Corrosion Science 72 (2013) 64–72

Contents lists available at SciVerse ScienceDi rect

Corrosion Sc ience

journal homepage: www.elsevier .com/locate /corsc i

The high-temperature oxidation of bulk nanocrystalline 304 stainless steel in air

0010-938X/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.corsci.2013.03.008

⇑ Corresponding author at: Shenyang National Laboratory for Material Sciences,Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, 110016 Shenyang, PR China. Tel.: +86 24 83978750; fax: +86 24 23891320.

E-mail address: sgwang@imr.ac.cn (S.G. Wang).

S.G. Wang a,b,c,⇑, M. Sun d, H.B. Han e, K. Long b, Z.D. Zhang a,b,c

a Shenyang National Laboratory for Material Sciences, 72 Wenhua Road, 110016 Shenyang, PR China b Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, 110016 Shenyang, PR China c International Centre for Materials Physics, Chinese Academy of Sciences, 72 Wenhua Road, 110016 Shenyang, PR China d School of Sciences, Shenyang University of Technology, Shenyang 110870, PR China e Shenyang Ligong University, 110168 Shenyang, PR China

a r t i c l e i n f o

Article history:Received 10 October 2012 Accepted 4 March 2013 Available online 16 March 2013

Keywords:A. Stainless steel B. XPS B. SEM C. Oxidation

a b s t r a c t

The high-temperature oxidatio n of bulk nanocrystalli ne 304 stainless steel (BN-SS304) and its conven- tional polycrystalline counterpart (CP-SS304) in air at 900 �C for 24 h were studied by thermogravimetric analysis, X-ray photoelectron spectroscopy and scanning electron microscope. We studied the valence electron configurations of BN-SS304, CP-SS304 and their oxide scales by ultra -violet photoelectr on spec- troscopy. The high-temperature oxidation resistance of BN-SS304 was enhanced in both initial and iso- thermal oxidation, which was attributed to its larger work function and more chemical stability, its more chemically stable and compact oxide scale, its weaker O2 adsor ption and diffusion, its weaker Cr and Mn atoms diffusions.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

FeCr- and FeCrNi-based stainless steels are employed in a very wide range of modern applicati on due to their abilities to simulta- neously withstand corrosive environments and to maintain good mechanical properties. The performanc e of these alloys in resisting high temperature oxidation generally depends on their abilities to form and maintain protective Cr 2O3 scales. Oxidation resistance of a metal depends on the penetrati on resistance and adhesion of its oxide scale [1]. From a thermodyna mic point of view, oxidation is the function of the chemical potential of components of environ- ment and the chemical potentials of metallic components in alloy [2]. In order to enhance the oxidation resistance at high tempera- tures, some rare-earth elements, such as Ce and Y, were added in alloys and pure metals [3–7]; alloy grains were refined to nano- crystalline or ultrafine ones [3,8–11]; different types of coatings were applied to 304 stainless steel surface [12–14]. The grain boundaries that intersect the surface of Cr-bearing alloys act as site for the preferent ial formatio n of Cr 2O3. As the grain size of alloy is reduced, the distance between the Cr 2O3 nuclei on the grain boundary is reduced, and hence, less time is required for sideways growth of nuclei to form a continuous layer Cr 2O3. The grain boundary may act a path for rapid diffusion of Cr from bulk to

the surface of the alloys to speed up the process of sideway growth [15–17]. Most high-temperat ure oxidation reactions that form pro- tective scales are controlled in the long term by the diffusion of oxide-for ming species through the scale. It seems unlikely that the microstructur e of alloy would have any significant effects on the compositi on and mechanical properties of scales [18]. From 400 to 600 �C, cold-worked Fe oxidized faster than annealed Fe;the oxide formed on annealed Fe was porous and separates from the metal; cold work has no effect and solid scale only was formed at 650 �C [19]. Ultrafine-grained Ni 3Al exhibited a greatly increased cyclic oxidation resistance in comparis on with its polycrystall ine counterpar t at 900 �C [11]. The surface layer of oxide scale was firstrich in Cr and then became rich in Ni as immersion time increased for type 304 stainless steel [20]. Minor solute additions signifi-cantly improved oxidation resistance, but could also reduce inter- stitial solubility resulting in formation of chromium carbides for air oxidation of Fe–18Cr–18Mn [21].

The popular point of view on high-temperat ure oxidation for stainless steels is that grain-bou ndary diffusion provides a means of effectively increasing the flux of solute to oxide/metal interface by providing a fast diffusion path to interface and decreasing alloy grain size should increase solute flux to the interface and hence the oxidation of base metal decreases while promoting the earlier for- mation of protective scale [15,17,18 ,22–25]. In this work, we stud- ied the high-temperat ure oxidation of bulk nanocrystal line 304 stainless steel (BN-SS304) produced by severe rolling technique and its conventi onal polycrystall ine counterpar t (CP-SS304) in lab- oratory air at 900 �C for 24 h by X-ray photoelectr on spectroscop y

S.G. Wang et al. / Corrosion Science 72 (2013) 64–72 65

(XPS), thermogravi metric analysis and scanning electron micro- scope (SEM). We studied the valence electron configurations of BN-SS304, CP-SS304 and the oxide scales on their surfaces by ul- tra-violet photoelectron spectroscopy (UPS). We explained our re- sults of high-temperat ure oxidation of BN-SS30 4 and CP-SS304 in term of the valence electron configurations of BN-SS304, CP- SS304 and their oxide scales. The enhanced high-tempe rature resistance of BN-SS30 4 did not depend on the faster Cr diffusion rate from its grain boundaries and the rapidly forming compact oxide scale in initial stage of oxidation.

