- Email: [email protected]
Electron spectroscopy studies of the oxidation behaviour and the electronic properties of some MgIn and MgAl alloys
Journal
of the Less-Common
Metals,
92(1983)
253
253-263
ELECTRON SPECTROSCOPY STUDIES OF THE OXIDATION BEHAVIOUR AND THE ELECTRONIC PROPERTIES OF SOME Mg-In AND Mg-Al ALLOYS
N. SHAMIR, M. H. MINTZ*, J. BLOCH and U. ATZMONY* Physics Department, Nuclear Research Centre-
Negev, P.O. Box 9001, Beer-Sheva (Israel)
(Received November 26,1982)
Summary The electronic properties and surface characteristics of some Mg-In and Mg-Al alloys were studied by electron spectroscopy techniques (Auger electron spectroscopy and X-ray photoelectron spectroscopy). It was found that the electrochemical interactions between magnesium and indium are stronger than those between magnesium and aluminium. It was further found that the surface oxidation behaviour of indium-containing alloys is different from that of aluminium-containing alloys. In the former case only the magnesium atoms are oxidized at the surface, leaving the indium atoms in their metallic state, while in the latter case both magnesium and aluminium are oxidized at the surface. Both types of alloy display oxidation-enhanced surface segregation of the magnesium atoms. The hydrogenation behaviour of the various magnesium-base alloys is discussed in the light of their surface properties.
1. Introduction Magnesium and its alloys have much technological importance as low density structural materials and as potential high weight content hydrogen storage materials. In the latter application it has been found that the addition of small amounts (i.e. about 1%) of either aluminium or indium to magnesium metal catalyses the hydrogenation reaction [l, 21. The purpose of the present study was to investigate the surface properties of such dilute alloys and to study the mechanisms that control the initial stages of the reactions of these alloys with oxygen and hydrogen. 2. Experimental
details
Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) measurements were performed on polycrystalline samples of Mg,,,,In,,,,, *Also at the Ben Gurion University of the Negev, Beer-Sheva, Israel. 0022-5088/83/$3.00
QElsevier
Sequoia/Printed
in The Netherlands
254
together with samples of pure magnesium, indium Mgo.95Alo.05 andMgo.55Alo.45 and aluminium for reference. The alloys were prepared by melting stoichiometric amounts of the constituents in an induction furnace using a magnesia crucible. The samples were cleaned using Ccl, and were introduced into a PHI 548 surface analysis system (base pressure, 2 x lo-” Torr). Measurements were performed on each sample in the three following states: (i) the as-received (AR) state; (ii) the sputtered (SP) state which was obtained after sputtering the sample with 4.5 keV Ar+ ions for a long period until all the contaminants had been completely removed or had reached a steady state; (iii) the 100 L 0, state which was obtained by exposing SP samples to 100 langmuirs of oxygen (Matheson 5N grade) introduced to the chamber through a leak valve. 3. Results and analysis
The Auger spectra of Mg0,951n0,05 and of pure indium in the three states are presented in Fig. 1. Several features can be seen by inspecting and comparing the spectra. (1) Although indium is only 5% of the alloy its spectral line is relatively strong. This is due to its high cross section for the Auger process [3]. (2) The sputtering process strongly enhanced the indium peak relative to the magnesium peak, indicating an increase in the indium concentration on the surface. As will be proved later, the magnesium segregates at the surface during oxidation. Therefore the AR state of the alloy, which is naturally oxidized, has a segregated layer on its surface. This layer is removed by the sputtering. In addition to the removal of this layer there may be preferential sputtering of the magnesium ions. (3) The ratio of the magnesium peak to the indium peak in the lOOLO, state is much higher than in the SP state (about a factor of 4 greater). This confirms the segregation of the magnesium at the surface during the oxidation process. (4) The indium AES lines in all the spectra shown in Fig. 1, except that of pure indium in the AR state, have a shape characteristic of indium in its metallic state [3]. The magnesium line in the AR state is typical of magnesium oxide, in the SP state it is typical of magnesium metal and in the 100 L 0, state it exhibits an intermediate state [S]. It is therefore concluded that there is selective oxidation of the magnesium ions during exposure of the Mg,,,,In,,,, alloy to oxygen and that its surface in the AR state consists of magnesium oxide and metallic indium. The corresponding X-ray photoelectron spectra of Mg,,,,In,.,, and pure indium (Fig. 2) reveal the following additional features. (5) The In 3d lines in the alloy spectra are shifted by about 0.5 eV to a lower binding energy relative to their positions in the pure indium spectra. This indicates electron charge transfer from the magnesium to the indium in the alloy.
