- Email: [email protected]
The interaction of oxygen with niobium studied by XPS and UPS
Solid State Communications, Printed in Great Britain.
Vol. 71, No. 10, pp. 849-852,
THE INTERACTION
OF OXYGEN
WITH
0038-1098189 $3.00 + .OO Pergamon Press plc
1989.
NIOBIUM
STUDIED
BY XPS AND
UPS
Z.P. Hu and Y.P. Li Department
of Materials
Science,
University
of Science and Technology
of China,
Hefei, P.R.C.
and M.R. Ji and J.X. Wu Central
Lab. of Structure
and Element
Analysis,
(Received
IApril
University P.R.C.
of Science and Technology
of China,
Hefei,
1989 by W. Y. Kuan)
The adsorption and desorption of O2 on Nb (110) surface are studied by XPS and UPS The changes of structure in the electron energy distribution curves for different oxygen exposures and desorption temperatures are correlated with formation of different oxides. The result indicates a reversible conversion of Nb, NbO and Nb*O, in the absorption and desorption process.
2. EXPERIMENT
1. INTRODUCTION
AND
RESULTS
The sample studied is in the form of single crystal disk approximate 8 mm in diameter and 0.5 mm thick. It is cut from a single crystal rod of Nb and is oriented to within 2’ of the Nb (110) surface using the back reflection Laue X-ray technique. A mechanical polish process and a chemical polish method are used to get the mirror like surface. Then the sample is mounted on a Ta holder. The sample heating and oxygen exposing are undertaken in the sample preparation chamber with the pressure of 5 x lOpro r. The treated sample can be transferred into the main chamber of the pressure of 8 x lo-” r through a vacuum tunnel to do XPS, UPS and AES measurements. The XPS set-up uses Mg K, line (1253.6 eV) and the UPS set-up uses HeI ultraviolet light (21.22 eV) for exciting photoelectron. First, the sample was gradually heated in the preparation chamber by electron bombardment to high temperatures till 1700°C. XPS measurements shows that the original contamination on the surface are oxygen nitrogen and carbon. After heating to SOO”C, there is only oxygen left. The ratio of 0 to Nb on the surface determined by XPS quantitatively is about 0.21. Further heating decreases the ratio to 0.14, then no obvious change until 1700°C. Limited by the experimental condition, we did not heat the sample to 2000°C for an hour as Haas and Farrell did [1, 21. An annealing and ion bombardment method is used many times circularly to get clean surface. As oxygen dissolved in the sample is in NbO
NIOBIUM has been the subject of extensive studies for many years, its refractory property, extensibility and its property of absorbing and desorbing Oz, N,, etc. has stimulated a number of studies [l-5]. Haas and Farrell et al. used low energy electron diffraction (LEED) technique to study the structure of oxygen and nitrogen adsorption on Nb (110) and Nb (100) surfaces [ 1, 21. Dawson used Auger electron spectronscopy (AES) and secondary ion mass spectroscopy (SIMS) [3], and Grundner used photoelectron spectroscopy (XPS) [4] to study the interaction of oxygen with niobium. Landau and Spicer used UPS to study the initial stages of oxidation of niobium. They correlated the changes of structure in the electron distribution curves (EDC) for different oxygen exposures and temperatures with the formation of different types of Nb oxides. Many interesting problems are discussed in their paper [5]. Despite a lot of excellent work that has been done, many important questions remain unclear and some of the results contradict each other. For example: Under what condition do Nb, NbO and Nb,05 form? Can these oxides convert each other? Is the oxidation process reversible? The present study has been performed at different oxygen exposure and different desorption temperatures. The XPS, UPS and AES measurements were undertaken. The results are correlated with theoretical calculation and complemented each other. Some meaningful explanations are given. 849
850
THE INTERACTION
manner, the NbO segregates to the surface when the sample is cooling down [3]. We bombard the sample with Ar ion of 2 keV as it is cooling. The 0 segregated on the surface is sputtered off and the oxygen amount in the sample obviously decreases. Repeat annealing and bombardment many times, the ratio of 0 to Nb on the surface decreases to less than 0.02. The oxygen adsorption can begin. The oxygen exposure is performed by leaking in ultrahigh-purity oxygen through a leak-valve in a controlled manner. The oxygen partial pressure in the preparation chamber is kept at 1 x 10-l r for a certain period of time. When it is turned off, the pressure quickly returns to 5 x lo-“2. XPS and UPS measurements are taken after every exposure. After the exposure is greater than 6OL, the XPS and UPS spectroscopys do not change drastically