, specieseffusing from the plasma, through a small orifice, aretrapped and isolated in a solid matrix of an inert gas atcryogenic temperatures (4—15°K) and are subsequentlyidentified by absorption spectroscopy. The problem ofplasma radiation is avoided and the whole spectral regionfrom u.v. to i.r. can be utilized for spectoscopicmeasurements. A large number of papers demonstrate thatfree radicals, atoms, as well as ions can be isolated andstored in the matrices.69'70 The problem of fragmentation byelectronimpactis avoidedbythe applicationof this method.

The matrix isolation apparatus used for the study of theC/H2-system is schematically shown in Fig. 8 (for moredetails see66). The high frequency discharge took place ina quartz tube (1) which is shown in cross section. Theconstruction of the metal "double" Dewar is similar tothat described elsewhere.7072 The molecular beam effusesfrom a part of the discharge tube which contains graphiteinsert (6) through a small orifice (2) (4 150—200 t). It istrapped on a liquid helium cooled, optically polishedcopper block (3) simultaneously with a suitably chosenmatrix gas M (e.g. Ar, Kr, Xe, N2). After the requiredmatrix has been deposited, the copper block, togetherwith the reservoirs for the liquefied gases, is rotatedthrough 180° and the absorption spectrum of the matrix isscanned through the KBr window (5).

A series of experiments has shown that there is nosignificant concentration of methane in the plasma underthe conditions of carbon transport. Since no fragmenta-tion could take place by this method, the second of theabove mentioned mechanisms was ruled out and themechanism originally suggested was supported.

Hence, the transport of carbon in hydrogen dischargeinvolves the following steps. In the charge region,H-atoms react rapidly with carbon and simple hydrocar-bon radicals are formed.

Fig. 8. Dewar assembly for diagnostic investigation of the lowpressure plasmas by means of the matrix isolation spectroscopy.(1) discharge tube; (2) orifice; (3) copper block at °"lO°K; (4) heat

shield at '°77°K; (5) KBr window; (6) graphite insert.

Non-isothermal low pressure plasmas 171

Due to the gas flow, the hydrocarbons reach the zone ofhigh plasma energy where they are decomposed by thedissociative de-excitation of the hydrogen metastables.65Solid carbon is deposited since its equilibrium pressure istoo low at the neutral gas temperature of 1300°K which isfound in the deposition zone.

Coulon and Bonnetain have recently shown, that thereactivity of H-atoms towards solid carbon has a localmaximum around 900°K (charge zone) and a minimumaround 1300°K (deposition zone) (Ref. 73, for furtherRefs. see also 56). Therefore, the higher temperature inthe deposition zone is favourable for the transport to takeplace and the necessary concentration of hydrogenmetastables can be lower than the upper limit estimatedearlier.65 The very interesting questions about the natureof species leaving the carbon surface as well as about theradial distribution of the hydrocarbons in the dischargeremain, however, still unresolved.t

The interaction of solid carbon with hydrogen,deuterium, tritium and other particles within a wideenergy range (lO_2_105 eV) is now being studied in severallaboratories because of its importance in further progresstowards controlled fusion (e.g. Ref. 74). It has beenrecently reported that the presence of relatively smallamounts of impurity ions in a hot plasma (energy about10 keV) constitutes one of the most severe impedimentsto achieving ignition conditions.75-79 These impurities arereleased from the first wall of the reactor and from thelimiter by plasma-wall interactions such as sputtering,blistering, thermal shocks and chemical reactions. Sincelow z impurities are more desirable than high z ones,various carbon materials, carbides and nitrides of lightelements are proposed as construction materials for thefirst wall and limiter, in the large Tokamak machineswhich are now being planned.

Some recent, preliminary investigations56 have shown,that the chemical erosion of silicon carbide and of aproperly chosen graphite (e.g. pyrolytical graphite attackalong the c-axis) should not inhibit ignition, provided thatthe reaction probability of hydrogen atoms with theparticular material is not increased as a result of radiationdamage to the surface. It is possible, however, thatradiation damage to the surface due to a bombardment byhigh energy particles (above 102 eV) will enhance thereaction probability and then chemical erosion couldproduce serious problems (for further discussion see Ref.56). These questions are now being investigated in manylaboratories including our own.1

The control of impurities appears to be the most serious

tRemark added in proof. More recent measurements utilizingbeams of thermally produced H-atoms'°3"°4 have revealed thedecreasingreactionprobabiityathigher temperatures (above about800°K) under these conditions. In an active discharge, however, nosuch decrease could be observed, but the reactivity increasesmonotonically in the temperature region measured so far.'°5 Twoeffects may be responsible for this discrepancy:105 (a) Since thesurface chemistry involves nonlinear processes of differentorders'°4 the overall mechanism canbe different atthehighdensitiesof the primary fluxes of H-atoms underplasmaconditions, fromthatat the low densities in the molecular beams. (b) The high energywhich is transformed to the chemisorbed layer due to therecombination of ions and atoms, and due to the de-excitation ofvarious species can enhance the surface diffusion (see Ref. 105).

tRemark added in proof. The effect of the radiation damage hasbeen demonstrated in severalpapers presented at the recent "Conf.on Surf. Effects in Controlled Fusion Devices" (San Francisco,February 1976) (see Ref s. 105—108).

obstacle in the development of the next generation ofTokarnaks and consequently the present research activityis very intense. By the time this article is published a largeamount of new information will probably be available andhopefully, some suitable material will have been found.Nevertheless, future progress towards controlled ther-monuclear reactors will give rise to many new problemsfor plasma chemists. Let us bear in mind that, in view ofthe present state of science and technology, controlledfusion appears to be the only way to solve the energydemands of the not-too-distant future, even within thescope of a reasonably limited economic growth.°

4. SYNTHE$IS OF NITRIDES

The high dissociation energy of nitrogen causes seriousproblems in the preparation of nitrides at low tempera-tures. For example, a number of heterogeneous systemsinvolving solid nitrides and HCI or iodine possess quitesuitable equilibrium compositions below 1000°C but thedeposition of nitride does not take place because of thehigh activation energy for the reaction of molecularnitrogen with the particular metal.6' A low pressureplasma with a high degree of dissociation is an idealmeans of achieving such a synthesis.

A number of papers on the preparation of thin nitridefilms by chemical vapour deposition (CVD) or by a directnitridation of metals in low pressure plasmas is available(e.g. Refs. 3, 19, 54, 81).

