The structure of nitrided iron-titanium alloys

The structure of nitrided iron-titanium alloys

THE STRUCTURE OF NITRIDED IRON-TITANIUM ALLOYS D. H. JACK Department of Metallurgy, The Cm\ersity of Leeds. England Abstract-The structure of gas-nit...

1MB Sizes 4 Downloads 20 Views

THE STRUCTURE OF NITRIDED IRON-TITANIUM ALLOYS D. H. JACK Department of Metallurgy, The Cm\ersity of Leeds. England

Abstract-The structure of gas-nitrided dilute Fe-Ti alloys has been investigated by electron microscopy, electron and X-ray diffraction, and measurements of nitrogen content. The observations are interpreted in terms of a periodic array of equiatomic Ti-N mono-layer plates on the cube planes of the matrix. The plates distort the surrounding matrix and allow the total accommodation of up to three times the nitrogen content appropriate to ‘TiS’ stoichiometry. The adsorption enthalpy is - 25 kcal mol- ’ and the excess nitrogen can be removed by hydrogen reduction at 575’C or heating in an inert atmosphere at 75O’C and above. After hydrogen reduction the excess nitrogen can be replaced by re-nitriding. The structure is resistant to coarsening. and only after several hours at 825’C does appreciable softening and plate growth occur. The model of mono or multi-layer substitutional atom-nitrogen plates is considered to be a special Ease of a mixed substitutional-interstitial atom cluster. but it can be extended to other nitriding systems and is compatible with previous observations. R&sum&On a Ctudii- la structure d’alliages dilues Fe-Ti nitruti-s B l’azote gazeux, par microscopic Clectronique, diffraction d’4ectrons et de rayons X et mesure de la teneur en azote. Les observations peuvent etre interprettes g partir d’un arrangement pkriodique de plaques monocouches de composition tquiatomique suivant les faces de la maille cubique de la matrice. Les plaques dCforment la matrice et permettent une accommodation totale allant jusqu’8 trois fois la teneur en azote correspondant Q la .stoechiomitrie ‘TIN’. L’enthalpie d’absorption est 6gale g -25 kcal mol-’ et l’on peut retirer I’azote en cxcts par reduction sous hydrogbne & 575’C ou par chauffage en atmosphere inerte a 75O’C et au dessus. Apres riduction par l’hydrog&e, on peut rtintroduire un exces d’azote par renitruration. La structure est assez stable vis a vis d’un grossissement, et ce n’est qu’aprts plusieurs heures B 825’C qu’il se produit un adoucissement et une croissance des plaques appriciables. On considtre le modele des plaques atome substitutionnel-azote d’une ou de plusieurs couches atomiques comme un cas particulier d’amas mixte substitutionnel-interstitiel, mais on peut l’etendre g d’autres systtmes nitrutants, il est compatible avec Ies observations antirieures. Zusammenfassung-Die Struktur in der Gasphase nitriertec verdiinnter Fe-Ti-Legierungen wurde mit Elektronenmikroskopie, Elektronen- und RGntgenbeupung. und mit Messungen des Stickstoffgehaltes untersucht. Die Beobachtungen werden mit einkr periodischen Anordnung von Platten monoatomarer Dicke aus Titan- und Stickstoffatomen nleicher Anzahl auf den WiirfelflIchen der Matrix eedeutet. Diese Platten verzerren die umgebende &rix und eestatten eine vollstindige Anpassung fir SLickstoffgehalte bis dreimal dem stijchiometrisch fiir *[email protected] notwendigen. Die Adsotptionsenthalpie betrQt -25 kcal mol-‘, der ijberschul3 an Stickstoff kann durch Wasserstoffreduktion bei 575’C oder durch Aufheizen auf !SO’C und haher in einer inerten AtmosphLre beseitigt werden. Nach Wasserstoffreduktion kann der UberschuB an Stickstoff durch eine weitere Nitrierung wieder erzeugt werden. Die Struktur ist gegeniiber Vergraberung widerstandsftihi g; nur nach mehrstiindiger Gliihung bei 825’C findet Entfestigung und Plattenwachstum in nennenswertem Urnfang statt. Das Model1 ein- oder vielatomarer Schichten von substitutionellen Atom-Stickstoff-Platten wird als ein Spezialfall der gemischt substitutionell-interstitiellen Atomanhgufungen angesehen. Es kann aber auf andere Nitrierungs-Systeme ausgedehnt werden und ist vertdglich mit friiheren Beobachtungen.

1. NTRODUCTION It is only recently that high-resolution microscopy has been used to study the nature and the morphology of the precipitates which are responsible for the high hardnesses of nitrided irons and steels. The microstructures of such alloys usually consist of a dense homogeneous precipitation occurring as thin plates on the { 100) planes of ferrite. e.g. in nitrided Fe-Cr [ 1,23, Fe-MO [3,4]. Fe-Nb [3], Fe-W [Sj and Fe-V [6]. Electron diffraction patterns are characterised by marked (100) streaking.