255
.O-;300
500
700
900
1100
ELECTRON KINETIC ENERGY (eV1
Fig. 1. Auger spectra of (a) Mg,,,,In,,,, intensity scales are only approximately
and (b) pure indium in the AR, SP and 100L 0, states. The equivalent.
(6) The In 3d,,, XPS line of Mgo,,SIno,,5 lies at the same energy for all three states, but for the AR state of pure indium it is shifted by about 1 eV to a higher binding energy which is typical of the oxidized state [4]. The centre of the Mg 2p however, moves from 50.5 eV in the AR state to 49.5 eV in lines ofMg,,,,In,.,,,
Mg 2s
1
455
I
I
440
100
ELECTRON BINDING ENERGY
I
I,
I,,
80
60
Mg 2P
In 2s
I
I
, 40
, 20
,
4
0
(ev)
Fig, 2. X-ray photoelectron spectra of (a) the In 3d lines of Mg,,,,In,.,, and pure indium in the AR, SP and fOOL 0, states, (b) the Mg 2p line of Mg,,,,In,.Os in the same states and (c)the binding energy region (O-100 eV) of Mg,,,,En,~,,. The long vertical lines indicate the energy corresponding to a metallic state of Mgo,ssIn,,,, while the short vertical lines (except for(c)) correspond to oxide states (ox). The intensity scales are different far each spectrum.
the SP state and to an intermediate energy for the lOOLO, state, which is characteristic of an oxidation process. Evidence for the preferential sputtering of the magnesium ions and their surface segregation during oxidation (features (2) and (3)) is also shown by the X-ray photoelectron spectra.
The relative contributions of the two effects cannot be calculated directly. The peak ratios of the oxygen, magnesium and indium lines in the Auger and X-ray photoelectron spectra of the various states do not give quantitative results because, owing to the different kinetic energies, the detected electrons yield information from various depths. Analysis of the Auger spectra gives better relative estimates of the effects than analysis of the X-ray photoelectron spectra because a much longer time is required to accumulate the XPS data so that partial oxidation of the SP state by the residual gases in the system can occur. The magnesium-to-indium intensity ratios of the Auger spectra are a factor of 9 less in the SP state than in the AR state. As the kinetic energy of the indium line is less than that of the magnesium line this factor should be regarded as an upper limit on the combined effect. This ratio is a factor of 4 greater in the 100 L 0, state than in the SP state. This is the lower limit on the segregation effect due to oxidation by 100 L 0,. 3.2. Mg,,,A 10., The cross sections of aluminium for the Auger and photoelectron processes are small compared with that of indium. It was therefore more difficult to obtain meaningful information from the Auger and X-ray photoelectron spectra of the sample. Therefore the spectra are not shown. Mg,.,,Al,.,, The following features corresponding to the Mg,,,,A10,05 spectra should, however, be noted. (1) The weak aluminium AES line in the AR state is typical of neither pure aluminium metal nor its oxide and can be regarded as a superposition of both states. In the SP state it is characteristic of pure aluminium and it shows evidence of partial oxidation again in the 100 L 0, state. (2) It is even more difficult to obtain any information about the aluminium from the X-ray photoelectron spectra as unfortunately its XPS lines coincide with a magnesium plasmon satellite. This satellite varies from state to state which makes it harder to detect the changes in the aluminium XPS lines (which are very small in any case). However, it can be seen that the Al 2s peak lies at about 120 eV (which is the position for oxidized pure aluminium [4]) in the AR state and at about 118 eV (its position for pure aluminium metal [4]) in the SP state. This result shows that no shift in the Al 2s lines between the alloy and pure aluminium, which would have indicated electron charge transfer, occurs. 3.3. Mg,,,A &.,, The measurements on this sample were performed to enable us to obtain more reliable results concerning the behaviour of aluminium in Mg-Al alloys. and of pure aluminium in the three The Auger spectra of Mg,,,,Al,,,, states are shown in Fig. 3. The following features can be seen. (1) The aluminium AES line is almost absent in the spectrum of the alloy in the AR state, which indicates marked segregation of magnesium at the surface. It is strongly enhanced (by a factor of 34 compared with the magnesium line) by the sputtering process. This shows that the combined effect of preferential sputtering of magnesium ions and removal of the segregation layer of magnesium is stronger here than for the Mg,,,,In,.,, sample.
