Low-pressure, high-density plasma nitriding: mechanisms, technology and results

Surface and Coatings Technology 108–109 (1998) 182–190

Low-pressure, high-density plasma nitriding: mechanisms, technology andresults

a , a b*T. Czerwiec , H. Michel , E. Bergmanna ´ ´ ´Laboratoire de Science et Genie des Surfaces (Unite Mixte de Recherche Associee au CNRS 7570), Institut National Polytechnique de Lorraine,

´Ecole des Mines de Nancy, Parc de Saurupt, 54042 Nancy Cedex, Franceb ´ ` `Ecole d’Ingenieur de Geneve, 4, Route de la Prairie, 1202 Geneve, Switzerland

Abstract

This paper reviews the low-pressure (,10 Pa), high-density plasma-assisted nitriding processes recently developed for metallurgicalsurface modification to improve wear, hardness and fatigue resistance of ferrous and non-ferrous materials. For that purpose, plasmageneration is most frequently ensured by d.c. glow discharges at relatively high pressure (100–1000 Pa) with the underlying limitationsassociated with this technology. Nevertheless, more flexibility and control are required for plasma nitriding of promising non-ferrousmaterials such as titanium, aluminium and their alloys. These requirements are fulfilled by the recently developed enhanced or intensifiedplasma nitriding processes that operate at lower pressures (,10 Pa) such as: thermionically assisted d.c. triode arrangements (TAT),plasma immersion ion implantation (PIII) or plasma source ion implantation (PSII), electron cyclotron resonance (ECR) systems andthermionic arc discharges (TAD). The purpose of this paper is to review these new nitriding processes from both technological andfundamental points of view. Plasma parameters and plasma–surface interactions are considered for these processes. 1998 ElsevierScience S.A. All rights reserved.

Keywords: Low-pressure /high-density plasma-assisted nitriding processes; Surface modification; Plasma-surface interactions; Diagnostics

1. Introduction processes which operate at pressure less than 10 Pa. Westart with an analysis of the plasma–surface interactions

Plasma nitriding, also called ion nitriding is now widely relevant for PAN processes. The currently used low-pres-used for surface hardening of ferrous and non-ferrous sure PAN processes are then described in section III. Thesematerials in the manufacturing industry [1–5]. The recent processes are then analysed in terms of the plasmadevelopment of a new generation of low-pressure, high- parameters relevant to them in section IV. A tentativedensity plasma sources in the electronics industry [6] has model of these low-pressure PAN processes is finally givenled to the development of an enhanced plasma-assisted in the last part of this paper.nitriding (PAN) concept. A common feature of these high-density sources is the uncoupling of the plasma generationfrom the substrate surface, often even from the section of 2. Plasma–surface interactions relevant for plasma-the reactor loaded with the substrates. In this case, the assisted nitriding at low pressureplasma generated by a ‘remote’ source are fed into theprocess chamber housing the substrates. Independent sub- The analysis of the interactions of the numerous speciesstrate biasing provides a means for independent ion flux found in a nitriding plasma with a solid surface is a veryand ion energy control. So far, this new PAN concept has complex problem. Among the large number of possiblebeen realised with four different configurations: therm- interactions, the following have been identified as relevantionically assisted d.c. triode (TAT) arrangements, plasma to PAN processes: (1) ion implantation and radiationimplantation (PI), electron cyclotron resonance (ECR) damage; (2) adsorption of nitrogen species; (3) sputteringsystems and thermionic arc discharges (TAD). and ion induced desorption of particles; (4) recombination

The aim of this paper is to review these new nitriding of nitrogen atoms.The first two mechanisms are source terms for nitriding,

* whilst the latter two mechanisms are loss terms forCorresponding author. Tel.: 133-3-83584252; fax: 133-3-83534764;e-mail: [email protected]. nitriding. These effects can occur simultaneously and in

0257-8972/98/$ – see front matter 1998 Elsevier Science S.A. All rights reserved.PI I : S0257-8972( 98 )00555-6

T. Czerwiec et al. / Surface and Coatings Technology 108 –109 (1998) 182 –190 183

favourable cases result in a net nitrogen mass transfer from implantation by the fact that initial nitride formation takesthe plasma to the solid. place at the surface followed by diffusion of the nitrogen

to the bulk in contrast to ion implantation, where thenitride formation takes place throughout the superficial

2.1. Ion implantation and radiation damage layer affected by ion implantation [12,13,15].Tibbetts [16] was the first to come to the conclusion that

1 1In the process of nitrogen ion implantation, N or N2 nitrogen atoms are the active species in PAN processes. Heions enter the solid provided that their kinetic energies are observed nitriding of samples protected against nitrogensufficient (.1 keV). In the solid, these incident ions ion bombardment by a biased grid. These results were laterundergo electronic and core collisions. These result in confirmed by Marchand et al. [17] and Ricard et al. [18],energy loss to the free electrons and lattice atoms called who came to the conclusion that not only are nitrogenelectronic and nuclear stopping [7]. Energy loss is fol- atoms the active species in plasma nitriding, but the

1lowed by trapping of the nitrogen species. N ions can2 nitrogen ions and the vibrational excited nitrogen mole-also undergo collision induced dissociation if their kinetic cules N (X,v) formed in the glow discharge play probably2energy lies sufficiently above the limit for molecular only the role of precursors in the nitrogen atom formation.dissociation (9.7 eV). At kinetic energies exceeding 100 However, we have recently shown that the metal surfaceeV, the collision-induced dissociation processes dominate nitriding in a flowing post-discharge reactor is caused by N

