An Analysis of Static Recrystallization During Continuous, Rapid Heat Treatment

S.L. SEMIATIN, I.M. SUKONNIK, and V. SEETHARAMAN Because of its widespread use in a number of industrial processes, the static recrystallization of cold-worked metals is a phenomenon which has received extensive experimental and theoretical attention.V-sI Much of this work has dealt with isothermal heat treatment behavior in which recrystallization occurs at a nominally constant temperature for a fixed length of time. Laboratory work of this type has resuited in a number of so-called engineering laws of recrystallization which describe the effects of starting grain size of the deformed metal, level of prior cold work, and rate of heating to the annealing temperature on the recrystallization temperature and time and the recrystallized grain size. These observations have been complemented by theoretical analyses which have sought to explain recrystallization kinetics in terms of classical nucleation and growth phenomena as well as, in more recent times, the MonteCarlo and cellular automaton approaches which treat the statistical aspects of metallurgical transformations. Over the last 25 years, a number of primary metals producers have introduced (or evaluated on a laboratory scale) various types of continuous annealing processes in which the metal is heated to some peak temperature and then cooled without soaking at this temperature. This method has found wide application in particular for the recrystallization of cold-rolled sheet metals such as low carbon steel, stainless steels, copper alloys, and titanium alloys.[6I Heating techniques for sheet materials range from those based on gas-fired convection and indirect-resistance furnaces, with annealing times of the order of one-half to several minutes, to radiant and induction heating techniques which reduce heat treatment duration by one or several orders of magnitude, i.e., to times of the order of seconds or fractions of a second. The selection of the processing parameters for continuous lines is often done through an extensive series of laboratory trials in which the effects of variables such as heating rate and peak temperature on recrystallization are documented. The objective of the present work was to develop a theoretical basis for the description of recrystallization kinetics during continuous heat treatment operations. To this end, the recrystallization kinetics during very rapid heat treatments involving induction or direct-resistance heating were analyzed. To a great extent, these processes allow the study of the recrystallization process per se without the mitigating effects of grain growth following the completion of recrystallization. The starting point for the analysis of recrystallization

during continuous, rapid heat treatment is the classical Johnson-Mehl-Avrami-Kolmogorov expressiontT,8,9] for isothermal phase transformation kinetics controlled by nucleation and growth: X = 1 - exp {(-In 2) (t/to.s)"} [1]

where X denotes the fraction recrystallized, t is time, to.5 is the time to achieve 50 pct recrystallization, and n is a constant known as the Avrami exponent. Because recrystallization is a thermally activated process, to.5typically follows an Arrhenius dependence on temperature, i.e., to.5 = .4 exp (Q,/RT), where A is a constant, Qr is the apparent activation energy for recrystallization, R is the gas constant, and T is temperature. Rearranging Eq. [1] and taking the natural logarithm of both sides, the following is obtained upon defining B = (In 2)"/A: { - I n (1 - AT)}1/" = Bt exp (-Qr/RT) [2]

For continuous heating cycles, the differential form of Eq. [2] can be used to study recrystallization kinetics, assuming that the material coefficients (n, Q, and B) are not temperature dependent: d { - I n (1 - X)} ~/" = B {exp (-QJRT)} dt [3]

For a constant heating rate, 0 = dT/dt, the right-hand side of Eq. [3] is readily integratedVO,"~to obtain an expression for { - l n (1 - X)}" in terms of an infinite series in the variable exp (-QJRT). As suggested by Bourell and Kaysser,V0] for Tj > > T~, where Tiand T, are the initial and final (peak) temperatures, series like this can be truncated to the following without the introduction of substantial errors: { - l n (1 - XI)}'/" - { - l n (1 - X~)}'/" [4] BR = OQ~--~{T? exp (-Qr/RTI) - ~ exp (-Qr/RT~)} in which Xs and X, denote the fraction recrystallized at temperatures T and T~, respectively. For Ti - ambient temperI ature, X, = 0. In addition, for TI > T~ + 200 ~ = 225 ~ which is typical for annealing of most engineering alloys, the term in T~ on the right-hand side of Eq. [4] becomes negligible with respect to the term in T~ and Eq. [4] reduces to the following expression: BR { - l n (1 - XI)}'/" = OQ-'--~{ ~ exp (-Qr/RTI) } [5]

