Upload
nicolas
View
215
Download
0
Embed Size (px)
Citation preview
Study by differential thermal analysis
of reverse spinodal transformation in 15-5 PH alloy.
Emilie Herny1,a, Eric Andrieu2,b, Jacques Lacaze2,c,
Frédéric Danoix3,d, Nicolas Lecoq3,e
1 – Département Qualité Matériaux et Procédés, Microturbo - Groupe Safran, BP 62089, 31019 Toulouse Cedex
2 - CIRIMAT, Université de Toulouse, ENSIACET, BP 44362, 31432 Toulouse, France 3 –UMR 6634 CNRS, GPM, Université de Rouen, 76801 Saint-Etienne de Rouvray, France
[email protected], [email protected], [email protected], [email protected], [email protected]
Key words: martensitic stainless steels, ageing, spinodal decomposition
Abstract. Alloy 15-5 PH is a stainless steel with 15 wt.% Cr and 5 wt.% Ni that is precipitation
hardened by addition of Cu. In its semi-finished state, this alloy consists in Cu-supersaturated soft
martensite; its high specific properties come from a final tempering consisting in a heating to 550-
600°C, holding for 4 hours, and then air cooling. This treatment leads to nanometric Cu
precipitation that hardens the material and to transformation of some martensite to reverted
austenite which is then stable and provides ductility. While α' embrittlement of such steels is known
to occur at temperature in the range 450-520°C, it has been reported that they can be sensitive to the
same phenomenon after long term ageing at temperature as low as 300°C, with a significant loss of
ductility and an increase of the ductile-to-brittle transition temperature. Atom probe studies showed
that this degradation is related to demixtion of martensite into Fe-rich and Cr-rich phases.
Depending on the ageing temperature, demixtion can proceed through a nucleation and growth
precipitation or by spinodal decomposition of the martensitic matrix. The present study reports
differential thermal analyses (DTA) performed upon heating samples of material held at various
temperatures (290-525°C) for various times (410 h to 8500 h) that have been characterized by atom
probe. A clear DTA signal is obtained upon the reverse spinodal transformation that is further found
to depend on ageing conditions.
Introduction
Improvement of stainless steels led to the development of numerous Fe-base alloys with various
amounts of chromium and nickel, ranging from austenitic to ferritic and martensitic stainless steels.
Among these, martensitic stainless steels can be further precipitation hardened (PH) by addition of
aluminium, molybdenum or copper [1] that lead to the formation of Ni3Al, Mo2C carbide or copper
precipitates. Alloy 15-5 PH contains 15 % Cr and 5 % Ni, 4 % Cu, 0.04 % C (all compositions
given in wt. %) and small amounts of Si, Mo, Nb and Mn. This alloy is very much similar to alloys
15-4 and 17-4 PH that have been studied in more details as reviewed by Danoix et al. [2]. Semi-
finished products consist essentially in a workable soft martensite that is supersaturated in copper.
The high specific properties of these alloys are obtained by means of a final tempering treatment
consisting in a slow heating up to 550-600°C, holding at that temperature for 4 hours, and then air
cooling. This treatment leads to the formation of numerous copper precipitates a few nanometers in
size that harden the material, as well as to a small amount of reverted austenite which is then stable
at room temperature and provides a significant ductility level [3, 4]. Depending on the alloying
species, low amount of carbides or nitrides may also precipitate during this heat treatment [3].
Solid State Phenomena Vols. 172-174 (2011) pp 338-343Online available since 2011/Jun/30 at www.scientific.net© (2011) Trans Tech Publications, Switzerlanddoi:10.4028/www.scientific.net/SSP.172-174.338
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP,www.ttp.net. (ID: 132.174.255.116, University of Pittsburgh, Pittsburgh, USA-18/12/14,11:31:40)
While α' embrittlement of these steels, also known as 475°C embrittlement in binary Fe-Cr
alloys, occurs at temperature in the range 450°C to 520°C, it has been widely reported that they are
sensitive to the same phenomenon after long term ageing at temperature as low as 300°C. Ageing
leads to increased hardness and yield strength together with a significant loss of ductility and a
dramatic increase of the ductile-to-brittle transition temperature (TDBT) [3]. Atom probe studies
showed convincingly that the degradation of the material is related to demixtion of the martensite
into an iron rich (α) and a chromium rich (α') phases [2]. Depending on the ageing temperature,
demixtion can either proceed through a nucleation and growth precipitation process or through
spinodal decomposition of the matrix. From works on 17-4 and 15-4 PH alloys, ageing at 450°C
leads to α’ precipitation [3] whereas at 400°C, and at lower temperatures, embrittlement is due to
spinodal decomposition of martensite [4]. The transition temperature between spinodal
decomposition and nucleation and growth is very much dependant on alloy composition, and the
level of embrittlement as well as ageing kinetics depend on both the exposure temperature and time.