as before. The adsorbed oxygen has reached saturation and a protective Nb oxide layer is formed on the surface [5]. An X-ray induced AES is accompanied with the XPS. Figure 1 shows the Nb MNV (1086eV) Auger peak at 0, 13, 80, 3000 L respectively. A new Nb MNV Auger peak (1090eV) which represents a new chemical state of Nb grows up as exposure increases. The 4eV big chemical shift mainly comes from the great increase of binding energy of valence band in oxidation. In principle the sample’s work function could be got by subtracting the length of EDC from hv directly. However the absolute value got from it does not have any meaning because of the unknown contact potentials of other conductors in the measurement circuit. On the other hand, the relative work function changes after different oxygen exposures could be measured from the changes of length of EDC’s if we keep all experiment conditions constant. In our experiment,
OF OXYGEN
Vol. 71, No. 10
Nb exposed h=21.22
to
O2
eV
IOL
IL
1
00
40
80 BindIng
12 0
160
energy(eV)
Fig. 2. EDC’s from Nb (110) exposed 10 L of oxygen.
to 0, 0.3, 1, 5,
the work function change reaches 0.8 eV as exposure increases from 0 to 10 L. In Fig. 2 a sharp increase of work function could be observed for exposure up to 1 L. After then it increases slowly. For exposure larger than 10 L, work function remains nearly constant. As there is a trace of oxygen on the original surface, which could be identified from the peak at around 6eV in the EDC of 0 L (this will be discussed in Section 3), the real work function change from clean Nb (110) surface to heavy oxygen exposure should be greater than that of our experiment. The heavy oxygen exposed sample is heated to a series of high temperatures for several minutes. After cooling, XPS and UPS measurements are taken for every desorption circle. The structure of EDCs changes back gradually. When temperature exceeds lOOO”C, the feature of EDC becomes similar to that of before oxygen exposure.
3. DISCUSSION
Bdng
energy (eV)
Fig. 1. Nb NMV Auger peaks from X-ray induced AES for 0.13, 80, 3000 L of oxygen exposure.
A theoretical calculation result of the electronic density of state (DOS) of metallic Nb based on the LCAO method [6] is shown on the bottom of Fig. 3. The EDC of UPS of near clean Nb (110) surface is above it. Since the theoretical DOS is from our original data base of bee structure of Nb, we did not specify the feature of (110) plane. Detailed analysis will be given by computation research with simulated models in another paper. In band structure of metal Nb (not appearing in this paper), the unfilled 4d shell were hybridized with valence and conduction parts
THE INTERACTION
Vol. 71, No. 10
r
1
EDC of
?, I
I 00
40
2.0 Bmdmg
Theoretlcal
Yenergy
Nb
851
OF OXYGEN
r
1 Nb 3d I
198
DOS
200
202
I
Bmding
6.0
(eV)
Fig. 3. (a) Theoretical DOS of Nb based on LCAO calculation. (b) EDC from near clean Nb (110) surface (hv = 21.22eV). from these k points along the (110) plane. A rough comparison also tells us that the electronic state of pure Nb extends to 4 eV below Fermi level, so the peak at 6 eV in the EDC is obviously due to the influence of oxygen atoms. The energy of 2p orbit of a free oxygen atom is about 14 eV [7]. Subtracted by the work function of the adsorbent Nb (110) surface, the binding energy of oxygen interacting with Nb takes the range of 4-10eV below Fermi level. This estimation coincides with our experimental date. When oxygen exposure amount increases from 0 to 10 L, the amplitude of the main peak from 4d level of metallic Nb from Fermi level to 2 eV decreases and that from 02p peak at about 6eV expands drastically (Fig. 2). Another noticeable change in the EDC’s for exposure increasing is at 0, peaks: it not only grows higher, but also expands to a wider energy range, and the peak for heavy exposure is composed of three peaks. One is the original peak at around 6eV, the other two are at about 4.6 and 7.8eV respectively. It is obvious that the chemical states of oxygen changes as exposure increases, and some new states corresponding to the 4.6 and 7.8eV peaks appear. As more oxygen is adsorbed on Nb, more oxygen atoms share the electrons of Nb, its binding energy must drift, in other words, the configurations of niobium oxides are different. Figure 4 shows the Nb 3d peaks from XPS of 3000 and OL respectively. Their difference is shown on the bottom, from which it could be seen clearly that two pairs of peaks for two Nb oxidation states appear at heavy exposure. One pair is that at around 203.5
Fig. 4. Nb (a) 3000L difference NbO and
204
206
energy
208
210
(eV)
3d peak from XPS of Nb (110) Surface of oxygen exposure. (b) 0 L exposure. The (a)-(b) is shown below and indicates the Nb,O, on the exposed surface.