Fig. 1. The oxidation kinetics of BN-SS304 and CP-SS304 in initial stage of oxidation from room temperature to 900 �C.

Fig. 2. The oxidation kinetics of BN-SS304 and CP-SS304 in isothermal oxidation at 900 �C for 24 h.

2. Materials and methods

BN-SS304 was produced by severe rolling technique with CP-SS304. The details of this preparation technique and the micro- structure characterizati ons of BN-SS304 and CP-SS304 were reported in our previous work respectively [26,27]. The dimensions of specimens for high-tempe rature oxidation and UPS (ESCA-LAB250, He I radiation hm = 21.22 eV) measure ment at temperature range from 80 �C to 500 �C were 12.0 � 12.0 � 1.0 mm3 and10.0 mm � 10.0 mm � 15 lm respectivel y. The rate of tempera- ture rise was 30 �C per minute from room temperat ure to 900 �Cduring initial stage of oxidation, and then isothermal oxidation was carried out at 900 �C in laborator y air for 24 h, the velocity of air flow was 160 ml per minute during both initial stage and iso- thermal oxidation processes. The growth kinetics of oxide scales formed on BN-SS304 and CP-SS304 were established by measuring the mass gain per unit area versus oxidation time during initial stage and isothermal oxidations by thermogravi metric analysis (SETARAM Setsys Evolution 18, precision mass 0.4 lg). The sur- faces and cross-sec tion morphologies of the two oxide scales were observed by SEM (SSX-550). For cross-sec tional SEM investigatio n,the surfaces of oxide scales were protected in advance by electro- less Ni plating. We investiga ted the valence electron configurationsof BN-SS30 4 and CP-SS304 at temperature range from 80 �C to 500 �C, the oxide scales on BN-SS30 4 and CP-SS304 after isother- mal oxidation for 24 h at room temperature by UPS. The samples for UPS and high-temperat ure oxidation were polished using SiC papers of successive grades up to 2000#, cleaned by distilled water, degreased with acetone, and then dried with hot hair in or- der to eliminate the effect of different surface microstructures on their oxidation and UPS. We obtained the atomic percentages of O (O0 + O2�), Cr (Cr0 + Cr 3+), Mn (Mn0 + Mn 4+), the binding energies of Cr2p3/2 (Cr0 and Cr 3+) and Mn2p3/2 (Mn0 and Mn 4+) in the oxide scales on BN-SS304 and CP-SS304 along depth profile by XPS (ESCALAB250, Al Ka line, 1486.6 eV). Depth profiling on the two oxide scales was performed over an area of 1.5 � 1.5 mm2 byAr+ sputtering (2 � 103 eV). Pure Au and Ag standard samples were used to calibrate the binding energies by setting the Au4f7/2 and Ag3d5/2 peaks at of 83.98 ± 0.02 eV and 368.26 ± 0.02 eV respec- tively. The Fermi edge was calibrated using pure Ni and setting the binding energy at 0.00 ± 0.02 eV. The base pressure of ESCA- LAB250 system was better than 2.8 � 10�10 Pa. We analyzed quan- titatively the XPS and UPS of the two oxide scales by XPSPEAK 4.1 software. The chemical compositions of BN-SS30 4 and CP-SS304 were shown in our previous work [28].

3. Results and discussion

3.1. Initial stage and isothermal oxidation of BN-SS304 and CP-SS304

Fig. 1 shows oxidation kinetics of BN-SS304 and CP-SS304 dur- ing the initial stage oxidation from room temperature to 900 �C.Although the grain size of BN-SS304 is evidently smaller than that of CP-SS304 and the volume fraction of grain boundari es of BN-

SS304 is larger than that of CP-SS304 [27], the oxidation rate of BN-SS30 4 was obviously less than that of CP-SS304 during initial stage oxidation. Fig. 2 presents the oxidation kinetics of BN- SS304 and CP-SS304 during the isothermal oxidation at 900 �Cfor 24 h. The oxidation rate of BN-SS30 4 was also evidently less than that of CP-SS304 during isothermal oxidation. None of oxide scales spalled away from BN-SS304 and CP-SS304 on cooling to room temperature. The oxidation kinetics of BN-SS304 and CP- SS304 follows a parabolic law given by [29]

ðDm=SÞ2 ¼ aþ kpt ð1Þ

where Dm, S, a, kp and t denote mass gain, the surface area of spec- imen, a constan t, the parabolic rate constant (g2 cm�4 s�1) and oxi- dation time respective ly. We used the parabolic rate law to interpret the rate data as shown in Fig. 3. Fig. 3 is not perfectl y par- abolic nature . It is quite acceptable to regard these as approxim ately parabolic. The upper graph is gently curved and the lower graph is more strongly curved, which may result from the factor that the two oxide scales did not cover the entire surfaces of BN-SS3 04 and CP-SS304 at the initial stage of isotherma l oxidation. kBN

p andkCP

p present the parabolic rate constants of BN-SS3 04 and CP- SS304 respectivel y, they are 2.236 � 10�6 and 1.004 � 10�5 mg2 -cm�4 s�1 respective ly. kBN

p is only about one fifth of kCP p . kBN

p andkCP

p were obtained from the data of isotherma l oxidati on range from 4 h to 24 h and from 2 h to 24 h for BN-SS304 and CP-SS304 respec- tively. Fig. 4a and b denote the surface morph ologies of the oxide

Fig. 3. The quadratic plot of mass variation versus time for BN-SS304 and CP-SS304 in isothermal oxidation at 900 �C for 24 h.