By means of a typical CYD technique, volatilecompounds of the particular metal are mixed with nitrogenor a nitrogen containing gas (NH3, Ar + N2) and introducedintothe reactionvesselwhere adischarge takes place. Inthisway nitrides of boron,'9 aluminum and gallium8' weredeposited from metal halides and nitrogen. Silane-ammoniamixture is preferredto siiciumhalides if silicon nitride ifimsare deposited for semiconductor devices.'°"82 In a givensystem, better conditions for nitride deposition are to beexpected if iodide is used instead of chloride, since theformer is generally less stable.2

4.1. Growth of crystalsChemical transport of metal with a simultaneous

nitridation appears to be a suitable method for the growthof crystals.6'48 The experimental arrangement is shown inFig. 9. The discharge took place in a quartz tube (1)between two outer electrodes (4) which were not watercooled in this particular case. The typical dischargeparameters were: frequency 80 MHz, discharge current2—3 A, pressure 1—2 ton, gas flow around 100 torrcm3 sec' and a N: Cl2-molar ratio about 12: 1.

Metal chlorides are formed by a reaction of chlorinewith the metal sheet (5) in the zone of alow plasma energyat a temperature of about 500°C. They are transported intothe insert (2) where a high plasma energy is obtained (high

r(6

Fig. 9. Experimental arrangement for growth of nitride crystals.(1) quartz discharge tube, 4 30cm; (2) insert (Al203, A1N, BN,etc.); (3) h.f. generator (80 MHz, 1.2 kW); (4) electrodes; (5) metal

sheet; (6) corundum tube.

N2

)IqaV 2 v1

3gn1-i1orl 3th lo m13i ni b3ni1qx3 d n i113 itfT 10 \hni3iv 3111 ni iqq 1131(1w '13fi3 £m1q 10 in3ibi

3(11 iol 'noi3ffl 3ii'1sI3 f1if1 gnill3loq biIo wohv lo noirii3x3-3b bns 2moi lo noiinidmo331 i m3ia 2-11D\WT 3111 m11q tuodliW 3i33q2 bIi3x3 bI1ir12 i muhdiliup3 13im31f 3 3111 bn 31m13111ox3 1noiTh

dgiif 1o131dT .3bhoft{3 muinslil thiswoi isi •

Ai ol MiT lo noiri2oq3b iol i2233fl Ii 'g13n3 smlsIq .331q

23223220 3bnhin muim1ii 13i31&13 3i1b313m ti ol 3IJU -idmo3i 33Ehu2 iol '3n3i3ffl3 3iI'1I13 i1id 'I3vii1131 11

23133q2 b1i3x3 lo noiIii3x3-3b bn 2moi lo Iioiifl 13fl3 fn2E1q 3(11 1Iff3Up32f10D 2 .gil ni d n3 3V113E 3111 moll b3n3i i r1idw qe 3(11 flI 2133b

3111 lo wolloil 3(11 3bilni imu1q on i 13r1i bn 391&132ib

3111 b3iS1fJ313 EIf13331 3vIrf 3W .(dOl .giI) 21i213 3(11 ni 13qq (131f1w noitth3oeib lo 331b di lo 21n3ibi iii ba1q hu2 b3n331 Ion .3.1 Iff lo fli3iV

mthnsjil ,cjj iol 211o11s1U3!3 is1iini m21q n3o1b11 -si3o221b 10 tb 3111 1r wo112 m3i2 no11in\3bho1d3

3(11 11mm 1.0 1uod lo 33ni2ib r O1 'd 231b noil .3.1 shu2 3(11 no lmoIs lo o1 lo iiIidEdo1q IBiol 10 101 lo 13b10 di lo i noi1u31 bnE floii&iidmo33l lo bn noiIi3oib 1o 3313b 3111 r&fi 2fl3 thIT

eilin3i W31 lo 33ni2ib i 1noiI 2323133b noi1Ii3x3 3(11 moil b3n33w 31E Il3iflw 2floi31 Ri 3lI3millIrn i 10 lo 33133b r1tc1 3fflu1 3ffI2 3111 lu 4ud ,gl&132ib 3vii3

1i331ib 31 (131(1W shtje no b3ni11do 21 noiii3oib 131m3r13 3111 ur1T £mlBIq 3Vii3 3(11 01 b3loqx3 in331b6 3bi2 di no 3biioIcl3 3(11 ol b311i112 21 mulldlIiup3

bnB wo11or1 3311 mu1q di ni '11rn33q23 3IE1I2dU2 3111 ol 12uo3nS11iJfffl2 Ii ithdw 313111 b3v1oeib 21 11213 3111

3i1i21J11i d bnIl 10i 231u8i1 .233hu2 131uo 3111 no ewoi ol '(1121m3113 £m1q moi1 noiii2nslI gnuI23i3ini 21111

.muridlllup3 3imEnbom13(1i 3(11 3fl 112im3(13 iunibio (111111w 21113E I(i2'(13 3(11 1o )bfl 3111 i11 2331iofl 3(10

in3m331th 21 113111w 3li3mullim11Io 2111(131 w3111 1o n1112ib .3vod11 n3v1 noii&ni123 3111 1111w

c1ic1 3(11 1o 2111i213 b3ni11ido IIn3331 211(1 iiW . sIft gni2u i1 lo noii113ftibom- 31U11113qm31

Q*.E .giI ni nwod2 in3msnsTI11 1111n3m115qx3

211u231 3111 1o wsiv iii 4iHi bi112 sd m ii noilul3no3 flu

ni 11oq2n1111 1831(113113 2msi22 523(11 1111w 1111 02 b3ni11ldo 2111i213 sbiiiin 1o diwoi ml bodism ni21mo1q 1121 11m2111q

'g13rz3 r1W1 21111 01 suU .(1n311u3 3111f132ib 3111 1o '1i2n3b (D°0001 ) slulfllsqmsi 211 11111u3n wol s1vi1s1si 11 bn11

gnthnoqs11o3 3(11 bn8 b3lfli3o22ib 3111 23b1101113 3111

1111w isnni sill no 211 113W 211 () sdiil no b3iiloqsb 21 3brllm bn11 13\M1A 3(11 lo fl0i221132ib 1113113103(11 A .() tislni 1o . .1351 ff1 ff3V1 211W 2m3j2'2 1D\MiT

33111q ni)l111 1331b noi1311i1uq 3(11 21 gifinoilfism iltioW no1i 3m gnini11lno3 munifnul11 li :2i2s(11n'2 3111 nnub

3g111113 sift ni bs2u 21 13O3 bais mui3ngm b11s1 no3i1i2

n113 mui23ngsm bn11 no3ili2 lo 2331111 1111m2 13V Ino 3no mthn18 sill ni bnuol 3d

211w icl\mrn .0 bnu irl\mm 0.0 luod 1o 31111 11lwoi A .13vf133q2s1 sbnlin muirnthl bri11 munimul11 101 bsvls2do

51u13u112 31112om 11 wod2 2111l213 sbnlin munimuls silT RI .21X11- 3(1110 noil33lib sill ni noil11lnsho2ib 3ffl02 illiw

2111I21D .bsvl32do 02111 211w rDwo1 111W! 232113 5ffl02 13113d 11 svs(1 11n3111qq11 2iX11 sdul 3(11 1115n nwoi ffWO1 320111 n11rft 213313b 33111111 l3wsI bn11 1l3ffloi113i0l2

lo 21n3ib111111ib111 oi 3ub '1d11do1q 21 2i11T .1111w 5111 1115(1 bnfl 2nothsls lo noil11llns3no3 211 113u2 21313m11111q 111n2111q