It has been further found that long term ageing of these alloys leads to co-precipitation of the G
phase which is rich in Ni and Si [2, 4] when the nominal amount of Si is high enough.
The present results are part of a study devoted to the investigation of microstructure and
mechanical evolution during ageing of alloy 15-5 PH. For this, samples have been aged for different
times at various temperatures where demixtion is expected to occur. Some of these aged materials
were first characterized by atome probe analysis, and then differential thermal analysis (DTA) has
been used to characterize phase transformations occurring when reheating the aged materials to the
austenitic state.
Materials and method
The 15-5PH alloy used in this study has been manufactured by Aubert & Duval using argon
oxygen degassing (AOD) and electro-slag remelting (ESR). It corresponds to the ARMCO PH 15-5
denomination or AFNOR X5CrNiCu15-5 standard, its chemical composition is given in table 1.
The overall hardening thermal treatment of the alloy consists in solution treatment at 1050°C for 30
minutes, air cooling to room temperature (RT) and finally tempering at 550°C for 4 hours after
heating at a rate of 2.5 K/min. The as-received alloy, i.e. before tempering, will be named
subsequently BM/NT for non-treated base material. The tempered alloy will be denoted BM/T, for
treated base metal; it contains a dispersion of nanometric Cu precipitates [5]. BM/T materials were
then aged at 290°C for 8500 h, 350°C and 400°C for 1000 and 5000 h, and 525°C for 360 hours.
Table 1: Chemical composition of 15-5 PH alloy [wt. %].
C Si Mn S P Ni Cr Mo Cu Nb Ta Fe
0.022 0.38 0.8 <0.002 0.015 4.91 14.83 0.25 3.02 0.19 <0.05 balance
Some of the aged materials were investigated by atom probe tomography carried out on a
CAMECA energy compensated tomographic atom probe, equipped with an advanced delay line
detector [6]. Analyses were conducted at 40 K, with a pulse repetition rate of 2 kHz and a pulse
fraction of 20%. Specimens were prepared using the classical two stage method [7], with a final
polishing in 2% perchloric acid at 7 volts. Three dimensional reconstructions were obtained using
the software package developed at the University of Rouen, based on the algorithm described in [8].
Typical analyzed volumes were 10x10x100 nm3.
DTA were performed with a SETARAM apparatus under Ar atmosphere and alumina as
reference. The size of the samples was 4 mm in diameter and 5 mm in length, their weight around
200 mg. In the records described later, negative (respectively positive) change of the DTA signal
relates to endothermic (respectively exothermic) transformation. DTA have been recorded for
various heating rates ranging from 0.5 to 15 K/min.
Solid State Phenomena Vols. 172-174 339
Atom Probe analyses
Atom probe tomography analyses are illustrated with the 3D maps in Fig. 1. They confirm that
chromium concentration fluctuations develop during ageing at 290°C (Fig. 1, top row) and 400°C
(Fig. 1, middle row). With increased ageing time, these fluctuations evolve both in terms of
amplitude and wavelength as seen by comparing the middle and bottom rows of Fig. 1 for ageing at
400°C. Because chromium concentration fluctuations were not observed in the BM/NT [5], Fig. 1 is
thus consistent with a spinodal decomposition mechanism of the unstable Fe-Cr solid solution. In
addition, the maps in Fig. 1 show that Cu-rich nanoclusters could be observed in some of the
investigated samples.
Figure 1 – 3D reconstructions of atom probe analyses on samples aged 8500 h at 290°C (top row),
500 h at 400°C (middle row) and 5000 h at 400°C. Light grey (green) dots represent Cr atoms, dark
grey (red) dots represent Cu atoms. The size of the boxes is 10x10x100 nm3.
DTA records.
Fig. 2 compares the DTA records obtained upon heating at 5 K/min untreated BM/NT and
tempered BM/T samples. A tentative base line has been superimposed on each of the curves to help
differentiating the thermal arrests. In the temperature range of interest for heat-treating the alloy,
namely between 400°C and 1000°C, four thermal arrests could be identified on the DTA record of
the BM/NT material that have been labeled H1 to H4 in the figure. While peaks H1 and H2 are
sharply defined, peaks H3 and H4 are much smoother. Note that peaks H2 and H4 are endothermic
while peaks H1 and H3 are exothermic. After tempering (BM/T material), it is seen that the DTA
record is essentially similar apart for peak H1 which is totally absent.