Fig. 4. The surface morphologies of oxide scales on BN-SS304 (a) and CP-SS304, (b)after isothermal oxidation at 900 �C for 24 h.

Fig. 5. The cross-sectional morphologies of oxide scales on BN-SS304 (a) and CP- SS304, (b) after isothermal oxidation at 900 �C for 24 h.

66 S.G. Wang et al. / Corrosion Science 72 (2013) 64–72

scales on BN-SS3 04 and CP-SS30 4 respective ly. Fig. 5a and b show the cross-sectio nal SEM morph ologies of the oxide scales on BN- SS304 and CP-SS304 respec tively. Obvio usly, the oxide scale on BN-SS304 was thinner than that on CP-SS304, the densities of void and defect inside the oxide scale on BN-SS304 were less than those on CP-SS304, which agrees with the results of initial and isotherma loxidations in Figs. 1 and 2. These results mean that O2 adsorptio n

and diffusion , Mn and Cr atoms diffusion in the oxide scale on BN-SS3 04 were more difficult than those on CP-SS30 4.

3.2. The XPS of oxide scales on BN-SS304 and CP-SS304

Fig. 6a–d denote the XPS of Fe2p and Ni2p on the two oxide scales during different Ar + sputter times. There were no Fe, Ni atoms and their cations during depth profiling by Ar + sputteringfor 1 h due to no XPS peaks of Fe2p and Ni2p in Fig. 6, which means that the two oxide scales did not entirely leave their surfaces after 1 h Ar + sputtering and that no oxidation and diffusion of Fe and Ni atoms occurred during oxidation. Fig. 7a and b show the XPS of Mn2p and Cr2p on the oxide scales of BN-SS304 and CP-SS304 after Ar+ sputtering for 1 h respectively . In the two oxide scales, there are Mn 0, Mn 4+, Cr 0 and Cr 3+ during different Ar + sputtering times accordin g to Fig.7. Cr and Mn atoms in the two oxide scales came from BN-SS30 4 and CP-SS304, and O2 came from air. These results indicate that Cr and Mn atoms diffused into the two oxide scales during oxidation. During different Ar + sputtering times, the binding energies of Mn 02p3/2, Mn 4+2p3/2, Cr 02p3/2 and Cr 3+2p3/2 are 640.74 eV and 640.65 eV, 641.92 eV and 641.89 eV, 575.01 eV and 574.94, 576.81 eV and 576.72 eV for BN-SS304 and CP-SS304 with error less than 0.11 eV respectively . We think that the binding energies of Mn 02p3/2, Mn 4+2p3/2, Cr 02p3/2 and Cr 3+2p3/2 on the oxide scale of BN-SS304 were the same as those of the correspond- ing elements on CP-SS304 within the error of ESCALAB250. Fig. 8a–c present the atomic percentages of O (O0 + O2�), Cr (Cr0 + Cr 3+), Mn

Fig. 6. The XPS of Fe2p and Ni2p elements on oxide scales on BN-SS304 and CP-SS304 (a–d) after isothermal oxidation at 900 �C for 24 h.

Fig. 7. The XPS of Mn2p (a) and Cr2p (b) elements on oxide scales on BN-SS304 and CP-SS304 after isothermal oxidation at 900 �C for 24 h.

S.G. Wang et al. / Corrosion Science 72 (2013) 64–72 67

(Mn0 + Mn 4+) in the two oxide scales along depth profiling. On the original surfaces of oxide scales without Ar + sputtering, the atomic percentages of Cr and Mn elements on the oxide scale of BN-SS30 4were less than those of CP-SS304, while the atomic percentage of Oelement on the oxide scale of BN-SS304 was larger than that of CP- SS304. Fig. 9a–c are the atomic ratios of the ion to its correspond- ing element (atom + ion) for O, Cr and Mn elements respectively.Cr3+/(Cr0 + Cr 3+), Mn 4+/(Mn0 + Mn 4+) and O2�/(O0 + O2�) on the ori- ginal surface of oxide scale on BN-SS304 were about 4% less than,nearly the same as and 8.9% larger than those on the original sur- face of oxide scale on CP-SS304 respectively, which indicates that O2 adsorption on the oxide scale of CP-SS304 was stronger than that on the oxide scale of BN-SS304. The atomic ratio of O2�/(O0 + O2�) in the oxide scale of BN-SS30 4 was larger than and nearly the same as (near 100%) that of CP-SS304 before and after

Ar+ sputtering for 800 s respectively from Fig. 9a. This result indi- cates that O2 adsorption on the oxide scale of CP-SS304 was stron- ger than that on the oxide scale of BN-SS30 4 due to the more chemical activity of the oxide scale on CP-SS304 and the large number of void and defect in the oxide scale on CP-SS304, and that it was more difficult for O2 to diffuse into the oxide scale on BN- SS304 and oxide scale/BN-SS 304 interface as shown in Fig. 5. The atomic ratios of Mn 4+/(Mn0 + Mn 4+) and Cr 3+/(Cr0 + Cr 3+) in the oxide scale of BN-SS30 4 were about 2.5% and 2.0% smaller than those of CP-SS304 respectively . The thickness of oxide scale on BN-SS30 4 was about one half of that on CP-SS304 from Figs. 2and 5. Therefore, the diffusion rates of Cr and Mn atoms through its oxide scale and from bulk alloy to oxide scale/alloy interface,the oxidation rates of Cr and Mn atoms for BN-SS30 4 were smaller than those for CP-SS304.

Fig. 8. The atomic percentages of O (O0 + O2�) (a), Cr (Cr0 + Cr 3+) (b) and Mn (Mn0 and Mn 4+) (c) on two the oxide scales after isothermal oxidation at 900 �C for 24 h.