.313 s1ul111sqms1 noii6i3o22ib lo 3313b 23i33q2 b31i3x3 d 111i213 niwoi sill lo noil11nlm11lno3 31d122oq 11 01 bn11

.1132n1 3(11 ìó 111i13111m 3111

02111 2tn3ib111 11m2111q hlibfll 5111 11w 1111irni2 u ni silT .21si213 stiin muinth lo IlIwoIg 3111 33nsuftni

135112 muiln11lil 11 ni2u 'd b3biovE 3d (1113 flOillSfli(fl611103

3q11(12 31(1113 a lo 2l1112TJ .noililoqsb 101 3111112d112 11 211

noil1131tin11m 11111 lllOom2 sd 01 nwodl 233E1 l11Iugni2 illiw .b3nisldo 513W 23mil 0000 01 qu lo

13(11111 211W 3flO nolti2oq3b 3(11 ni 13fl3 amaIq 311111

Isnistni isia1 11 10 lnsnu3 31111132ib 131111m2 11 1311113) 2231

313W 21111213 niwo 3111 (1132(11 sill 10 lslsmath sill 01 1ff3311(bfl sbi2 3111 fl0 bsvloa2ib 112uosn11l1umi2

3111 Ii b33nuonolq siom 211W 133113 idT .3111112du2

211W 313(11 .3.1 l1s2ni 3(11 (11 '1no bs3slq 211w 133112 muinajil •(Q .giI 332) () 133(12 11113m on

isns 11ic1 3111 ni 135112 muin11lii lo noilamlol2nail 3rlT 11oq2n1111 3n111-I1o1l2 11 01 sub 33111q Aool sbiilin olni 3flO

lo noil11i1111i21331 3111 101 bsdil32sb nssd 21111 113i11w

.1sq11q 1n3331 1110 ni ffl1Jff3bd'1Off1

sdoiq 3111 RO bsxfi 111l2s13 11 113112 2wo112 1101 31u1q dOl 31ugi1 .Sqo32ol3im gninn1132 no111s sill lo i3bIorl

ni illwoig sill gnnub lala'13 3111 lo noili2oq 3(11 23111112u1li

.11m2111q

4

YO3i.ia I-IOIH

AMAJ 014

(if)

(a) .31aiidu2 srll ol 1113311 jba sbil sill no bsf3afla 4iwoig sill nhub 2aW 113111w 18l2'i3 sblilinmuinaiiT .01 •jcJ •wsiv 31111m31132 (if) (3qo32o131m flfl832 nothsls) llqaigolodqol3im

(a)

Non-isothermal low pressure plasmas 173

at relatively low temperatures. In order to obtain bettercrystals a frequency higher than 3O MHz and h.f. power4 kW or more should be used. In addition, the plasmaenergy necessary for the deposition of nitrides should beless if iodine or bromine is employed instead of chlorine.2

4.2. The synthesis of phosphorus nitrideThis is another example which illustrates the advantage

of plasma chemistry. Recent thermodynamic calculationshave shown that the formation of phosphorus nitriderequires the use of high temperatures.83 On the otherhand, nitrogen reacts with phosphorus at a lowtemperature under plasma conditions and solid phos-phorus nitride of various compositions between P3N3 andP3N5 can be obtained.8 This compound is used fordoping of semiconductors (e.g. silicon). Our recent studyof thin films of phosphorus nitride P3N5 has revealed theexcellent dielectric properties of this compound.9

The films have been prepared by a direct synthesis fromthe elements in a high frequency low pressure nitrogendischarge (for further details see Ref. 9). At a temperatureof about 265°C, X-ray amorphous thin layers could bedeposited on various substrates. They are transparent inthe visible and near i.r. regions. An absorption edge around350 nm and two strong, broad absorption bands withmaxima at 8.2 j and 11.2 m have been found.

Permittivity of 4.4 / (±0.4) at room temperature isindependent of frequency within the limits of accuracy ofmeasurement between 1 kHz and 20 MHz and increaseslinearly with temperatures between 77 and 525°K. Thecorresponding temperature coefficient is 3.6 x i0 deg'.The films show relatively low dielectric losses,tan 10-2 at 1 kHz, and a high dielectric strength of upto lxlO7Vcm'.

All these values are quite comparable with those oftypical dielectric films such as Si02 (see Table 2) andpossibly they can still be improved by further preparativework. Phosphorus nitride might therefore find someapplications in semiconductor technology, especially forcompound semiconductors (e.g. gallium phosphide).Further work on the preparation and relevant character-ization of this material is now being done in ourlaboratory.

5. TRANSPORT WITH HYDROGEN

A number of binary compounds—commonly called"hydrides"—are formed by reactions of discharge acti-vated hydrogen with corresponding elements or corn-pounds."87' Many of these hydrides are volatile and theydecompose at higher temperatures. In such systemschemical transport of the element can take place.

This phenomenon was first described around the year1920 and, since it appeared as an anomalous strongcathode sputtering of some elements by hydrogen, it was

called "chemical cathode sputtering".89'° However,GUntherschulze had already shown that "sputtering" ofAs, Sb and Bi took place even if they were not electricallyconnected with the cathode, but only placed in thepositive column of glow discharge at floating potential.

GUntherschulze originally thought that this phenorne-non was due to an interaction of hydrogen ions with thesolid,°'° but later work of other authors showed that theformation of volatile hydrides can take place by reactionsof H-atoms at thermal energies (e.g. Refs. 1, 88).Whatever the particular mechanism of the process maybe, one has to distinguish between real sputtering andchemical interactions with the surface. In the former casethe primary particle penetrates into the solid and an atomof the solid is sputtered due to a momentum transfer fromthe primary particle to the crystal lattice. This phenome-non is observed at high energies and does not take placebelow some threshold energy which depends on theparticular system. The ion energy in the low pressureplasmas of the positive column of a glow discharge is farbelow these threshold energies even if ambipolar diffusionis considered (e.g. Ref. 56). Moreover one can calculatethat the total flux of ions towards the surface of the solidis too small to be able to cause the observed rate of thereaction (e.g. Refs. 27, 66).

High energy sputtering is theoretically well understoodat the present time,9' but only very little is known aboutthe interactions of gaseous particles with solid surfaces atenergies below 100 eV.26"°9 The transport phenomenawhich will be discussed in this section occur at thermalenergies.

The transport of a number of elements with hydrogenplasma have been described previously: As, Sb, Bi, Se,Te,89'92 C,27'89 Ge, Si,93 p 8 carbon transport has alreadybeen discussed in Section 3.2. There is no doubt that thistransport takes place due to the formation and subsequentdecomposition of volatile hydrocarbon compounds, andthat the reaction of electrically neutral species dominate.