1and almost all primary N ions will end up as atoms [8].2 atoms and not by N (X,v) molecules [19].21In the case of plasma implantation, the dose for N222implantation, D (atoms cm ) is given by [9]:

2.3. Sputtering and ion-induced desorption of particles2]D 5 E j dt (1)i Physical ejection of particles from a solid under ione

bombardment (sputtering) is treated extensively in theliterature [7,20]. Two reactions are relevant for PAN: ionwhere e is the charge on the electron and j is the currenti bombardment-induced desorption of adsorbed layers,density of the molecular nitrogen ions. 21which takes place for ion energies in the range of 10 toThe nuclear collisions displace atoms from their lattice10 eV; sputtering, which becomes important at ion energiessites creating interstitial vacancy pairs. Accumulation ofin excess of the surface binding energy (at least 50 eV forthese Frenkel defects and of the implanted nitrogen, whichmost metals and alloys). Sputtering and/or ion-inducedin many cases will form Schottky defects, can lead to andesorption during PAN can improve the nitriding rate if theamorphisation of the regions adjacent to the surface.material to be treated has strong tendency to form oxidesTheoretical estimates of the range of implanted ions andwith residual oxygen in the reactor, e.g. the case ofdamage distribution can be obtained straightforwardly withaluminium nitriding [21].computer codes such as TRIM (transport of ions in

The importance of sputtering in d.c. diode nitriding hasmatter): 0.8 and 3.5 nm are the mean implantation depths1 been emphasised many times [5,17]. This reaction isobtained with TRIM for N ions incident on titanium at

incorporated in the numerical models of plasma nitriding100 eV and 1 keV, respectively, [10]. When the incident ioncurrently under development [22–24].energy is higher than 10 keV a rough estimate of 2 nm.

21keV can be made for the mean depth of ion implantation[10,11].

2.4. Recombination of nitrogen atoms

2.2. Adsorption of nitrogen species Surface recombination of nitrogen atoms is a further lossmechanism in PAN processes, as suggested by Renevier et

Several nitrogen adsorption reaction paths have been al. [25].Very little is known on the heterogeneous recombi-propounded to explain PAN processes. Physisorption of nation probability g of nitrogen atoms on surfaces relevantnitrogen molecules on metal surfaces followed by low to PAN, in particular its temperature dependence [26].energy (0.3–8 keV) ion bombardment-induced surface Work has been undertaken in our laboratory [27] tonitride formation has been claimed to be very efficient for measure g as a function of temperature on differenttransition metals. This mechanism was reported by Baba surfaces by a combination of optical emission spectroscopy

1and co-workers [12,13] to explain their data on Ar ion measurement and two-dimensional hydrodynamic, thermalbombardment of transition metals exposed to nitrogen with and kinetic modelling of Ar–N microwave post-dis-2

23a partial pressure of 10 Pa, and by Lancaster and charges. Furthermore, superficial destabilization of a ni-1Rabalais [14] who used N ions at the same nitrogen trided layer has been observed in low-temperature argon2

pressure. This mechanism of ion-induced chemical re- plasmas [28]. This has been explained by nitrogen re-actions differs from the chemical modification by ion trodiffusion from the solid phase to the gaseous phase.

184 T. Czerwiec et al. / Surface and Coatings Technology 108 –109 (1998) 182 –190

2.5. Diffusion of nitrogen atoms into the metal lattice 3.2. Plasma immersion ion implantation (PIII) orplasma source ion implantation (PSII)

The mechanisms of nitrogen species dissolution in the3bulk to form interstitial nitrogen atoms are not known at Plasma immersion ion implantation (PIII or PI ) or

present. Some works have been published on single-crystal plasma source ion implantation (PSII) are new hybridFe surfaces [29,30]. After the nitrogen atoms transfer into technologies combining elements from ion implantationthe surface layer, the nitrogen atom concentration gradient with elements from conventional ion nitriding for theconstitutes the driving force for the interstitial diffusion of surface modification of materials. Both designations arenitrogen in the solid bulk. The interstitial diffusion of the currently used in the literature to designate similar pro-N atom is temperature dependent. The activation energies cesses developed independently in Australia [9,48–57] andfor the diffusion in many compounds are known. The in the USA [58,59] over the last 10 years. Plasmadiffusion of nitrogen from the surface layer towards the Implantation (PI) will be used in this paper as designationbulk is accompanied by precipitation reactions. In some for both technologies. In PI, the substrate to be nitrided ismaterials they can be limited to the alloying elements immersed in a plasma, and ions are extracted directly frompresent. the plasma and accelerated by applying a series of negative

high-voltage triode pulses (10–50 kV; pulse length, 2–100ms; repetition rate, 100–200 Hz) to the substrate. A plasmasheath forms around the target and ions bombard the target

3. Low-pressure plasma-assisted nitriding techniques normal to the surface, thus minimising material sputteringcurrently employed caused by oblique incidence in diode nitriding. The pulsing

of the voltage is essential to achieve spatial uniformity andOne of the first realisations of low-pressure PAN was a uniform implantation depth. The pulse width must be

made by Brokman and Tuler [31,32]. They used a d.c. chosen short enough to limit the plasma sheath expansiondiode discharge in an external magnetic field to increase during the voltage pulse and to prevent the sheath fromthe electron path and thus to maintain a discharge at low interacting with the wall of the processing chamber. Apressure (1–10 Pa). The currently used plasma-assisted complete description of the experimental results and atechniques for low-pressure nitriding (,10 Pa) can be review of theoretical treatments of sheath dynamics can beclassified under the following four categories (Fig. 1): (1) found in Refs. [6,57]. In PSII, the plasma is generated by athermionically assisted d.c. triode (TAT); (2) plasma conventional filament discharge (Section 3.1), whereas it isimmersion ion implantation (PIII) or plasma source ion created by an inductively coupled radio-frequency dis-implantation (PSII); (3) electron cyclotron resonance charge in PIII. Currently, several industrial PI prototypes(ECR) systems; (4) thermionic arc discharge (TAD). are under development [60,61].