Fig. 9. The atomic ratios of O2�/(O0 + O2�) (a), Cr 3+/(Cr0 + Cr 3+) (b) and Mn 4+/(Mn0 + Mn 4+) (c) on two the oxide scales after isothermal oxidation at 900 �C for 24 h.

68 S.G. Wang et al. / Corrosion Science 72 (2013) 64–72

Fig. 10. The UPS of BN-SS304 and CP-SS304 (a) and (b) at different temperatures from 80 �C to 500 �C.

Fig. 11. The UPS of oxide scales on BN-SS304 (a) and CP-SS304 (b) after isothermal oxidation at 900 �C for 24 h.

Table 1The binding energies and their weights of valence electrons of the two oxide scales on BN-SS304 and CP-SS304.

Valence electron Binding energy (eV) Weight (%)

BN-SS304 CP-SS304 BN-SS304 CP-SS304

Cr–O 13.59 13.43 7.68 10.63 Mn–O 13.19 12.75 10.59 25.29 Mn 03d 12.42 11.48 28.04 20.01 Cr 03d 10.69 8.97 24.95 21.22 Mn 0 and Cr 0 4s 5.52 4.48 28.72 22.84

S.G. Wang et al. / Corrosion Science 72 (2013) 64–72 69

3.3. The UPS of BN-SS304 and CP-SS304 at different temperatures and the UPS of oxide scales at room temperature

Fig. 10 a and b present the UPS of BN-SS304 and CP-SS304 at temperature range from 80 �C to 500 �C. UPS measureme nt cannot be carried out at higher temperature because it may damage ESCA- LAB250 system. The UPS of BN-SS304 and CP-SS304 at room tem- perature were measured in our previous work, the work function of BN-SS30 4 was 0.2 eV larger than that of CP-SS304 and the weight of 4s–4s valence electrons of BN-SS30 4 was 11% less than that of CP-SS304 [30]. Work function is defined as the minimum potential barrier for one electron leaving the surface, which can di- rectly determine the resistance of valence electrons to transfer from metal bulk to oxidation interface. The state densities of va- lence electrons with smaller/larger binding energy of BN-SS304 were smaller/larger than those of CP-SS304 during different tem- peratures according to Fig. 10 , which means that the valence elec- trons of BN-SS304 were more chemically stable than those of CP- SS304 at temperatures range from 80 �C to 500 �C. These results agree with the high-tem perature oxidation results in Figs. 1–3and Fig. 5. Fig. 11a and b show the UPS of the two oxide scales at room temperature respectivel y. The two oxide scales were made up of Cr 0, Mn 0, Cr 2O3 and MnO 2 due to Cr 0, Cr 3+, O2�, Mn 0 andMn4+ in the two oxide scales, there were the valence electrons in Cr–O and Mn–O bonds (Cr2O3 and MnO 2), 3d and 4s (Mn0 and

Cr0). ECr—Ob—BN, EMn—O

b—BN , ECr—3db—BN , EMn—3d

b—BN , ECr—Mn—4sb—BN and ECr—O

b—CP, EMn—Ob—CP , ECr—3d

b—CP ,

EMn—3db—CP , ECr—Mn—4s

b—CP denote the binding energies of valence electrons in Cr–O, Mn–O, 3d and 4s (Mn0 and Cr 0) in the two oxide scales

on BN-SS304 and CP-SS304 respectivel y. Therefore, ECr—Ob—BN>EMn—O

b—BN >

EMn—3db—BN >ECr—3d

b—BN >ECr—Mn—4sb—BN and ECr—O

b—CP>EMn—Ob—CP >EMn—3d

b—CP >ECr—3db—CP >ECr—Mn—4s

b—CP

are due to the following facts: (1) Cr 2O3 is more chemically stable than MnO 2; (2) Mn atom is more chemically stable than Cr atom;(3) 4s valence electrons of Cr and Mn atoms are difficult to be dis- tinguishe d because of the less difference of binding energy be- tween them in BN-SS304 and CP-SS304. The binding energies and weights of these valence electrons in the two oxide scales

are shown in Table 1. ECr—Ob—BN EMn—O

b—BN , ECr—3db—BN , EMn—3d

b—BN , ECr—Mn—4sb—BN are

0.16 eV, 0.44 eV, 0.94 eV, 1.72 eV and 1.04 eV larger than ECr—Ob—CP,

EMn—Ob—CP , ECr—3d

b—CP , EMn—3db—CP , ECr—Mn—4s

b—CP respectively. It is well known that the larger binding energy of valence electron, the larger resistance for valence electrons to separate from its atom. According to Table 1, the weights of valence electrons in Cr–O and Mn–O bonds in the oxide scale on BN-SS304 are less than those in the oxide

70 S.G. Wang et al. / Corrosion Science 72 (2013) 64–72

scale on CP-SS304 respectively , the weights of 3d and 4s valenceelectrons of Cr 0 and Mn 0 in the oxide scale on BN-SS304 are larger than those in the oxide scale on CP-SS304 respectively because of the following two facts: (1) the atomic percentages of Cr 3+ andMn4+ in the oxide scale on BN-SS304 are less than those in the oxide scale on CP-SS304 respectivel y in Fig. 9b and c; (2) the less atomic ratios of Cr 3+/(Cr0 + Cr 3+) and Mn 4+/(Mn0 + Mn 4+) must re- sult in the larger atomic ratios of Cr 0/(Cr0 + Cr 3+) and Mn 0/(Mn0 + -Mn4+) in the oxide scale on BN-SS304. Therefore, the results in Fig. 11 and Table 1 are consistent with the results in Fig. 9b and c. The UPS results of the two oxide scales in Fig. 11 demonst rate that the oxide scale on BN-SS304 was more chemically stable than that on CP-SS304 as shown in Figs. 1–3, and 5. Fig. 11 is consisten twith the results of high-temperat ure oxidation in Figs. 1–3 and 5.