The transport of germanium and silicon took place onlyin the direction of increasing neutral gas temperatureT,— T2, but the discharge current density was nearlyconstant between the charge and the deposition zone.93The experimental arrangement was similar to that shownin Fig. 11, but the deposition of germanium and silicontook place only if the deposition zone was heated up to400 and 600°C respectively. In view of the very lowdeposition rate of about 20—50 nm hf', a reactionmechanism involving ions cannot be completely ruled outin this case. However, experimental findings of otherauthors indicate that the hydrides are formed rather byreactions of H-atoms.11°

For example, Radford observed formation of gaseousmonohydride radicals like SeH, TeH, etc. during areaction of the corresponding element with hydrogen

QUARTZ TUBE/H.E GENERATORit____

ELECTRODES

TO PUMP

DEPOSIT

Fig. 11. Experimental arrangement for transport of red phosphorus with hydrogen.

)Iaq3V 2

goibd q .5wb3301q Io tikIqcni2 3di 3d b!11o3

or 31qm12 bnE qs3d3 2i b32U f3iffl3uI3 \Ino 3f11 2ifl3ifiw i uoignb i ibifw uTodq2odq irIw on o2!A .3Ibnsñ 3di Io 2r33q2 noii3fihIJq 3ñT .oiq 3d3 iii b3vlovth

uo iii b3igii3vth gni3d won 31E hoq2nnlr .2mth1&1m ioq2nsti 31dioq 3111 22iJ32ib won 211 13J

nnxli 31dn12 3lom IIn3imsnsbom13c1i 21 2LflofIq2ofiq b5I lo l3dmun s ni i21x3 rin ii bnn 3flO 311(1W 311t

-V32 noiin1oqI1v 10q 2flO1ln3ftibOffl 3iI1qn1o11nl213 1s1b3d1i3i-,q n bnn n3Aold 3d lum 2bnod thrri q_q 113

noilnvibn gin1 n 231iupi 2iflT .b3fnlol 31u331om 21J032ng 21 noilnloqnv3 cfi 1in3up32no3 bnn \13ff3

* nollnloqnv3 nA .( lom In1 2 = (b31 wok 8.bmjo1 n33d 2&l Ol ((Sf11 2231 1f13i3ffl303

m3i22 21J03f13g0131311 3(11 1o ioqnv mtthdillup3 3ff!' 21 2lno(iq2orlq bilo2 11 .51b3f151131 q 3111 lo 2i212no3

3111 1no noiiibno3 muildiliup3 •153n is b3iioq3b b3m1o1 21 2u1o(1q2offq 31111w cf1 ffo!1n31hbom 3Idni2ni3m

323(11 lo be2oqmo3 33it151 11112\f13 151113310m s gnivild

S 23vIovrn noiin3flibom 21111 lo noilnloqsv3 3(IT q 2k 1n3ifflo noiiLnoqJw3 3(11 1)115 13fl3 noitnvibn wol

83f1O b31 lo loq2fu(1i nmnIq 3(11 (11 15111 3o1313f11 fl332 211!

rlir1 3(11 421iT :b3vlovni 315 2133113 owl 2uIO(Iq2orlq 21 2ulorIq2orIq b31 lo noiinloqnv3 3rIi iol 'g13n3 noi1nvi1n

3(11 bno2 bns 2233O1 335111J2 mo d b3in3vmu3li3 3Idi2oq 11n3ii3rnA 3bllm 21 JriocIqlocIq bi lo noiii2oq3b 8.32&Iq ng 3(11 111 23hJ3310ffl q lo 33fl32d5 3(11 01 3111)

13(113(1w 1n nm 3n0 .nolinloqnv3 3(11 13b12003 211 13J 01 3ub 3d bIlJo3 noiinloqnv3 Ia 3151 dgiil b3vl3do 3(11 d ionnnz ii1idi22oq n c13u .gniil3m 331311112 311102

3m 201015 H 33012 Iooiq 3v12u13no3 Iuodiiw b3bul3x3 .m3 332 Ol—°O1 x lo 3151 sin 335111123111 ff0 mniqmi — H 21n3v3 nobnnidmo33l 01_E0j 12531 is 23V1 2IfIT

S 2W0112 f 31111q .332 13 3112 33Eh1J2 (1353 no niTuI33o 13\n1 2IrrofIq2ofIq b31 3111 lo 335111123(1110 rlqsio1odqoioim luods is 3g1&132ib n3go1b11 3(11 ol b32oqx3 1133d 2511113111w

ni113m 33511112 lo noiln3ibni on 21 31311T .nim U lol D°02 1sr1315m 21111 no 3Idsninido 00111110231 123cli(1 31111111w 113V3

3(11 lo 2103m3111253m 11113153 noi1ibbs nI .(xOOOO luods) rn1s3(fl3vo 1ns3flrn12 on 1id 3nob 113311 3vs11 31u1513qm31

3110n3 ni113m 33S11112 31o131311T .bnuol 2SW 21fl011q2011q10

(13311 2nd (131(1W 2l3S1 lUlOfIqlodq 1)31 3111 10 33nh1J .1 00113313) nim El luodn iol D°OE In 311u132ib n3go1b(f1 ol b5loqx3

.(xQèE qoo13im n1nrrna2

—bnuol 2SW 431(1w noiislln33003 dgiñ 311'!' .wo1913i1s 23bu13x3—l3v1I33q231 H3T bnn H3 101 I bnn luods

ls3ilqo d b3v132do 213W Hi .2001 1 3101 lnn3lhngi2 'cnn n33Wi3d 3351q 9(11)151 3915f132ib 315 ff5 111 'qo32o113q2

21 31(51)2 II .(V .q Q .1351) 23bo113313 (103(112 owl lo 1)1102 13 3g1nf132ib 31112231q wol )153W n olni b33ubolini

daidw b31i2oq3b 21 (V.1—.l = ) Hi nobi2oqmo3 31k m (.1 = ) Hi 215311)51 3bnbr1 15111 231531brn nisgs

1311311115 232oqm033b 1)1102 3(IT .sm2sIq 3111 ni b3miol bnn no3ili2 bilo2 lo noiinmiol 3(11 dIiw 31111513qrn31

.q Q .1351) (13301b'f1 (1031112 lo lloq2nnll 3111 in1k 31n3rbr11 291111)011 323111 hA

1n311p32du2 hon nolismiol 341 23vIovni muinnfni3g hon noi1sqi3iiis .21531bs1 3bnbfI 311151ov lo noin2oqrno33b

lud b3bu13x3 1313hqmo3 3d lonnn3 2233O1 21(1101 2noi lo .3IdsdOiqrfIi 3d 01 201332 Ii

1O11112O(1c o l1oq%nu1 911T .12 23VlOVfli 11 33ni 8.'1in3331 1no b3ibui2 n33d 2511 (jIlT