It has been proposed [3,7] that in these systems there exists a precipitation series, analogous to that occurring during the ageing of Al-4 wt.:; Cu, namely:

zones + intermediate phase(s) -+ equilibrium phase In the case of Fe-5 wt.% MO the initial zones are thought of as a mixed substitutional-interstitial atom cluster of Fe. MO and N on the {100) planes. The first intermediate phase, which arises due to nitrogen b.c. ordering. is essentially molybdenum-substituted tztraponal Fe,,N2 and is succeeded bl; f.c.c. y-Mo2N. 137





Not all stages of the series are observed in all systems. however. and in Fe-Cr the equilibrium f.c.c. CrK has been identified even in the earliest stages of nitriding [I. 21. The f.c.c. intermediate or equilibrium nitrides invariably precipitate in ferrite with a Bain relationship to the matrix, i.e. ~~~~~,.~.,.;‘/~O~~~,.,,~,C~oOl~,~.,.l!’~~ lOIf,,.,.. The respective unit-cell sizes are such that the mismatch;

in the plane of the precipitate plate is ~50,. whereas perpendicular to the habit plane it is z 337,. The microstructure of nitrided Fe-Ti alloys has not been unequivocally established. Electron micrographs of F4 wt.?/, Ti ion-nitrided at 55O’C have been described as a random distribution of ‘dots’ some 15A in dia. [l]. and in alloys gas-nitrided at 6OO’C and subsequently over-aged at 8jO’C fine plates of f.c.c. TiN have been identified [S, 91. The present work complements a study of the kinetics of nitriding [lo], and seeks to determine the asnitrided structure, and to relate this to the structures observed in other nitriding systems.

2. EXPERMEIQAL 2.1 Macerids


The compositions of vacuum induction melted alloys used in the investigation are given in Table 1. The ‘active’ titanium concentration is that calculated by assuming all carbon to be present as Tic. Alloys were received in the form of hot-rolled f” dia. bar which was descaled before cold rolling to the desired thickness. Before nitriding, the samples were



annealed in dr\ hydrogen for 4 hr at 9OO’C. abraded. degreased and brietly electropolished. 2.2 Sitriding procdrwes Specimens. typically 2 x 1 cm x 150,~tm and weighing _ 200 mg were weighed immediately before and after nitriding on a balance which gave direct readings to lo-’ mg and estimated readings to 10e3 mg. Gas-nitriding was carried out in the temperature range -WI-575’C in a horizontal furnace titted with a gas-tight recrystallised alumina work tube. Hydrogen and anhydrous ammonia were purified by standard methods [ 1l] and passed through constant pressure-head capillary flolvmeters [ 121 to regulate the gas composition on entry to the furnace. The exit gas was analysed periodically using an ammonia absorption burette and in all cases corresponding to the inlet composition. Specimens were introduced into the hot-zone of the furnace using a magnetic throw-rod. The ammonia content of the gas mixtures was always below that required to form iron-nitrides. and the specimens kvere nitrided long enough to achieve constant weight. In low ammonia content mixtures at 4OO’C the nitriding rate is so slow that values of nitrogen uptake were obtained by first nitriding in 30% NH, : 70:; H3 and then re-nitriding to constant weight in the loiver ammonia content mistures. 2.3 H$rogen

refllrcriorl cud high-tettrperatwe


After nitriding. certain specimens were subsequently treated in dry hydrogen at 575’C. Others were aged at temperatures up to 9OO’C in argon filled fused silica capsules. Selected specimens xere also renitrided after hydrogen reduction or ageing. 2.4 Merhods of rsatt~itxxiot~ Nitrogen gains and losses were determined by weighing. Electra-thinned specimens were examined by transmission electron microscopy using a Philips EM 300 operating at 100 kV. Electrothinned wires. from the electron microscopy specimens. ivere used to obtain X-ray diffraction photographs using filtered Co-K% radiation and a 11-Imm dia. poader camera. Unit cell dimensions were corrected by the NelsonRiley extrapolation method using the 211. 220 and 310 reflections. and are believed to be accurate to 00005 A. 3. RESULTS 3.1 Electron tnerallogmph~ 3.1.1 As-nitrided, reduced and re4trirled

Fig. 1. Fe-2Ti nitridcd at 575-C in 89bNH :9?dH.: Bright field micrograph showing (1 IO) tweed 3striation<. electron beam parallel to [OOI] and g = 200. Inset shows associated diffraction pattern.


Bright and dark field micrographs of as-nitrided alloys, nitrided and reduced alloys. and reduced and re-nitrided alloys exhibit a fine irresolvable random ‘dot’ contrast under general imaging conditions, but contrast under two-beam dynamical conditions is similar to that observed by .Armitage et al. [l;] and by Tanner [14] in aged Cu-2 wt.“,: Be. The micrographs of Figs. 1 and 2 Lvere taken from Fe-2YATi

nitrided ,It 5’5 C in s” .*UH,:91”,,H2. X chquerctd L ‘:ussd’ pattern is obtained u ha g = 360 u-ith tneed 5triationS parnllsl t0 ’ I IO, (Fig. I). n-hereas for g = I10 onI> striations paralkl to g xc obwrvtd (Fig. 31. but strxin contraht Iran individual plates with sdges parallcl to [loI?] can 3150 be made out. The edge kn~th is its. 40 .A. These ctturacteristics of the tweed rattzrn \\erz obtained with the electrott-beam paraM TV both : lCx7 and ! I IO directions of the matrix. Ths apparent period&it) of the tw:ed pattern = 3%) RX &wed irom bright and da& field : ii crographs and is &mated to bs 50 1 Xd along 10~. The prssencc of grain boundark. or of dislocations a specimai cold-rolisd before nitridinp. has no lbs