3.4. Discussion

Our present results demonst rate that the oxidation rates of BN- SS304 during both initial stage oxidation and isothermal oxidation at 900 �C for 24 h were less than those of CP-SS304 respectively.From XPS results, there were no C and N elements in the two oxide scales, the high-temperat ure oxidations of BN-SS304 and CP-SS304 at air were mainly selective oxidation of Cr and Mn. The first step of SS304 oxidation is air (mainly O2) adsorptio n on their surfaces.The less oxidation rate of BN-SS30 4 in initial stage oxidation should be due to the weaker O2 adsorption on its surface in Fig. 8a and Fig. 9a. The diffusion rates of Cr and Mn atoms from bulk to oxidation interface for BN-SS304 were less than those for CP-SS304 during initial stage oxidation accordin g to Figs. 1, 2, 5,8 and 9. Our present results of initial stage oxidation for BN- SS304 were evidently different from other previous results: (1)the short-term, transient oxidation rate was found to decrease with decreasing alloy grain size due to the rapid grain boundary transport of Al and Cr to the oxide/metal interface which promoted to the formation of Cr 2O3 and Al 2O3 [19]; (2) the highest oxidation rate occurred at onset of oxidation, after which the oxidation rate abated [18]; (3) the abundant grain boundaries of nanocrys talline 304 stainless steel greatly enhanced the Cr diffusion to guarantee the stable growth of Cr-rich scale [31].

The high-tempe rature oxidations of BN-SS30 4 and CP-SS304 should be the direct interactio n between O2 and preferential nucle- ating elements (Cr and Mn atoms) before the two oxide scales did not cover their whole surfaces during initial stage oxidation.Therefore, the work function and valence electron configurationsof BN-SS30 4 and CP-SS304 are the main factors to determine their oxidation resistances. The state densities of valence electrons with high binding energies of BN-SS304 were larger than those of CP- SS304, and the state densities of valence electrons with low bind- ing energies of BN-SS304 were less than those of CP-SS304 from 80 �C to 500 �C in Fig. 10 , which means BN-SS304 was more chem- ically stable than CP-SS304 at different temperatures. Therefore,we can explain the reason why the oxidation rate of BN-SS304 was less than that of CP-SS304 during initial stage oxidation by the difference of valence electron configurations between BN- SS304 and CP-SS304 at atomic scale.Accor ding to conventi onal point of view, an effective diffusivit y (Deff) of preferential nucleat- ing elements is a summation of lattice diffusivity DL) and the diffu- sivity of grain boundary (DGB) as given as below [31]

Deff ¼ ð1� f ÞDL þ fDGB ð2Þ

where f is the area proportio n of grain boundar y. Assuming the grain are cubic, f ¼ 2d=d and DGB � DL d and d are grain size and width of grain boundary respective ly). Eq. (2) should be [31]

Deff ¼ DL þ2dd

DGB ð3Þ

Deff increases with the decrement of d. However, the diffusion rates of Cr and Mn atoms of BN-SS304 were less than those of CP-SS304 due to the more chemic al stabilit y of BN-SS3 04 than that of CP- SS304 according to Fig. 10 . Our present results did not agree with Eqs. (2) and (3), which may result from the following facts: (1)the diffusion of Cr and Mn atoms depends on not only the area pro- portion of grain boundary but also the microstruc ture of grain boundar y; (2) the microstruc tures of grain boundar y depend on the product ion technique of materials. Accordin g to Fig. 3, the oxi- dation kinetics of BN-SS304 and CP-SS304 follow a paraboli c law,which indicate s that the growth of oxide scale was controlled by diffusion . For BN-SS304, the diffusion of reactants only includes Cr and Mn atoms diffusion from bulk alloy to oxide scale/bulk alloy interface and through the oxide scale because O2 was mainly ab- sorbed on the surface of its oxide scale in Fig. 8 due to its more com- pact oxide scale in Fig. 5. Meanwhile, for CP-SS304 , the diffusion of reactants includes O2 diffusion throug h its oxide scale due to the larger densities of void and defect in its oxide scale, Mn and Cr atoms diffusion from bulk alloy to oxide scale/bulk alloy interface and through the oxide scale in Figs. 5 and 8. Therefore, the oxidation resistanc e of BN-SS3 04 was enhanced at the angle of the different diffusion processe s of reactants between BN-SS304 and CP-SS304.

The chemical stability of oxide scale is another important factor to be able to affect oxidation resistance. From Fig. 11 and Table 1,the binding energies of valence electrons of the oxide scale on BN- SS304 were larger than those of their correspondi ng valence elec- trons of the oxide scale on CP-SS304. The more chemical stability of the oxide scale on BN-SS304 can result in not only its more dif- ficulty to evaporate and adsorb O2, but also the larger resistance to prevent form oxide scale growing further. The more chemical sta- bility of the oxide scale on BN-SS304 also manifested the following facts: (1) its smaller diffusion and oxidation rates of Cr and Mn atoms from Figs. 1, 2, 4, 8 and 9; (2) the weaker O2 adsorptionand diffusion on the oxide scale of BN-SS304 according to Figs. 8and 9. After oxide scales covered their whole surfaces, the oxida- tion processes of BN-SS304 and CP-SS304 were associated with (1) O2 adsorption and diffusion on oxide scale; (2) the diffusion of preferential oxidation elements through oxide scale and from al- loy to the oxide scale/alloy interface; (3) the oxidation of Mn and Cr atoms. In Fig. 8, the atomic ratio of O2�/(O2 + O2�) of the oxide scale on BN-SS304 was larger than that of the oxide scale on CP- SS304 for Ar + sputtering for less than 800 s, which means the less densities of void and defect in the oxide scale on BN-SS304 in Fig. 5,and the weaker O2 adsorptio n (Figs. 8 and 9) inside the oxide scale on BN-SS30 4.

