- Email: [email protected]
Investigation of the interaction of oxygen with an Al(100) surface by work function methods
L689
Surface Science 203 (1988) L689-L694 North-Holland, Amsterdam
SURFACE
SCIENCE
INVESTIGATION WITH
LEIT’ERS
OF THE INTERACTION
AN Al(100) SURFACE
U. MEMMERT
BY WORK
OF OXYGEN
FUNCTION
METHODS
and P.R. NORTON
Departments of Physics and Chemistry, Interface Science Western, University of Western Ontario, London, Ontario, Canada N6A 587 Received
4 April 1988; accepted
for publication
18 May 1988
The change in work function (A+) of an Al(100) surface due to oxygen adsorption at different pressures (10-6-10~5 Torr) and different crystal temperatures (115-600 K) has been investigated. The data show an initially positive A+ at low temperatures and negative A+ values for temperatures above 160 K. The results are pressure independent during the first 500 L oxygen exposure but become pressure dependent at higher exposures. Measurements of A+ in vacuum after exposure to 1040 L 0, show a nearly exponential decrease of A+ with time. The time constants of this exponential behaviour are temperature dependent and vary between 1.5 min for 370 K and 33 min at 115 K. These time dependent effects are believed to be related to the movement of adsorbed oxygen to sites below the aluminum surface.
The work function change (A+) of aluminum surfaces due to the adsorption of oxygen has been the subject of numerous studies over the last two decades (e.g. refs. [l-5]) because of the technological importance of this system and because it has often been regarded as a theoretical model adsorption system [6]. However, theoretical studies always predict an increase in A+ upon oxygen adsorption, while the measurements (performed at room temperature) showed only a slightly positive initial A+ for the Al(111) surface, and adsorption on Al(100) and Al(110) surfaces display a negative A+. The Al(111) surface also showed exposure rate dependencies and time dependent changes of the work function after stopping the exposure. Such effects have never been reported for the Al(100) and Al(110) surfaces at room temperature. A A+ study of the adsorption of oxygen on evaporated Al films [3] showed a positive A+ for low temperatures and negative A+ for room temperature and higher. It is the aim of this Letter to present measurements at different temperatures on Al(100) and to demonstrate that the Al(100) surface shows at lower temperatures, very similar behavior to that exhibited by the Al(111) surface at room temperature. The measurements were performed in a UHV system equipped with a Kelvin probe to measure A$, a differentially pumped mass spectrometer, LEED optics, ion gauge and argon sputter gun. The chamber is pumped by oil 0039-6028/88/$03.50
(North-Holland
Physics
0 Elsevier
Publishing
Science Publishers Division)
B.V.
Fig. 1. A+ versus exposure at 300 K for different oxygen pressures.
diffusion pumps with LN, cooled traps. The base pressure was < 3 x 10-i’ Torr. The Al(100) crystal was polished with diamond paste followed by vibratory polishing with 0.05 pm alumina. It was sputtered at 570 K with 3 keV Ar+ ions at 1 PA cme2 Torr about 50 h. After this procedure sharp LEED patterns were obtained. The oxygen exposures were carried out using ~atheson Research purity oxygen. After each exposure to oxygen the surface was cleaned for about 30 n-tin by Ar+ ion sputtering using the above conditions. The exposures were measured by integrating the total pressure in the chamber as a function of time. The uncertainty of the ion gauge calibration leads to a concomitant uncertainty in the oxygen exposures given in this Letter, of a factor of approximately two. Fig. 1 shows A$ as a function of the oxygen exposure at 300 K for oxygen pressures in the range 10-6-10-5 Torr. The data clearly indicate that the oxidation process can be divided into two regions. During the first 500 L (region I) the A# versus exposure curves are pressure independent and Arp -250 mV, while region II displays different curves at different reaches pressures. This indicates that in region II the slope of the A+ versus exposure curves is no longer controlled only by the oxygen exposure but also by surface processes. It also has to be noted that the saturation values are different for different pressures. Lower pressures lead to smaller A#I saturation values. This fact may also have been responsible for the different saturation values published in the literature (a review can be found in ref. [7]) since these measurements were performed at very different oxygen pressures. Fig. 2 shows A+ versus exposure for different temperatures in the range from 115 to 570 K. At temperatures above 160 K, A+ is negative in both regions I and II, while for T < 160 K, A+ first increases in region I before decreasing again in region II to negative values. The fact that A+ changes
U. Memmerr,
P.R. Norton / Interaction
Exposure
Fig. 2. A+ versus exposure at 3.5
X
of oxygen with A&100)
L691
CL)
10m6 Torr for different temperatures.
t (min) Fig. 3. A+, at 163 K, as a function of time after an exposure to 1040 L at 8x10-
shown linearly
in the upper
part
and
logarithmically
in the lower
part
of the
-’ Tom. A+ is figure.
