Decomposition kinetics of expanded austenite with ...

A

Applied

T. Christiansen, M. A. J. Somers: Decomposition kinetics of expanded austenite with high nitrogen contents

Thomas Christiansen, Marcel A. J. Somers Technical University of Denmark, Department of Manufacturing Engineering and Management, Lyngby, Denmark

© 2006 Carl Hanser Verlag, Munich, Germany

www.ijmr.de

Not for use in internet or intranet sites. Not for electronic distribution.

Decomposition kinetics of expanded austenite with high nitrogen contents This paper addresses the decomposition kinetics of synthesized homogeneous expanded austenite formed by gaseous nitriding of stainless steel AISI 304L and AISI 316L with nitrogen contents up to 38 at.% nitrogen. Isochronal annealing experiments were carried out in both inert (N2) and reducing (H2) atmospheres. Differential thermal analysis (DTA) and thermogravimetry were applied for identification of the decomposition reactions and X-ray diffraction analysis was applied for phase analysis. CrN precipitated upon annealing; the activation energies are 187 kJ/mol and 128 kJ/mol for AISI 316L and AISI 304L, respectively. Isothermal stability plots for expanded austenite developed from AISI 304L and AISI 316 were obtained. Keywords: Expanded austenite; Nitriding; Austenitic stainless steel; Thermal stability

1. Introduction Low-temperature nitriding of stainless steel by plasmabased methods has been the subject of several recent publications [e. g. 1 – 3]. The possibility of improving the wear and tribological properties of stainless steel by dissolving nitrogen in the surface-adjacent region of the steel is the driving force for this interest. Nitriding at temperatures below 723 K (450 °C) brings about a conversion in the surface-adjacent region of stainless steel: the development of expanded austenite (also called S-phase [1]). In fact expanded austenite, containing a significant quantity of nitrogen in solid solution, is thermodynamically metastable and tends to develop chromium nitride. Expanded austenite (cN) is responsible for the highly favourable properties associated with the low-temperature nitriding treatment: a significant increase in surface hardness, presumably due to solid solution hardening, and an associated improvement of the tribological performance, as reflected in lower friction coefficients and improved wear resistance [4]. Although of crucial importance for designing the process window of surface engineering stainless steel and the subsequent practical/industrial application of low-temperature nitrided stainless steel, so far little attention has been given to the thermal stability of expanded austenite. Pertinent experiments for a quantitative assessment of the thermal stability of homogeneous cN have hitherto been hindered by the practical difficulties encountered in obtaining homogeneous cN with a well-defined composition. The published investigations concern plasma-nitrided samples containing an inhomogeneous surface layer of cN, which has been subZ. Metallkd. 97 (2006) 1

 Carl Hanser Verlag, München

jected to various isothermal annealing experiments [5 – 7]. It is generally agreed upon that thermal exposure of cN leads to the development of CrN. For a surface layer of cN, decomposition occurs concurrently by the diffusion of nitrogen from the cN layer into the unnitrided substrate [6]. Moreover the development of austenite or ferrite accompanying the CrN formation has been reported [6, 8]. In the present work the decomposition kinetics of expanded austenite was investigated simultaneously with thermogravimetric analysis (TGA) and differential thermal analysis (DTA) on homogeneous cN powders, obtained from through-nitriding of thin foils of austenitic stainless steel.

2. Kinetic analysis In the present chapter the theory according to Mittemeijer et al. [9 – 11] is adopted. It is assumed that the thermal history of the investigated samples is fully described by a path variable. 2.1. Isothermal and non-isothermal transformations The fraction transformed, f, is fully described by the path variable b: f ¼ FðbÞ

ð1Þ

This path variable can be interpreted as being proportional to the number of atomic jumps. Since the temperature, T, determines the atomic mobility and time, t, defines the duration of the process considered [9], this implies that b can be expressed as: R b ¼ kðTðtÞÞ dt ð2Þ The rate constant k(T(t)) obeys the Arrhenius expression: Q ð3Þ kðTðtÞÞ ¼ k0 exp RTðtÞ where k0 is the pre-exponential factor, R is the gas constant and Q is the overall effective activation energy [9]. For isothermal annealing kðTÞ is independent of t. Hence it holds (cf. Eq. (2)): b ¼ kðTÞ t

ð4Þ

The assumption of the path variable b as a dependent variable for the degree of transformation conforms to the Johnson – Mehl – Avrami (JMA) equation for heterogeneous 79

A

Applied T. Christiansen, M. A. J. Somers: Decomposition kinetics of expanded austenite with high nitrogen contents

(solid-state) reactions: f ¼ FðbÞ ¼ 1 expðb n Þ

ð5Þ

Table 1. Composition of the stainless steels in atomic %. Composition of AISI 304L and AISI 316L were certified by Sandvik Materials Technology; the composition of AISI 304 was determined with EDS taking the other foils as a reference.

where n denotes the JMA exponent [9]. The equations applied for describing the degree of transformation for isothermal and non-isothermal annealing are identical as long as they are expressed in terms of b [10].