We usually study high-temperat ure oxidation at the angle of conventi onal parameters (such as dislocation, temperat ure, cold work, grain boundary, preferential nucleating elements, grain size,oxygen pressure , and the addition of rare earth elements). For example, dislocations and grain boundari es provide special diffu- sion paths for vacancie s and solute, the decrement of grain size can influence resulting oxide composition and morphology , and thereby alter oxidation behavior [32]. The adherence of oxide scale to alloy may be increased considerabl y by the presence of a fine,uniform dispersion of precipitates in alloy [33]. The composition of oxidizing atmosphere had no effect on AIS439 oxidation behav- ior, contrary to the case of AISI 304, the growth mechanis m of oxide scales did not depend on oxygen pressure [29]. Different Ce additions in 25Cr20Ni alloy can result in positive or negative ef- fect on its oxidation resistance at 950 �C [5]. The cold-worked stainless steels accelerated oxidation due to highly dislocated microstru cture [34]. From 400 to 600 �C in oxygen, cold-work ed Fe oxidized faster than annealed Fe, cold work had no effect and solid scale only was formed at 650 �C [19]. The oxidation resistance of cold-worked pure metals deceased due to the dislocations in metal which acted as sinks for vacancies generated by oxidation

S.G. Wang et al. / Corrosion Science 72 (2013) 64–72 71

reaction [35]. Cold-worked simple alloys increased their oxidation resistance because dislocatio ns increased diffusion of solute to oxi- dation front [36]. As the oxide scale thickens the effect of the underlying substrate grain size becomes less important and even- tually negligible, substrate grain size has a significant influenceonly on transient oxidation stage [18]. Ultrafine-grained Ni 3Alexhibited increased cyclic oxidation resistance at 900 �C in com- parison with its polycrystall ine counterpar t [11]. The long-term oxidation behavior of fine-grained NiAlY was found to be indepen- dent of the grain size of underlying alloy. The short-term, transient oxidation rate was found to decrease with decreasing alloy grain size due to the rapid grain boundary transport of Al and Cr atoms to oxide/metal interface [18].

In fact, the nature of high-temperat ure oxidation is the ex- change process of valence electrons among different atoms or ions at atomic scale. The four factors that can directly affect this process are (1) the work function and binding energies of valence electrons of metal materials; (2) O2 adsorption and diffusion; (3) the diffu- sion of preferential nucleating elements through oxide scale and from metal base to oxidation interface; (4) the microstru cture and chemical stability of oxide scale. The conventional paramete rs mentioned in the former paragraph can influence high-tempe ra- ture oxidation resistance through their effects on the four factors.It is possible that one conventi onal parameter acts as the positive and negative effects on the high-temperat ure oxidation resistances rather than monotonous effect with the continuous variation of this conventi onal parameter because this conventional parameter cannot directly or monotonous ly influence the work function of metal, the exchange of valence electrons among different atoms or ions, the diffusion of O2 and preferential nucleating elements.Therefore, it is reasonabl e to understand that these conventional parameters mentioned above cannot monotonous ly affect high- temperature oxidation resistance [11,18,19,29,32 –36], which indi- cates that these conventional parameters cannot be considered as the intrinsic parameters associated with high-tempe rature oxida- tion. We think that the work function and valence electron config-urations of metal, the compactness and chemical stability of oxide scale, the O2 adsorption and diffusion, the diffusion and chemical stability of preferential nucleating elements can be defined as the intrinsic parameters associate d with high-temperat ure oxidation.Therefore, we should investiga te the effects of these conventional parameters on the intrinsic parameters associate d with high-tem- perature oxidation if we want to understa nd the effects of these conventional parameters on high-tem perature oxidation further.According to our present results in Figs. 1–11, Table 1 and above discussion, we found that the results about the valence electron configurations of BN-SS304, CP-SS304 and the two oxide scales agree with the results of high-tem perature oxidation of BN- SS304 and CP-SS304, which means that it is reasonable and reliable to understand the high-tem perature oxidation of metal materials at the angle of their valence electron configurations at atomic scale.

According to our present results, the enhanced high-tem pera- ture resistance of BN-SS30 4 did not result from the faster diffusion and oxidation of preferential nucleating Cr and Mn atoms during initial oxidation from room temperature to 900 �C and isothermal oxidation at 900 �C from Figs. 1–3, 5, 8 and 9. Our present results were different from the popular point of views that the enhanced oxidation resistance was due to rapidly forming compact oxide scale in initial stage of oxidation and the faster Cr diffusion rate from grain boundaries for stainless steels [15,17,18,22–25] . The oxidation resistances of BN-SS30 4 and CP-SS304 depend on their chemical stability, the microstructur es and chemical stability of their oxide scales, the diffusion of O2, Cr and Mn atoms. The chem- ical stabilities of BN-SS304, CP-SS304 and their oxide scales de- pend on their valence electron configurations (Figs. 10 and 11).The diffusion of O2, Cr and Mn atoms depend on the compactness