Table 1. Composition of alloys used dlioys








Fe-l Ti





as-nikidred reduced renitrided cged 25h 82SpC






A iI X +

at.: c



Fe-O.3 Ti

Fe-2 Ti



Fe-l Ti:


a as-nitrided fJ reduced


l as-nitndred i


2.08 <0.02 4

be almost continuous in a (100) zone, see Fig. 1. Tilting experiments show, however, that the streaks occur on the low angle side of matrix 200 reflections, and in as-nitrided specimens they have a half-peak length of approx. 0.7ja* (a* is the 100 reciprocal lattice vector of ferrite), indicating that the feature giving rise to the streaks are appr0.x. 4A thick. Such streaks are also observed in nitrided-and-reduced specimens. and in specimens nitrided and aged at 750 and 825’C. The streaks in specimens aged 2.5 hr at SYC show an intensity maximum at the position corresponding to the 200,,,, reflection. (see Fig. 3) and the half-peak streak length is somewhat shorter (a*/?) than in asnitrided specimens. Detailed tilting experiments show that close to the matrix reciprocal lattice points there are additional

diffuse diffraction effects. Around the 200 reflection there are short diffuse streaks in + [I lo] and F: [IiO]. see Fig. 4(a). Around 110 there are diffuse intensity regions along + [ITO]. see Fig. 4(b). As well as in the as-nitrided and nitrided-andreduced conditions. the diffuse diffraction effects in the vicinity of ma&Lx reciprocal lattice points are observed in specimens aged at 750°C and for 2.5 hr at 825°C. 3.3 X-ray


Debye-Scherrer photographs of all specimens show sharp reflections and with the experimental system used, no evidence of any anomalous diffraction effects. The unit-cell dimensions obtained by different treatments are given in Table 2 and in Fig. 5 the increase in cell dimension over that of annealed material is plotted against the nitrogen content of the specimen (see Section 3.4). By least-squares analysis, the data are represented by Au = [email protected] + OGO15A;‘wt.:; N Table 2. Unit cell dimensions (A) of nitrided Fe-Ti determined by X-ray powder diffraction using CoK, radiation

/ 4 001 / I









t Nl/wt.%

Fig. 5. Variation of increase in unit-cell dimensions. Il\a. with nitrogen content of the sample.

The weight increase of variously treated Fe-?Ti varies with the nitriding potential of the gas mixture. (pNH#(pH2)3 ‘. where p’s are partial pressures in atmospheres. and can reach N:Ti atomic ratios of up to 3:1. On hydrogen reduction, however. no matter what the initial nitriding treatment, the nitrogen uptake corresponds closely to that which would be obtained by the formation of stiochiometric TIN (05235 wt.‘?,). After hydrogen reduction and then renitriding the nitrogen uptake reverts to that obtained on initial nitriding. The removal and replacement of the ‘excess’ nitrogen is therefore reversible. On nitriding Fe-2Ti at 575C in S”,; NH3:920/d H2 the N:Ti atomic ratio is cc. 2:1 but on ageing above 75O’C, and after -PI hr at 7jO’C, there is a total loss of the excess nitrogen above the stoichiometric 1:l ratio and this cannot be replaced by subsequent renitriding at lower temperatures. Nitrided Fe-l Ti and Fe-05 Ti also take up more nitrogen than is required to form TIN, and one expression of the amount of the excess is the N,Ti atomic ratio, obtained by first subtracting the normal Table 3. Excess nitrogen content and N:Ti ratio in alloys nitrided at 57YC in 8”,NH,:92‘?~H2. The ‘excess’ nitrogen is the weight gain. less the normal lattice solubility of nitrogen. less that required for stoichiometric TN The normal lattice solubility was subtracted from the weight gain before calculating the N:Ti ratio



I Alloy


I Ti

, Excess Nitrogen wt.%


I N:Tl



















there is an equilibrium for which the equilibrium constant : (5) ~to-a = Q.v(&cfl.vtOP where the excess nitrogen activity (I.~,~)is assumed constant for the same proportion of saturation at the three temperatures considered. Such an assumption is implicit in. e.g. the Langrnuir isotherm model. T’he activity in ordinary sites, a,Y,o,,is related to the nitriding potential by equations (1 and 3). whence. from the Clapzyron equation. i.e. -AH

d(ln &o-~,) d(l:T)



the molar enthalpy for reaction (1) is calculated from the curves of Fig. 6 to be - 25 & 5 kcal in the range 40%575’C.

