HAL Id: jpa-00226891
https://hal.archives-ouvertes.fr/jpa-00226891
Submitted on 1 Jan 1987
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
INITIAL STAGES OF OXIDATION OF COPPER AND COPPER-IRON ALLOY
K. Hono, T. Sakurai, H. Pickering
To cite this version:
K. Hono, T. Sakurai, H. Pickering. INITIAL STAGES OF OXIDATION OF COPPER AND COPPER-IRON ALLOY. Journal de Physique Colloques, 1987, 48 (C6), pp.C6-505-C6-510.
�10.1051/jphyscol:1987683�. �jpa-00226891�
INITIAL STAGES O F OXIDATION O F C O P P E R AND COPPER-IRON ALLOY
K. Hono, T. ~ a k u r a i * a n d H.W. Picketing
D e p a r t m e n t of Materials S c i e n c e & Engineering, T h e Pennsylvania S t a t e University, University Park, PA 16802, U.S.A.
*The I n s t i t u t e of Solid S t a t e Physics, The University of Tokyo, Roppongi, Minato-ku, Tokyo 106, J a p a n
Abstract
The initial oxidation stages of copper and copper-iron alloys have been studied by the Atom- Probe Field Ion Microscope (AP-FIM). The depth profiles of copper oxide formed in-situ were successfully determined. The composition of the oxide was close to the value of Cu20, but a large concentration fluctuation was found. The interface between the oxide and the metal was found to be diffuse, the width of which was approximately 2 nm. Annealing the supersaturated Cu-1.2at%Fe alloy tip in-situ a t 1000 K in UHV caused enrichment of the surface in iron atoms due to heterogeneous precipitation. No indication of surface segregation was found in Cu-O.OGat%Fe or aged Cu-1.2at%Fe alloys. In-situ oxidation of the supersaturated Cu-l.2at%Fe alloy led to the formation of both external and internal oxides of iron.
I. Introduction
Much interest has been devoted to the initial oxidation stage of various metals and alloys, because the surface oxide plays .a dominant role in many technologically important surface and interface phenomena such a s catalytic reactions, corrosion susceptibility, and electron transport phenomena. Thus many modern surface analytical techniques have been employed to characterize the surface oxide layers, such a s Auger Electron Spectroscopy (AES), Secondary Ion Mass Spectroscopy (SIMS), X-ray Photoelectron Spectroscopy (XPS), and Low Energy E!ectron Diffraction (LEED). Atom-probe Field Ion Microscopy (AP-FIM) was first employed for studying oxide formation on metal surfaces by Ng et al. [ l l , and it was shown that the instrument is suited to obtain concentration depth profiles with excellent spatial resolution.
Later, several works were done to investigated the initial stage of oxidation of metals and semi-conductors by AP-FIM 12-61. However, because of the limited number of collected atoms, the d a t a did not seem to be statistically meaningful. Although a recent attempt to analyze the metal oxide using the imaging atorn- probe gave a statistically meaningful result [31, the study only showed the stoichiometry of the oxide.
The merit of the atom-probe is that it can determine the concentration depth profile for any atomic species with excellent spatial resolution. Thus, its application to alloy oxides should give more useful information a s to the initial oxidation stage. In the present study, the initial oxidation stages of copper and copper- iron alloys have been studied by AP-FIM.
Pure copper, 99.999 5% Cu wire with a diameter of 0.5 mm, was purchased from Johnson Matthey Inc.
The wire was further drawn to 0.2.5 mm and annealed in high vacuum a t 1223 K for 30 min.
Cu-O.OGat%Fe and Cu-1.2at%Cu alloys were supplied from AMAX specialty Metals Corporation in rod form, which was made from oxygen free electric copper and high purity iron. The alloys were also drawn to wire of 0.25 mm in diameter, annealed in high vacuum a t 1223 K and quenched into water.