The main characteristics of these processes are compiled A new modified PI device can be used for implantationin Table 1. of inner surfaces [62]. A metal anode connected with the

chamber wall is set up inside a cylindrical aluminiumtarget. Large size, scalable plasmas up to square or cubic

3.1. Thermionically assisted d.c. triode (TAT) metres are a necessary criteria for industrial development.Such plasma dimensions can be obtained by using uni-

Adding a thermionic electron source to a conventional formly distributed (UD) ECR sources (see Section 3.3,diode system leads to a ‘triode’ plasma, with a thin cathode below).sheath at the substrate surface. One of the earliest success-ful applications of the TAT process was reported by 3.3. Electron cyclotron resonance (ECR) systemsKorhonen and co-workers [33,34].

The great advantage of this process lies in the possibility ECR employs the coupling between an injected micro-22of adjusting the substrate current density (2–4 mA cm ) wave (radian frequency v 52.45 GHz) and an axially

independently from the substrate potential by changing the varying magnetic field B that locally fulfils the electrondiode parameters of the electron emitter. This discharge cyclotron resonance frequency v 5eB /m , where m isce e e

can be operated at low pressures (1–10 Pa in N or Ar–N the electron mass. For microwaves at v 52.45 GHz, the2 2

gas mixtures), because the supply of electrons is no longer required magnetic field for resonance is 875 G. Although adependent on the rate of ionic bombardment of the great number of configurations achieve this basic feature,cathode. However, electrostatic probe [35] and optical two different implementations in PAN can be distinguishedemission spectroscopy (OES) [36] measurements reveal [63]. The first uses an axial magnetic field [64–70], whilenon-uniform distribution of plasma density for this pro- the second uses a multipolar magnetic confinementcess. Most of the work on TAT nitriding was performed by [71,72].two groups: Matthews and co-workers [15,36–42] and In axial magnetic field ECR devices, the microwaveMeletis and co-workers [43–47]. power is injected axially through a dielectric window into


Fig. 1. Schematic view of a thermionically assisted d.c. triode (a), plasma immersion ion implantation or plasma source implantation (b), axial magneticfield electron cyclotron resonance (c) and thermionic arc discharge (d) for plasma-assisted nitriding.

a chamber, where it excites the right-hand part of a towards the distributed ECR (DECR) and finally thecircularly polarized wave that propagates to a resonance uniform DECR (UDECR) concepts [71]. An array of 24zone, created by one or several magnetic coils, where the tubular magnets, microwave power feeds and wave prop-wave is absorbed [6,73,74]. As a result, electrons gain a agators are positioned in between each magnet pair leadingsufficient energy to allow ionisation of the background gas to a peripheral plasma source that produces a uniformat pressures less than 1 Pa. This plasma streams or diffuses distributed plasma. In this configuration, the substrates canalong the magnetic field lines into a processing chamber be biased independently by applying d.c. or r.f. voltages.with the substrates [64–70]. Combined with low pulsednegative bias (0.4–3 keV; pulse length, 50–500 ms; 3.4. Thermionic arc discharge (TAD)repetition rate, 100–200 Hz) applied to the substrate [64–67] and an auxiliary heater, this technique is known as The first results obtained by an arc-assisted nitriding‘plasma source ion nitriding’. process have been reported by Bergmann and Hummer

Microwave plasma excitation in a multipolar magnetic [76] and by D’Haen et al. [77,78] in the late 1980s. Theseconfinement, was introduced by Pelletier, Arnal and co- authors have used a thermionic arc of the type widely usedworkers [75]. This technique has been further developed in activated reactive ion plating processes (ARIP) [79–81].


Table 1Plasma parameters for plasma assisted nitriding processes at low pressure

Thermionically assisted Plasma implantation systems Axial magnetic field ECR UDECR devices Arc discharges

d.c. triode systems

discharges

Pressure (Pa) 0.5–15 0.01–1 0.1–10 0.1–0.2 0.4–4

Gas mixture Ar–x%N N N N Ar–x%N2 2 2 2 2

20#x#100 20#x#55

Substrate bias (V) 290→2200 210 000→250 000 Floating→22000 Floating→245 000 Floating→280 V

Type of substrate bias D.c. Pulsed:(pulsed D.c. or pulsed: (pulsed Pulsed: (pulsed D.c.

length 2–100 ms, length 50–500 ms, length up to 100 ms,

repetition rate 0–400 Hz) repetition rate |100 Hz) repetition rate up to 10 Hz)

Substrate current 0.2–3 1–4 0.44–0.6322density (mA cm )

Plasma source Not precisely described 200–300 W r.f. power Microwave power Microwave power, Arc voltage, 25–40 V,

parameters at 13.56 Hz in PIII 100–300 W 1300 W Arc current, 100–300 A

Electronic temperature, 2.6–4 1–10 0.1–10 1.2 3–4

T (eV)e8 9 9 10 10 12 10 10 11Electronic density, 10 –5310 10 –10 10 –10 2.10 10 –10