X “excess”



after reduction






575X -renitrided


005 01 nitriding


points of equal she saturation





0.15 0.2 0.25 0.3 potenhal pNH,/(pH,IM

Fig. 6. Variation of excess nitrogen concentration in Fe-2 Ti with nitriding potential pNH,!‘(pH1)’ :. Error bars are two standard deviations of the msan. lattice solubility of nitrogen in ferrite from the weight gain, and converting the net weight gain into an atomic fraction, see Table 3. In Fig. 6 the excess nitrogen. above that for TIN stoichiometry and the normal ferrite solubility, is plotted against the nitriding potential for the Fe-2.Ti alloys. The normal ferrite lattice solubility is calculated from the data of Podgurski and Knechtel [LSJ for the reaction:

NH 3 %zN (in solution in Fe) + t Hz K, = (wt.% N) . (p&.I(PNHJ is given by: -9270 = + 10.27. T (The standard state for nitrogen activity is defined such that, in dilute solutions, activity = wt.“,;.) The isotherms in Fig. 6 resemble those of the Langmuir adsorption type in that they tend to a saturation limit, but detailed consideration shows that they do not obey the mathematical requirements for a Langmuir isotherm. The most probable explanation for the discrepancy is that the enthalpy of reaction is changing with degree of saturation of available sites for the excess nitrogen. In the almost linear portions of the isotherms the enthalpy of reaction can be estimated, provided that account is taken of the change in nitrogen potential at which saturation occurs at different temperatures. The nitriding potentials at which the same fraction of saturation occurs are marked in Fig. 6. For the reaction : N (Ordinary sites) + N (Excess) (4) lnK,

= - R

4.1 I~uturr

of‘ fhr



The tweed contrast in electron micrographs of Cu-2 wt.?, Be is known to arise from an arrangement of closely spaced thin plates on cube planes of the matrix L-161.The observation of tweed contrast in micrographs of nitrided Fe-Ti implies a similar structure, and this is borne out by the marked (100> matrix streaking which suggests that the plates are approx. 4a thick. The plates could be either thin f.c.c. TiN or some form of metastable zone or mixed Fe-Ti-N ‘cluster’. The reversible removal of excess nitrogen to a level appropriate to the formation of stoichiometric TiN implies that the Ti:N ratio in the plate is unity. On this basis the excess nitrogen would not be regarded as an integral part of the plate and must be accommodated at the interface. or in the matrix. Krawitz and Sinclair [ 171 have recently discussed the effects of pre-precipitation and precipitation on marrix unit-cell dimensions. They emphasise the conclusion of Guinier [18] that if a supersaturated random solid-solution decomposes to give a non-random distribution of solute on the same lattice sites as the matrix, and coherent with the matrix, then the Bragg reflections u-ill not change position or sharpness. They propose that such an invariance in the Bragg 001, J- G’Oa 100,

4 2~07.i




4 d


! 3&G



CJ i&&i .--_- 287i --.s Olstorlea



-_.--2.75i --.-. Reqmr



Fig. 7. Dimensions of distorted unfilled and regular filled octahedra in the b.c.c. iron structure [?I].




reflections is a definition of a metastable prs-precipitale or zone. In the present work the increase in unit-cell dimensions of nitrided Fe-Ti is consistent with both titanium and nitrosen being effectively in solution. i.e. present as some sort of zone on the above definition. The introduction of nitrogen into the interstices of b.c.c. iron distorts the surrounding atoms into regular octahedra, see Fig. 7. If all three orientations of octahedra are occupied equally. the isotropic distortion in a concentrated solution results in an increase in unit-cell dimensions. The observed increase: a = no -t [email protected]:; N


is slightly less than that predicted from results on very dilute solutions of nitrogen in ferrite [19]. but. as has been pointed out [ZOJ extrapolations from such dilute solutions are probably inadequate and the observed distortion should be compared with those in nitrogen martensite [Zl, 221 and r”-Fe,,N1 [23J. from which the relation a = a0 f 0.028 + OQO2&‘wt?A N is predicted, in excellent agreement with the observed result. Such an agreement is taken to imply the existence of zones [20] but in the present work this is not necessarily considered to be the case. Nitrogen atoms in f.c.c. TiN also occupy pctahedral interstices and the f.c.c. structure is no more than an ordered arrangement of the co-ordinating metal atom octahedra, see, e.g. Azaroff [24]. The dimensions of the Ti-N filled octahedron is 95% greater than the nitrogen filled octahedron in iron. Hence any proposed structure for the plates should take account of the octahedral co-ordination of nitrogen, the N:Ti ratio of one, and, due to the strong Ii-N interaction, the probability that the mutual coordination of N:Ti is a maximum. Figure 8 illustrates a possible plate structure one Ti-N layer thick. Within the plane of the plate there is 4:4N:Ti co-ordination and the atomic arrangement is the same as that on the cube planes of b.c.c.


Ti atoms


[email protected]



L---i%. --__---_ _----1-





2:s _h_




Fig. 9. Model of a three layer ‘TiN’ plate, which contains a complete unit cell of kc. 7-W. Metal atoms occur at the intersections of lattice lines.

ferrite and f.c.c. TIN. The nitrogen atoms are also co-ordinated by iron atoms above and below the plane of the plate. and these iron atoms could tx regarded as an integral part of an f.c.c. tetragonal FezTiN *zone’ for which II = 4.25 A, c = 3.B A (Fig.