Alloy

Cr

AISI 304 19.6 AISI 304L 19.45 AISI 316L 18.93

Ni

Mo

Mn

Si

Fe

8.7 9.49 13.55

0 0 1.69

1.7 1.17 1.76

1.3 0.98 0.62

balance balance balance


www.ijmr.de


2.1.1. Determination of kinetic parameters In order to evaluate transformation kinetics the degree of transformation as a function of time/temperature is monitored [10]. Describing the progress of the transformation in terms of b n makes it possible to determine the kinetic parameters n, k0, and Q. For isothermal annealing it holds [10]: b n ¼ ðkðTÞ tÞn

ð6Þ

The parameters k(T) and n can be determined from one isothermal experiment comprising at least two sets of data (t, b n). Determination of Q and k0 requires isothermal experiments at (at least) two different temperatures [10]. For non-isothermal isochronal annealing it holds [10]: n k0 uRt2 Q n exp ð7Þ b Rut Q The time parameter t* is given by: t ¼ t þ

T0 u

ð8Þ

where T0 is the starting temperature and u is the heating rate. Values of n, Q, and k0 can in principle be determined from a single isochronal annealing experiment comprising at least three data sets (t*, b n) which gives 3 equations with 3 unknowns. However, at least two isochronal annealing experiments with different heating rates are recommended in order to ensure more accurate data [10]. The value for the activation energy of the decomposition reactions occurring during isochronal annealing was determined by applying a Kissinger-like method based on the equation [9]: ln

Tf20 E ¼ þ constant RTf 0 u

ð9Þ

where u is expressed in K/min and Tf' is the temperature at a fixed degree of transformation, f'. In the present work Tf 0 is approached by the peak maximum in DTA experiments and the temperature for a fixed degree of transformation in TGA experiments.

3. Experimental 3.1. Sample preparation Foils of stainless steel AISI 304L and AISI 316L were used for investigating the stability of expanded austenite. The AISI 304L and the AISI 316L foils were provided in thicknesses 20 lm and 100 lm, respectively. In addition, a 5 lm-thick foil of stainless steel AISI 304 (Goodfellows) was applied. The compositions of the three materials are 80

given in Table 1. The composition for AISI 304L and AISI 316L were certified by Sandvik; the composition of the AISI 304 foil was determined with energy-dispersive X-ray analysis (EDS), taking the certified foils as reference materials. AISI 316L was thinned to a thickness of 20 – 30 lm by electrochemical polishing to ensure that saturation with nitrogen was accomplished within a reasonable nitriding time (to prevent the development of CrN during nitriding; cf. Discussion). For recrystallisation and austenitisation the stainless steel thin foils were heated to 1343 K at a heating rate of 0.333 K/s and upon reaching this temperature immediately cooled in pure H2. During austenitisation the deformation-induced martensite, which was introduced upon cold-rolling during the manufacturing of the foils, was totally transformed to austenite. After heat treatment, the foils’ surfaces were activated to enable gaseous nitriding [12]. 3.2. Gaseous nitriding Gaseous nitriding was performed in a Netzsch STA 449C thermal analyzer, which allows simultaneous thermogravimetric analysis (TGA) and differential thermal analysis (DTA). Gaseous nitriding was performed in an atmosphere of ammonia and nitrogen until saturation with nitrogen was obtained in the stainless steel (stationary weight). A flow of nitrogen gas was led via the measurement compartment of the balance, to prevent corrosive attack by ammonia, and was mixed with ammonia in the furnace chamber of the equipment. The flows of ammonia and nitrogen were adjusted with mass flow controllers: 0.833 ml NH3/s and 0.0833 ml N2/s. A detailed description of adjusting the nitrogen content in stainless steel through controlled gaseous nitriding was given in [13]. The AISI 304L and AISI 316L foils were nitrided for 28 hours at 703 K; the much thinner AISI 304 foils were nitrided for 12 hours at 693 K. This difference in experimental conditions was necessary to prevent the development of CrN during nitriding. Preliminary work had shown that CrN develops easier in AISI 304 than in AISI 316. The thin foils were fully transformed into expanded austenite powder. AISI 304 was also reduced in pure H2 after nitriding in order to retract “loosely” bound nitrogen [13]. A saturation and reduction cycle was performed within 12 hours at 693 K; the remaining nitrogen content after reduction is given in Table 2.1 The nitrogen contents vary slightly due 1

Hydrogen reduction of as-nitrided foils was not possible for the thicker AISI 304L and AISI 316L foils as the time for nitrogen saturation was markedly longer and, hence, the risk for CrN formation during subsequent reduction too high.

Z. Metallkd. 97 (2006) 1

A

Applied


Table 2. Nitriding treatment and nitrogen contents in synthesized cN. Material


www.ijmr.de


AISI 304 AISI 304 reduced AISI 304L AISI 316L

Duration Temperature N : Cr (h) (K) ratio 12 12 28 28

693 693 703 703

3.14 0.83 3.05 2.93

yN

[N] (at.%)

0.616 0.162 0.594 0.554

38.1 13.9 37.3 35.6

to the fact that different material thicknesses and thermal treatments were applied. Hence the foils were not fully equilibrated with respect to the imposed nitriding potential. The nitrogen contents in AISI 304 and AISI 304L were determined directly from the weight gain as recorded with TGA. The nitrogen content in AISI 316L was determined from the lattice parameter obtained by X-ray diffraction [13]. All nitrogen contents determined in the as-nitrided and in the as-reduced materials as well as their thermal history are given in Table 2. 3.3. Thermal analysis Isochronal annealing of the expanded austenite powders was performed in the Netzsch STA 449C thermal analyzer, using various heating rates. Both inert (N2) and reducing atmospheres (N2 + H2) were applied for the investigation of the decomposition kinetics of expanded austenite. Constant heating rates within the range 0.0833 K/s to 0.333 K/s were applied, thus enabling determination of activation energies. The starting temperature was 303 K and the end temperature was 1173 K for all isochronal annealings. An empty crucible (alumina) was used as a reference for the DTA experiments. The gas flows for the inert atmosphere were 0.167 ml/s N2 and for the reducing atmosphere 0.667 ml/s H2 + 0.0833 ml/s N2. The typical sample mass used for each experiment was within the range 50 – 100 mg. For AISI 304 a much smaller amount of sample was used, because of practical problems associated with obtaining sufficient amounts of cN powder from 5 lm thin foils. Consequently, the results obtained with this material are at best semi-quantitative. A baseline was obtained in order to correct for instrumental effects during heating. This was performed in the following way: The thermal analyzer was cooled to the starting temperature (303 K) after the sample-run and a similar heating program was performed – now with the decomposed sample in the sample crucible. All DTA/TGA results presented are baseline corrected and pertain to the sample.