221132ib 11542 3W 232233o1q 33Eh112 9ni123131n1 13V 30102 .1)5131) 31001 flI Ii

31ff .11 .gi1 (11 itwotk (1 1n3m39n5115 lsin3mn3qx3 3dT oslo 36u1 thSUp s (11 3351q )1001 391511321b '31I31Jp31l dgi11

1315W 15fl131X3 owl 1133W13d mm Ol lo 1313m51b 1311111

313w 21313m1315q 33'151i32th 1S3iq\1 3flT .23bo113313 b3Ioo3 2133:11011 31112231q Islol ;HM 1 3n3up311 :2wolloI an

1fl3TIJJ3 39154321b ;'332 Em3 nol OM luods 31s1 wofi 23bo113313 3(11 n33w13d 330s121b bnn A I bun .O n33w13d

A313M) 2111o(Iqoflq 1)31 21Jor1qIoff1s 'si-X .mm OI 13fl3d 'li1uq lo n33o1b'11 bnn (dnd OVV hA .Inup .0(1331

1inUp S RI 3915(13 (TJlo(Iq2O(lq A .b32u 313W. nsdi owl 3111 n33W13d 3fl0 3111 oini b31132n1 (SW mod

.(8 .1351 332) b322591uo '11U131n3 bnn 23bo113313

1513V3( 2fl011ibfl03 3915(1321b b311133q2 3vods 3111 13bnU

25 iuo4 as iii b311oq2n511 3d bluo3 (LJlOfIq2oflq lo (10519

2139 1511U3n 3f1T .11 .giT m (WOllE 3(11 d b3ls3ibni 3111 Is D°OVF bun O?. n33w13d 3ulnv n b3i13s31 311115i3qm31 341 01 °OO bus O1 1133w13d 31111w s bns 3915(13

2Iflodq2odq 1o 31131 rIIwol3 ls3nil A .3flO noiii2oq3b dgiil 3(11 ol 311(1 .b3v132do (SW 11(1 mm .O bnuois 213s1

-(orlq b31 2uO11q1oms si-X 3Al1-2anI3 31S1 noIil2oq3b '1ivii2i231 311i33q2 m3 g 'Ii2n3b) b3mloI (I 2IrIo(1q

.(m3O1 I 3bo113313 1o 1131 3111 (10 36111 3315(13211) 31k 1(311W

(SW Wloflq2Odq on D°O? 3vods b31n34 11Sn131x3 (SW 3(11 01 13(111111 33131q Aooi noilianq3b 3rlT .31341 b3lkoq3b

.b3linv3iq D°OU bnuois 31u1s13qm31 5 31311W 1131 b3Ioo3 (SW 31J1J1 3915(1321b 3(11 bnsil 13(110 3(11 no 1I

23b0113313 owl 3111 fl33W13d 1315w 9111W011 lo (1(5301 'td th1kiw ((01331 isdith b311u33o 2inodqzoflq b31 lo nolil2oq3b

.91111003 10 9nInrn33d 3(11 13115 2bn0332 W31 S 1)321125w 110915 3ulq 1(311W b3vl3do (SW lioqansil 31111.1

owl nncll 31001 (SW 3151 hoq2flsI1 311'!' .113901b\(d lo bs3tni 21(111)115 n33o1b\f1 411w mill nsdl ((31 3blJlln3Sm lo 2l3blo

01 3kIioqIm 2111331(12 n33o1bf1 lo 23351101 311b 3d 1)11103

2U1O(lqOlIq 15311111331 341 ff011 3iIJi2iom 3vom3113i31qmo3 .36111 thsup 3111 moil bns

3(11 3nh111b (ln3qqn (13111W 133113 noiiS3ftnuq 91(0112 A 1snimi131q .212313101 1n311351q lo 3d 053 tloqnnli

lo 11(31(103 thuqmi 15111 nwoil 3V511 21101i59ii23Vffl d b3deinimib 2k I nM JD sD 3M sM fl 3)111 21(1301313

3mo .13V31 mqq 3111 01 qu 3bulm3sm lo (13b1o 1513V32 (UlollqoHq Ia noiinnimnino3 3111 moil 3(115 m3ldolq

.(mqq 1513v3( ol qu) 3d111 39151132kb 341 moil (1031112 411w (1(11 9nibiovn lo ii1idi2oq boog n (I 313(11 22313I1t13v3M

.3dlJi 3915113(ib 3(11 10 ngi23b iqoiq s d noilnnimnlno3 bo1k3m nOiiS3ftruJq 5(5 lloq(nSli smaslq lo 3951n5vb5 3(1'!'

Non-isothermal low pressure plasmas 175

= P + PH(g)+ 112(g)

H226 ± 5 kcal mol'

P—P—3H----*P = P• ± PH3—24 ± 5 kcal mol'

P+H—PHP H—*P = P-H

excluded and some surface chemical processes must be kcal mo11) a direct desorption of PH2 is much slower thaninvolved in the transport mechanism. the sticking rate of H-atoms.t

A detailed analysis of surface processes belongs to the Therefore, another H-atom will probably impinge onmost difficult problems in chemistry and one is not able to the strongly chemisorbed PH2-group (5.2) and recombinegive a final solution of this problem now. However, a step with one of the H-atoms there:by step discussion of all possible processes, which will ,lead to an interesting model of the chemical evaporationof phosphorus will be given.

The primary flux of H atoms between 5 x 1020 and

,/H + H—*p / H(5.3)

H H21021 cm2 sec' corresponds to about 5 x 105_106 atomsimpinging everysecond on each surface site. It is evident,that a PH species strongly bonded to the phosphorussurfacewiibe immediatelyformedwithanyP surface atompossessing a free electron (e.g. atom No.21 in Fig. 2 of Ref.99):

The H-atom can come either directly from the gas phase,or from a weakly bonded, mobile ("liquid") layer which isadsorbed on the first one.

The step (5.3) is then followed by the formation eitherof gaseOus PH and H2 species (5.4),

or of the PH3 molecule

I=P+H—P=P-H (5.1)

The desorption frequency, Vdes, of the PH species isextremely low at a temperature of about 250 to 315°C ascan be calculated from the formula Vdes =

vo exp (— ES/RT).'°° Assuming vo = 10-10' sec',and that E(P—H) E?0(P, red) 50 kcal mol',

iO sec1.The overall surface density of H-atoms which are,

adsorbed on the surface can be estimated only roughly. Inanalogy with other systems (e.g. Ref. 101) one can expecta desorption energy E,(H) 20 kcal mol' which gives adesorption frequency of about iO—iO sec1. This isequal to, or smaller than the number of H-atoms sorbedon one site per second if a sticking probability larger than,or equal to l0_2 is assumed for the impinging H-species(cf. Ref 102). Thus, the phosphorus surface is almostcompletely covered with a chemisorbed layer of atomichydrogen.