8). The ‘zone’ thickness estimated from diffraction streaking (5 4 a) includes the two iron-atom planes, which, together with the Ti-N plane. give a three layer ‘zone’ with two 1.948\ interplanar spacings. Two and three layer Ti-N pfates can also be envisaged and a triple layer zone (Fig. 9) is sffectivel> a f.c.c. TiN plate one unit-cell thick, but is stili flanked by two iron atom planes separated from the Ti-N layers by spacings of ISA. At this stage the ‘zones’ could reasonably be described as thin f.c.c. TN plates. Additional Ti-N layers can be added without an) change in the atomic arrangement. until eventualI> the plate will give the bulk diffraction effects characteristic of f.c.c. TiN. The length of the diffraction streaking indicates that the as-nitrided plates are most probably a single titanium-nitrogen layer. and are most aptly described as such. The spacing of the iron atom planes on either side of the mono-layer determines the length of the diffraction streaks but it is considered inappropriate to describe the chemical composition of the plates as other than TIN’. In whatever way it is described. a plate causes a tetragona1 distortion of the matrix. but plates on all three (100~ matrix planes will cause the overall distortion to be approximately isotropic (account must be taken of the elastic anisotropy of the matrix. and this is considered below) and if the plats diffract coherently with the matrix the assembly will behave towards the X-rays like a random solid-solution and give the observed increase in the averugr unit-cell dimension. 4.2 Effect ofngeing

Fig. 8. Model of a mono-layer ‘TN plate; (a) showing arrangements within the layers; fb) (I IO) matrix section through the plate, showing ideal dimensions (.A).

011 the as-nifrided strwmre

The specimen aged 2.5 hr at 825’C exhibits (100) diffraction streaking. but the streak length is reduced




from its as-nitrided baluz and is centred on the 3M(TiS) reflection position (Fig. 3). The streak length implies that the plates are S-4 layers thick and at this stage they are behaving towards ckctrons like thin kc. TiS. But towards X-raqs the same specimen behases like a solid-solution. see Fig. 5. implying that in X-raq diffraction the matrix and precipitates diffract coherently whereas in electron diffraction the) diffract separately. Invariance of the positions of the S-ray Bragg reflections during decomposition of a solid-solution cannot always be taken as evidence. therefore. of a metastable pre-precipitate stage. Continued oserageing causes further plate growth. and for an alloy aged 7.5 hr at 9oOC the msdsured unit-cell dimension is that of pure iron. This is interprsted to mean that the plates have grown to such a thickness that they no longer ditlract coherently \rith the matrix.






Fig. I I. (al Projection of the arrangement oi Fig. 10 onto plane. Plates depicted b! dashed lin:~ lie at L 2

a 1001)

abore and below the plane of thz prqection. 131Schemaric tweed contra3 from o~erl~~~~i~~ biach-\\!ute images. Completer simll~~tions have shown that ‘tweed contrast can arise from a twa-dimensional array of g = 30. beam direction [OOI]. IS) Schematic tweed contrast. g = 110. beam direction [ix!;]. two mutualI) perpendicular loops (the -stair step‘ array) ivhich produce overlapping black+ hite strain contrast [35]. The three-dimensional nrrabs which electron-microscope image contrast due to mEurin et ~i. (271 have considered the elastic interacmetric. black-\\ hits images. Calculzarion of the tion between precipitate plates which cause tetragonal detailed image shape from each plate needs to take distortions and conclude that they arrange themselves into account the elastic anisotropy of the matrix. and in a pseudo-periodic array to maximise the number is beyond the scope of this paper. but ti-.c nature of of edge-face configurations. The most probable ar- the final image can be seen in principle by superimrangement for the ordered nuclei in CoPt which the>- posing the simple black-\\ hit< strain contast from consider is shown in Fig. IO, and is considered to each plate and in Fig. I I(b). (c) the owlapping conbs the ~ir~angement of plates in nitrided Fe-Ti. trast for g = 200 and g = 1IO is sketchA schematiin Fig. 1l(a) the ~~rrangement of Fig. 10 has been call) and corresponds to the weed striations actually projected onto the (001) cube plane and in calculating observed (Figs. 1. 2). The sketched irnLg?s are very similar to computer simulations of contrast from the two-dimensional ‘stair-step’ array iden:izr?l to the projectsd array of Fig. I l(a). and situated J! 0.3 of an extinction distance from the foil surhcr [3]. .A periodic arrangement oi tetrugonn! zntres might be espccted to lead to the appearance ci side-bands on X-ray diffraction patterns and the2 would be expected to occur predominantly on the high-angle side of matrix rstlections in a system i:: bvhich the volume fraction of loiver J-spacing matrix is much greater than that of the larger t/-spacing precipitate [ZS]. In the present work side-bands have not been observed. Kirkuood t’r CI/.[9] interpret diffuse intcnsitk maxima on the high angle side of sharp fundamentai reflections from nitrided Fe-f: in terms of tetragonal distortions of the matris. bur that implies that precipitnts plates only occur on 011s jet of : 100: matrix planes. and this is not obserwd. The results Fig. IO. Disposition of plates in a b.c.c. lattice of edge can be better interpretsd as side-ban& [29], and if lsngth L. The arrow indicate the direction of thz particular plate norma’ the diffuse intensity maxima are repnrd& as such then




Kirkwood’s data can be evaluated by means of the Daniel-Lipson relation [30]. L = Z/z. tan @:i(h’ + k” + I’).dB),

where 9 is the angle of the Bragg reflection Ml and LiBis the angular separation between the side-band maximum and the Bragg reflection. L is the periodi-, city expressed as a number of interplanar spacings in the ( 100; direction. The data indicates a periodicity of 4 35 A in Fe-l.56 wt.% Ti nitrided at 56O’C. Assuming that in the present investigation the plates are 4OA square and a Ti-N mono-layer thick, the dimensions of the periodic array can be calculated and for Fe-2Ti the cell edge length, L, in Fig. IO is 55 A. The white-white image distance along (110) in a cube plane tweed structure with g = 200 would be L\i’2,‘2 = 60A, consistent with the observed spacing of 50 F 20A. These dimensions are also consistent with the side-band spacing of 358, calculated from the data of Kirkwood provided that the periodicity is determined by the ~roj~cre~ interplate spacing. 4.4 Surtrce of anonmious electron diffaction