As the difference in microstructure between the BM/NT and BM/T materials is that Cu has
precipitated during tempering, peak H1 in BM/NT material may thus be related to this
transformation. The very specific shape of peak H2 allows associating it with the magnetic
transition of martensite at the Curie temperature, about 640°C. However, the spreading of this peak
over a very large temperature domain as observed in the case of the BM/T material, starting at about
340°C, suggests that martensite undergoes some continuous evolution upon heating. With
temperature further increasing above the Curie temperature, martensite should transform to
austenite with concomitant dissolution of copper precipitates. These transformations are assumed to
give rise to the smooth endothermic peak H4 between 750 and 950°C. Finally, peak H3 could relate
to precipitation of carbides as expected for such alloy [3].
340 Solid-Solid Phase Transformations in Inorganic Materials
-4
-2
0
2
4
6
8
10
300 400 500 600 700 800 900 1000
DTA signal
Temperature (°C)
BM/NT
H1
H2
H3
BM/T
H4
endothermic
-25
-20
-15
-10
-5
0
5
10
15
300 400 500 600 700 800
DTA signal
Temperature (°C)
BM/T
290 °C, 8500 h
350 °C, 5000 h
400 °C, 5000 h
525 °C, 360 h
Figure 2 – Comparison of the DTA record obtained during heating at 5 K/min of BM/NT and
BM/T materials. Dotted lines are tentative hand-drawn base lines. The records have been shifted
along the Y axis for clarity.
Figure 3 – Comparison of the DTA records obtained at 5 K/min on BM/T and aged materials (the
curves are labeled with the corresponding ageing treatment).
Fig. 3 shows the DTA records obtained at 5 K/min on the BM/T material (as a reference) and
four aged materials. The Curie transition is observed on all records, at about 640°C except for the
material aged at 525°C for which it is slightly higher. As expected, there is no exothermic peak
related to copper precipitation and instead materials aged at 290, 350 and 400°C, show a marked
endothermic peak below the Curie temperature. This peak may certainly be related to spinodal
decomposition that has occurred in the material during ageing, and more precisely be associated to
the smoothing of the chemical heterogeneities built up during ageing as the temperature increases
during the DTA run. Such a process can be defined as reverse spinodal transformation. This
correlation is ascertained by the observation that ageing at 525°C gives a record essentially
equivalent to the one obtained on the BM/T material, i.e. without endothermic peak, as expected
from the fact that this temperature is higher than the two-phase alpha–alpha prime domain for the
alloy's composition [3]. Also, it has been checked that the endothermic peak does not appear during
the second heating when repeating the DTA cycle. This further ascertains the relation between the
endothermic peak and spinodal decomposition of the material after long-term ageing.
Fig. 3 shows also that the amplitude of the endothermic thermal arrest and the peak temperature
are both sensitive to the ageing temperature. Fig. 4 illustrates again the effect of ageing temperature
(350 and 400°C) and shows that the endothermic peak is also strongly sensitive to ageing time
(1000 and 5000 h). In order to emphasize these effects, the records have been shifted along the
ordinate axis in Fig. 4 so as to superimpose each other in the low temperature domain.
Further use of the DTA records was the evaluation of the peak temperature of the endothermic
arrest as a function of the heating rate for the various aged materials. Because of heat transfer
resistance between the sample and the DTA sensor, the peak temperature is expected to increase
linearly with the square root of the heating rate [9]. The values obtained are plotted in Fig. 5, where
it is seen that a linear evolution of the peak temperature with the cubic root of the scanning rate and
not its square root could be obtained. Such a relation has been reported previously for solid state
transformation [10], this shows an effect of the heating rate on the reverse spinodal transformation.
It is seen in Fig. 5 that the temperatures obtained by extrapolation to a zero scanning rate, that
should be the characteristic temperatures of the transformation, are much higher than the
corresponding ageing temperatures. Interestingly enough, the series of measures obtained after
different ageing times at a given temperature may be extrapolated to the same temperature for a
Solid State Phenomena Vols. 172-174 341
zero scanning rate, 428°C±3°C (respectively 489°C±1°C) after ageing at 350°C (respectively
400°C). It is postulated that these temperatures relates to the alpha-alpha prime two phase boundary
for the composition of the alpha prime phase achieved as a result of the spinodal decomposition.
However, this result calls for an appropriate modeling of reverse spinodal decomposition to clearly
identify the relationship between the ageing temperature and the characteristic temperature
determined from Fig. 5.