and 206.5 eV for NbO, the other pair is that at around 207.4 and 209.7 eV for Nb,O,. There might be some other niobium oxides, e.g. NbO,. It needs to be further confirmed. The XPS result confirms that there are at least two kinds of Nb oxides on the heavy oxygen exposed sample surface: NbO and Nb,O, . Figure 5 shows three EDC’s for (a) heavy oxygen exposed Nb (110) surface, (b) after heating to 200°C for 5 min, (c) after heating to 400°C for 5 min. Comparing it with Fig. 2, it could be observed that in oxygen adsorption and desorption processes, the structure of EDC changes in opposite ways. The chemical state of oxygen in Nb transfers back as temperature rises. I
I
I 40 Blmng
120
80 energy
(ev!
Fig. 5. EDC’s from (a) heavy oxygen exposed Nb (1 lo), (b) after heating to 200°C for 5 min, (c) after heating to 400°C for 5min.
852
THE INTERACTION
Vol. 71, No. 10
OF OXYGEN 4. CONCLUSION
Nb 3b
Our results lead to the following suggestion: When Nb is exposed to O,, oxidation happens, both NbO and Nbz05 form on the surface at heavy exposure; heating the sample with oxidation surface at around 500°C in vacuum, the Nb,O, convert into NbO, and the oxygen amount in the sample decreases. The adsorption and description process form a reversible circle: Nb -+ NbO + Nb?O, + NbO + NbO + Nb.
Btnding
energy
kv)
Acknowledgements - This work is supported by the National Natural Science Foundation of China, and the Laboratory of Structure and Element Analysis, University of Science and Technology of China.
Fig. 6. Nb 3d peaks corresponding to Fig. 5 (a) and (b). On the bottom is their difference.
REFERENCES 1.
Figure 6 shows the ccrresponding Nb 3d peaks of XPS. The curve on the bottom is the difference of Nb 3d peaks in the cases of before and after heating to 200°C. The two hollows in the curve at 207.4 and 209.7 eV indicates that Nb,O, decreases after heating, and the two peaks at 203.7 and 205.8eV indicates NbO increasing. The Nb 3d peaks of XPS of the sample after heating to 500°C indicates that there is no Nb,O, left on the sample. This result agrees with that of UPS fairly well.
2.
3. 4. 5. 6. 7.
T.W. Haas, A.G. Jackson & M.P. Hooker, J. Chem. Phys. 46, 3025 (1967). H.H. Farrell & M. Strongin, Surf. Sci. 38, 18 (1973); H.H. Farrell, H.S. Isaacs & M. Strongin, Surf. Sci. 38, 38 (1973). P.H. Dawson & Wing-Cheung Tam, Surf. Sci. 81, 464 (1979). M. Grundner & J. Halbritter, Surf. Sci. 136, 144 (1984). I. Lindau & W.E. Spicer, J. Appl. Phys. 45,372O (1974). Yong Ping Li, Unpublished data. F. Herman & S. Skillman, Atomic Structure Calculation, Prentice-Hall, New Jersey (1963).