S.L. SEMIAT1N, Senior Scientist, is with the Metals and Ceramics Division, Wright Laboratory Materials Directorate, WL/MLLN, WrightPatterson Air Force Base, OH 45433-7817. I.M. SUKONNIK, Technical Staff Member, is with the Materials and Controls Group, Texas Instruments, Inc., Attleboro, MA 02703. V. SEETHARAMAN is Senior Scientist, UES, Inc., Dayton, OH 45432. Manuscript submitted December 4, 1995. METALLURGICALAND MATERIALSTRANSACTIONS A

The applicability of the result in Eq. [5] was established using continuous, rapid recrystallization data for a coldworked low carbon steel and commercial-purity titanium reported by Kuznetsov[121 and Gordienko et al., [~31respectively. The material used by Kuznetsov was a Soviet sheet steel comparable to AISI 1010 with a measured composition in weight percent of 0.12 carbon, 0.31 manganese, 0.012 phosphorous, 0.028 sulfur, and balance iron. The steel had been cold-rolled to an 80 pct reduction in thickness to a final gage of 0.50 ram. Samples of the steel were induction heated to various peak temperatures at various rates and then water quenched. Following heat treatment, the samples were examined metallographically to determine the temperatures for initiation and completion of recrystallization as a function of heating rate. In similar work, Gordienko et al. employed a direct resistance heating technique
VOLUME 27A, JULY 1996---2051

U.S. GOVERNMENTWORK NOT PROTECTED BY U.S. COPYRIGHT

700

I
COMPLETIONOF RECRYSTALLIZATION

[{Ti2 exp (-Q,/RT))/Olh = [{T~ exp (-Qr/RTj)}/Oll2

[6]

650 --

~600

o.

550

in which the bracketed terms are evaluated for two different heating rates,0t and02. Taking the natural logarithm of both sides of Eq. [6] and rearranging yields an expression from which Q]R can be determined from measurements of the peak temperatures ~2 and Tli needed to obtain the prescribed ~. at the two heating rates02 and0,:

Ii1 I-5OO
.~/~ r/

[7]
I--500 HEATING RATE ANALYTICAL FFr I 1000

45o

1500

(~

Similarly, for a given value of the heating rate 0, Eq. [5] may be used to obtain the following expression for n: { n = (-In(1 In -In (1 X/~)/J { (T2 e x p ( - Q J R ~ 2 ) ] ' ~ T26 exp (-QflRTfl)/J
[8]

Fig. l--Comparison of experimental data of KuznetsovI~Zl and the analytical fit of Eq. [5] for recrystallization during continuous heat treatment of a cold-rolled low carbon steel.

800

750
~" o

700

RECRYSTALLIZATION

UJ rr 650
k<

J m

,,al, 600
n m 550 l500 - START OF RECRYSTALLIZATION

m
Mf ASUREMENTS ANALYTICAL FIT

450o

100

200

300

I
400

500

HEATING RATE (~

Fig. 2--Comparison of experimental data of Gordienko et al. tul and the analytical fit of Eq. [5] for recrystallization during continuous heat treatment of cold-rolled commercial purity titanium.

to rapidly anneal sheet samples of a Soviet grade of commercial purity titanium (referred to as VT1-0) which had been cold-rolled to a 40 pct reduction in thickness. Again, samples were heated to various peak temperatures at various rates followed by water quenching to establish the kinetics of recrystallization under continuous, rapid heating conditions. The experimental recrystallization data of KuznetsovE 2 ~1 and Gordienko et a/. [13] are plotted in Figures 1 and 2, respectively. To apply Eq. [5] for the interpretation of these results, it was assumed that the curves for the start of recrystallization correspond to ~- = 0.05, and those for the completion of recrystallization correspond to Xj = 0.95. By this means, undefined quantities corresponding to ~ = 0 and ~. = 1.0 are avoided. In addition, even if the measured curves for the start and completion of recrystallization do indeed correspond precisely to ~ = 0 and X~ = 1.0, the curves for Xj = 0.05 and ~ = 0.95 would probably not vary much from these measurements. Assuming them to be temperature-independent constants, Q~ and n were obtained by the application of Eq. [5] to the measurements. For a fixed value of ~ , this relation yields the following:
2052--VOLUME 27A, JULY 1996