It is proposed that in electron diffraction the precipitates and the matrix diffract separately, and hence the diffuse diffraction effects around matrix reflections are presumed to be due to elastic distortions in the matrix. The primary matrix displacement can be regarded as a wave with a wave vector (100) and a paralfei displacement vector. In an elastically anisotropic matrix such a displacement causes shear waves with wave vector of the type (1 IO) and displacement vectors (IiO}. These latter have been discussed by Tanner [14] and can give rise to the (110) type streaks observed in the vicinity of matrix reflections in electron diffraction patterns. The displacement waves will cause regions of diffuse intensity to appear around each reciprocal lattice point in the direction of the wave vector, provided the displacement vector is not perpendicular to the reciprocal lattice vector, g. There is agreement between the predicted and observed diffuse diffraction. 4.5 Sitesfir rize ‘excess’ nitrogen Since excess nitrogen can be removed without any apparent change in structure, it is not considered that N:Ti ratios greater than one imply the existence of some Fe-Ti-N zone or cluster. Support for this assumption is provided by the variation in N:Ti ratio with alloy content, see Table 3. Possible sites for the excess nitrogen are regions of tensile strain at the edges of plates, or the broad faces of the plates where nitrogen is co-ordinated by one titanium and five iron atoms. It is well established that excess nitrogen can dissolve in deformed ferrite [31] but the amount which can be accommodated at screw and edge dislocation sites is small. 004 and OGO3% respectively at a true strain of 2.0. In nitriding Fe-2 wt.% Al [32] a total


of CLJ.0.25?; excess nitrogen can he taken up, of which O.OSP,is thought to be adsorbed at the non-coherent precipitate matrix interface and cannot be removed by hydrogen reduction: O.OYOis held at dislocation cores sites with an adso~tion enthaipy of - I2 kcal mol- i, and up to O.l?i is accommodated in the stress fields of the dislocations which are necessarily generated during the growth of non-coherent AIN. In nitrided Fe-Ti up to 0.9 wt.“, excess nitrogen must be accounted for in Fe-2Ti nitrided at 400°C. There is no evidence for dislocation formation accompanying the formation of the coherent plates but a Ti-N mono-layer on (OOt) is analogous to an interstitial-type edge dislocation loop with Burgers vector approx. a(, 1 - 1) [OOl] and line segments (100) and (010). The nitrided specimens are subject toan externally applied nitrogen potential which equals thechemical potential of nitrogen throughout the solid, so in tensile stressed regions where the activity coet?icient of nitrogen is lowered, the solubility is enhanced. Li and Chou [33] have calculated that O-125% excess nitrogen~nbea~ommodatedinasp~imencontaining 7 x IO” cm/cm’ of (u/2) [ 1111 dislocations and nitrided at 500°C in a nitrogen potential, pNH,/pH:‘” = 0.25. For the plate dimensions established in Section 4 the total length of effective a(J2 - 1) [OOl) dislocation line is lOi4 cm,‘crn3. The stressed edge regions of coherent precipitates would appear, therefore, to have the required capacity for the observed excess nitrogen. In the case of the coherent broad Faces of the plates, the driving force for possible adsorption is the strong chemical interaction between nitrogen and titanium, since the first adsorbed nitrogen layer will be coordinated by one titanium and five iron atoms. For a mono-layer plate, the maximum attainable N:Ti ratio by such adsorption is 3:1, and even for a triple Ti-N layer plate the attainable ratio is 5:3. The maximum observed N:Ti ratio is indeed 3:l and it is tempting to conclude that this occurs due to adsorption on the broad faces of Ti-N mono-layer plates. and that lower N-Ti ratios are the result of two and three-layer plate formation at higher nitriding temperatures or Iower titanium contents. But on this basis there seems no reason why re-nitriding of overaged specimens should not result in re-adsorption of a measurable excess. The effective dislocation density is, however, inversely proportional to the edge length of the plates, assuming the thickness remains constant. If plates thicken, the equivalent Burgers vector increases, but the effective dislocation density is correspondingly reduced. Hence, even if plates remain coherent as they coarsen, the effective dislocation density will decrease, conceivable to such an extent that no excess nitrogen will be detected on re-nitriding. Once coherency is lost, no excess nitrogen at all can be adsorbed by this mechanism. In more dilute alloys the plates could well be slightly larger, thus reducing the effective dislocation density.





It is considered. thersfore. that the stress fields of the cohersnt plates ars the most important sites for the ewss nitrogen and that ths molar snthalpy for ths adsorption reaction is -35 kcal. Internal friction studies have been performed on nitrided Fe-Ti [33]. but have not been interprsted in terms of the structures currently observed. Two abnormal peaks in the ranges 116-166’C and “l-‘A’-C at 1 Hz u’sre found. with activation ener___ gies of -25 and - 31 kcal mol-‘. There was a complets absence of the normal intsrstitial nitrogen peak at around room temperature. In the light of the present vvork. the high temperature peaks can be associatzd with nitrogen adsorbed in the stress fields of coherent plates, but an exact correlation is not possible since in the internal friction esperimsnts the specimens were nitridsd without control of nitrogen potential. which makss the appropriate structures impossibls to predict. The absence of the normal nitrogen peak is expected. since at lovver temperatures. and on rsmovul of the ssternally applied nitriding potential. ths high adsorption enthalpy will promote the almost complet2 transfer of nitrogen in normal lattice sites to the stmss- field sites.