on a glass-plate and, after evaporation of ethanol, analysed with a Bruker AXS D8 X-ray diffractometer, equipped with a Cr anode and a set of Göbel mirrors in the incident beam.

4. Results 4.1. Annealing in N2 The results obtained with simultaneous TGA and DTA of AISI 304 with the nitrogen contents corresponding to saturation in pure NH3 and after reduction in H2 are depicted in Fig. 1a. The differentiated TGA results for the as-nitrided sample are compared with the DTA results for both samples in Fig. 1b. The sample containing 38.1 at.% N begins to loose weight at approximately 750 K, i. e. about 60 K above the nitriding temperature. This is attributed to the release of nitrogen from the sample powder by association of the atomically dissolved N to N2 molecules, which desorb from the surface. Apparently, the release of nitrogen proceeds in two successive steps, as clearly reflected by the two minima in Fig. 1b. The sample containing 13.9 at.% N does not loose weight; all nitrogen participates in the exothermic reaction

Fig. 1a. DTA and TGA curves simultaneously obtained during isochronal annealing of AISI 304 with different nitrogen contents corresponding to saturation in pure NH3 (38.1 at.% N) and reduction in H2 (13.8 at.% N).

3.4. X-ray diffraction X-ray diffraction was applied for the identification of the decomposed material. The decomposed cN powder was (further) crushed in an ultrasonic bath while submerged in ethanol, thus avoiding deformation-induced transformations, like martensite formation, from mechanical processing, e. g. milling2. The powder-ethanol slurry was smeared 2

The extreme brittleness of expanded austenite was lost after isochronal annealing. Hence, a mixture of flakes and powder of decomposed expanded austenite was used.

Z. Metallkd. 97 (2006) 1

Fig. 1b. DTA and differentiated TGA curves during isochronal annealing of AISI 304 with different nitrogen contents corresponding to saturation in pure NH3 and reduction in H2.

81

A


with a (DTA) maximum at approximately 850 K. Comparing the two DTA curves with the differentiated TGA curve in Fig. 1b shows that the positions of the two minima in the TGA curve for the as-nitrided sample coincide with the exothermic peaks in the DTA curve of the reduced sample. For the DTA curve of the as-nitrided sample no such coincidence occurs with the TGA “minima”. Instead the minimum at about 950 K coincides with an endothermic reaction, whilst an exothermic reaction is observed at about 870 K. This discrepancy is attributed to the occurrence of more than two reactions in the temperature range investigated, implying that the DTA peak at 870 K for the saturated sample is the net result of several contributions. A kinetic analysis of this DTA peak in terms of activation energy from several heating rates is therefore meaningless; indeed no meaningful values were obtained. The DTA peak for the saturated sample is shifted to a higher temperature as compared to the reduced sample; this cannot altogether be ascribed to overlap with the desorption reaction. Similar isochronal annealing experiments in an N2 atmosphere were performed for AISI 304L and AISI 316L (Fig. 2). For the samples a different nitriding treatment was applied than for the samples in Fig. 1: both a higher nitriding temperature and a longer nitriding time (cf. Experimental). The relatively large quantities of AISI 304L and AISI 316L used for the thermal stability experiments allow a quantitative analysis of the results. The same trends as observed for AISI 304 nitrided at lower temperature and shorter time also apply for AISI 304L and AISI 316L. Loosely bound nitrogen is released


www.ijmr.de


Applied

from the sample at temperatures above approximately 750 K. The temperatures at which maximum nitrogen release rates occur, reflected as minima in the differentiated TGA curves, are identical for AISI 304L and AISI 316L. This indicates that desorption of N2 is not influenced by the slight difference in composition between these two alloys. A step-like temperature dependence of N2 desorption is obvious (Fig. 2). The maximum nitrogen release rates in the two steps, i. e. the minimum values in the differentiated TGA curves, are equal for AISI 316L, whereas the second nitrogen release rate for AISI 304L is significantly slower than the first. In contrast with the results in Fig. 1 the discrepancy between the temperatures at which the first minimum in the differentiated TGA curves and the maximum in the DTA curves occurs is less pronounced and largest for AISI 316L, meaning that the temperature at which the exothermic peak maximum occurs is higher for AISI 316L than for AISI 304L for the same heating rate. The decomposed powders investigated with X-ray diffraction had been heated to 1193 K twice, once for the actual decomposition and once for the recording of the baseline. It was verified that the second heating cycle did not influence to any noticeable degree the appearance of the X-ray diffraction patterns. X-ray diffraction patterns of asnitrided and decomposed AISI 304L and AISI 316L are given in Fig. 3 for annealing in N2. In AISI 304L the decomposed powder contained mainly CrN and ferrite (α) after annealing; only a very small amount of austenite (c) appears to be present. The X-ray diffraction patterns of decomposed AISI 316L contain mainly CrN and austenite (c); only a very small fraction of ferrite (α) is present. The above results are explained as follows. Molecular nitrogen development occurs over the entire temperature range beyond about 750 K, which should result in a broad endothermic peak in the DTA curve. In the lower temperature range an exothermic peak associated with the formation of CrN (and possibly ferrite for AISI 304L; but see Section 4.2 in this respect) in cN is superimposed onto this endothermic peak. This may explain the shape of the DTA curve as constituted of two overlapping processes. The consumption of Cr during the development of CrN reduces the solubility of nitrogen in the austenitic matrix. Consequently, the driving force for nitrogen release is higher in