Field Ion Microscopic observation was performed using either Ne or W2 as the imaging gas. The details of
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1987683
C6-5 06 JOURNAL DE PHYSIQUE
the Atom-Probe FIM used for the present study are d scribed elsewhere [71. The atom-probe analyya was performed under a hydrogen atmosphere of lXlO-' Torr or under Ultra High Vacuum of 1x10- Tom. The pulse ratio (VpNdc) was between 0.10 and 0.15.
5 Results
3.1 Oxidation of Pure C o p p e r
It was often observed that the oxide layer spalled from the sample, so it is doubtful that the ions were evaporated smoothly from the field ion tip. Actually, when a n oxidized sample was analyzed under UHV, field evaporation seemed to proceed irregularly. The oxide scale often evaporated in a cluster, and very high mass was registered with the timer. Also the chance of tip rupture was much higher in the oxidized sample. In order to avoid such cases, a low pressure of hydrogen gas was used during the analysis of the copper oxide layer. I t is well known that hydrogen adsorption can decrease the evaporation fie d of most a t a m E81. The atom-probe mass spectrums of pure copper and oxidized copper under lXlO-' Torr H2 are shown in Fig. I-a and 1-b, respectively. As shown in Fig.1-a, more than 99.5% of th Cu ato were
#$tected a s ingly charged ions, most of which were hydrider The detected ions were S3Cut ' k u r , CUH$, "CuH$, and a small number of CuH, CuHg, and CuH4 singly charged ions. NO; that the
~t is not background noise is negligibly smaU and neither OH; nor OH$ are detected a t all.
Although+
shown here, more than 50 % of the Cu atoms were detected a s doubly charged ions, i.e., a s Cu
,
when Cu wa$ analyzed in UHV. The mass spectrum of the oxidized copper (Fig.1-b) shows strong peaks of CuOHX and OH:. As noted above, no OHx (x= 1 to 3) peaks were detected for pure Cu analyzed in the same condition. Therefore, it can be concluded that these oxygen and hydroxygen peaks a r e due to the oxygen ions of the copper oxide which were bound with hydrogen during either the field ionization or evaporation processes. Thus, when the data were analyzed, these ions were counted a s oxygen.An integral concentration profile of Cu oxidized a t 573 K for 1000 sec is shown in Fig. 2. Oxygen signals were detected throughout a relatively long data chain, approximately 2000 total detected signals. The average concentration of oxygen in this oxide layer was determined to be approximately 42 at%, which is between 5 0 at% for CuO and 33.3 at% for CuzO. It is well documented that C u t 0 is formed a t this temperature (9); thus one would expect that this oxide layer is also Cu2O. The concentration profile of this oxide layer which is formed a t 573 K under 1 Torr 0 2 is shown in Fig. 3. Each layer consists of 20 atoms, which roughly corresponds to the average number of signals per layer. Therefore, each layer corresponds to about 0.2 nm in the depth direction. The width of the error bars is two times the standard deviation of each layer. I t is seen from this figure that the oxide has a large concentration fluctuation of oxygen. The fluctuation is larger than the statistical fluctuation, which means there is actually a concentration fluctuation in the oxide. If this oxide is stoichiometric, the oxygen concentration should be 33.3 at%, but the experimental value was 42 at%. There is a possibility that the atom-probe did not give the actual value because of miscounting copper signals. But a large fluctuation of oxygen concentration can not be explained by this artifact. The details will be discussed later.
The oxygen concentration gradually decreases a t the interface. Since the spatial resolution of atom-probe analysis is very high, it is impossible that the broadening of the interface is due to some artificial effect a s in the case of other modern surface techniques which use spattering for obtaining depth profile.
Therefore, this broad interface must be a real feature of the CuICu2O interface. Since a data point roughly corresponds to a n atomic layer on the { I l l ) plane, the width of the interface is in a range of 2 nm.
To check the quantitativeness of the atom-probe analysis, many unsuccessful attempts were made to analyze thick oxide layers formed a t high temperature by the AP-FIM. The problem was that the oxide layers formed a t high temperature were so brittle that the oxide layer spalled before the analysis.