in which Tr2 and ~ denote the values of the peak temperatures that produce recrystallized volume fractions Xr2 and X/;, respectively, at the given heating rate. After determining Qr and n using Eqs. [7] and [8], Eq. [5] can then be applied directly to a selected datum to estimate the value of B. Equations [5], [7], and [8] were used in the analysis of the two sets of experimental data using various combinations of peak temperatures, heating rates, and recrystallized volume fractions. For the low carbon steel, these calculations resulted in Q]R = 22,900 K, n = 1.02, and B = 8.27 10 ~2s- ~. For the commercial purity titanium alloy, values of QJR = 49,800 K, n = 0.71, and B = 1.82 1022 s -~ were found. Upon placing these values into Eq. [5], the analytical fits shown in Figures 1 and 2 were obtained. The excellent fit between the measurements and the fitted curves in Figures 1 and 2 establishes the suitability of the modified Avrami approach, as embodied in Eq. [5] for the description of recrystallization kinetics under continuous, rapid heating conditions. Moreover, the good agreement adds credibility to the assumed constancy of Q]R, n, and B over the range of heating rates and peak temperatures in these experiments. The constant values of n found for the rapid heating data contrast to the not infrequent observations of variable n in long time, isothermal grain growth kinetics studies. However, the present analysis could be modified to investigate variable n behavior through the application of the pertinent equations over a series of small regimes of peak temperature and heating rate. The values of Qr and n obtained in the present work provide insight into the metallurgical processes controlling recrystallization under continuous, rapid heating conditions. The Qr for the low carbon steel (Q, = 22,900R = 191 kJ/mol) is comparable to that reported for self-diffusion in alpha iron in Reference 14, i.e., 239 kJ/mol. By contrast, the Q~ found here for recrystallization of commercial purity titanium under rapid heating conditions (Qr = 49,800 R = 414 kJ/mol) is considerably higher than the activation energy for self-diffusion in alpha titanium reported in the literature for high purity titanium,E~5~i.e., 169 kJ/mol. The high value of Q~ for the recrystallization of commercialpurity titanium may perhaps be attributed to the effect of
METALLURGICAL AND MATERIALS TRANSACTIONS A

Table I. The lnterphase Spacing, Droplet Spacing, and Cylindrical Rod Radius Obtained from the Geometric Relation with Phase Spacing for Various Experimental Growth Conditions of Zn-Bi Alloy
G 10 3 V X 10 .6 h M X 10 .6 Ro 10 6 h 10 -6

(K/m) 22.0 23.0 22.0 24.0 30.0 30.0 30.0 30.0

(m/s) 3.0 8.0 17.0 25.0 3.0 8.0 17.0 25.0

(m) 4.2 + 0.30 2.1 ___ 0.30 1.8 + 0.15 1.4 _ 0.15 4.3 + 0.40 3.5 + 0.15 2.0 ___ 0.15 1.6 + 0.10

(m) 0.56 + 0.04 0.28 + 0.04 0.24 ___ 0.02 0.18 + 0.02 0.57 + 0.05 0.46 ___ 0.02 0.27 + 0.04 0.21 + 0.01

(m) 4.8 + 0.1 3.5 + 0.3 2.9 + 0.3 1.9 _.+ 0.1 4.6 ___ 0.2 3.8 _+ 0.1 2.4 + 0.2 2.3 + 0.3

12. B.A. Kuznetsov: MetalL Obr. Met., 1958, vol. 2, p. 28. 13. A.I. Gordienko, M.Ya. Brun, V.V. Ivashko, and L.A. Elagina: in Titanium 1990: Products and Applications, Titanium Development Association, Dayton, OH, 1990, p. 726. 14. N.L. Peterson: J. Nucl. Mat., 1978, vol. 69, p. 3. 15. F. Dyment: in Titanium '80: Science and Technology, H. Kimura and O. Izumi, eds., TMS, Warrendale, PA, 1980, p. 519. 16. V. Seetharaman and R.L. Goetz: UES, Inc., Dayton, OH, unpublished research, 1995.

G = temperature gradient; V = growth velocity, A = interdroplet spacing; Ro = cylindrical rod radius, and AM = phase spacing.