(a)

(b) Fig. 2. TGA, differentiated TGA and DTA of (a) AISI 304L and (b) AISI 316L during isochronal annealing at 0.417 K/s. The change in weight (TGA) is expressed as the N : Cr ratio.

82

Fig. 3. X-ray diffraction patterns of cN and decomposed cN in AISI 304L and AISI 316L. The diffraction patterns correspond to a heating rate of 0.417 K/s and to H2 and N2 atmospheres.

Z. Metallkd. 97 (2006) 1

A

Applied


4.2. Annealing in H2 The purpose of choosing a reducing H2 atmosphere was to investigate the thermal stability of expanded austenite without the concurrent N2 desorption, which obscures the kinetic information associated with the actual decomposition of cN. The DTA and (differentiated) TGA curves of both AISI 304L and AISI 316L are given in Fig. 4 for a heating rate of 0.417 K/s. Nitrogen retraction by the development of NH3 occurs from approximately 525 K (Fig. 4). The maximum nitrogen release rate of loosely bound nitrogen occurs for the same temperature for AISI 304L and AISI 316L. The temperature for which the maximum nitrogen release rate occurs coincides with the peak position of the DTA curve, implying that these exothermic DTA peaks are associated with the desorption of NH3. Strong asymmetry of the differentiated TGA curves and the corresponding DTA peaks is attributed to the geometry of the samples, i. e. flakes and grains of varying dimensions.


www.ijmr.de


the region where precipitation of CrN occurs. Thus the first minimum in the differentiated TGA curve is explained. Presuming that all Cr has precipitated, the relatively high nitrogen content is explained from adsorption of N atoms at the CrN-matrix interface. Upon continued heating the CrN precipitates coarsen, thereby reducing the interfacial area between precipitates and matrix and thus the capacity for N atoms at such interfaces. Coarsening of CrN is not associated with a distinct DTA peak, but leads to the second minimum in the differentiated TGA curve.

(a)

Provided that solid state diffusion of nitrogen atoms towards the surface, where NH3 forms, plays a determining role (as is likely to be the case at these low temperatures), this implies that N atoms do not reach the surface of all parts and pieces at the same time. Consequently, the development of NH3 spreads over a temperature range. The second DTA peak is not associated with a change of weight and can therefore be ascribed solely to the decomposition of expanded austenite. As for the reduced AISI 304 sample (Fig. 1) all nitrogen participates in this decomposition reaction, since the sample maintains its weight. A significant difference between the positions of the main exothermic DTA peaks for the two materials can be observed: decomposition is shifted to higher temperatures for AISI 316L. For AISI 304L a small exothermic peak was distinguished on the high-temperature side of the main peak. This peak was observed for all 7 heating rates for AISI 304L and did not occur for AISI 316L. X-ray diffraction patterns of as-nitrided and decomposed AISI 304L and AISI 316L are given in Fig. 3 for annealing in H2. In AISI 304L the decomposed powder contained only CrN and ferrite (α) after annealing; decomposed AISI 316L contains only CrN and austenite (c). Therefore, arbitrarily, the main peaks in the 870 K to 920 K range of the DTA curves in Fig. 4 are ascribed to the development of CrN. The small exothermic peak for AISI 304L at higher temperature (at approximately 955 K) is associated with the ferrite-austenite conversion (cf. Section 5.2.1). Interestingly, the nitrogen contents maintained in the samples after full reduction in H2 are higher than corresponding with a stoichiometric ratio N : Cr = 1 : 1. The higher nitrogen content in AISI 304L as compared to AISI 316L (cf. Table 2) was maintained after reduction. This is ascribed to the development of other nitrides than CrN. In this respect Si3N4 is the most likely candidate. In ferrite this hexagonal nitride is known to nucleate with great difficulty [14] and amorphous Si3N4 develops instead [15]. In an austenitic matrix the nucleation barrier is expected to be much lower, because of the possibility of a favourable orientation relationship between the hexagonal nitride and the fcc matrix. Comparing the decomposition of the samples in the two environments N2 and H2, a remarkable difference is observed. In H2 the temperature at which a maximum appears in the DTA curve for AISI 316L is shifted to considerably higher temperatures (a shift of about 50 K) as compared to its position for annealing in N2. 4.3. Kinetic analysis 4.3.1. Ammonia release

(b) Fig. 4. Isochronal annealing of (a) AISI 304L and (b) AISI 316L cN in H2 at a heating rate of 0.417 K/s.