3.2. Oxidation of Copper-Iron Alloys
Two types of Cu-Fe alloys were used for this study. Since no indication of surface segregation of iron was found for the Cu-O.OGat%Fe alloy, the effect of the small amount of Fe atoms on the oxidation behavior was not clearly detected by the atom-probe. Therefore, only the result from the Cu-1.2atBFe alloy is presented.
The integral concentration depth profile of the alloy mnealed in-situ a t 1000 K for 30 sec in C'HV is
concluded that Fe enrichmen on the surface extends to several layers in Fig. 4. The number of Fe enriched surface layers varied from one to several depending on the annealing conditions. However, the composition of the Fe enriched surface layer was always determined to be 100 a t % Fe. Note that a precipitate of Fe is also observed in the bulk region and that the composition is 6 0 at% Fe on average, which is significantly lower than that of the surface layer. Since, surface segregation of Fe in Cu-Fe alloy is not theoretically expected [lo], interpretation of this surface enrichment must be done carefully. Thus, a sample tip was prepared from a n aged alloy in which thermal equilibrium was attained. Then, there was no indication of surface enrichment from the matrix of the aged alloy; in other words, F e does not segregate on the surface from the thermally equilibrated solid solution. Therefore, the surface enrichment of Fe shown in Fig. 4 is not the usual surface segregation which is driven by a decrease in the surface energy due to the enrichment of the solute Fe atoms on the surface. Rather, it is a sort of heterogeneous precipitation of iron, the driving force of which is to decrease the bulk free energy of the alloy.
A typical example of a n i y g r a l depth profile obtained from a Cu-1.2at%Fe alloy oxidized in-situ a t 1000 K for 1 min under 1x10' Torr 0 2 is shown in Fig. 5. On the surface, Fe and 0 a r e mainly detected although copper ions a r e also present in the surface phase. The ratio of Fe and 0 is roughly one to one, and if the average is taken in the surface phase, the phase composition is roughly 26at%Cu-33at%Fe-4lat%O. But when the depth profile is carefully examined, it is seen that F e and CU are not associated with each other, i.e. where F e ions are detected, Cu ions a r e not detected. From this fact, it can be concluded that the surface oxide layer consists of two phases: one is FeO and another is Cu2O. Very small CugO islands would exist in the FeO layer. By annealing the Cu-1.2at%Fe alloy in UHV, nearly pure iron phase was formed on the surface a s shown if Fig. 4. On the other hand, CU is incorporated into the surface layer when the alloy tip is annealed in oxygen.
Several concentration depth profiles were taken in the same manner. It was not always, but often the case, that iron oxide was observed not only on the surface but also in the bulk. Such a case is shown in Fig. 6. Approximately 20 ions correspond to 0.2 n m in depth along the < 1 1 1 > direction. In addition to the oxide phase on the surface, another iron oxide is observed a t 3700 signals, i.e. at about 37 nm depth from the surface. Note that a n iron cluster a t 1500 counts (15 nm) is not oxidized although it is located closer to the surface than the other iron cluster. I t is interesting to obtain the concentration depth profile throughout the entire region of internal oxides. Unfortunately, however, the sample tip always ruptured when the internal oxides were detected. Thus, the composition of the internal oxide is not quite clear, but it is expected to be FeO from the partial slope of the integral concentration profile.
4. Discussion
The average composition of the copper oxide formed a t 573 K was measured to be 42 at% of 0 by the atom-probe. One possibility is that the atom-probe miscounted some of the copper atoms during analysis, in which case the apparent oxygen composition would be higher than the actual value. The measured value, 42 at% 0, was reproducible throughout several runs. The copper oxide formed a t room temperature and a t 473 K also had the same oxygen content. In addition to the possibility of a systematic miscounting of copper atoms, it should also be noted that Cu2. 0, where y is 0.0 to 0.5, is
y .
reported to form below 473 K. Wieder and Czanderna [I11 suggested that t h ~ s off-stoichiometry can be explained by a gross defect structure of Cu20, in which one vacancy is introduced in every unit cell.