W t 0
-r 500
SP
’
’
700
-
’
900
1
*
‘
1100
*
*
1300
ELECTRON KINETIC ENERGY (eV)
Fig. 3. Augerspectraof (a) Mg0.,&0.4, and (b) purealuminiumin the AR, SP and lOOLO, states. (2) The segregation of magnesium during the oxidation process is confirmed here, as it was confirmed for Mg,.,, In0.05, by the fact that the ratio of the magnesium AES line to the aluminium AES line is about three times larger in the 100 L 0, state than it is in the SP state. (3) The shapes of the magnesium and aluminium AES lines in the AR state are typical of the oxides of the metaIs [3], and the shapes in the SP state are
259
typical of the pure metals [3]. These two characteristic shapes are superposed in the 100 L 0, state. The corresponding X-ray photoelectron spectra are shown in Fig. 4. As for MgO.&O.O,, the features of these spectra agree with those of the Auger spectra. The additional features are as follows. (4) The magnesium XPS lines of the AR state are characteristic of the oxide type whereas the XPS aluminium peaks are composed of superposed metal and oxide lines of about the same relative intensity. This can be explained either by the existence of metallic aluminium islands on the AR samples or by the presence of a thin aluminium oxide layer. (5) The XPS aluminium lines of the SP state are almost purely metallic aluminium whereas the magnesium lines are composed of both metal and oxide contributions of about equal intensities. The additional oxide contribution is observed in both the aluminium and the magnesium peaks.
Mq 2s
Al 2s M Al 2~
ELECTRON
BINDING
ENERGY
ox
2~ metal
(eV)
Fig. 4. X-ray photoelectron spectra of the 40-140 eV binding energy region of Mg,,,,Al,,,, in the AR, SP and 100 L 0, states. The long vertical lines indicate the metallic states and the short vertical lines indicate the oxide states. The intensity scales are only approximately equivalent.
260
for
(6) As Mg0.95&.05tno energy shifts are observed for the aluminium or magnesium lines (in either the metal or the oxide state) compared with the energies of the pure metals [4]. This indicates that there is no detectable charge transfer in the Mg-AI alloys. The segregation of magnesium during the oxidation process and the preferential sputtering of the magnesium ions is also observed in this alloy. The magnesium-to-aluminium intensity ratios in the X-ray photoelectron spectra of the AR, SP and 100 L 0, states are in the ratio 8: 1: 1.8. As discussed earlier these values are less reliable than those determined from the Auger spectra (34: 1: 27) owing to the partial oxidation of the sample in the SP state by the residual gases during the measurement period. It is again impossible to separate quantitatively the preferential sputtering of magnesium from the oxidation-induced segregation effect. However, the magnesium-to-indium intensity ratio of the Auger spectra of AR Mg,,,,In,,,S is a factor of 9 greater than that of the SP state, whereas in AR Mg0.ssAlo,45 it is a factor of 34 greater than that of the SP state. This observation, together with the comparison of the magnesium-to-indium and magnesium-to-al~inium intensity ratios in the AR states of both samples, suggests that the segregation effect due to atmospheric oxidation is more pronounced for the Mg-Al alloys than for the Mg-In alloys. (7) A pronounced peak appears at 103 eV (13.5 eV refative to the Mg 2s line) in the X-ray photoelectron spectrum of AR Mg,.,SA1,,4,. This peak is almost completely removed by sputtering and is barely detectable in the spectrum of the SP state. The source of this peak is not clear. It may be a plasmon satellite of the magnesium line, although it is unlikely that it would be associated with only one X-ray photoelectron line. To clarify this point, we performed an electron loss spectroscopy examination of this sample in the AR and SP states (a I.5 keV electron beam was used to sample at about the same depth as sampled by the outgoing photoelectrons). The electron loss spectra are shown in Fig. 5. It is immediately obvious that the plasmon intensity (it is well established that the satellites of the magnesium and aluminium lines are plasmons [S]) in the SP state is greater than that in the AR state. The energy of the SP plasmon is 14.5eV, whereas that of the AR plasmon is 13.7 eV. The corresponding energies for the plasmons in the SP states of magnesium and aluminium (not shown) are 11 eV and 15.5eV respectively. This excludes the possibility that the peak in the X-ray photoelectron spectrum of AR Mg,,,,Al,,,, is a plasmon satellite. 4. Discussion 4.1. Bonding characteristics of the alloy Electron charge transfer from magnesium to indium takes place in MgO.&o.Os (Section 3.1) but not in the Mg-Al alloys (Sections 3.2 and 3.3). This suggests that the electrochemical interactions between magnesium and indium are stronger than those between magnesium and aluminium. These observations are consistent with Pauling’s electronegativity scale, according to which
261
I
t
1470
,
t
1480 ELECTRON
?500
:490 KINt:IC
E&ERG"
I
(eV)
Fig. 5. Electron loss spectra of Mg,,,,Al,,,, for a primary electron energy of 1.5 keV: curve a, elastic peak (dN/dE mode) obtained in the AR state; curve b, satellites obtained in the AR state enlarged by a factor of 20; curve c, satellites obtained in the SP state enlarged by a factor of 20. The intensity scales of curves b and c are equivalent.