Ths nitriding reaction in Fe-Ti occurs by the formation of a hard sub-scale which advances progressively into the bulk material. The kinetics of sub-scale advance are compatible with theories for internal oxidation [lo]. The nitrogen concentration at the nitriding front increases until the critical solubility product, pi] [S]. for coherent precipitation is exceeded. and thereafter any nitrogen removed from solution by precipitation is quickly rsplenished via the gas phase. The critical supersaturation for precipitation is constantly matched. until the titanium is exhausted. and the reaction front moves on. The evidence points to a periodic distribution of plates. and this strongly suggests a spinodal mechanism for the reaction. If such a mechanism does occur it means that as the nitrogen concentration at the reaction front gradually increases the point is reached first whers the system is unstable to large Huctuations. and only if these do not occur can the nitrogsn content reach ths cohzrent spinodal of the free-sn2rgy composition surface. at which point decomposition will taks place immediately. If the mechanism is spinodal decomposition then the reaction is an example of uhat occurs very close to the spinodal point. Altzrnatively. th2 elastic interactions betvveen conventionally nucleated coherent tetragonal prscipitatrs may cause the regular arrangement. Such interactions have been invoked in order to sxplain the periodic arrangement of ordered nuclei in Co-Pt [273. and with tetragonal distortions a critical particle size and separation for stabilisation ars also predictzd [;A]. and the centre centre

to edge length


of 1.3:1




is close to that estimated for nitrided Fe-2Ti. This could well explain the marked resistance to coarsening of the nitrided allovs. i.e. times greater than 2.5 hr at YX’C are required in order to cause appreciable softening and plate growth. Ko figures are available for th2 solubility product of bulk TiN in fsrrite. still less the mono-layer Ti-N plates. but from the data of Leslie [35] a 2’, Cr alloy, and chromium is a relativclv weak nitride former. is saturated at 575C by IO-’ wt.“,, N. 30 supersaturations of IO3 are easily attainable and classical coherent homogeneous nucleation is a possibility.

It is proposed that the model of single or multilayer ‘MY or ‘MZW plates can also sxplain the observed metallographic and diffraction effects in nitrided Fe-V’. Fe-Cr. Fe-M0 and Fe-W. For example, the as-nitrided structure of Fe-MO has been investigated by elsctron microscopy [3.36]. field-ion microscopy [-I] and atom-probe F.I.M. [36]. Precipitates occur as thin plates and have been characterised by atom-probe F.I.\l. as F2,MoJN1 [36]. This is not inconsistsnt with prssent model since the interfacial iron atom planes are both distorted and partly coordinated by nitrogen and hence might be expected to contribute to the image of a ‘zone’. The overall compositions expected from I-. 2- and 3-lay:r ‘Mo,N plates are Fe,hlo,N, Fe,Mo,NI and FeJlo,N,, the average of which gives ths Fe:Mo ratio observed experimentally by atom probe. In essence. the two interprstutions are similar, and the presznt mod21 is no doubt oversimplified. but the reversible removal of excess nitrogen to form stoichiometric ‘TN strongly implies that substitutional soluts and nitrogen co-ordinate as fully as possible in the plates and that iron is not a chemical component of any ‘zone’ in the Fe-Ti-N system. except in so far as the interfacial iron atoms are 1 6 coordinated by nitrogen. The growth of the plates to form thin f.c.c. MN or SI,N precipitates is regardsd as a continuous process. without any structural discontinuity- except. at som2 stage. a loss of coherency. 5. CONCLL.SIONS 1. The structure of nitrided Fe-Ti consists of equiatomic mono-layers of titanium and nitrogen occurring as plates 4013, square on ( 100; planes of ferrite. The iron-atom planes on either side of the mono-layer are distorted. but these planes do not contain nitrogen and are not regarded as an integral part of the plate. An analogous structure is thought to occur in other nitriding systems and is consistent with ths observations of other authors. ’ The plates are arrangsd in a periodic array _. which minimises the elastic strain energy. and gives rise to ‘tweed‘ structures in electron microscopy. : Towards X-rays. the matrix and the plates dif_. fract coherently. but in electron diffraction they do





not. The matrix distortions

due to the plates cause diffuse streaks on electron diffraction patterns in < 110) directions. -1. The plate morphology is stable on prolonged nitriding or hydrogen reduction at 575’C. and only after 44 hr ageing at SZYC is marked coarsening and softening observed. Such stability is attributed to the effects of elastic interaction between periodically arranged plates. On coarsening the plates are identified as Ccc. TiN which precipitates with a Bain refationship to the matrix. 5 Up to three times the nitrogen required to form equiatomic Ti-N mono-layers can be accommodated in nitrided Fe-Ti. The excess nitrogen is situated mainly in the stress-fields of the plates with an adsorption enthalpy of -25 kcal mol- ‘. This excess nitrogen can be removed by hydrogen reduction at 575°C or by ageing at 75O’C and above. The excess removed by hydrogen can be replaced completely by re-nitriding at 57K. 6. The observed structure is compatible with a spinodal decomposition mechanism close to he spinodal. or with classical coherent homogeneous nucleation accompanied by periodic alignment as a result of tetragonal distortion. .-lskno,virdlletne~ts-The author wishes to acknowledge the many fruitful discussions he has had with members of the iVolfson Research Group for High Strength Materials, cniversity of Newcastle-upon-Tyne, where the work was started during the tenure of an I.C.I. Research Fellowship, and where it formed part of a continuing programme on mixed substitutional-interstitial atom clusters. Subsequently the work has been supported by the Science Research Council, to whom due acknowledgement is made.