Another explanation is given by O'Keeffe and Stone [12,13] that the oxide is a mixture of CuO and CupO, i.e. (CuOxCu201-,I. In this case, the valency of Cu should be one and two. Although the valency of a n ion detected by the atom-probe does not necessarily correspond to the valency of a n ion in the sample, it is a n interesting fact that the ratio of C U ~ + / C U + increased when the copper was oxidized a s shown in Fig.
1. Furthermore, the fluctuation of the concentration depth profile shown in Fig. 3 is larger than the statistical errors. From these facts, it may be true that the oxide is a mixture of Cu2O and CuO and has a n off-stoichiometric value of 42 at% 0. To be more conclusive, further work is warranted.
It is reported that a metal rich layer or an oxygen deficient layer is formed on the topmost layer of Fe.
Co, and Ni oxide by Ng et al. [I]. In the present study, such a layer was not found in any of the oxide layers analyzed by the atom-probe. A broad interface between oxide and metal is found in the present study a s well a s in the papers by Ng r t nl. [ I ] and Crnnstoun et nl. (21. The strain due to a difference of the lattice constants a t the metalloxide interphace could be ~ccommodated by the broad (about 2nm) oxygen deficient region a t the interface.
C6-508 JOURNAL DE PHYSIQUE
Based on the theories of segregation, it is not expected t h a t Fe segregates on the surface in Cu-O.O6at%Fe. Instead, the theories for segregation predict Cu segregation on the surface. Thus, the enrichment of Fe on the surface is nothing more than heterogeneous precipitation of Fe on the surface a s is the case for Pt-C alloy [141. The surface layer was found to be always 100 % Fe. On the other hand, an iron precipitate in the bulk region has lower concentration of iron as shown in Fig. 4. As will be reported elsewhere, the composition of precipitates in the bulk a t very early stages of the precipitation w a s found to be lower than 100 % Fe. This means that on the surface the kinetics of the precipitation is much faster than in the bulk, so that the surface precipitate was nearly 100 % Fe from the beginning a s expected from the equilibrium phase diagram. Another possible explanation is that since there is less restriction of interfacial energy in the nucleation stage of iron precipitate on the surface, the nucleus could have the equilibrium value from the earliest stage.
After annealing in vacuum, the surface precipitate was analyzed to be 100 % Fe, so it was expected that the surface oxide would be a n iron oxide. When the alloy was annealed in oxygen, however, the composition of the surface layer was roughly Cu:Fe:O= 1:l:l. The incorporation of Cu atoms in the oxide layer suggests that the surface precipitaion does not precede the formation of the oxide. If so, the surface oxide would be FeO excluding Cu inside the phase. In other words, oxide formation and precipitaion of F e are independent processes. So, precipitation of iron particles and nucleation of oxides proceed almost simultaneously. This idea is also supported by Fig. 6. In spite of the fact that the F e oxide is obsemed a t 3700 counts, the Fe precipitate a t 1500 counts are not oxidized. On the surface oxide layer, Cu2O was incorporated within FeO. Since these phases a r e insoluable in each other, the later oxidation process would lead to the migration of Cu cations, and FeO and Cu2O would form in each layer.
The atom-probe is not suited to study such later stages. Interesting results have been reported by Rapp et al. [15].
5. Conclusion
The Atom-Probe Field Ion Microscope was employed to study the initial oxidation stage of copper and copper-iron alloys.
1) The Cu oxide layer that formed a t 573 K under 1 Torr 0 2 was analyzed by the atom probe. The apparent oxygen concentration in this oxide was 42 at%, which is between the value of the two stoichiometric oxides, CuO and Cu20. A possible explanation for this value was discussed.