23n (cm )e

Floating and plasma 10#V #16 1#V #3 V 51 V 57 and V 510 V 515p p p f p p

potential, V –V (V)f p

Starting from the same ARIP device as Bergmann and synthesis. In the PAN reactor, an additional resistive heaterHummer, Renevier et al. [21,25,82,83] have designed a produces a uniform temperature over a large pay load. Thisnew PAN reactor. A high-current (100–300 A), low-volt- technique has been called the ‘arc-assisted nitriding’age (25–40 V) TAD is generated in argon in an ionization process. The fact that nitriding in this process is caused bychamber mounted on the top of the nitriding reactor itself. nitrogen atoms, rather than nitrogen ions is confirmed byWhile in an ARIP process where the arc plasma is more or the observation that the substrate bias potential has no orless confined to an anodic crucible and all substrates are insignificant influence on the nitriding rate. Therefore theplaced off the beam centre, where the plasma density is substrates can be left floating. The low plasma potential oflevelling off, in the PAN reactor one tries to spread the this discharge (25–15 V) ensures that sputtering isplasma emanating from the ionisation chamber uniformly negligible on the substrate surface. This feature allows, forover the nitriding chamber. This is achieved by suitable the first time, nitriding without degradation of the surfacemagnetic field configurations and segmented anodes lining finish and is the key to integrated duplex processes. Thethe reactor wall. Substrates are loaded freely between process has been extended to cold cathode arcs [86,87],anodes and the ionisation chamber(s). These segmented where it has been called ‘the arc-enhanced glow discharge’anodes and the hot filaments in the ionisation chamber (AEGD). PAN has since been performed in a cathodic arccreate a uniform low-pressure plasma in the reactor (0.4– evaporation device [88].0.8 Pa). The mechanisms in this type of discharge can beanalysed simply in terms of dissipated energy [84]. Rough-ly 2 kW are used to extract the electrons from the

4. Plasma parameters relevant in low-pressureionisation chamber. The remainder is shared between the

plasma-assisted nitridinganode fall and the plasma column. The sharing is de-termined by the magnetic field configuration, which is

It seems that each group was successfully guided by itsvariable. The energy of the anode fall can be used to heat

own pet model about nitriding species and nitridingdirectly the substrates or to heat the chamber linings for

processes; we will try in this section to analyse theindirect substrate heating [85]. This latter configuration is

different PAN processes with respect to the physical andused in PAN. Most of the energy dissipated in the positive

chemical characteristics of their plasmas: ionisation, nitro-column is used to dissociate nitrogen. Indeed, it has been

gen atom production, substrate bias, etc.shown in the context of CVD synthesis of diamond, wherethis same type of plasma is used successfully, that dis-sociation of binary molecules is the main mechanism of 4.1. Nitrogen atoms production in PAN processescooling of an arc. This energy is then released on thesubstrates or the walls, where the atoms recombine. This At high pressure (P.100 Pa), vibrational N (X,v)2

dissociation /cooling cycle produces the copious flow of molecules would contribute to nitrogen dissociation bynitrogen atoms impinging on the alloy surfaces for nitrid- vibration–vibration (V–V) and vibration–translation (V–ing, and the copious flow of hydrogen atoms for diamond T) energy exchange processes [89]:


1N (X,v) 1 N (X,v 5 45) → N (X,v 2 1) 1 2N (reaction 1) N generation in PAN processes occurs by the follow-2 2 2 2

ing reaction whose cross-section is shown in Fig. 2 [92]:N 1 N (X,v) → N 1 2N (reaction 2)2 2 2 1e 1 N → 2e 1 N (reaction 8)2 2but these processes are only of importance at high pressure(P.100 Pa). 4.2. Ionisation in PAN processes

In the d.c. cathode sheath [36,37,39], the dissociativecharge exchange mechanism is important: In PAN processes, electrons are instrumental for the

1 1 production of N atoms and ionic species. The rate constantN 1 N → N 1 N 1 N (reaction 3)2,fast 2,slow 2,fast slow slow K5ksv l for electron particle collisions is given by [6]:e

The energy threshold for this reaction is around 15 eV [90]. `]2 ]Electron impact dissociation of molecular nitrogen will Œ]ksv l 5 E s(E) E f(E) dE (3)e mproduce nitrogen atoms at low pressures and high plasma œ e

E0densities:Where s and E are, respectively, the cross-section and the0e 1 N → e 1 2N (reaction 4)2energy threshold for the considered reaction and v and me e

1 are, respectively, the electron velocity and mass. Thise 1 N → 2e 1 N 1 N (reaction 5)2equation is only valid for isotropic velocity distributions.

The energy dependence of the cross-sections for reactions The electron velocity contributions in the low-pressure(4) [91] and (5) [92] are shown in Fig. 2. Especially, PAN systems considered in this paper are always highlyreaction (4) should be dominant in low-pressure PAN anisotropic as one exploits always a drifting plasma.processes [11], but the following electron–ion recombina- Therefore electron temperature, T , is not a very meaning-etion reaction is also a possible source of N atoms ful parameter. The fact that we will use it, nevertheless, in

1 the subsequent discussion is simply pragmatic.e 1 N → 2N (reaction 6)2 Using Langmuir probes in an argon triode dischargeThe total rate constant for this reaction is given by [93]: (filament bias current, 0.5–2 A; substrate bias voltage, 0–2

kV) between 0.3 and 2 Pa, Swarnalatha et al. [35] found1 / 23003 21 27 ]k (cm s ) 5 4.810 (2) that the electron density n increases with increasingS D6 eTe filament bias current with a maximum around 0.5 Pa. Greatwhere T is the electron temperature. discrepancies, in particular with respect to the publishede