Two H-atoms adsorbed on adjacent sites can break thecorresponding P—P bond with either the formation of twoP=P—H groups (e.g. atoms No. 19 and 20 in Fig. 2 of Ref.99), or one P=P—H and one

(5.2)

group (e.g. atoms No. 21 and 17 or 21 and 18 in Fig. 2 ofRef. 99). The latter process occurs if a P—P bond between aPH-group (see eqn 5.1) and a neighbouring P-atom isattacked. Considering the energies of P—H and P—Pbonds (about 77 kcal mol' and 48—58 kcal mol'respectively86) it is evident that both those steps arestrongly exothermic. Due to the low desorption frequencyof the PH species, each formation of a PH group isprobably followed by the formation of (5.2).

Also the PH2-species are chemisorbed relativelystrongly to the phosphorus surface as shown by thefollowing consideration: A desorption frequency of thePH2 species equal to the lowest value of the sticking rateof H atoms, Vdes 5 X iO sec', is obtained if a desorptionenergy Ees(PH2) 20 kcal mol' is assumed. Since thisvalue is muchless thanthebondingenergy(E(P—P) 48—58

(5.4)

(5.5)

The latter process, although exothermic, seems to beless likely beôause of steric reasons. In addition, if thetransport were to take place by the formation andsubsequent decomposition of the PH3 species, thedeposition of phosphorus should be enhanced at a highertemperature (see eqn 5.5). However, the opposite effect hasbeen observed experimentally and an insignificant amountof gaseous products was condensed in a cold trap (77°K)attached to the outlet of the discharge tube.

In conclusion, the proposed mechanism of the chemicalevaporation of red phosphorus can be summarized as.follows:

(5.6a)

(5.6b)

P = P-H + (5.6c)

H

+ H—P = P. ± PH(g)+ H2(g) (6d)

Considering the known crystal structures of red andblack phosphorus,86'' it is easy to see, that thismechanism produces more and more active sites P = P'.Therefore, the overall reaction rate (which might be verylow for an ideal surface) will increase up to a steady statevalue corresponding to a surface with a high density of theactive sites.

The previous estimations indicate that the rate limitingstep should be the desorption of PH and H2, (5.4). Anactivation energy for this process of Ees(PH + H2) H(5.4) = 26 kcal mol along with a surface site density ofthe order of iO' cm2 gives a total evaporation rate of

5 x i0' P-atoms cm2 sec'.An experimental value between about 5 x 1016 and

5 x 10 cm2 sec' has been estimated from the weightloss of the phosphorus charge. The uncertaintity is due tothe unknown ratio of the active surface and the geometricarea of the phosphorus powder used.

tRemark added in proof. As pointed out by Olander (D. R.Olander, Univ. of Berkley, Private communications), the rate offormation and desorption of the PH2 group could be enhanced by ahighactivationentropy(increaseofthenumberofrotationaldegreesof freedom). SuchaneffectwasreportedrecentlybyMadix etal.(seealso Ref. 26a) for Co desorption from Nickel.11'

176 S. VEPEK

A better estimation could be obtained from the lineardeposition rate of about 5 x 1O_2 cm hr1 when thedischarge tube was cooled by flowing water at a placeclose to the charge (see above). The absolute temperatureof such a layer during deposition is about half that ofphosphorus charge, which results in a decrease of thefactor exp (—E (5.4)/RT) by orders of magnitude. Itmeans that the evaporation from this layer is negligible,and the deposition rate at the cooled place is equal to theevaporation rate at the charge providing, that diffusionlimitations canbe omitted. Thelatteris approximately validif the cooled surface is small. The deposited layers of redphosphorus are glass like and the ratio of activesurface : geometric area estimated from microphotographsis between 1 and 3. Thus, the calculated deposition rate ofabout 2—6 x iO' cm2 sec1 can be directly compared withthe theoretically estimsted evaporation rate of

5 x 10 cm2 sec.The agreement supports the proposed mechanism.

Moreover, the considerations of the last paragraph alsoshow why the transport takes place in the direction ofdecreasing temperature even at a constant concentrationof H atoms.

Such an agreement, of course, gives no conclusiveevidence for the correctness of the proposed model. Onlyfurther experiments will allow a better understanding ofthe transport mechanism. The actual significance of themodel consists in the fact that it allows one to formulate theproblems to be studied and enables experiments to be donemore precisely.

6. CONCLUSIONS

Plasma chemistry is being developed in a regionsomewhere between physics and chemistry. The choice,generation and, characterization of a suitable plasmapresumç a deep knowledge of plasma physics but theirapplication to interesting and important chemical prob-lems requires the experience and imagination of chemists.

The success of plasma technique in various fields suchas semiconductor technology, deposition of organicpolymer films, plasma spraying, preparation and modifica-tion of powder materials, justifies the efforts of manyscientists and technologists in this field. In contrast tothese results, there has been relatively little progresstowards the understanding of the basic processesoccurring in the plasma. The present article summarizespart of the recent investigation of these latter problems.

Low pressure plasma chemistry is a method "parexcellence" for doing "high temperature chemistry" atlow temperatures. It is especially suitable for depositionof thin films and crystal growth since the reaction yield isnot limited by the low pressure, as it is in manyhomogeneous systems.

An energy of reaction, as well as an activation energy ofup to 100 kcal mol', can be supplied by the plasma attemperatures of less than 1000°C for both the neutral gasand solid. Thus, strongly endothermic reactions can takeplace at low temperatures and the rates of many processes,including the evaporation and deposition of solids, whichinvolve some kinetically hindered steps, canbe significantlyenhanced by using plasma.

There are many basic problems to be solved in thefuture work. Particularly, more attention has to be paid tothe interactions of plasmas with solid surfaces, which alsoinvolves a detailed study of surface chemistry. A betterunderstanding of the basic processes will surely open newpossibilities for plasma applications.

The erosion of solids by atoms, radicals, excited speciesand ions, with energy varying over a wide range, is nowbeing investigated because of the impurity problems in theTokamak devices for controlled thermonuclear fusion.This appears to be one of the most serious impedimentstowards achieving ignition conditions. Similar kinds oferosion also arise in the exhaust of rockets and during theentry of satellites and spacecraft into the atmosphere of theearth and other planets.

Chemical transport of solids in low pressure plasmas isa promising preparative method, as well as a simplediagnostic technique for studying the behaviour ofheterogeneous systems under plasma conditions. Thepossible use of plasma transport for solid purificationrequires further investigation into this effect.

It has not been possible in this article to cover the wholefield of heterogeneous reactions under plasma conditions.We have attempted rather to illustrate with severalselected examples the unique properties of lowpressure plasmas and the feasibility of their applicationsto preparative solid state chemistry.

Acknowledgement—The author wishes to thank Professor Dr. H.R. Oswald (Univ. of ZUrich) for active support of this work and toDr. D. L. Cocke (Fritz—Haber—Inst., Berlin), Dr. M. R. Haque,dipl. chem. R. Wild (Univ. of ZUrich) for valuable remarks andadvice on preparation of the manuscript. Mr. Erwin Wirz haskindly allowed us to mention some of his results here.