Z. Metallkd. 97 (2006) 1

Kinetic analysis was carried out for the two identified reactions i. e. retraction of loosely bound nitrogen (NH3 reaction) and the decomposition reaction involving the develop ment of CrN (CrN reaction). Plots of ln Tm2 =uÞ versus 1/T for the NH3 reaction are given in Figs. 5a, b for AISI 304L and AISI 316L. From the slopes of the fitted straight lines (cf. Eq. (9)) the activation energies collected in Table 3 were obtained. The values for AISI 304L and AISI 316L by DTA and for a fixed transformation (TGA) are of comparable magnitude. The activation energy value obtained taking the differentiated TGA minima is systematically higher by about 15 kJ/mol. 83

A


www.ijmr.de



(a)

Fig. 6. Plots of ln Tm2 =uÞ versus 1/T for the decomposition of cN (CrN reaction) in AISI 304L and AISI 316L. The vertical bars are the reciprocal of the weighting factor and do not represent the standard deviation.

4.3.3. Coupling of isothermal and non-isothermal annealing

(b) Fig. 5. Plots of ln Tm2 =uÞ versus 1/T for the NH3 reaction for a) AISI 304L and b) AISI 316L.

4.3.2. Decomposition of expanded austenite Plots of ln Tm2 =uÞ versus 1/T for the decomposition reaction are given in Fig. 6 for AISI 304L and AISI 316L. The intensity of the DTA peak is strongly influenced by the applied heating rate: for low heating rates the integrated area of the DTA peak was relatively low (pertinent only for the decomposition reaction). This necessitated the use of a weighting factor. The integrated peak area per quantity of sample was used as the weighting factor. The values obtained for the activation energy for the decomposition of expanded austenite are 128 (± 9.9) kJ/mol and 187 (± 17.7) kJ/mol for AISI 304L and AISI 316L, respectively.

Table 3. Activation energies for the NH3 reaction for AISI 304L and AISI 316L.

DTA peak Differentiated TGA minimum 10 % transformation (TGA) 50 % transformation (TGA)

84

AISI 304L

AISI 316L

87 kJ/mol (± 3.0) 100 kJ/mol (± 2.9)

83 kJ/mol (± 4.0) 98 kJ/mol (± 6.7)

82 kJ/mol (± 6.0)

82 kJ/mol (± 5.3)

84 kJ/mol (± 3.2)

76 kJ/mol (± 6.9)

In order to relate the isochronal annealing experiments to isothermal annealing of cN a kinetic analysis was performed. The outline presented in Section 2 was followed. The DTA peak from the decomposition of cN (CrN reaction) in H2 was studied for AISI 304L and AISI 316L. The degree of transformation was approximated by the cumulative DTA peak area. The applicability of JMA kinetics was assumed and the state variable b n was introduced. Fitting of Eq. (7) to the data for three different heating rates was performed and the average values of the three fits were used (heating rates of 0.333, 0.417 and 0.5 K/s.). The values for the kinetic parameters are given in Table 4. It is noted that the value for the effective overall activation energy Q for AISI 304L is commensurate with the value obtained by the traditional analysis according to the Kissinger-like method (128 ± 9.9 kJ/mol). However, for AISI 316L the value obtained for the effective overall activation energy Q is significantly lower than the value obtained by the Kissinger-like method (187 ± 17.7 kJ/mol).

5. General discussion Heating of expanded austenite in an inert atmosphere does not bring about N2 desorption until a temperature of approximately 773 K is reached. Hence, the nitrogen contents achieved during nitriding and denitriding (693 – 703 K) can be conceived as a state of (metastable) equilibrium between gas mixture and solid state, rather than a stationary state between nitriding (i. e. nitrogen dissolution in the solid state) and N2 development (as for the case of ferritic nitriding), because a concurrent development of molecular nitrogen can be neglected at the nitriding temperature [16]. Table 4. Kinetic values for AISI 304L and AISI 316L. Kinetic parameter n Q [kJ/mol] k0 [1/s]

AISI 304L

AISI 316L

2.64 (± 0.09) 2.20 (± 0.10) 130 (± 0.6) 139 (± 0.1) 574 103 (± 34 103) 576 103 (± 9 103)

Z. Metallkd. 97 (2006) 1

A

Applied



www.ijmr.de


5.1. NH3 desorption Retraction of nitrogen, by development of NH3, occurs for the isochronal experiments carried out in a reducing atmosphere of H2. The development of NH3 (gas) entails transport of dissolved nitrogen in cN. Firstly, the morphology of the cN material has to be addressed; a mixture of flakes and powder (grains) was used3. For grains of cN diffusion of N atoms to the surface of grains has to occur. Volume diffusion of N may be facilitated by diffusion along defects in the solid state; this effectively lowers the activation energy. For flakes of cN volume diffusion and grain boundary diffusion of N atoms are anticipated to be the prevailing mechanisms. Formation of NH3 at the surface follows a stepwise hydrogenation of the nitrogen atom (and a stepwise dehydrogenation of the ammonia molecule for the reverse process, i. e. nitriding) [17]: [N] , Nads

(10)

Nads + Hads , NHads

(11)

Nads + NHads , NH2,ads

(12)

Nads + NH2,ads , NH3,ads

(13)

The adsorbed atomic hydrogen in reactions (11) – (13) stems from the following reaction: H2(g) , 2Hads

(14)