REFERENCES Philips V. A. and Seybolt A. U. Trans. A.I.M.E. 242, 2415‘(1968). Mortimer B.. Grieveson P. and Jack K. H. Scund. J. Alet. 1, 203 i1972). Speirs D. L., Roberts W., Grieveson P. and Jack K. H. Proc. 2nd [nrrrnationnl Corlference on Srrength of .Cleta/s and .4Ilo5 Monterey. California. Aug-Sept. 1970. A.S.M. (1970). Driver J. H. and Papazian J. M. Aern &frr. 21. 1139 (1973). Stephenson A.. Grieveson P. and Jack K. H. Sand. J. Mer. 2, 39 (1973). Pope M.. Grieveson P. and Jack K. H. Scnnd. J. Mer. 2, 19 (1973). Jack K. H. Scared. J. Met. 1. 195 (1971).



8. Chen F. D. H. Ph.D. Thesis. Rensselaer Polytechnic Inst. (1965). 9. Kirkwood D. H.. Atasoc_ 0. E. and Keow-n S. R. .2frr. Sci. 8. 49 (1974). 10. Jack D. H.. Lidstzr. P. C.. Grieveson P. and Jack K. H. C~~e~~ic~~ ‘~fera~~~~rg~ in irorr and Steel. p. 374. Iron and Steef Inst., London (1973). 11. Schwerdtfeger K. and Turkdogan E. T. Tcchnir[ues for &let& Rrsenrcl~. Vol. -I. Part I. p. 346. Interscience. New York (1970). 12. Darken L. S. and Gurry R. W. J. Am. Chm~. Sot. 67. 1398 (19453. 13. Armitage W. K.. Kelly P. M. and Nutting J. Ptoc. 5th lntrrncffional Corzaress 011 EIecrron 2ficroscow.. Philadelphia,.K4 (196s. 13. Tanner L. E. Phil. Msg. 14. 1I L (1966). !5. Podgurski H. H. and Knechtel H. E. Tratfs. .4.I.fLI.E. 245, 1595 (1969). 16. Phillips V. A. and Tanner L. E. Acta .!lrr. 21. 41 (1973).

17. Krawitz A. and Sinclair R. Phil. Mug. 31. 697 (1975). 18. Guinier A. Soliri St. Phvs. 9. 293 (1959). 19. Wriedt H. A. and Zweil L. Trans.’ A.f.i\I.E. 224, 1211 ( 1962). 20. Jack K. H. Hent Treut~~re~r ‘73. Metals Sot.: London (1975). 21. Jack K. H. hoc. R. Sot. A208, 200 (1951). 2’. Bell T. and Owen W. S. J. lrort Steel Inst. 205. 428 (1967). 23. Jack K. H. Proc. R. Sue. A208, 216 (1951). 14. Azaroff t. V. ~~~tro~luctio~lto Solids, p. 70. McGraw Hill. New York (1960). Fillingham P. J., Leamy H. J. and Tanner L. E. Electron %ficroscopJ and .%-trcmre of Materials. p. 163. Universitv of Cal. Press. CA 11972). 26. Gavrilovi A. V., Tyapkin Yg. D. ‘and Csikov M. P. Sot. Pizys. Dokl. 12, 970 (1968). 27. Eurin P. H., Penisson J. IM. and Bourret A. .&tn Mer. 21, 559 (1973). 28. de Fontaine D. Local Atomic Arrangemenrs Srudied bj X-Ray Diffraction. p. 51. Gordon & Breach. New York ( 1966). 29. Henderson S. Ph.D. Thesis. University of Newcastle Upon Tyne (1974). 30. Daniel V. and Lipson H. Proc. R. Sot. A182, 378 (1943). 31. Wriedt H. A. and Darken L. S. Tmns. A.I..CJ.E. 233. 111 (1965). 32. Podgurski H. H.. Oriani R. A. and Davis F. N. with Appendix by Li J. C. M. and Chou Y. T. Trans. A.I.M.E. 245, 1603 (1969). 33. Szabo-Miszenti G. Acta .Cfet. 18, 477 (1970). 34. Brown L. M.. Cook R. H.. Ham R. K. and Purdv G. R. Scripta Alet. 7, Sl5 (1973). 35. Leslie W. C. ?,%rogerl in Ferritic Sreek p. 36, U.S. Steel Co. (1964). 36. Driver J. H., Unthank D. C. and Jack K. H. Phil. &lag. 26, 1227

( 1972).

37. Brenner S. S. and Goodman S. R. Scripta -Vet. 5, 865 (1971).