the electronegativity of aluminium (1.5) lies between those of magnesium (1.2) and indium (1.7). However, it contradicts Miedema’s scale [S], which is based on the heat of formation of various binary alloys and intermetallic compounds, according to which aluminium is the most electronegative of the three metals. The above observation might also be regarded as a contradiction of the phase diagrams of the Mg-Al and Mg-In systems [I’?]. Intermediate ordered phases which are stable up to the melting point are present in the Mg-Al system. In the Mg-In system, however, the ordered phases transform to disordered solid solutions before reaching the melting point. It is believed that the absence of intermediate ordered compounds near the melting point indicates weaker electromagnetic interactions. However, it may be due to other properties of the constituents such as the size factor. A favourable size factor between the metallic constituents of an alloy will promote the formation of a solid solution, particularly at high temperatures, and may override the electrochemical interactions. The size factor of the Mg-In system is more favourable than that of the Mg-Al system [7]. 4.2. ConducGon electron densities Plasmon satellites were present in the X-ray photoelectron spectra (Fig. 4) and electron loss spectra of Mg,.,,Al,,,$ as well as in the X-ray photoelectron spectra of pure magnesium and aluminium (not shown). Magnesium and aluminium plasmons have been investigated by Powell and Swan [SJ. The satellites are assumed to be due to collective excitation of the conduction electrons [8]. The energy of the elementary plasmon is given by [5] :
262
where W, is the plasma oscillation frequency and n is the density of free electrons (electrons per unit volume) in the material. Powell and Swan calculated h W, for both aluminium (assuming three free electrons per atom) and magnesium (assuming two free electrons per atom). Their experimental and calculated values together with the results of the present study are given in Table 1. The agreement between our results for the pure metals and those of Powell and Swan (both experimental and theoretical) is good. TABLE 1 Experimental
and calculated
Sample
energies of the elementary
Plasmon energy (eV) This work
Al Mg Mg,,,, Al,.,, W’f Mg,,,,Al,.,, (AR)
bulk plasmons
Ref. 5 Experimental
Calculated
15.6
15.5
11.0 14.5 13.7
10.6
15.8 10.9 -
The delisity of Mg,,,,A10.45 is 2.125 g cm -3 [9]. By using this value together with the value of n derived from eqn. (l), we obtain the unexpected result that there should be three free electrons per formula unit (Mg,,,,Al,,,& If we assume that magnesium donates two free electrons to the conductivity band and aluminium donates three, there should be effectively 2.45 free electrons per formula unit. This result is not fully understood. The application of eqn. (l), which is derived from simplified plasmon theory, is probably not justified for the intermetallic Mg-Al compound. The decrease in the plasmon energy in the AR state relative to the SP state is in accordance with the presence of an oxide layer. MgO and Al,O, are known to be insulators. However, the fact that the change in the plasmon energy is small shows that the oxide layer is thin and that the surface still has a metallic character. This is further supported by the fact that no charging effects were detected in the X-ray photoelectron and Auger spectra of the AR state. 4.3. Oxidation b~ha~iou~ Both constituents of the Mg-A1 alloys oxidized during their exposure to oxygen (Sections 3.2 and 3.3), whereas in the MgO.,,In,,,, alloy only the magnesium was oxidized and the indium remained in the metallic state (Section 3.1). The oxygen affinities of magnesium and aluminium in the alloys appear to be similar to those of the pure unalloyed metals. However, the oxidation resistance of indium in Mg-In alloys is greater than that of pure metallic indium as shown by the oxidized AR state of pure indium. The higher oxygen affinities of magnesium and al~ini~ relative to that of indium are consistent with the greater stability of their oxides [8] as measured by the enthalpy AHof formation
263
of the oxides. The relevant values are [lo] AH,,, = 287 kcal (mol 0,)-l, and AH,,,o, = 148kcal (mol 0,)-l. AHAM,~ = 267 kcal (mol0,))’ An additional explanation for the enhanced oxidation resistance of indium in Mg,,,,In,~,, relative to that of pure indium is probably the strength of the electrochemical interactions between magnesium and indium as shown by the presence of electron charge transfer (Section 3.1). It has been reported that the addition of small amounts (less than about 1 at.%) of group IIIa metals to magnesium increases its rate of reaction with gaseous hydrogen. This catalytic activity was similar for various group IIIa additives investigated. The results of the present study indicate that the addition of aluminium and indium to magnesium produces different surface characteristics. It is thus reasonable to conclude that the catalytic effect of the various group IIIa metals is not due to gas-solid interactions at the surface. It had been suggested that bulk diffusion properties are responsible for this catalytic effect [2].