L692
U. Memmert,
of oxygenwith
P.R. Norton / Interaction
30-
A&100)
t
2520-
”
2
+ t+
.E 15v
t t
lot 5-
T(K)
Fig. 4. Time constants, 7, for the change of A+ after exposure, as a function of temperature
from a positive value in the beginning to a negative value for longer times (at least for T < 160 K) leads to the conclusion that the underlying process may be the migration of oxygen from initial adsorption sites on top of the surface to subsurface and/or oxide sites. To investigate the processes leading to the pressure dependencies in region II in fig. 1 in more detail, we exposed the surface at different temperatures to 1040 L at 8 X 10e6 Torr, pumped the oxygen away and followed A+ as a function of time. The upper part of fig. 3 shows as an example, a measurement at 163 K on a linear scale, while the lower part of fig. 3 displays the same data on a semilogarithmic scale. These data show that AI#J approaches the final value almost exponentially with elapsed time after the removal of 02; i.e. according to a first-order rate law: d A+/d t = constants
x ( A+, - A#,,,,
).
From this exponential plot one obtains a time constant, r = 22 min. Similar measurements were performed in the range from 130 to 370 K and the resulting time constants are shown as a function of temperature in fig. 4. Fig. 5 depicts the same data in an Arrhenius plot. The range over which In A# is nearly linear with time is very surprising, particularly since we are not certain that A+ and coverage on top of the surface are linearly inter-related. This relationship is being investigated by synchrotron radiation photoemission methods, nuclear reaction analysis and Auger spectroscopy. Since the values of 7 in fig. 5 are not on a straight line, it is not possible to extract a single activation energy from these data, but is does appear as though there is a change in slope at T > 330 K. It should also be noted that if the process is simply the activated hopping of adsorbed O-atoms from surface to subsurface sites, then the assumption of a “normal” pre-exponential of lOI SC’ would
U. ~emme~t, P.R. Norton f I~teruction of oxygen with Al(ltN)
L693
Fig. 5. Logarithmic plot of the time constants, I, as a function of the reciprocal temperature.
lead to a much larger change in r over the measured T-interval; i.e. a larger apparent activation energy than is actually observed. Preliminary measurements by Auger spectroscopy have provided additional evidence for the above picture of migration of oxygen from atop sites to subsurface sites. The Auger spectra exhibit increasing metallic character as a function of time after the evacuation of oxygen following a low temperature exposure [8]. In summary, the data show that the oxidation of an Al(100) surface at low temperatures results in a positive A+, while exposure at high temperatures produces only negative A+ values. This leads to the conclusion that the oxygen first adsorbs into sites on top of the surface and then migrates to subsurface sites. At low temperatures the migration process is very slow, leading to an accumulation of oxygen on top of the surface and hence to a positive A$. The migration is much faster at higher temperatures, leading directly to oxygen incorporation and hence negative A$ values. The low temperature behaviour is very similar to the behaviour which has been observed on the Al(111) surface at room temperature [4]. At high pressures the Al(100) surface also shows pressure dependencies like those observed on Al(111) surfaces [4]. These similarities between low temperature behaviour on Al and the room temperature behaviour on Al(111) leads to the conclusion that the observed strong anisotropies of the oxidation of Al surfaces may mainly be caused by higher rates for the migration of oxygen atoms from their initial adsorption sites on top of the surface to subsurface sites. In this picture this process is faster on Al(100) than on Al(111) at the same temperature. A simple explanation would be that the Al(100) surface provides more space between the aluminum atoms for the oxygen to migrate to subsurface sites because of its lower density (1.22 x lOI atoms cme2 for Al(100) compared to 1.41 X lOI4 atoms cm-* for the Al(111) surface).
L694
U. Memmert,
P.R. Norton / Interaction
ofoxygets
with AlffOO)
The authors gratefully acknowledge the provision of crystals and financial support by Alcan International Ltd. and helpful discussions with Dr. Dave Creber (Alcan) and Steve Bushby. U.M. also acknowledges the support of the Natural Sciences and Engineering Research Council.
References fl] [2] [3] [4] [5] [6] [7] [8]
V.K. Agarwala and T. Fort, Jr., Surface Sci. 45 (1974) 470. P.O. Gartfand, Surface Sci. 62 (1977) 183. C. Benndorf, H. Seidel and F. Thieme, Surface Sci. 67 (1977) 469. P. Hofmann, W. Wyrobisch and A.M. Bradshaw, Surface Sci. 80 (1979) 344. R. Michel, J. Gastaldi, C. Allasia, C. Jourdan and J. Derrien, Surface Sci. 95 (1980) 309. N.D. Lang and R.W. Williams, Phys. Rev. Letters 34 (1975) 531. I.P. Batra and L. Kleinman, J. Electron Spectrosc. Related Phenomena 33 (1984) 175. U. Memmert and P.R. Norton, to be published.