Vol. 71, No. 10, pp. 849-852,
THE INTERACTION
OF OXYGEN
WITH
0038-1098189 $3.00 + .OO Pergamon Press plc
1989.
NIOBIUM
STUDIED
BY XPS AND
UPS
Z.P. Hu and Y.P. Li Department
of Materials
Science,
University
of Science and Technology
of China,
Hefei, P.R.C.
and M.R. Ji and J.X. Wu Central
Lab. of Structure
and Element
Analysis,
(Received
IApril
University P.R.C.
of Science and Technology
of China,
Hefei,
1989 by W. Y. Kuan)
The adsorption and desorption of O2 on Nb (110) surface are studied by XPS and UPS The changes of structure in the electron energy distribution curves for different oxygen exposures and desorption temperatures are correlated with formation of different oxides. The result indicates a reversible conversion of Nb, NbO and Nb*O, in the absorption and desorption process.
2. EXPERIMENT
1. INTRODUCTION
AND
RESULTS
The sample studied is in the form of single crystal disk approximate 8 mm in diameter and 0.5 mm thick. It is cut from a single crystal rod of Nb and is oriented to within 2’ of the Nb (110) surface using the back reflection Laue X-ray technique. A mechanical polish process and a chemical polish method are used to get the mirror like surface. Then the sample is mounted on a Ta holder. The sample heating and oxygen exposing are undertaken in the sample preparation chamber with the pressure of 5 x lOpro r. The treated sample can be transferred into the main chamber of the pressure of 8 x lo-” r through a vacuum tunnel to do XPS, UPS and AES measurements. The XPS set-up uses Mg K, line (1253.6 eV) and the UPS set-up uses HeI ultraviolet light (21.22 eV) for exciting photoelectron. First, the sample was gradually heated in the preparation chamber by electron bombardment to high temperatures till 1700°C. XPS measurements shows that the original contamination on the surface are oxygen nitrogen and carbon. After heating to SOO”C, there is only oxygen left. The ratio of 0 to Nb on the surface determined by XPS quantitatively is about 0.21. Further heating decreases the ratio to 0.14, then no obvious change until 1700°C. Limited by the experimental condition, we did not heat the sample to 2000°C for an hour as Haas and Farrell did [1, 21. An annealing and ion bombardment method is used many times circularly to get clean surface. As oxygen dissolved in the sample is in NbO
NIOBIUM has been the subject of extensive studies for many years, its refractory property, extensibility and its property of absorbing and desorbing Oz, N,, etc. has stimulated a number of studies [l-5]. Haas and Farrell et al. used low energy electron diffraction (LEED) technique to study the structure of oxygen and nitrogen adsorption on Nb (110) and Nb (100) surfaces [ 1, 21. Dawson used Auger electron spectronscopy (AES) and secondary ion mass spectroscopy (SIMS) [3], and Grundner used photoelectron spectroscopy (XPS) [4] to study the interaction of oxygen with niobium. Landau and Spicer used UPS to study the initial stages of oxidation of niobium. They correlated the changes of structure in the electron distribution curves (EDC) for different oxygen exposures and temperatures with the formation of different types of Nb oxides. Many interesting problems are discussed in their paper [5]. Despite a lot of excellent work that has been done, many important questions remain unclear and some of the results contradict each other. For example: Under what condition do Nb, NbO and Nb,05 form? Can these oxides convert each other? Is the oxidation process reversible? The present study has been performed at different oxygen exposure and different desorption temperatures. The XPS, UPS and AES measurements were undertaken. The results are correlated with theoretical calculation and complemented each other. Some meaningful explanations are given. 849
850
THE INTERACTION
manner, the NbO segregates to the surface when the sample is cooling down [3]. We bombard the sample with Ar ion of 2 keV as it is cooling. The 0 segregated on the surface is sputtered off and the oxygen amount in the sample obviously decreases. Repeat annealing and bombardment many times, the ratio of 0 to Nb on the surface decreases to less than 0.02. The oxygen adsorption can begin. The oxygen exposure is performed by leaking in ultrahigh-purity oxygen through a leak-valve in a controlled manner. The oxygen partial pressure in the preparation chamber is kept at 1 x 10-l r for a certain period of time. When it is turned off, the pressure quickly returns to 5 x lo-“2. XPS and UPS measurements are taken after every exposure. After the exposure is greater than 6OL, the XPS and UPS spectroscopys do not change drastically as before. The adsorbed oxygen has reached saturation and a protective