Acknowledgment The authors samples.
would
like to thank
Mr. H. Mamman
for preparing
the
References 1 2 3
4
M. H. Mintz, Z. Gavra and Z. Hadari, J. Inorg. NucE. Chem., 40 (1978) 765. M. H. Mintz, S. Malkiely, Z. Gavra and Z. Hadari, J. Inorg. Nucl. Chem., 40 (1978) 1949. L. E. Davis, N. C. MacDonald, P. W. Palmberg, G. E. Riach and R. E. Weber, Handbook of Auger Electron Spectroscopy, Physical Electronics Inc., Eden Prairie, MN, 1976. C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E. Muilenberg (eds.), Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corporation (Phys-
ical Electronics Division), Eden Prairie, MN, 1979. 5 C. J. Powell and J. B. Swan, Phys. Rev., 115(1959) 869; I16(1959) 81. 6 A. R. Miedema, J. Less-Common Met., 32(1973) 117. 7 8
9 10
G. V. Raynor, The Physical Metallurgy of Magnesium and its Alloys, Pergamon, Oxford, 1959. D. Pines and D. Bohm, Phys. Rev., 85 (1952) 338. D. Pines, Rev. Mod. Phys., 28(1956) 184. P. Nozieres and D. Pines, Phys. Rev., 109 (1958) 1062. E. S. Makarov, Dokl. Akad. Nauk S.S.S.R., 74(1950)935. 0. Kubachevski and B. F. Hopkins, Oxidation of Metals and Alloys, Butterworths, London, 1962.
of the Less-Common
Metals,
92(1983)
253
253-263
ELECTRON SPECTROSCOPY STUDIES OF THE OXIDATION BEHAVIOUR AND THE ELECTRONIC PROPERTIES OF SOME Mg-In AND Mg-Al ALLOYS
N. SHAMIR, M. H. MINTZ*, J. BLOCH and U. ATZMONY* Physics Department, Nuclear Research Centre-
Negev, P.O. Box 9001, Beer-Sheva (Israel)
(Received November 26,1982)
Summary The electronic properties and surface characteristics of some Mg-In and Mg-Al alloys were studied by electron spectroscopy techniques (Auger electron spectroscopy and X-ray photoelectron spectroscopy). It was found that the electrochemical interactions between magnesium and indium are stronger than those between magnesium and aluminium. It was further found that the surface oxidation behaviour of indium-containing alloys is different from that of aluminium-containing alloys. In the former case only the magnesium atoms are oxidized at the surface, leaving the indium atoms in their metallic state, while in the latter case both magnesium and aluminium are oxidized at the surface. Both types of alloy display oxidation-enhanced surface segregation of the magnesium atoms. The hydrogenation behaviour of the various magnesium-base alloys is discussed in the light of their surface properties.