Surface Science 203 (1988) L689-L694 North-Holland, Amsterdam
SURFACE
SCIENCE
INVESTIGATION WITH
LEIT’ERS
OF THE INTERACTION
AN Al(100) SURFACE
U. MEMMERT
BY WORK
OF OXYGEN
FUNCTION
METHODS
and P.R. NORTON
Departments of Physics and Chemistry, Interface Science Western, University of Western Ontario, London, Ontario, Canada N6A 587 Received
4 April 1988; accepted
for publication
18 May 1988
The change in work function (A+) of an Al(100) surface due to oxygen adsorption at different pressures (10-6-10~5 Torr) and different crystal temperatures (115-600 K) has been investigated. The data show an initially positive A+ at low temperatures and negative A+ values for temperatures above 160 K. The results are pressure independent during the first 500 L oxygen exposure but become pressure dependent at higher exposures. Measurements of A+ in vacuum after exposure to 1040 L 0, show a nearly exponential decrease of A+ with time. The time constants of this exponential behaviour are temperature dependent and vary between 1.5 min for 370 K and 33 min at 115 K. These time dependent effects are believed to be related to the movement of adsorbed oxygen to sites below the aluminum surface.
The work function change (A+) of aluminum surfaces due to the adsorption of oxygen has been the subject of numerous studies over the last two decades (e.g. refs. [l-5]) because of the technological importance of this system and because it has often been regarded as a theoretical model adsorption system [6]. However, theoretical studies always predict an increase in A+ upon oxygen adsorption, while the measurements (performed at room temperature) showed only a slightly positive initial A+ for the Al(111) surface, and adsorption on Al(100) and Al(110) surfaces display a negative A+. The Al(111) surface also showed exposure rate dependencies and time dependent changes of the work function after stopping the exposure. Such effects have never been reported for the Al(100) and Al(110) surfaces at room temperature. A A+ study of the adsorption of oxygen on evaporated Al films [3] showed a positive A+ for low temperatures and negative A+ for room temperature and higher. It is the aim of this Letter to present measurements at different temperatures on Al(100) and to demonstrate that the Al(100) surface shows at lower temperatures, very similar behavior to that exhibited by the Al(111) surface at room temperature. The measurements were performed in a UHV system equipped with a Kelvin probe to measure A$, a differentially pumped mass spectrometer, LEED optics, ion gauge and argon sputter gun. The chamber is pumped by oil 0039-6028/88/$03.50
(North-Holland
Physics
0 Elsevier
Publishing
Science Publishers Division)
B.V.
Fig. 1. A+ versus exposure at 300 K for different oxygen pressures.
diffusion pumps with LN, cooled traps. The base pressure was < 3 x 10-i’ Torr. The Al(100) crystal was polished with diamond paste followed by vibratory polishing with 0.05 pm alumina. It was sputtered at 570 K with 3 keV Ar+ ions at 1 PA cme2 Torr about 50 h. After this procedure sharp LEED patterns were obtained. The oxygen exposures were carried out using ~atheson Research purity oxygen. After each exposure to oxygen the surface was cleaned for about 30 n-tin by Ar+ ion sputtering using the above conditions. The exposures were measured by integrating the total pressure in the chamber as a function of time. The uncertainty of the ion gauge calibration leads to a concomitant uncertainty in the oxygen exposures given in this Letter, of a factor of approximately two. Fig. 1 shows A$ as a function of the oxygen exposure at 300 K for oxygen pressures in the range 10-6-10-5 Torr. The data clearly indicate that the oxidation process can be divided into two regions. During the first 500 L (region I) the A# versus exposure curves are pressure independent and Arp -250 mV, while region II displays different curves at different reaches pressures. This indicates that in region II the slope of the A+ versus exposure curves is no longer controlled only by the oxygen exposure but also by surface processes. It also has to be noted that the saturation values are different for different pressures. Lower pressures lead to smaller A#I saturation values. This fact may also have been responsible for the different saturation values published in the literature (a review can be found in ref. [7]) since these measurements were performed at very different oxygen pressures. Fig. 2 shows A+ versus exposure for different temperatures in the range from 115 to 570 K. At temperatures above 160 K, A+ is negative in both regions I and II, while for T < 160 K, A+ first increases in region I before decreasing again in region II to negative values. The fact that A+ changes
U. Memmerr,
P.R. Norton / Interaction
Exposure
Fig. 2. A+ versus exposure at 3.5
X
of oxygen with A&100)
L691
CL)
10m6 Torr for different temperatures.
t (min) Fig. 3. A+, at 163 K, as a function of time after an exposure to 1040 L at 8x10-
shown linearly
in the upper
part
and
logarithmically
in the lower
part
of the
-’ Tom. A+ is figure.