REFERENCES

'F. K. McTaggart, Plasma Chemistry in Electrical Discharges.Elsevier, Amsterdam (1967).

25 Vepfek, A theoretical approach to heterogeneous reactions innon-isothermal low pressure plasma, In Topic in CurrentChemistry. Vol. 56, Springer, Berlin (1975).

3J. R. Hollahan and A. T. Bell, Techniques and Applications ofPlasma Chemistry. Wiley, New York (1974).

4E. Kaldis, Principles of the vapour growth of single crystals, InCrystal Growth, Theory and Techniques (editor C. H. L.Goodman) Vol. 1, Plenum Press, London (1974).

5J. P. Hirth and G. M. Pound, Condensation and Evaporation.Pergamon Press, Oxford (1963).

65 Vepiek, C. Brendel and H. Schafer, J. Crys. Growth 9, 266(1971).

7R. S. Berg and R. D. Nasby, .1. Vac. Sci. Tech 12, 188 (1975).85 Vepiek and H. R. Oswald, Z. anorg. allg. Chem. 415, 190(1975).

95, Vepek and J. Roos, .1. Phys. Chem. Solids 37, 554 (1976).'°A. M. Mearns, Thin Solid Films 3, 201 (1969)."R. W. Kirk, Applications of plasma technology to thefabrication of semiconductor devices, In Techniques andApplications of Plasma Chemistry. Vol. 56, p. 347, Wiley, NewYork (1974).

'2E. Geittner, D. KUppers and H. Lydtin, lEE-Meeting, London,Post-Dead-Line Paper (16/18 Sept. 1975).

'3J. Koenings, D. Kuppers, H. Lydtin and H. Wilson, Vth mt.Conf. on CVD, Stoke Poges (G.E.) (21/25 Sept. 1975).

'4R. Kassing and H. R. Deppe, Thin Solid Films 13, 27 (1972).'5R. F. Bunshah and A. C. Raghuram, .1. Vac. Sci. Tech 9, 1385

(1972).'6A. C. Raghuram and R. F. Bunshah, Ibid. 9, 1389 (1972).'7M. Venugopalan, (ed), Reactions Under Plasma Conditions.

Vols. 1, 2, Wiley, New York (1971).'8B. D. Blaustein (ed), Chemical reactions in electrical dis-

charges, Adv. Chem. 80, Am. chem. Soc.: Washington D.C.(1969).

'9L. D. Locker, Materials produced by electrical discharges,, InModern Materials. (editors B. W. Gonser) Vol. 7, AcademicPress, London (1970).

2O5 M. L. Hamblyn and B. G. Rebuen, Use of radio-frequencyplasma in chemicai synthesis, In Advances in Inorganic

Non-isothermal low pressure plasmas 177

Chemistry and Radiochemistry. (editors H. J. Emeléus andA. G. Sharpe) Vol. 17, Academic Press, New York (1975).

21J. F. O'Hanlon, .1. Vac. Sci. Tech. 7, 330 (1970).22D. M. Mattox, .1. Vac. Sci. Tech. 10, 47 (1973).23R. D. Bland, G. J. Kominiak and D. M. Mattox, J. Vac. Sci.

Tech. 11, 671 (1974).24M. Hudis, Plasma treatment of solid materials, in Ref. 3, p. 113.25H. Wise and B. J. Wood, Reactive collisions between gas and

surface atoms, In Advances in Atomic and Molecular Physics.(editors D. R. Bates and I. Estermann) Vol. 3, p. 291. AcademicPress, New York (1967).

25&G. A. Melin and R. J. Madix, Trans. Farad. Soc. 67, 198, 2711(1971).

26S. G. Davison (ed) Progress in Surface Science. Vol. 5, Part 3 and4, Pergamon Press, Oxford (1974, 75).

26&RJ Madix,Reactive scatteringfrom solid surfaces, in: PhysicalChemistry of Fast Reactions (editor D. 0. Hayward) Vol. 2,Plenum Press (1976). In press.

275• Vepkk, J. Cryst. Growth 17, 101 (1972).285 Vepkk, IEEE Trans. Plasma Sci. P8—2, 25 (1974).29K. H. Zaininger and A. S. Waxman, IEEE Trans. El. Devices

ED 16, 333 (1969).30F. B. Micheletti, P. E. Norris and K. H. Zaininger, RCA Rev. 31,

330 (1970).31H. Carchano, J. Chem. Phys. 61, 3634 (1974).32D. Sanchez, M. Carchano and A. Buii, J. AppL Phys. 45, 1233

(1974).33M. Millard, Synthesis of organic polymer films in plasmas, in

Ref. 3, p. 177.34P. K. Tien, G. Smolinsky and R. J. Martin, App!. Optics 11,637

(1972).35T. Wydeven and J. R. Hollahan, Semipermeable membranes

produced by plasma polymerization, in Ref. 3, p. 215.36 M. Irving, Proc. Kodak Photoresist, 2, 26 (1968).37H. Abe, Y. Sonobe and T. Enomoto, Japan J. AppL Phys. 12,

154 (1973).38H. Hosokawa, R. Matsuzaki and T. Asamaki, Japan J. App!.

Phys. Sup!. 2(1) 435 (1974).39P. Belland, A. I. Cuira, D. Véron and L. B. Whitbourn, mt.

Conf. Submi!!imetre Waves and Their App!. Locations, IEEECat. No. 74 Cho 856—5 MTT (1974), p. 75.

40W. Chantry, Submillimetre Spectroscopy. Academic Press,London (1971).

41A. K. Levine (ed), Lasers. Vol. 2, Marcel Dekker, New York(1968).

42H. J. Schötzau and S. Vepkk, AppL Phys. 7, 271 (1975).435 Kunstreich and J. P. Lesieur, Opt. Commun. 9, 146(1975).S. Kunstreich and J. P. Lesieur, Proc. mt. Conf. Inrared Phys.

ZUrich (Switzerland) P. C 200 (11/15 August 1975).45Rapport interne CERCEM-LDC No. 74/123, 93350 Le Bourget,

France.46J P. Shkarofsky, T. W. Johnston and M. P. Bachynski, The

Particle Kinetics of Plasmas. Addison-Wesley, Reading, Mass.(1966).

475 Verpek and E. Wirz, ZUrich 1974, unpublished results.48E. Wirz, PhD-Thesis, University of ZUrich (in preparation).49E. Wirz, H. R. Oswald and S. Vepek, to be published.50S Vepkk and P. Kocian, He!vectica Phys. Acta 46, 447

(1973).51V. E. Golant, Izv. Akad. Nauk USSR, Ser. Fyz. 23, 947 (1959)

(English translation).52A. von Engel, Ionized Gases. 2nd Edn, Clarendon Press,

Oxford (1965).535 Vepek, PhD-Thesis, University of ZUrich (1972).54A. N. Wright and C. A. Winkler, Active Nitrogen, Academic

Press, New York (1968).55H. J. Schötzau and F. K. KneubUhl, App!. Phys. 6,25(1975).