Argon in nitrogen plasmas may also cause the following electron temperatures, exist in the electron distribution datadissociative charge exchange reaction: for PI [48,95]. It has been demonstrated both by experi-

ments [96] and by simulation [11] that the n and T rise1 1 e eAr 1 N → Ar 1 N 1 N (reaction 7)2 rapidly after switching on the bias voltage. T values ase

high as 15 eV are reached at the onset of the pulse,The energy threshold for this reaction is around 20 eVlevelling off to a stationary value around 1 eV. The results[94]; therefore this reaction would be expected to occurfor the electron distribution parameters in axial magneticunder the same conditions as for reaction (3).field ECR [64,70] are included in Table 1. In UDECRdevices, the plasma characteristics without substrate bias

10 23show a radially uniform plasma with n 510 cm ande

T 51.2 eV [71,72]. The two TAD studies [97–99] at 0.15e10 11 23Pa found T around 3 eV and 10 –10 cm , and that thee

dependence of n on arc current I is stronger than the onee d

of T . But measurements performed by the second deriva-e

tive of Langmuir probe characteristics on a hollow cathodearc (15,I ,400 A, 18,V ,50 V in argon), whered d

similar conditions prevail, show the existence of beam-likeelectrons at 7 cm from the cathode [99,100]. In this case,the mean beam energy appears at 19 eV, close to the firstionization potential of the background gas confirming thenon-maxwellian character of the electron energy distribu-tion function f(E).

Optical emission spectroscopy is a non-intrusive methodthat can be easily set up on any plasma reactor and hasbeen used in d.c. ion-nitriding reactors [5]. The opticalFig. 2. Cross-sections versus electron energy for reaction (4) [91],

reaction (5) [92] and reaction (8) [92]. emission spectrum obtained in PAN reactors is dominated


by the first and second positive systems of the N molecule triode current. At low bias potential, it becomes a zone of21and the first negative system of the N molecular ion ion drift: this is the case in arc and ECR nitriding, which2

[51,70,82,83]. Some atomic N lines have also been iden- work under such high plasma densities, that high voltagetified by Fancey et al. [36,42]. He determined the relative would require very short pulse lengths to avoid overheat-proportions of atomic and molecular nitrogen in the ing.cathode sheath and plasma region of a TAT in Ar–N gas (2) In the PI, the extension and composition of this zone2

mixtures and found that the presence of argon makes a will be time dependent. Ion depletion and ion enrichmentsignificant contribution to the dissociation of nitrogen, by avalanche processes take place.particularly at low nitrogen partial pressures. This was Few studies exist on the composition of this zone for theattributed to a reduction of the quenching of electron situation of plasma nitriding, because it is often difficult toenergy by nitrogen. This phenomenon has also been access for spectroscopic studies. In the solid (a few latticeclearly demonstrated in TAD nitriding [82,83]. The distances from the surface) we will find a region whosequenching of the electron energy by nitrogen results from constitution is at any time determined by diffusion andthe very large cross-sections for inelastic collisions of precipitation phenomena, which have been investigatedelectrons with N molecules [101]. Hence, argon in extensively by standard metallurgical techniques. The2

addition to nitrogen in low-pressure PAN processes can transport envelope and the bulk diffusion zone are sepa-promote nitrogen dissociation by reactions (4) and (5) rated by an ‘interphase’, where the reactions will differ forthrough an increase of the electron mean energy. the four different nitriding routes shown (Fig. 3): (1)

In PI, no reliable method is available for the determi- physisorption route; (2) direct chemisorption route; (3)nation of the impinging ion flux. The total current de- bulk phase dissociation route; (4) ion implantation route.livered by the high-voltage pulse generator is only a We propose to analyse all plasma-nitriding systemsqualitative indication of the ion flux density on the target, under the hypothesis that only these four routes occur, butbecause it also contains the contribution from the electron that the extent of their contribution will vary for differentemission from the substrates. These secondary electrons systems.contribute to a large extent to the total current through the Nitrogen molecules will be adsorbed by the surface tosubstrate and can substantially modify the ion composition form a physisorbed layer. The coverage as a function ofof the nitrogen plasma by reactions as those described in pressure for different temperatures is a family of isotherms,Section 4.1. Therefore, considerable efforts are actually which can and have been determined for most metals.under way to perform time-resolved ion flux measurements They are a measure of the type of interaction between

1in PI processes. Tang et al. [59] have determined N and metal and molecular nitrogen [102]. Suitable models exist1N current measurements in a PI system with a thermionic to describe these interactions in terms of localisation [103].2

21 22filament (10 ,P,10 Pa). At low discharge current We assume that nitrogen molecule diffusion in the metal is1(30 mA) they found that 70–80% of the ions are N . This negligible. Different reaction paths then link this2

percentage is decreasing with an increasing discharge physisorbed nitrogen to a layer of chemisorbed nitrogen:current or an increasing filament heating current. However, unimolecular decomposition, catalytic decomposition, plas-

1 1Le Coeur et al. [72] have found a N /N ratio of 7 /3 in ma-induced dissociation [12–14]. Direct chemisorption2

an UDECR device at 0.13 Pa.

5. A simple model for plasma-assisted nitriding

In view of the results and observations described inSection 2 Section 3 Section 4, we tentatively propose thefollowing model for PAN. In the bulk plasma we have acertain composition of molecular, atomic and ionic species,which depends on the plasma condition. This distributionwill be modified in a zone adjacent to the surface as afunction of the transport mechanism and reaction with thesecondary electrons. The following modifications occur inthe different processes:

(1) In all low-pressure triode arrangements, i.e. TATand TAD and ECR, the composition of this zone can bemodulated by the bias potential. At high bias potential, itcan become a zone of increased ion density; this seems tobe the case in many triode ion-nitriding systems. In factthis corresponds to abnormal triode operation; the correct Fig. 3. Schematic representation of the model proposed for plasma-approach would be to increase the diode current to increase assisted nitriding.