2) Fe enriched on the surface when Cu-1.2at%Fe supersaturated solid solution was annealed in UHV a t 873 K for 30 sec, which is interpreted a s a result of heterogeneous precipitation of Fe on the surface.
No indication of surface segregation of Fe from the thermally equilibrate solid solution was obtained.
3) On annealing the Cu-1.2at%Fe a l b y a t 873 K for 30 sec under lXIOd Torr 0 2 , surface oxidation and internal oxidation of Fe was observed. The surface oxidation layer was basically FeO with small islands of CugO.
Acknowledeement
This research has been supported by the International Copper Research Association (INCRA), Contract
#363.
References
1. Y. S. Ng, S. B. McLane, and T. T. Tsong, J. Apl. Phys., 49, 2517 (1978).
2. G. K. L. Cranstoun, D. R. Pyke, and G. D. W. Smith, Appl. Surf. Sci., 2, 375 (1979).
3. G. L. Kellog, J. Catalysis, 92, 167 (1985).
4. T. Adachi, M. Tomiya, T. Kuroda, and S. Nakamura, J. de Phys., Coll. C7 suppl. 11, 47, C7-315.
6. G. Vauer and M. Leishch, J. de Phys., Coll. C7, suppl. 11, 47, C7-189 (1986).
6. I. Kamiya, T. Hashizume, A. Sakai, T. Sakurni and H. W. Pickering, J. de Phys., Coll. C7, suppl.
11, 47, C7-195 (1986).
7. D. R. Hess, Ph. D Thesis, The Pennsylvania State University, 1886.
8. E. W. Muller and T. T. Tsong, Fielrl Ion Microscopy, Elsevier, New York, 1869. p. 166.
9. A. Ronnquist and H. Fiscmeister, J. Inst. Met., 61, 8 9 (1960).
10. F. F. Abraham and C. R. Brundle. J. Vat. Sci. Tech., 18, 306 (19Sl).
11. H. Wieder and A. W. Czundera, J. Chem. Phys., 66, Y 16 (1962).
12. M. O'Keeffe and F. S. Stone, J. Chem. Phys., 36, 501 (1862).
15. Rapp, A. Ezis, and G. J. Yurek, Met. Trans., 4, 1283 (1973).
min
(b)
Fig. 1 Atom-probe mass spectrums of pure copper and oxidixed copper analyzed under 1 ~ 1 0 - ~ Tom H2 Hydrogen ions are excluded.
Fig. 2 Integral oxygen concentratio profile of a copper sample oxidized a t 573 K for 1000 sec in 1 Torr 0 2 .
Number o f D e t e c t e d Cu + 0
Fig. 3 Concentration depth profile of the copper oxide film formed by oxidation a t 573 K for 1000 sec in 1 Torr 0 2 .
L a y e r Number
JOURNAL DE PHYSIQUE
I . - -
9 I
i I
I Fig. 4 An Integral concentration
'
profde of Cu-1.2at%Fe alloy annealed1
m . _ . s~tu a t 873 K for 30 sec in UHV.1
I
0 500 1000 1500 2000 2500 3000 WOO
Number o f D e t e c t e d Cu
+
FeFig. 5 Integral concentration profiles of Cu-l.Oat%Fe annealed in-s'tu at 1000 K for under 7x10-
4
TOIT oxygen.depth alloy 1 min
Oxide l ~ e t a l
I
Number o f D e t e c t e d Cu
+
Fe+
00
1
4
I
i i
1,
z at I
u
U Fe
al
I
C,
a, 301
I
Q I,J
u-
i
20t 0 I
L
I
n E
'
iOi
0
1
1
I
! Fig. 6 Integral concentration depth1
profes of Cu-l.Zat%Fe alloy anneaied in- 'tu at 1000 K for 1 rnin under IXtO-' Torr 0 29
0 500 1000 1500 2000 2500 MOO 3500 4000
Number o f D e t e c t e d Cu