1. Introduction Magnesium and its alloys have much technological importance as low density structural materials and as potential high weight content hydrogen storage materials. In the latter application it has been found that the addition of small amounts (i.e. about 1%) of either aluminium or indium to magnesium metal catalyses the hydrogenation reaction [l, 21. The purpose of the present study was to investigate the surface properties of such dilute alloys and to study the mechanisms that control the initial stages of the reactions of these alloys with oxygen and hydrogen. 2. Experimental
details
Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) measurements were performed on polycrystalline samples of Mg,,,,In,,,,, *Also at the Ben Gurion University of the Negev, Beer-Sheva, Israel. 0022-5088/83/$3.00
QElsevier
Sequoia/Printed
in The Netherlands
254
together with samples of pure magnesium, indium Mgo.95Alo.05 andMgo.55Alo.45 and aluminium for reference. The alloys were prepared by melting stoichiometric amounts of the constituents in an induction furnace using a magnesia crucible. The samples were cleaned using Ccl, and were introduced into a PHI 548 surface analysis system (base pressure, 2 x lo-” Torr). Measurements were performed on each sample in the three following states: (i) the as-received (AR) state; (ii) the sputtered (SP) state which was obtained after sputtering the sample with 4.5 keV Ar+ ions for a long period until all the contaminants had been completely removed or had reached a steady state; (iii) the 100 L 0, state which was obtained by exposing SP samples to 100 langmuirs of oxygen (Matheson 5N grade) introduced to the chamber through a leak valve. 3. Results and analysis
The Auger spectra of Mg0,951n0,05 and of pure indium in the three states are presented in Fig. 1. Several features can be seen by inspecting and comparing the spectra. (1) Although indium is only 5% of the alloy its spectral line is relatively strong. This is due to its high cross section for the Auger process [3]. (2) The sputtering process strongly enhanced the indium peak relative to the magnesium peak, indicating an increase in the indium concentration on the surface. As will be proved later, the magnesium segregates at the surface during oxidation. Therefore the AR state of the alloy, which is naturally oxidized, has a segregated layer on its surface. This layer is removed by the sputtering. In addition to the removal of this layer there may be preferential sputtering of the magnesium ions. (3) The ratio of the magnesium peak to the indium peak in the lOOLO, state is much higher than in the SP state (about a factor of 4 greater). This confirms the segregation of the magnesium at the surface during the oxidation process. (4) The indium AES lines in all the spectra shown in Fig. 1, except that of pure indium in the AR state, have a shape characteristic of indium in its metallic state [3]. The magnesium line in the AR state is typical of magnesium oxide, in the SP state it is typical of magnesium metal and in the 100 L 0, state it exhibits an intermediate state [S]. It is therefore concluded that there is selective oxidation of the magnesium ions during exposure of the Mg,,,,In,,,, alloy to oxygen and that its surface in the AR state consists of magnesium oxide and metallic indium. The corresponding X-ray photoelectron spectra of Mg,,,,In,.,, and pure indium (Fig. 2) reveal the following additional features. (5) The In 3d lines in the alloy spectra are shifted by about 0.5 eV to a lower binding energy relative to their positions in the pure indium spectra. This indicates electron charge transfer from the magnesium to the indium in the alloy.
255
.O-;300
500
700
900
1100
ELECTRON KINETIC ENERGY (eV1
Fig. 1. Auger spectra of (a) Mg,,,,In,,,, intensity scales are only approximately
and (b) pure indium in the AR, SP and 100L 0, states. The equivalent.
(6) The In 3d,,, XPS line of Mgo,,SIno,,5 lies at the same energy for all three states, but for the AR state of pure indium it is shifted by about 1 eV to a higher binding energy which is typical of the oxidized state [4]. The centre of the Mg 2p however, moves from 50.5 eV in the AR state to 49.5 eV in lines ofMg,,,,In,.,,,
Mg 2s
1
455
I
I
440
100
ELECTRON BINDING ENERGY
I
I,
I,,
80
60
Mg 2P
In 2s
I
I
, 40
, 20
,
4
0
(ev)
Fig, 2. X-ray photoelectron spectra of (a) the In 3d lines of Mg,,,,In,.,, and pure indium in the AR, SP and fOOL 0, states, (b) the Mg 2p line of Mg,,,,In,.Os in the same states and (c)the binding energy region (O-100 eV) of Mg,,,,En,~,,. The long vertical lines indicate the energy corresponding to a metallic state of Mgo,ssIn,,,, while the short vertical lines (except for(c)) correspond to oxide states (ox). The intensity scales are different far each spectrum.
the SP state and to an intermediate energy for the lOOLO, state, which is characteristic of an oxidation process. Evidence for the preferential sputtering of the magnesium ions and their surface segregation during oxidation (features (2) and (3)) is also shown by the X-ray photoelectron spectra.