L692
U. Memmert,
of oxygenwith
P.R. Norton / Interaction
30-
A&100)
t
2520-
”
2
+ t+
.E 15v
t t
lot 5-
T(K)
Fig. 4. Time constants, 7, for the change of A+ after exposure, as a function of temperature
from a positive value in the beginning to a negative value for longer times (at least for T < 160 K) leads to the conclusion that the underlying process may be the migration of oxygen from initial adsorption sites on top of the surface to subsurface and/or oxide sites. To investigate the processes leading to the pressure dependencies in region II in fig. 1 in more detail, we exposed the surface at different temperatures to 1040 L at 8 X 10e6 Torr, pumped the oxygen away and followed A+ as a function of time. The upper part of fig. 3 shows as an example, a measurement at 163 K on a linear scale, while the lower part of fig. 3 displays the same data on a semilogarithmic scale. These data show that AI#J approaches the final value almost exponentially with elapsed time after the removal of 02; i.e. according to a first-order rate law: d A+/d t = constants
x ( A+, - A#,,,,
).
From this exponential plot one obtains a time constant, r = 22 min. Similar measurements were performed in the range from 130 to 370 K and the resulting time constants are shown as a function of temperature in fig. 4. Fig. 5 depicts the same data in an Arrhenius plot. The range over which In A# is nearly linear with time is very surprising, particularly since we are not certain that A+ and coverage on top of the surface are linearly inter-related. This relationship is being investigated by synchrotron radiation photoemission methods, nuclear reaction analysis and Auger spectroscopy. Since the values of 7 in fig. 5 are not on a straight line, it is not possible to extract a single activation energy from these data, but is does appear as though there is a change in slope at T > 330 K. It should also be noted that if the process is simply the activated hopping of adsorbed O-atoms from surface to subsurface sites, then the assumption of a “normal” pre-exponential of lOI SC’ would
U. ~emme~t, P.R. Norton f I~teruction of oxygen with Al(ltN)
L693
Fig. 5. Logarithmic plot of the time constants, I, as a function of the reciprocal temperature.
lead to a much larger change in r over the measured T-interval; i.e. a larger apparent activation energy than is actually observed. Preliminary measurements by Auger spectroscopy have provided additional evidence for the above picture of migration of oxygen from atop sites to subsurface sites. The Auger spectra exhibit increasing metallic character as a function of time after the evacuation of oxygen following a low temperature exposure [8]. In summary, the data show that the oxidation of an Al(100) surface at low temperatures results in a positive A+, while exposure at high temperatures produces only negative A+ values. This leads to the conclusion that the oxygen first adsorbs into sites on top of the surface and then migrates to subsurface sites. At low temperatures the migration process is very slow, leading to an accumulation of oxygen on top of the surface and hence to a positive A$. The migration is much faster at higher temperatures, leading directly to oxygen incorporation and hence negative A$ values. The low temperature behaviour is very similar to the behaviour which has been observed on the Al(111) surface at room temperature [4]. At high pressures the Al(100) surface also shows pressure dependencies like those observed on Al(111) surfaces [4]. These similarities between low temperature behaviour on Al and the room temperature behaviour on Al(111) leads to the conclusion that the observed strong anisotropies of the oxidation of Al surfaces may mainly be caused by higher rates for the migration of oxygen atoms from their initial adsorption sites on top of the surface to subsurface sites. In this picture this process is faster on Al(100) than on Al(111) at the same temperature. A simple explanation would be that the Al(100) surface provides more space between the aluminum atoms for the oxygen to migrate to subsurface sites because of its lower density (1.22 x lOI atoms cme2 for Al(100) compared to 1.41 X lOI4 atoms cm-* for the Al(111) surface).
L694
U. Memmert,
P.R. Norton / Interaction
ofoxygets
with AlffOO)
The authors gratefully acknowledge the provision of crystals and financial support by Alcan International Ltd. and helpful discussions with Dr. Dave Creber (Alcan) and Steve Bushby. U.M. also acknowledges the support of the Natural Sciences and Engineering Research Council.
References fl] [2] [3] [4] [5] [6] [7] [8]
V.K. Agarwala and T. Fort, Jr., Surface Sci. 45 (1974) 470. P.O. Gartfand, Surface Sci. 62 (1977) 183. C. Benndorf, H. Seidel and F. Thieme, Surface Sci. 67 (1977) 469. P. Hofmann, W. Wyrobisch and A.M. Bradshaw, Surface Sci. 80 (1979) 344. R. Michel, J. Gastaldi, C. Allasia, C. Jourdan and J. Derrien, Surface Sci. 95 (1980) 309. N.D. Lang and R.W. Williams, Phys. Rev. Letters 34 (1975) 531. I.P. Batra and L. Kleinman, J. Electron Spectrosc. Related Phenomena 33 (1984) 175. U. Memmert and P.R. Norton, to be published.