The Rayleigh Instability and the Origin of Rows of Droplets in the Monotectic Microstructure of Zinc-Bismuth Alloys
B. MAJUMDAR and K. CHATTOPADHYAY A well-aligned rod morphology can be observed during the cooperative growth of monotectic alloys with the liquid phase taking the form of rods. However, under certain processing conditions, rows of droplets instead of rods are observed during unidirectional solidificationY-41 Various models have been proposed to explain the origin of these microstructures in terms of growth interface instabilities.t5] On the other hand, the experiments by Grugel and Hellawell t3] using transparent analog clearly suggest that in addition to oscillatory instability at the growth interface due to oscillation in the interface velocity, a Rayleigh-type ripening instability sets in well behind the stable growth front. Kamio et al.,t21 in the course of a detailed experimental study on Al-In monotectic alloy, found evidence of both the processes under two different growth conditions. However, no quantitative evaluation of either of the models exists. In the present article, we report the results of experiments on Zn-Bi monotectic alloy which yield a microstructure containing rows of droplets. We endeavor to explain the observed morphology quantitatively in terms of the Rayleigh instability of the rods formed during coupled growth behind the growth front. A Zn-0.6 at. pct Bi monotectic alloy was unidirectionally solidified using a Bridgman-type apparatus. The details are available elsewhere.t6j The experiments were performed at different growth velocities at two temperature gradients of approximately 22 103 K/m and 30 103 KJm. The actual conditions are given in Table I. Typical low magnification aligned microstructures of the longitudinal and cross sections are given in Figure 1. Rods, when viewed at a higher magnification, clearly reveal that each is composed of row of droplets rather than a continuous rod (Figure 2). The droplets are often faceted (Figure 3(a)). The evidence that they have originated in the liquid state is given in Figure 3(b). The dendritic structure is clearly resolved inside the droplets. The observed interphase spacings (AM) and interdroplet spacings (A) within the row at different growth ve-

solutes in this material on pinning of grain boundaries during the recrystallization process. The values of the Avrami exponent n found in the present work for the continuous, rapid recrystallization kinetics are approximately equal to unity for both the low carbon steel and the commercial purity titanium. Values of n for isothermal recrystallization processes are typically between 1 and 3. [4I Furthermore, recent simulations of static recrystaUization using a cellular automaton approachtt6] suggest that n's of approximately one should indeed be expected in which nucleation dominates the phase transformation process. This latter finding provides a plausible explanation for the observation of substantially finer grain sizes in rapidly annealed metals (resulting from limited growth) as compared to the grain sizes in furnace heat treated materials.

This work was conducted as part of the in-house research activities of the Metals and Ceramics Division, Air Force Materials Directorate, Wright Laboratory. The assistance of L. Farmer, L. Osterloh, and J. Paine in preparing the manuscript is appreciated.

REFERENCES
1. P.G. Shewman: Transformation in Metals, McGraw-Hill, New York, NY, 1969. 2. J.G. Byrne: Recovery, Recrystallization, and Grain Growth, McMillan, New York, NY, 1965. 3. J.E. Burke and D. Turnbull: in Progress in Metal Physics, B. Chalmers, ed., Interscience Publishers, Inc., London, 1952, vol. 3, p. 220. 4. T. Furu, K. Marthinsen, and E. Nes: Mater. Sci. Technol., 1990, vol. 6, p. 1093. 5. H.W. Hesselbarth and I.R. Gobel: Acta Metall. Mater., 1991, vol. 39, p. 2135. 6. S.L. Semiatin, D.N. Williams, T.G. Byrer, and R.E. Hook: Adv. Mater. Processes, Incorporating Met. Progress, 1987, vol. 131, p. 28. 7. W.A. Johnson and R.F. Mehl: Trans. AIME, 1939, vol. 135, p. 416. 8. M. Avrami: J. Chem. Phys., 1939, vol. 7, p. 1103. 9. A.E. Kolmogorov: Akad. Nauk SSSR, Izv. Ser. Mat., 1937, vol. 1, p. 355. 10. D.L. Bourell and W. Kaysser: Acta MetalL Mater., 1993, vol. 41, p. 2933. 11. J.C. Soper and S.L. Semiatin: Wright Laboratory Materials Directorate, Wright-Patterson AFB, OH, unpublished research, 1993. METALLURGICAL AND MATERIALS TRANSACTIONS A

B. MAJUMDAR, Research Associate, and K. CHATTOPADHYAY, Professor, are with the Department of Metallurgy, Indian Institute of Science, Bangalore 560 012, India. Manuscript submitted June 1, 1995. VOLUME 27A, JULY 1996~2053