The relative contributions of the two effects cannot be calculated directly. The peak ratios of the oxygen, magnesium and indium lines in the Auger and X-ray photoelectron spectra of the various states do not give quantitative results because, owing to the different kinetic energies, the detected electrons yield information from various depths. Analysis of the Auger spectra gives better relative estimates of the effects than analysis of the X-ray photoelectron spectra because a much longer time is required to accumulate the XPS data so that partial oxidation of the SP state by the residual gases in the system can occur. The magnesium-to-indium intensity ratios of the Auger spectra are a factor of 9 less in the SP state than in the AR state. As the kinetic energy of the indium line is less than that of the magnesium line this factor should be regarded as an upper limit on the combined effect. This ratio is a factor of 4 greater in the 100 L 0, state than in the SP state. This is the lower limit on the segregation effect due to oxidation by 100 L 0,. 3.2. Mg,,,A 10., The cross sections of aluminium for the Auger and photoelectron processes are small compared with that of indium. It was therefore more difficult to obtain meaningful information from the Auger and X-ray photoelectron spectra of the sample. Therefore the spectra are not shown. Mg,.,,Al,.,, The following features corresponding to the Mg,,,,A10,05 spectra should, however, be noted. (1) The weak aluminium AES line in the AR state is typical of neither pure aluminium metal nor its oxide and can be regarded as a superposition of both states. In the SP state it is characteristic of pure aluminium and it shows evidence of partial oxidation again in the 100 L 0, state. (2) It is even more difficult to obtain any information about the aluminium from the X-ray photoelectron spectra as unfortunately its XPS lines coincide with a magnesium plasmon satellite. This satellite varies from state to state which makes it harder to detect the changes in the aluminium XPS lines (which are very small in any case). However, it can be seen that the Al 2s peak lies at about 120 eV (which is the position for oxidized pure aluminium [4]) in the AR state and at about 118 eV (its position for pure aluminium metal [4]) in the SP state. This result shows that no shift in the Al 2s lines between the alloy and pure aluminium, which would have indicated electron charge transfer, occurs. 3.3. Mg,,,A &.,, The measurements on this sample were performed to enable us to obtain more reliable results concerning the behaviour of aluminium in Mg-Al alloys. and of pure aluminium in the three The Auger spectra of Mg,,,,Al,,,, states are shown in Fig. 3. The following features can be seen. (1) The aluminium AES line is almost absent in the spectrum of the alloy in the AR state, which indicates marked segregation of magnesium at the surface. It is strongly enhanced (by a factor of 34 compared with the magnesium line) by the sputtering process. This shows that the combined effect of preferential sputtering of magnesium ions and removal of the segregation layer of magnesium is stronger here than for the Mg,,,,In,.,, sample.
W t 0
-r 500
SP
’
’
700
-
’
900
1
*
‘
1100
*
*
1300
ELECTRON KINETIC ENERGY (eV)
Fig. 3. Augerspectraof (a) Mg0.,&0.4, and (b) purealuminiumin the AR, SP and lOOLO, states. (2) The segregation of magnesium during the oxidation process is confirmed here, as it was confirmed for Mg,.,, In0.05, by the fact that the ratio of the magnesium AES line to the aluminium AES line is about three times larger in the 100 L 0, state than it is in the SP state. (3) The shapes of the magnesium and aluminium AES lines in the AR state are typical of the oxides of the metaIs [3], and the shapes in the SP state are
259
typical of the pure metals [3]. These two characteristic shapes are superposed in the 100 L 0, state. The corresponding X-ray photoelectron spectra are shown in Fig. 4. As for MgO.&O.O,, the features of these spectra agree with those of the Auger spectra. The additional features are as follows. (4) The magnesium XPS lines of the AR state are characteristic of the oxide type whereas the XPS aluminium peaks are composed of superposed metal and oxide lines of about the same relative intensity. This can be explained either by the existence of metallic aluminium islands on the AR samples or by the presence of a thin aluminium oxide layer. (5) The XPS aluminium lines of the SP state are almost purely metallic aluminium whereas the magnesium lines are composed of both metal and oxide contributions of about equal intensities. The additional oxide contribution is observed in both the aluminium and the magnesium peaks.
Mq 2s
Al 2s M Al 2~
ELECTRON
BINDING
ENERGY
ox
2~ metal
(eV)
Fig. 4. X-ray photoelectron spectra of the 40-140 eV binding energy region of Mg,,,,Al,,,, in the AR, SP and 100 L 0, states. The long vertical lines indicate the metallic states and the short vertical lines indicate the oxide states. The intensity scales are only approximately equivalent.