[2] W. Rembges, J. Luhr, in: T. Spalvins, W.L. Kovacs (Eds.) Proc. ofand atomic adsorption produce chemisorbed nitrogenthe 2nd Int. Conf. on Ion Nitriding/Carburizing, Cincinnati, OH,directly. In many metals, in particular those forming18–20 September 1989, p. 147.interstitial nitrides, one should consider at least formally

[3] R. Grun, H.J. Gunther, Mater. Sci. Eng. A 140 (1991) 435.another phase between chemisorbed nitrogen atoms at the [4] F. Hombeck, in: T. Spalvins, W.L. Kovacs (Eds.), Proc. of the Int.surface and the nitrogen diffusing in the solid: absorbed Conf. on Ion Nitriding, Cincinnati, OH, 15–17 September 1986, p.

169.nitrogen atoms. The activation energy for the transfer of[5] H. Michel, T. Czerwiec, M. Gantois, D. Ablitzer, A. Ricard, Surf.nitrogen atoms from the surface to the interstitial sites in

Coat. Technol. 72 (1995) 103.the metal will not be zero in all cases. Rather it seems, that[6] M.A. Lieberman, A.J. Lichtenberg, Principles of Plasma Dischargesin systems operating with an ample supply of atomic

and Materials Processing, Wiley Interscience, New York, 1994.nitrogen, this becomes the rate-determining step. Details [7] R.G. Wilson, G.R. Brewer, Ion Beam and Applications to Ionwill depend on how well the lattice of the chemisorbed Implantation, Wiley-Interscience, New York, 1973.nitrogen matches the lattice of interstitial nitrogen. Ion [8] H.K. Hu, P.T. Murray, Y. Fukuda, J.W. Rabalais, J. Chem. Phys. 74

(1981) 2247.implantation produces absorbed nitrogen directly. We[9] G.A. Collins, R. Hutchings, K.T. Short, J. Tendys, C.H. Van Derassume that the absorbed nitrogen is the starting point for

Valk, Surf. Coat. Technol. 84 (1996) 537.the nitrogen diffusion. In view of the material published on[10] P. Martin, IEEE Trans. Plasma Sci. 18 (1990) 855.

the different processes, one would be tempted to propose [11] M.J. Bear, J.U. Guillory, J. Vac. Sci. Technol. A 15 (1997) 1963.the following conclusions: gas nitriding proceeds exclu- [12] Y. Baba, T.A. Sasaki, Mater. Sci. Eng. A 115 (1989) 203.sively via the physisorption route. Diode nitriding may [13] T.A. Sasaki, Y. Baba, H. Yamamoto, Vacuum 41 (1990) 185.

[14] G.M. Lancaster, J.W. Rabalais, J. Chem. Phys. 83 (1979) 209.combine contributions from all four routes. TAD, AEGD,[15] K.S. Fancey, A. Leyland, D. Egerton, D. Torres, A. Matthews, Surf.ECR and post-discharge nitriding proceed probably mainly

Coat. Technol. 76–77 (1995) 694.via the bulk phase dissociation route. But the only direct[16] G.G. Tibbetts, J. Appl. Phys. 45 (1974) 5072.evidence which may rule out the chemisorption route are[17] J.L. Marchand, D. Ablitzer, H. Michel, M. Gantois, in: T. Spalvins,

experiments which discount the effects of vibrational W.L. Kovacs (Eds.) Proc. of the 2nd Int. Conf. on Ion Nitriding/excitation in post-discharges [19]. PI has certainly a large Carburizing, Cincinnati, OH, 18–20 September 1989, p. 67.contribution from the implantation route. Although de- [18] A. Ricard, J.E. Oseguera-Pena, L. Falk, H. Michel, M. Gantois,

IEEE Trans. Plasma Sci. 18 (1990) 940.tailed proof is missing, one would assume the following[19] A. Ricard, T. Czerwiec, T. Belmonte, S. Bockel, H. Michel, in:ranking with respect to the rate for the different routes:

Proc. of the 1st Asian–European Int. Conf. on Plasma Surf.implantation.bulk phase dissociation.chemisorption.Engineering, Seoul, South Korea, 5–9 October 1997, p. 98.

physisorption. [20] G. Carter, J.A. Colligon, Ion Bombardment of Solids, Heimann,London, 1968.

[21] N. Renevier, T. Czerwiec, P. Collignon, H. Michel, Le Vide Sci.Tech. Appl 284 (1997) 11.6. Conclusions

[22] V.I. Dimitrov, J. d’Haen, G. Knuyt, C. Quaeyhaegens, L.M. Stals,Appl. Phys. A 63 (1996) 475.

We propose to reduce all variants of plasma-assisted [23] Y. Sun, T. Bell, Mater. Sci. Eng. A 224 (1997) 33.nitriding processes to a mix of four nitriding routes: [24] Y. Sun, T. Bell, Heat Treatment Metals 2 (1997) 43.physisorption, direct chemisorption, bulk phase dissocia- [25] N. Renevier, T. Czerwiec, H. Michel, P. Collignon, Surf. Coat.