The rate limiting step for the stepwise hydrogenation of atomic nitrogen, cf. (10) – (13), depends on the temperature and pressure of H2. For high pressures of H2 the rate limiting step is (12) [17]. However, for the present case the most likely rate determining step is solid state diffusion. This is substantiated by literature values for activation energies for nitrogen diffusion, which compare well with the obtained values of 80 – 100 kJ/mol. For a nitrogen containing c-Fe alloy (9.5 at.% N) the activation energy for nitrogen diffusion is 90 kJ/mol [18]. The activation energy for nitrogen diffusion in c0 -nitride (i. e. an fcc Fe host lattice with approximately 20 at.% N) is 91.4 kJ/mol [19]. For nitrogen diffusion in pure c-Fe the activation energy is 152 kJ/mol [20]. The activation energy depends strongly on the concentration of interstitially dissolved atoms; this was established for carbon in c-Fe, where an increase of the carbon content lowered the activation energy for solid state diffusion (a value of approximately 93 kJ/mol for 9.5 at.% C is stated) [18 and references therein], obviously as a consequence of a dilation of the lattice with increasing C content. Similarly, for the present case N dilates the fcc lattice ([13]) and the diffusion coefficient of N in cN increases with increasing N content (for cN < 0.45 [21]). 5.2. Decomposition of expanded austenite 5.2.1. Decomposition products A significant difference between the alloys AISI 304L and AISI 316L was found in the phase constitution after anneal3

Synthesized expanded austenite is extremely brittle.

Z. Metallkd. 97 (2006) 1

Fig. 7. Equilibrium diagram showing the thermodynamically stable phases as a function of temperature for AISI 304L and AISI 316L with a nitrogen content equivalent to the chromium content (Thermo-Calc). The calculations were performed considering only fcc and bcc phases.

ing: ferrite developed in AISI 304L and austenite was maintained in AISI 316L. Equilibrium diagrams for the investigated alloys, taking the nitrogen content equivalent to the chromium content, were calculated (Thermo-Calc) (see Fig. 7). The change in composition from AISI 304L to AISI 316L has a major influence on the temperature dependence of the equilibrium fractions of α and c. Experimentally, a temperature difference was observed for expanded austenite decomposition; decomposition occurs at approximately 50 K higher temperature in AISI 316L than in AISI 304L. The DTA peak at 870 K in Fig. 4a, attributed to the development of CrN lies within the three-phase (α + c + CrN) region. A two phase (c + CrN) region is first expected for temperatures beyond 948 K (675 °C), which is compatible with the temperature where a small peak is observed as a shoulder to the major exothermic DTA peak in Fig. 4a. This is interpreted as evidence for the transformation of ferrite to austenite; the ferrite being formed during (or preceding) the development of CrN. The development of ferrite or CrN (depending on which of the two nucleates first) is expected to proceed relatively easily, since the following energetically favourable orientation relationship is expected between CrN and α-Fe: {100}α-Fe//{001}CrN; [100]α-Fe // [110]CrN [22]. During cooling of the samples after thermal analysis a transformation of austenite into ferrite is expected to occur in AISI 304L. The DTA peak position associated with the decomposition of expanded austenite for AISI 316L (Fig. 4b) takes place within or close to the austenite region (cf. Fig. 7). This explains the absence of an exothermic peak on the high temperature shoulder of the major DTA peak for AISI 316L. During cooling a transformation into ferrite should occur at a significantly lower temperature as compared to AISI 304L. Evidently, this transformation does not take place, which could be ascribed to the presence of Mo. Summarizing, for AISI 304L expanded austenite decomposes according to the reaction cN ! α þ CrNðþ cÞ; for AISI 316L expanded austenite decomposes according to cN ! c þ CrN. It is suggested that in AISI 304L a eutectoid transformation occurs and in AISI 316L (discontinuous) precipitation. 85

A


Table 5. Activation energies for substitutional diffusion. Element

Activation energy for diffusion (kJ/mol) Fe-18Cr-8Ni


www.ijmr.de


Cr Ni Fe

Fig. 8. Stability plot (temperature-time) of cN in AISI 304L and AISI 316L.

5.2.2. Kinetics of the decomposition of expanded austenite The kinetic values obtained by description in terms of the state variable b (4.3.2.) allow a prediction of the isothermal annealing behaviour of cN. A stability plot (temperature – time plot) for isothermal annealing of homogeneous expanded austenite (in H2) is given in Fig. 8. The stability of expanded austenite depends strongly on time and temperature. Furthermore, a significant difference in stability occurs for the two different alloys: cN is more stable in AISI 316L than in AISI 304L. Expanded austenite in AISI 304L and AISI 316L are stable for many years at a temperature of, say, 473 K (200 °C). However, for a temperature of 800 K (527 °C) it takes 8 and 29 minutes to obtain 50 % transformation into CrN for AISI 304L and AISI 316L, respectively. In the nitriding range, say, 693 K (420 °C) it takes 12 hours to obtain 50 % transformation in AISI 316L but only 2.7 hours in AISI 304L.4 The possible steps involved in the decomposition of cN and the development of CrN are diffusion of N and Cr to nucleation sites and diffusion of other atoms away from the nucleation sites. This entails volume diffusion and/or diffusion of Cr (and other substitutional alloying elements) along defects. Nucleation of CrN occurs most likely heterogeneously on dislocations, stacking faults, grain boundaries, etc. The activation energies obtained by kinetic analysis (Kissinger-like method) do not resemble the values normally associated with nitrogen diffusion (as the rate limiting step). More likely, the obtained energies should pertain to the diffusion of the substitutional elements, viz. Cr, as volume diffusion or as short-circuit diffusion (grain boundary and pipe diffusion). Literature values for activation energies for volume and grain boundary diffusion of Cr, Ni, and Fe in stainless steel alloys are listed in Table 5 [23, 24]. Values for pipe diffusion in stainless steel are expected to be markedly lower than for volume diffusion. The activa-

4

86

The values obtained for annealing in H2 do not reflect the maximum allowable time for nitriding. The investigated samples were already exposed to a nitriding treatment for 28 hours at 703 K and the stability reflects the allowable temperature for exposure of the as-received stress-free powder.