Nb oxide layer is formed on the surface [5]. An X-ray induced AES is accompanied with the XPS. Figure 1 shows the Nb MNV (1086eV) Auger peak at 0, 13, 80, 3000 L respectively. A new Nb MNV Auger peak (1090eV) which represents a new chemical state of Nb grows up as exposure increases. The 4eV big chemical shift mainly comes from the great increase of binding energy of valence band in oxidation. In principle the sample’s work function could be got by subtracting the length of EDC from hv directly. However the absolute value got from it does not have any meaning because of the unknown contact potentials of other conductors in the measurement circuit. On the other hand, the relative work function changes after different oxygen exposures could be measured from the changes of length of EDC’s if we keep all experiment conditions constant. In our experiment,
OF OXYGEN
Vol. 71, No. 10
Nb exposed h=21.22
to
O2
eV
IOL
IL
1
00
40
80 BindIng
12 0
160
energy(eV)
Fig. 2. EDC’s from Nb (110) exposed 10 L of oxygen.
to 0, 0.3, 1, 5,
the work function change reaches 0.8 eV as exposure increases from 0 to 10 L. In Fig. 2 a sharp increase of work function could be observed for exposure up to 1 L. After then it increases slowly. For exposure larger than 10 L, work function remains nearly constant. As there is a trace of oxygen on the original surface, which could be identified from the peak at around 6eV in the EDC of 0 L (this will be discussed in Section 3), the real work function change from clean Nb (110) surface to heavy oxygen exposure should be greater than that of our experiment. The heavy oxygen exposed sample is heated to a series of high temperatures for several minutes. After cooling, XPS and UPS measurements are taken for every desorption circle. The structure of EDCs changes back gradually. When temperature exceeds lOOO”C, the feature of EDC becomes similar to that of before oxygen exposure.
3. DISCUSSION
Bdng
energy (eV)
Fig. 1. Nb NMV Auger peaks from X-ray induced AES for 0.13, 80, 3000 L of oxygen exposure.
A theoretical calculation result of the electronic density of state (DOS) of metallic Nb based on the LCAO method [6] is shown on the bottom of Fig. 3. The EDC of UPS of near clean Nb (110) surface is above it. Since the theoretical DOS is from our original data base of bee structure of Nb, we did not specify the feature of (110) plane. Detailed analysis will be given by computation research with simulated models in another paper. In band structure of metal Nb (not appearing in this paper), the unfilled 4d shell were hybridized with valence and conduction parts
THE INTERACTION
Vol. 71, No. 10
r
1
EDC of
?, I
I 00
40
2.0 Bmdmg
Theoretlcal
Yenergy
Nb
851
OF OXYGEN
r
1 Nb 3d I
198
DOS
200
202
I
Bmding
6.0
(eV)
Fig. 3. (a) Theoretical DOS of Nb based on LCAO calculation. (b) EDC from near clean Nb (110) surface (hv = 21.22eV). from these k points along the (110) plane. A rough comparison also tells us that the electronic state of pure Nb extends to 4 eV below Fermi level, so the peak at 6 eV in the EDC is obviously due to the influence of oxygen atoms. The energy of 2p orbit of a free oxygen atom is about 14 eV [7]. Subtracted by the work function of the adsorbent Nb (110) surface, the binding energy of oxygen interacting with Nb takes the range of 4-10eV below Fermi level. This estimation coincides with our experimental date. When oxygen exposure amount increases from 0 to 10 L, the amplitude of the main peak from 4d level of metallic Nb from Fermi level to 2 eV decreases and that from 02p peak at about 6eV expands drastically (Fig. 2). Another noticeable change in the EDC’s for exposure increasing is at 0, peaks: it not only grows higher, but also expands to a wider energy range, and the peak for heavy exposure is composed of three peaks. One is the original peak at around 6eV, the other two are at about 4.6 and 7.8eV respectively. It is obvious that the chemical states of oxygen changes as exposure increases, and some new states corresponding to the 4.6 and 7.8eV peaks appear. As more oxygen is adsorbed on Nb, more oxygen atoms share the electrons of Nb, its binding energy must drift, in other words, the configurations of niobium oxides are different. Figure 4 shows the Nb 3d peaks from XPS of 3000 and OL respectively. Their difference is shown on the bottom, from which it could be seen clearly that two pairs of peaks for two Nb oxidation states appear at heavy exposure. One pair is that at around 203.5
Fig. 4. Nb (a) 3000L difference NbO and
204
206
energy
208
210
(eV)
3d peak from XPS of Nb (110) Surface of oxygen exposure. (b) 0 L exposure. The (a)-(b) is shown below and indicates the Nb,O, on the exposed surface.