-25
-20
-15
-10
-5
0
5
10
300 400 500 600 700 800
350 °C
400 °C
DTA signal
Temperature (°C)
1000 h
5000 h
350
400
450
500
550
600
0 0.5 1 1.5 2 2.5 3
350 °C, ageing for 1000 h
350 °C, ageing for 5000 h400 °C, ageing for 1000 h400 °C, ageing for 5000 h
Peak temperature (°C)
(scanning rate)1/3
Figure 4 – Effect of ageing time, 1000 and 5000 h, and of ageing temperature, 350 and 400°C, on
the DTA signals recorded at 5 K/min.
Figure 5 – Effect of the scanning rate [K/min] on the peak temperature of the endothermic peak
observed in materials aged at 350 and 400°C for 1000 and 5000 h.
Discussion - conclusion
The sensitivity of the endothermic peak to both ageing temperature and ageing time is certainly
related to the advancement of the spinodal decomposition, exactly as is the case of the hardness of
the material [5]. The hardness (HV30) increase with respect to the value of the BM/T material is
plotted in Fig. 6 where it is compared to a few results from literature on 17-4 PH alloyed with
copper [1, 3]. As expected for a diffusion controlled transformation, the effect of ageing appears
more rapid with increased temperature.
One further point about the DTA records that could be discussed is the relative sharpness of the
endothermic peak. For checking that this may effectively relate to smoothing of the chemical
heterogeneities built-up by the spinodal decomposition, the value of 2L/Dt=α (where D is the
diffusion coefficient, t the time and L the characteristic distance over which diffusion should
proceed) could be calculated. Homogenization is expected to be completed for α values higher than
one. In the present case, α was calculated as the integral ∫
=α dt2L/)T(D because of the non-
isothermal process [11]. With L of the order of 10 nm and using the diffusion coefficient of Fe
measured by Ray and Sharma [12] for a Fe-Cr alloy with 10 wt.% Cr, the integration between
400°C and 550°C at a heating rate of 5 K/min gives a value of α of about 5. Such a high value
confirms that the material could get homogenized even at these low temperatures, and this further
ascertains the association of the endothermic peak with reverse spinodal transformation.
From the results presented here, reverse spinodal transformation investigated by DTA appears as
a useful tool for characterizing the progress of the ageing process and N' embrittlement in
martensitic steels. Further, this data provides a firm basis for checking the capabilities of any
modeling of spinodal decomposition.
342 Solid-Solid Phase Transformations in Inorganic Materials
0
20
40
60
80
100
0.1 1 10 100 1000 104
This work, ageing at 350°CThis work, ageing at 400°C[3], aging at 350°C[3], aging at 400°C[1], aging at 450°C
change in HV -30
Time (h)
Figure 6 – Evolution of the hardness of the material as function of ageing time and temperature.
Acknowledgments
The authors are indebted to Airbus for supplying BM/NT material.
References
[1] Meyzaud Y. and Cozar R., in: Topical Conference on Ferritic Alloys for Use in Nuclear Energy
Technologies, The metallurgical society of AIME (1984) p. 27
[2] Danoix F. and Auger P.: Mater. Characterization Vol. 44 (2000), p. 177
[3] Yrieixb B. and Guttmann M.: Mater. Sci. Technology Vol. 9 (1993), p. 125
[4] Murayama M., Katayama Y. and Hono K.: Metall. Mater. Trans. A Vol. 30A (1999) p. 345
[5] E. Herny, PhD thesis, Université deToulouse, 2006.
[6] G. DaCosta, F. Vurpillot, A. Bostel, M. Bouet and B. Deconihout: Review of Scientific
Instruments Vol. 76 (2005), 013304.1-01334.8
[7] M.K. Miller, A. Cerezo, M.G. Hetherington and G.D.W. Smith, in: Atom Probe Field Ion
Microscopy (Clarendon Press, Oxford Science Publication, Oxford, 1996)
[8] P. Bas, A. Bostel, B. Deconihout and D. Blavette: Applied Surface Science Vol. 87/88 (1995),
p. 298
[9] W.J. Boettinger and U.R. Kattner: Metall. Mater. Trans. A Vol. 33A (2002), p. 1779
[10] Y.T. Zhu, J.H. Devletian and A. Manthiram: J. Phase Equilibria Vol. 15 (1994), p. 37
[11] G.D. Merz, T.Z. Kattamis and A.F. Giamei, J. Materials Science Vol. 14 (1979), p. 663
[12] S.P. Ray and B.D. Sharma, Acta Mater. Vol. 16 (1968), p.981
Solid State Phenomena Vols. 172-174 343
Solid-Solid Phase Transformations in Inorganic Materials 10.4028/www.scientific.net/SSP.172-174 Study by Differential Thermal Analysis of Reverse Spinodal Transformation in 15-5 PH Alloy. 10.4028/www.scientific.net/SSP.172-174.338