56S Veplek and M. R. Haque, App!. Phys. 8, 303 (1975).575 Vepek and Z. Hauptman, Z. anorg. allg. Chem. 359, 313

(1968).585 Vepek and Z. Hauptman, Proc. myth Yugoslav Symp. Phys.

Ionized Gases, Herzeg Novi (Yugoslavia) (1968).595 Vepkk, Z. phys. Chem. N. F. 86, 95 (1973).605• VepIek, .1. Chem. Phys. 57, 952 (1972).

PAC vOL. 48 NO. 2—D

61H. Schafer, Chemische Transportreaktionen. Verlag ChemieWeinheim/Bergstr. (1962); English translation. Academic Press,London (1964).

62M. R. Haque, H. R. Oswald and S. Vepek, Proc. XmIth Int.Conf. Phenom. Ionized Gases (Part 1) Eindhoven (TheNetherlands) p. 42 (18/22 August 1975).

63M. R. Haque, H. R. Oswald and S. Vepek, To be published.ML. C. Brown and A. T. Bell, Ind. Eng. Chem. Fund. 13, 203, 210

(1974).655 Vepek and W. Peier, Chem. Phys. 2, 478 (1973).66S. Vepek, D. L. Cocke and K. A. Gingerich, Chem. Phys. 7, 294

(1975).67B. Lersmacher, H. Lydtin, W. F. Knippenberg and A. W.

Moore, Carbon 5, 205 (1967).W. F. Knippenberg, B. Lersmacher, H. Lydtin and A. W.

Moore, Philips Techn. Rev. 28, 231 (1967).69A. B. Bass and H. P. Broida, (Eds.), Formation and Trapping of

Free Radicals. Academic Press, New York (1960).70B. Meyer, Low Temperature Spectroscopy. Elsevier, New York

(1971).71D. L. Cocke, C. A. Chang and K. A. Gingerich, App!. Spect. 27,

260 (1973).72J. M. Hastie, R. H. Hauge and J. L. Margrave, High Temp. Sci.

3, 257 (1971).73M. Coulon and L. Bonnetain, .1. Chim. Phys. 71, 711, 717, 725

(1974).74Vth Conf. on Plasma Phys. and ControL Nuc!. Fusion Res.

Tokyo (Japan) (11/15 Nov 1974).75R. BehrischandB. B. Kadomtsev,P!asmaPhys. & Contro!Nuc!.

Fusion Res. 2 (1974).76R. W. Conn, W. A. Houlberg and J. Kesner, Report UW

FDM-l06, Nucl. Eng. Dept. Univ. of Wisconsin (1974). NuclearFusion (1975). In press.

77R. W. Conn and J. Kesner, Nuclear Fusion 15, 775 (1975).78G. R. Hopkins, P!asmaPhys. & Cont ro! Nuc!. Fusion Res. 2,275

(1974).79D. Eckhartt and G. Venus, JET Design Group, Tech. Note (74)9,

Cuiham (1974).80W. C. Gough, In The Chemistry of Fusion Techno!ogy. (editor D.

M. Gruen), Plenum Press, New York (1972).81J. Pastrñák and L. Souèková, Phys. Status Solidi 3, K 71(1973).82H. F. Sterling and R. C. G. Swann, Solid State El. 8,653(1965).83J. I. Krasnokutskij, S. N. Ganz, V. D. Parchomenko, A. M.

Alekseev and V. I. Fedorov, Zh. PrikL Khim. (Leningrad) 47,1735, 1972 (1974).

84Gmelins Handbuch der anorg. Chemie, Vol. 16/B, VerlagChemie, Weinheim/Bergstrasse (1964).

85Ibid. Vol. 16/C.D. E. C. Corbridge, The Structural Chemistry of Phosphorus.

Elsevier, Amsterdam (1974).87D. T. Hurd, An Introduction to the Chemistry of Hydrides.

Wiley, New York (1952).88B. Siegel, .1. Chem. Educ. 38, 499 (1961).89A. GUntherschulze, Z. Physik 36, 563 (1926).90A. GUntherschulze, Vacuum 3, 360 (1953).91H. H. Andersen, Proc. VmIth Yugos!av Symp. on Physics of

Ionized Gases, Rovinj (Yugoslavia) (16/21 Sept. 1974).92S W. Ing. Jr and Y. S. Chiang, .1. Electrochem. Soc. 113, 192

(1966).935 Vepek and V. Mareóek, So!id State EL 11, 683 (1968).H. E. Radford, .1. Chem. Phys. 40, 2732 (1964).95Gme!ins Handbuch der anorg. Chemie, Vol. 15/B, Verlag

Chemie, Weinheim/Bergstrasse (1959).H. Krebs, Grundzflge der anorganischen Krista!lchemie. Fer-

dinand Enke Verlag, Stuttgart (1968).97J. Brewer and J. S. Kane, J. Phys. Chem. 59, 105 (1955).98G. A. Somorjai and J. E. Lester, In Progress in Solid StateChemistry, (editor H. Reiss) Vol. 4, Pergamon Press, Oxford(1967).

H. Thurn and H. Krebs, Acta Cryst. B 25, 125 (1969).'°°S. Glastone, K. L. Laidler and H. Eyring, The Theory of Rate

Processes. McGraw-Hill, New York (1941).'°'B. McCarroll and D. W. McKee, Carbon 9, 301 (1971).'°2G. A. Beitel, J. Vac. Sci. Techn. 6, 224 (1969); 8 647 (1971).

178 S. VEPEK

'°3R. K. Gould, .1. Chem. Phys. 63, 1825 (1975).'°4M. Balooch and D. R. Olander, J. Chem. Phys. 63,4772 (1975).'°5s Vepfek, M. R. Haque and H. R. Oswald, mt. Conf. on Surf.

Effects in COntrolled Fusion Devices, San Francisco, J. NucI.Materials (1976). To be published.

'°6R. Wright, R. Varma and D. M. Gruen, Ibid., J. Nucl. Materials(1976). To be published.

'°7J. Roth,J. Bohdansky, W. PoschenriderandM. K. Sinha,Ibid.,J.Nucl. Materials (1976). To be published.

'°8S.K.Erents,C.M.BraganzaandG.M.McCracken,Ibid.,J.Nucl.Materials (1976). To be published.

'°9P. A. Finn, D. M. Gruen and D. L. Page, A study of aluminumoxide sputtering: Anew approachtounderstandingthe sputteringprocess forbinary targets, In: Adv. Chem., Proc. Am. Chem. Soc.In press.

"°K. M. Sancier, S. R. Morrison and H. U. D. Wiesendanger, .1.Catalysis 5, 361 (1966).

"1C. R. Helms and R. J. Madix, Surf. Sci. 52, 677 (1975).