and 206.5 eV for NbO, the other pair is that at around 207.4 and 209.7 eV for Nb,O,. There might be some other niobium oxides, e.g. NbO,. It needs to be further confirmed. The XPS result confirms that there are at least two kinds of Nb oxides on the heavy oxygen exposed sample surface: NbO and Nb,O, . Figure 5 shows three EDC’s for (a) heavy oxygen exposed Nb (110) surface, (b) after heating to 200°C for 5 min, (c) after heating to 400°C for 5 min. Comparing it with Fig. 2, it could be observed that in oxygen adsorption and desorption processes, the structure of EDC changes in opposite ways. The chemical state of oxygen in Nb transfers back as temperature rises. I
I
I 40 Blmng
120
80 energy
(ev!
Fig. 5. EDC’s from (a) heavy oxygen exposed Nb (1 lo), (b) after heating to 200°C for 5 min, (c) after heating to 400°C for 5min.
852
THE INTERACTION
Vol. 71, No. 10
OF OXYGEN 4. CONCLUSION
Nb 3b
Our results lead to the following suggestion: When Nb is exposed to O,, oxidation happens, both NbO and Nbz05 form on the surface at heavy exposure; heating the sample with oxidation surface at around 500°C in vacuum, the Nb,O, convert into NbO, and the oxygen amount in the sample decreases. The adsorption and description process form a reversible circle: Nb -+ NbO + Nb?O, + NbO + NbO + Nb.
Btnding
energy
kv)
Acknowledgements - This work is supported by the National Natural Science Foundation of China, and the Laboratory of Structure and Element Analysis, University of Science and Technology of China.
Fig. 6. Nb 3d peaks corresponding to Fig. 5 (a) and (b). On the bottom is their difference.
REFERENCES 1.
Figure 6 shows the ccrresponding Nb 3d peaks of XPS. The curve on the bottom is the difference of Nb 3d peaks in the cases of before and after heating to 200°C. The two hollows in the curve at 207.4 and 209.7 eV indicates that Nb,O, decreases after heating, and the two peaks at 203.7 and 205.8eV indicates NbO increasing. The Nb 3d peaks of XPS of the sample after heating to 500°C indicates that there is no Nb,O, left on the sample. This result agrees with that of UPS fairly well.
2.
3. 4. 5. 6. 7.
T.W. Haas, A.G. Jackson & M.P. Hooker, J. Chem. Phys. 46, 3025 (1967). H.H. Farrell & M. Strongin, Surf. Sci. 38, 18 (1973); H.H. Farrell, H.S. Isaacs & M. Strongin, Surf. Sci. 38, 38 (1973). P.H. Dawson & Wing-Cheung Tam, Surf. Sci. 81, 464 (1979). M. Grundner & J. Halbritter, Surf. Sci. 136, 144 (1984). I. Lindau & W.E. Spicer, J. Appl. Phys. 45,372O (1974). Yong Ping Li, Unpublished data. F. Herman & S. Skillman, Atomic Structure Calculation, Prentice-Hall, New Jersey (1963).