and chemical stability of oxide scale, the diffusion of Cr and Mn atoms depend on the work function of metal materials and the compactness of oxide scale. The nature of electrochemical corro- sion is also the exchange process of valence electrons among differ- ent atoms or ions at atomic scale. We can understa nd electrochemi cal corrosion at the angle of valence electron configu-rations of metal materials at atomic scale in our previous work [30,37]. Therefore, it is reasonabl e to understand high-tem perature oxidation at the angle of their valence electron configuration. BN- SS304 produced by severe rolling techniqu e also owns the en- hanced uniform corrosion and pitting corrosion resistances [28,30,38], the less thermal expansion from liquid nitrogen to room temperature [27], the enhanced biocorrosion property and well-beh aved in vitro cytocompati bility in comparison with CP- SS304 [39]. Severe rolling techniqu e can simultaneously improve several properties rather than one property at the cost of another property , we can produce large dimension bulk nanocrys talline materials by this technique with low cost [26,28], which is its advantag e different from other preparation techniques of nano- crystalline materials.

4. Conclusion s

The high-temperat ure oxidation resistance of bulk nanocrystal- line 304 stainless steel produced by severe rolling techniqu e was enhanced in comparison with its conventi onal polycrystall ine counterpart at 900 �C for 24 h. The enhanced high-temperat ure resistance of BN-SS30 4 displayed in both initial stage oxidation from room temperature to 900 �C and isothermal oxidation at 900 �C for 24 h rather than only in isothermal oxidation. The en- hanced oxidation resistance of BN-SS304 in initial stage was attrib- uted to its larger work function, its smaller/larger state density of valence electrons with smaller/lar ger binding at atomic scale. The enhanced isothermal oxidation resistance of BN-SS304 was due to its more compact oxide scale with more chemical stability, the weaker O2 adsorption and diffusion on its oxide scale, and the weak diffusion of Cr and Mn atoms with more chemical stability in BN-SS304 and its oxide scale. The constant of parabolic rate of BN-SS30 4 was only about one fifth of CP-SS04. The enhanced high-tem perature resistance of BN-SS30 4 did not depend on the faster Cr diffusion rate from its grain boundaries and rapidly form- ing compact oxide scale in initial stage of oxidation.

Acknowled gements

Authors are grateful to the financial support of Natural Sciences of Foundation of China, Contract Nos. 50771098 and 51171199,and the National Basic Research Program (No. 2010CB934603 ) of China, Ministry of Science and Technology China.

References

[1] L. Zhou, D.G. Lee, R.D. Arnell, D. Johnson, A. Chew, Growth behavior of oxide forming on a sputtered oxygen-enriched type 304 stainless steel, J. Mater. Eng.Perform. 4 (1995) 242–247.

[2] E.A. Gulbransen, K.F. Andrew, Oxidation studies on 304 stainless steel, J.Electrochem. Soc. 109 (1962) 560–564.

[3] P.Y. Hou, J. Stringer, The effect of reactive element additions on the selective oxidation, growth and adhesion of chromia scales, Mater. Sci. Eng. A – Struct.202 (1995) 1–10.

[4] J. Stringer, The reactive element effect in high-temperature corrosion, Mater.Sci. Eng. A – Struct. 120 (1989) 129–137.

[5] M.Z. Shao, L.H. Cui, Y.J. Zheng, L.L. Xing, Effect of cerium addition on oxidation behavior of 25Cr20Ni alloy under low oxygen partial pressure, J. Rare Earth 30 (2012) 164–169.

[6] X. Peng, J. Yan, Z. Dong, C. Xu, F. Wang, Discontinuous oxidation and erosion–oxidation of a CeO 2-dispersion-strengthened chromium coating, Corros. Sci. 52 (2010) 1863–1873.

72 S.G. Wang et al. / Corrosion Science 72 (2013) 64–72

[7] J.B. Yan, Y.M. Gao, L. Liang, Z.Z. Ye, Y.F. Li, W. Chen, J.J. Zhang, Effect of yttrium on the cyclic oxidation behavior of HP40 heat-resistant steel at 1373 K, Corros.Sci. 53 (2011) 329–337.

[8] C.T.J. Low, R.G.A. Wills, F.C. Walsh, Electrodeposition of composite coatings containing nanoparticles in a metal deposit, Surf. Coat. Technol. 201 (2006)371–383.

[9] F.J. Honey, E.C. Kedward, V. Wride, The development of electrodeposits for high temperature oxidation/corrosion resistance, J. Vac. Sci. Technol. A 4 (1986)2593–2597.

[10] X. Yang, X. Peng, C. Xu, F. Wang, Electrochemical assembly of Ni–xCr–yAl,nanocomposites with excellent high-temperature oxidation resistance, J.Electrochem. Soc. 156 (2005) C167–C175.

[11] X. Peng, M. Li, F. Wang, A novel ultrafine-grained Ni 3Al with increased cyclic oxidation resistance, Corros. Sci. 53 (2011) 1616–1620.

[12] M. Zandrahimi, J. Vatandoost, H. Ebrahimifar, Al, Si, and Al–Si coatings to improve the high-temperature oxidation resistance of AISI 304 stainless steel,Oxid. Met. 76 (2011) 347–358.

[13] R. Thanneeru, S. Patil, S. Deshpande, S. Seal, Effect of trivalent rare earth dopants in nanocrystalline ceria coatings for high-temperature oxidation resistance, Acta Mater. 55 (2007) 3457–3466.