Fe-17Cr-12Ni

volume

grain boundary

volume

grain boundary

244.5 – 280.5

– –

263.8 250.8 279.2

152.2 131.7 177.2

tion energies obtained in the present study are significantly lower than the values for volume diffusion and resemble those for grain boundary diffusion. The activation energy of 128( 9.9) kJ/mol, obtained for AISI 304L corresponds to short-circuit diffusion as the rate limiting step. The activation energy of 187( 17.7) kJ/mol for AISI 316L could be ascribed to a rate limiting step consisting of a combination of short-circuit diffusion and volume diffusion. It is anticipated that the relatively low activation energy for AISI 304L is associated with the simultaneous development of CrN and ferrite. The interface between ferrite and austenite is suggested as an easy diffusion path for substitutional elements. For AISI 316L no ferrite develops and the redistribution of substitutional elements has to occur along other defects or by volume diffusion. Accordingly, a higher activation energy applies than for AISI 304L. 5.2.3. Influence of the nitrogen content There appears to be an influence of the nitrogen content on the decomposition of expanded austenite. Evidently, a higher nitrogen content postpones the development of CrN (cf. Fig. 1a). Assuming that decomposition is governed by the mobility of Cr atoms to form CrN and not by nucleation, a high content of interstitially dissolved N may impede the Cr mobility and thereby delay the decomposition. Alternatively, the postponement of CrN precipitation in high nitrogen austenite can be explained in terms of coherency between the CrN and the (expanded) austenite lattice. See in this respect Fig. 3: the (111) reflections of CrN and cN deviate only about 2° 2h; those of CrN and c deviate 6 – 7° 2h). Possibly, CrN precipitates develop coherent or semi-coherent interfaces with (expanded) austenite at higher nitrogen contents and nucleate homogeneously, thus necessitating (slow) volume diffusion of substitutionally dissolved elements. Conversely, for low nitrogen contents the opposite could hold, i. e. CrN develops heterogeneously and has an incoherent interface with austenite. 5.3. Nitrogen desorption during annealing in N2 The desorption of nitrogen during annealing in N2 occurs in, what appears to be, two steps. The thermogravimetric curves for the annealing of expanded austenite in nitrogen gas showed the occurrence of a two-stage desorption (cf. Figs. 1a and 2). In the light of the discussion in Section 5, these observations can be explained as follows. The first development of N2 starts before the appearance of CrN precipitates. Therefore, the most likely rate determining step in the development of nitrogen gas is solid Z. Metallkd. 97 (2006) 1

A

Applied


www.ijmr.de



state diffusion of nitrogen atoms to the surface. Upon the nucleation of CrN the effective cross section for diffusion of nitrogen through austenite is reduced. The nitrogen content in the samples on development of CrN is considerably higher than for the samples investigated in a reducing atmosphere. Therefore, CrN is anticipated to develop (partly) homogeneously and has a (semi-) coherent interface with expanded austenite (cf. Section 5.2.3). On continued annealing Ostwald ripening of the CrN precipitates occurs or, as in ferritic Fe – Cr alloys [25], discontinuous precipitation, and the precipitates loose coherency with the matrix. Consequently, the effective cross-section for diffusion of nitrogen atoms increases and a new maximum N2 development rate occurs. Alternatively, this behaviour could be attributed to different levels of nitrogen bonding in the octahedral interstices as reflected by nitrogen absorption isotherms for cN 0.36 (cf. Ref. 13). The release of relatively loosely bound nitrogen occurs in the first step followed by release of more strongly bound nitrogen in the second step. 5.4. Relation to the thermal stability of layers of expanded austenite on stainless steel In the present investigations the homogeneous expanded austenite samples were stress-free foils/powders. The decomposition behaviour of expanded austenite layers on a substrate (bulk material) may be different from the behaviour observed here. The scarce data available is included in Fig. 8 [5]. Although the same trend is observed, i. e. the time to decomposition increases with reducing temperature, substantial deviation occurs. The time for decomposition is longer for a layer/substrate situation as compared to synthesized expanded austenite. However, care should be taken when comparing the data. First of all a strict criterion of 50 % transformation is imposed for the case of synthesized expanded austenite, which is not the case for the data for layer/substrate situation where each data point – apparently – represents the time for occurrence of precipitates without further definition of the amount. Secondly, the nitrogen contents are not necessarily the same and the nitriding conditions prior to annealing are not similar. Also, the stress situation is bound to have an impact on the decomposition kinetics; the synthesized expanded austenite in the present study is stress-free, as opposed to a layer/substrate situation where the presence of compressive stress (gradients) most likely affects the decomposition. Qualitatively, comparing the volume of 1 mol expanded austenite with Cr : N = 1 : 1 with that of 1 mol austenite wherein all Cr and N has reacted to CrN may indicate whether or not stresses promote decomposition. The transformation is associated with a volume decrease of approximately 3 %, which suggests that hydrostatic compressive stresses will promote decomposition. On the other hand, the density of heterogeneous nucleation sites in layers on a relatively thick substrate is probably higher due to large compressive stresses in the layer during growth. For the thin foil samples investigated here, stress relaxation can occur partly in the substrate and the stresses will never reach the level as in a thin layer on a thick substrate, because overlap of diffusion fields occurs relatively soon in the process. Z. Metallkd. 97 (2006) 1

Finally, for an expanded austenite layer, inward diffusion of nitrogen from cN into the austenitic parent phase occurs, which is impossible for synthesized homogeneous expanded austenite. The associated reduction of the nitrogen content in cN, lowers the driving force as well as the involved strain energy effects for CrN development.