Technol. in press.tion and ion implantation.[26] A. Ricard, A. Popescu, A. Baltog, C.P. Lungu, Le Vide 280 (1996)The available data do not yet allow the analysis of the

248.different methods in terms of contributions from these four `[27] T. Belmonte, L. LeFevre, T. Czerwiec, H. Michel, A. Ricard, in:routes, or even an unambiguous comparison of the differ- Proc. of the 1st Asian–European Int. Conf. on Plasma Surf.ent methods in terms of plasma parameters. We propose Engineering, Seoul, South Korea, 5–9 October 1997, p. 18.

[28] N. Dingremont, A. Pianelli, E. Bergmann, H. Michel, Surf. Coat.that future research should be directed in this direction.Technol. 61 (1993) 187.This will also improve the analysis of the plasma-nitriding

[29] F. Bozso, G. Ertl, M. Grunze, M. Weiss, Catalysis 49 (1977) 18.processes of different materials. [30] G. Ertl, S.B. Lee, M. Weiss, Surf. Sci. 114 (1982) 515.

[31] A. Brokman, J. Vac. Sci. Technol. 17 (1980) 657.[32] A. Brokman, F.R. Tuler, J. Appl. Phys. 52 (1981) 468.[33] A.S. Korhonen, E.H. Sirvio, Thin Solid Films 96 (1982) 103.Acknowledgements[34] A.S. Korhonen, E.H. Sirvio, M.S. Sulonen, Thin Solid Films 107

(1983) 387.The authors wish to express their appreciation to M. [35] M. Swarnalatha, C. Sravani, K.R. Gunasekhar, G.K. Muralidhar, S.

Renevier and P. Collignon (Balzers) for their contribution Mohan, Vacuum 48 (1997) 845.to this paper. [36] K.S. Fancey, A. Matthews, Vacuum 41 (1990) 2196.

[37] K.S. Fancey, A. Matthews, Surf. Coat. Technol. 33 (1987) 17.[38] K.S. Fancey, A. Matthews, IEEE Trans. Plasma Sci. 18 (1990) 869.[39] A. Leyland, K.S. Fancey, A.S. James, A. Matthews, Surf. Coat.

References Technol. 41 (1990) 295.[40] A. Leyland, K.S. Fancey, A. Matthews, Surf. Eng. 7 (1991) 207.

[1] W.L. Kovacs, in: T. Spalvins, W.L. Kovacs (Eds.), Proc. of the 2nd [41] A. Leyland, D.B. Lewis, P.R. Stevenson, A. Matthews, Surf. Coat.Int. Conf. on Ion Nitriding /Carburizing, Cincinnati, OH, 18–20 Technol. 62 (1993) 608.September 1989, p. 5. [42] K.S. Fancey, Vacuum 46 (1995) 695.


[43] S. Yan, E.I. Meletis, Surface Modification Technologies IV, TMS- [73] J. Margot, T.W. Johnston, J. Musil, in: M. Moisan, J. Pelletier (Eds.),AIME, Warrendale, PA, 1991, pp. 445–460. Principles of Magnetically Assisted Microwave Discharges in

[44] E.I. Meletis, S. Yan, J. Vac. Sci. Technol. A 9 (1991) 2279. Microwave Excited Plasmas, Elsevier, Amsterdam, 1992, p 181.[45] T.M. Muraleedharan, E.I. Meletis, Thin Solid Films 221 (1992) 104. [74] J. Musil, Vacuum 47 (1996) 145.[46] E.I. Meletis, S. Yan, J. Vac. Sci. Technol. A 11 (1993) 25. [75] Y. Arnal, J. Pelletier, C. Pomot, B. Petit, A. Durandet, Appl. Phys.[47] A.A. Adjaottor, E. Ma, E.I. Meletis, Surf. Coat. Technol. 89 (1997) Lett. 45 (1984) 132.

197. [76] E. Bergmann, E. Hummer, Verfahren zur thermochemischen Ober-[48] J. Tendys, I.J. Donnelly, M.J. Kenny, T.J. Pollock, Appl. Phys. Lett. flachenbehandlung von Werkstoffen in einem reaktiven Gasplasma,

53 (1988) 2143. Patent FR 2600082, GB 2192196, US 4762756, DE3702984, CH[49] G.A. Collins, R. Hutchings, J. Tendys, Mater. Sci. Eng. A 139 671407, SE 8702410-5, NL 465578, JP 93484/87, 1986.

(1991) 171. [77] J. D’Haen, C. Quaeyhaegens, G. Knuyt, L. De Schepper, L.M. Stals,[50] R. Hutchings, G.A. Collins, J. Tendys, Surf. Coat. Technol. 51 M. Van Stappen, Surf. Coat. Technol. 60 (1993) 468.

(1992) 489. [78] J. D’Haen, C. Quaeyhaegens, G. Knuyt, M. D’Olieslaeger, L.M.[51] G.A. Collins, R. Hutchings, J. Tendys, Surf. Coat. Technol. 59 Stals, Surf. Coat. Technol. 74–75 (1995) 405.

(1993) 267. [79] E. Moll, E. Bergmann, Surf. Coat. Technol. 37 (1989) 483.[52] M. Samandi, B.A. Shedden, D.I. Smith, G.A. Collins, R. Hutchings, [80] E. Moll, R. Buhl, E. Bergmann, Surf. Coat. Technol. 39–40 (1989)

J. Tendys, Surf. Coat. Technol. 59 (1993) 261. 475.[53] M. Samandi, B.A. Shedden, T. Bell, G.A. Collins, R. Hutchings, J. [81] E. Bergmann, Surf. Coat. Technol. 57 (1993) 133.