[14] J.B. Yan, Y.M. Gao, Y.D. Shen, F. Yang, D.W. Yi, Z.Z. Ye, L. Liang, Y.Q. Du, Effect of yttrium on the oxide scale adherence of pre-oxidized silicon-containing heat- resistant alloy, Corros. Sci. 53 (2011) 3588–3595.

[15] C.S. Giggins, F.S. Pettit, Oxidation of Ni–Cr alloys between 800 �C and 1200 �C,Trans. TMS-AIME 245 (1969) 2495–2502.

[16] C.S. Giggins, F.S. Pettit, Effect of alloy grain size and surface deformation on selective oxidation of chromium in Ni–Cr alloys at temperature of 900 �C and 1100 �C, Trans. TMS-AIME 245 (1969) 2509–2512.

[17] G.J. Yurek, D. Eisen, A. Garratt-Reed, Oxidation behavior of fine-grained rapidly solidified 18-8 stainless steel, Metall. Mater. Trans. A 13 (1982) 473–485.

[18] J.G. Goedjen, D.A. Shores, The effect of alloy size on the transient oxidation behavior of an alumina-forming alloy, Oxid. Met. 37 (1992) 125–142.

[19] D. Caplan, M. Cohen, Effect of cold work on the oxidation of iron from 400 to 650 �C, Corros. Sci. 6 (1966) 321–335.

[20] W.J. Kuang, X.Q. Wu, E.H. Han, L.Q. Ruan, Effect of nickel ion from autoclave material on oxidation behavior of 304 stainless steel in oxygenated high temperature water, Corros. Sci. 53 (2011) 1107–1114.

[21] J. Rawers, Oxidation characteristics of Fe–18Cr–18Mn-stainless steel alloys,Oxid. Met. 74 (2010) 167–178.

[22] Z. Huang, X. Peng, C. Xu, F. Wang, Effect of alloy nanocrystallization and Cr distribution on the development of a chromia scale, J. Electrochem. Soc. 156 (2009) C95–C102.

[23] M.D. Merz, Oxidation resistance of fine-grained sputter-deposited 304 stainless steel, Metall. Trans. A 10 (1979) 71–77.

[24] D.R. Baer, M.D. Merz, Differences in oxides on large-grained and small-grained 304 stainless steel, Metall. Mater. Trans. A 11 (1980) 1973–1980.

[25] X. Peng, Nanoscale assembly of high-temperature oxidation-resistant nanocomposites, Nanoscale 2 (2010) 262–268.

[26] S.G. Wang, C.B. Shen, K. Long, H.Y. Yang, F.H. Wang, Z.D. Zhang, Preparation and electrochemical corrosion behavior of bulk nanocrystalline ingot iron in HCl acid solution, J. Phys. Chem. B 109 (2005) 2499–2503.

[27] S.G. Wang, R.J. Huang, Y. Mei, K. Long, L.F. Li, Z.D. Zhang, The linear thermal expansion of bulk nanocrystalline Al and SS304 at low temperature, Physica B406 (2011) 2758–2762.

[28] N. Li, Y. Li, S.G. Wang, F.H. Wang, Electrochemical corrosion behavior of nanocrystallized bulk 304 stainless steel, Electrochim. Acta 52 (2006) 760–765.

[29] A.M. Huntz, A. Reckmann, C. Haut, C. Sévérac, M. Herbst, F.C.T. Resende, A.C.S.Sabioni, Oxidation of AISI 304 and AISI 439 stainless steels, Mater. Sci. Eng. A –Struct. 47 (2007) 266–2276.

[30] S.G. Wang, M. Sun, K. Long, The enhanced even and pitting corrosion resistances of bulk nanocrystalline steel in HCl solution, Steel Res. Int. 83 (2012) 800–807.

[31] X. Peng, J. Yan, Y. Zhou, F. Wang, Effect of grain refinement on the resistance of 304 stainless steel to breakaway oxidation in wet air, Acta Mater. 53 (2005)5079–5088.

[32] M.K. Hossain, Effects of alloy microstructure on the high temperature oxidation of an Fe-10% Cr alloy, Corros. Sci. 19 (1979) 1031–1045.

[33] I.M. Allam, H.C. Akuezue, D.P. Whittle, Influence of small Pt additions on Al 2O3

scale adherence, Oxid. Met. 14 (1980) 517–530.[34] S. Lozano-Perez, D.W. Saxey, T. Yamada, T. Terachi, Atom-probe tomography

characterization of the oxidation of stainless steel, Scripta Mater. 62 (2010)855–858.

[35] D. Caplan, G.I. Sproule, Effect of oxide grain structure on the high temperature Cr, Oxid. Met. 9 (1975) 459–472.

[36] G.H. Gilmer, H.H. Parrell, Grain-boundary diffusion in thin-films. 1. Isolated grain-boundary, J. Appl. Phys. 47 (1976) 3792–3798.

[37] S.G. Wang, M. Sun, K. Long, The electrochemical corrosion of bulk nanocrystalline ingot iron in HCl solutions with different concentrations,Mater. Chem. Phys. 127 (2011) 459–463.

[38] C. Pan, L. Liu, Y. Li, S.G. Wang, F.H. Wang, Passive film growth mechanism of nanocrystalline 304 stainless steel prepared by magnetron sputtering and deep rolling techniques, Electrochim. Acta 56 (2011) 7740–7748.

[39] F.L. Nie, S.G. Wang, Y.B. Wang, S.C. Wei, Y.F. Zheng, Comparative study on corrosion resistance and in vitro biocompatibility of bulk nanocrystalline and microcrystalline biomedical 304 stainless steel, Dent. Mater. 27 (2011) 677–683.