6. Conclusions The thermal stability of synthesized nitrogen expanded austenite, cN, in austenitic stainless steel was investigated for three different alloy compositions, i. e. AISI 304, AISI 304L and AISI 316L. Isochronal annealing experiments were conducted in both inert and reducing atmosphere. Annealing of cN in inert atmosphere gave rise to nitrogen desorption of loosely bound nitrogen at temperatures above approximately 750 K. In the same temperature regime precipitation of CrN occurred. Annealing in reducing atmosphere gave rise to nitrogen retraction of loosely bound nitrogen as NH3 at lower temperatures (510 – 560 K). The activation energy for the rate limiting step in the decomposition of cN into CrN was found to be higher for AISI 316L (187 kJ/mol) as compared to AISI 304L (128 kJ/mol). The obtained kinetic parameters from the isochronal annealing experiments enabled a prediction of the isothermal annealing behaviour of cN (stability plot): cN is significantly more stable in AISI 316L compared to AISI 304L. The decomposition products of cN in AISI 304L and AISI 316L differ: for AISI 316L cN decomposes into CrN and austenite, whereas for AISI 304L it decomposes into CrN and ferrite. Financial support by the Danish Research Agency under grant 26-01-0079 is gratefully acknowledged. The authors would like to thank Kristian V. Dahl for providing Thermo-Calc calculations and Finn T. Petersen (Sandvik) for providing stainless steel thin foil material.

References [1] Y. Sun, T. Bell, Z. Kolosvary, J. Flis: Heat Treatment of Metals 26 (1999) 9. [2] X. Xu, L. Wang, Z. Yu, J. Qiang, Z. Hei: Metall. Mater. Trans. A 31 (2000) 1193. [3] C. Blawert, B.L. Mordike, Y. Jirásková, O. Schneeweiss: Surf. and Coat. Techn. 116 – 119 (1999) 189. [4] X.B. Tian, Z.M. Zeng, B.Y. Tang, T.K. Kwok, P.K. Chu: Surf. and Coat. Techn. 128 – 129 (2000) 226. [5] X.Y. Li, Y. Sun, T. Bell: Z. Metallkd. 90 (1999) 901. [6] Y. Jirásková, C. Blawert, O. Schneeweiss: Phys. Stat. Sol. 175 (1999) 537. [7] X.Y. Li, H. Dong: Mater. Sci. Techn. 19 (2003) 1427. [8] E. Menthe, K.T. Rie: Surf. and Coat. Techn. 116 – 119 (1999) 199. [9] E.J. Mittemeijer: J. Mater. Science 27 (1992) 3977. [10] A.T.W. Kempen, F. Sommer, E.J. Mittemeijer: J. Mater. Science 37 (2002) 1321. [11] F. Liu, F. Sommer, E.J. Mittemeijer: J. Mater. Science 39 (2004) 1621. [12] M.A.J. Somers, T. Christiansen, P. Møller: Case hardening of stainless steel, Danish Patent DK174707 B1 and PCT/DK03/ 00497 [13] T. Christiansen, M.A.J. Somers: Metall. Mater. Trans. A, in press. [14] M.A.J. Somers: Ph.D. Thesis, T.U. Delft, 1989. [15] E.J. Mittemeijer, M.H. Biglari, A.J. Böttger, N.M. van der Pers, W.G. Sloof, F.D. Tichelaar: Scripta Mater. 41(6) (1999) 625. [16] E.J. Mittemeijer, M.A.J. Somers: Surf. Eng. 13(6) (1997) 483. [17] H.J. Grabke: Mater. Sci. Forum 154 (1994) 69.

87

A


[18] L. Cheng, E.J. Mittemeijer: Metall. Trans. A 21 (1990) 13. [19] M.A.J. Somers, E.J. Mittemeijer: Metall. Mater. Trans. A 26 (1995) 57. [20] H. Bester, K.W. Lange: Arch. Eisenhüttenwesen 43 (1972) 207. [21] T. Christiansen, M.A.J. Somers: Determination of concentration dependent diffusion coefficients of nitrogen in expanded austenite, in preparation. [22] M.A.J. Somers, R.M. Lankreijer, E. J. Mittemeijer: Phil. Mag. A 59(2) (1989) 353. [23] R.A. Perkins: Metall. Trans. A 4 (1973) 1665. [24] R.A. Perkins, R.A. Padgett, JR, N.K. Tunali: Metall. Trans. A 4 (1973) 2535. [25] S.S. Hosmani, R.E. Schacherl, E.J. Mittemeijer: Mater. Sci. Techn. 21(1) (2005) 113.

(Received July 1, 2005; accepted September 5, 2005)

Correspondence address Professor Marcel A. J. Somers Technical University of Denmark (DTU) Kemitorvet building 204, DK-2800 Kgs. Lyngby Tel.: +45 45 25 22 50 Fax: +45 45 93 62 13 E-mail: [email protected]


www.ijmr.de


Applied

88

Z. Metallkd. 97 (2006) 1