Tendys, J. Vac. Sci. Technol. B 12 (1994) 935. [82] N. Renevier, P. Collignon, H. Michel, T. Czerwiec, Surf. Coat.[54] G.A. Collins, R. Hutchings, J. Tendys, M. Samandi, Surf. Coat. Technol. 86–87 (1996) 285.

Technol. 68–69 (1994) 285. [83] N. Renevier, T. Czerwiec, P. Collignon, H. Michel, Surf. Coat.[55] G.A. Collins, J. Tendys, Plasma Sources Sci. Technol. 3 (1994) 10. Technol. 98 (1998) 1400.[56] G.A. Collins, R. Hutchings, K.T. Short, J. Tendys, X. Li, M. [84] E. Moll, US 4555611, 1982.

Samandi, Surf. Coat. Technol. 74–75 (1995) 417. [85] E. Bergmann, Proc. 5th Int Symp. on Trends and Applications in[57] S. Leigh, M. Samandi, G.A. Collins, K.T. Short, P. Martin, L. Thin Films (1–3 April 1996, Colmar, France), Le Vide Sci. Tech.

Wielunski, Surf. Coat. Technol. 85 (1996) 37. Appl. 279 (1996) 1.[58] J.R. Conrad, J.L. Radtke, R.A. Dodd, F.J. Worzala, N.C. Tran, J. [86] J. Vetter, A.J. Perry, Surf. Coat. Technol. 61 (1993) 305.

Appl. Phys. 62 (1987) 4591. [87] J. Vetter, T. Wallendorf, Surf. Coat. Technol. 76–77 (1995) 322.[59] B.Y. Tang, R.P. Fetherston, M. Shamim, R.A. Breun, A. Chen, J.R. [88] F. Sanchette, E. Damond, M. Buvron, L. Henry, P. Jacquot, N.

Conrad, J. Appl. Phys. 73 (1993) 4176. Randall, P. Alers, Surf. Coat. Technol. 94–95 (1997) 261.[60] C. Blawert, B.L. Mordike, Surf. Coat. Technol. 93 (1997) 274. [89] J. Loureiro, Chem. Phys. 157 (1991) 157.[61] C. Blawert, B.L. Mordike, Nucl. Instrum. Methods Phys. Res. B [90] A.V. Phelps, J. Phys. Chem. Ref. Data 20 (1991) 557.

127–128 (1997) 873. [91] P.C. Cosby, J. Chem. Phys. 98 (1993) 9544.[62] M. Sun, S.Z. Yang, B. Li, Nucl. Instrum Methods Phys. Res. B 111 [92] A. Crowe, J.W. McConkey, J. Phys. B 6 (1973) 2108.

(1996) 187. [93] I.A. Kossyi, A.Y. Kostinsky, A.A. Matveyev, V.P. Silakow, Plasma[63] S.R. Rossnagel, in: J.Vossen W. Kern (Eds.), Thin Film Processes II, Sources Sci. Technol. 1 (1992) 207.

New Jersey, Academic Press, Orlando, FL, 1991,p. 11. [94] W.B. Maier II, J. Chem. Phys. 41 (1964) 2174.[64] M.K. Lei, Z.L. Zhang, J. Vac. Sci. Technol. A 13 (1995) 2986. [95] Y. Chen, Y. Shi, H. Xie, Z. Wu, X. Jiang, T. Bell, H. Dong, Surf.[65] M.K. Lei, Z.L. Zhang, J. Mater. Sci. Lett. 16 (1997) 1567. Eng. 12 (1996) 137.[66] M.K. Lei, Z.L. Zhang, J. Vac. Sci. Technol. A 15 (1997). [96] J. Brutscher, R. Gunzel, W. Moller, Surf. Coat. Technol. 93 (1997)[67] M.K. Lei, Z.L. Zhang, Surf. Coat. Technol. 91 (1997) 25. 197.[68] P. O’Keeffe, H. Mutoh, S. Den, Y. Hayashi, S. Komuro, T. [97] M. Elmiger, E. Bergmann, Surf. Coat. Technol. 61 (1993) 300.

Morikawa, Thin Solid Films 281–282 (1996) 102. [98] M. Nesladek, C. Quaeyhaegens, S. Wouters, L.M. Stals, E. Berg-[69] P. O’Keeffe, K. Yamakawa, H. Mutoh, S. Den, Y. Hayashi, Jpn. J. mann, G. Rettinghaus, Surf. Coat. Technol. 68–69 (1994) 339.

Appl. Phys. 36 (1997) 4576. [99] A. Lunk, R. Basner, Mater. Sci. Eng. A 139 (1991) 41.[70] T. Hino, I. Fujita, M. Nishikawa, Plasma Sources Sci. Technol. 5 [100] A. Lunk, Vacuum 41 (1990) 1965.

(1996) 424. [101] Y. Itikawa, M. Hayashi, A. Ichimura, K. Onda, K. Sakimoto, K.[71] F.Y. Arnal, R.R. Burke, O. Lesaint, J. Pelletier, Surf. Coat. Technol. Takayanahi, M. Nakamura, H. Nishimura, T. Takayanagi, J. Phys.

93 (1997) 265. Chem. Ref. Data 15 (1986) 985.[72] F. Le Coeur, T. Lagarde, J. Pelletier, Y. Arnal, M. Brusel, R.R. [102] E. Bergmann, J. Phys. Chem. 78 (1974) 405.

Burke, Rev. Sci. Instrum. 69 (1998) 1. [103] R. Tenne, E. Bergmann, Phys. Rev. B 20 (1980) 702.

Documents

Low-pressure, high-density plasma nitriding: mechanisms, technology and results