260
for
(6) As Mg0.95&.05tno energy shifts are observed for the aluminium or magnesium lines (in either the metal or the oxide state) compared with the energies of the pure metals [4]. This indicates that there is no detectable charge transfer in the Mg-AI alloys. The segregation of magnesium during the oxidation process and the preferential sputtering of the magnesium ions is also observed in this alloy. The magnesium-to-aluminium intensity ratios in the X-ray photoelectron spectra of the AR, SP and 100 L 0, states are in the ratio 8: 1: 1.8. As discussed earlier these values are less reliable than those determined from the Auger spectra (34: 1: 27) owing to the partial oxidation of the sample in the SP state by the residual gases during the measurement period. It is again impossible to separate quantitatively the preferential sputtering of magnesium from the oxidation-induced segregation effect. However, the magnesium-to-indium intensity ratio of the Auger spectra of AR Mg,,,,In,,,S is a factor of 9 greater than that of the SP state, whereas in AR Mg0.ssAlo,45 it is a factor of 34 greater than that of the SP state. This observation, together with the comparison of the magnesium-to-indium and magnesium-to-al~inium intensity ratios in the AR states of both samples, suggests that the segregation effect due to atmospheric oxidation is more pronounced for the Mg-Al alloys than for the Mg-In alloys. (7) A pronounced peak appears at 103 eV (13.5 eV refative to the Mg 2s line) in the X-ray photoelectron spectrum of AR Mg,.,SA1,,4,. This peak is almost completely removed by sputtering and is barely detectable in the spectrum of the SP state. The source of this peak is not clear. It may be a plasmon satellite of the magnesium line, although it is unlikely that it would be associated with only one X-ray photoelectron line. To clarify this point, we performed an electron loss spectroscopy examination of this sample in the AR and SP states (a I.5 keV electron beam was used to sample at about the same depth as sampled by the outgoing photoelectrons). The electron loss spectra are shown in Fig. 5. It is immediately obvious that the plasmon intensity (it is well established that the satellites of the magnesium and aluminium lines are plasmons [S]) in the SP state is greater than that in the AR state. The energy of the SP plasmon is 14.5eV, whereas that of the AR plasmon is 13.7 eV. The corresponding energies for the plasmons in the SP states of magnesium and aluminium (not shown) are 11 eV and 15.5eV respectively. This excludes the possibility that the peak in the X-ray photoelectron spectrum of AR Mg,,,,Al,,,, is a plasmon satellite. 4. Discussion 4.1. Bonding characteristics of the alloy Electron charge transfer from magnesium to indium takes place in MgO.&o.Os (Section 3.1) but not in the Mg-Al alloys (Sections 3.2 and 3.3). This suggests that the electrochemical interactions between magnesium and indium are stronger than those between magnesium and aluminium. These observations are consistent with Pauling’s electronegativity scale, according to which
261
I
t
1470
,
t
1480 ELECTRON
?500
:490 KINt:IC
E&ERG"
I
(eV)
Fig. 5. Electron loss spectra of Mg,,,,Al,,,, for a primary electron energy of 1.5 keV: curve a, elastic peak (dN/dE mode) obtained in the AR state; curve b, satellites obtained in the AR state enlarged by a factor of 20; curve c, satellites obtained in the SP state enlarged by a factor of 20. The intensity scales of curves b and c are equivalent.
the electronegativity of aluminium (1.5) lies between those of magnesium (1.2) and indium (1.7). However, it contradicts Miedema’s scale [S], which is based on the heat of formation of various binary alloys and intermetallic compounds, according to which aluminium is the most electronegative of the three metals. The above observation might also be regarded as a contradiction of the phase diagrams of the Mg-Al and Mg-In systems [I’?]. Intermediate ordered phases which are stable up to the melting point are present in the Mg-Al system. In the Mg-In system, however, the ordered phases transform to disordered solid solutions before reaching the melting point. It is believed that the absence of intermediate ordered compounds near the melting point indicates weaker electromagnetic interactions. However, it may be due to other properties of the constituents such as the size factor. A favourable size factor between the metallic constituents of an alloy will promote the formation of a solid solution, particularly at high temperatures, and may override the electrochemical interactions. The size factor of the Mg-In system is more favourable than that of the Mg-Al system [7]. 4.2. ConducGon electron densities Plasmon satellites were present in the X-ray photoelectron spectra (Fig. 4) and electron loss spectra of Mg,.,,Al,,,$ as well as in the X-ray photoelectron spectra of pure magnesium and aluminium (not shown). Magnesium and aluminium plasmons have been investigated by Powell and Swan [SJ. The satellites are assumed to be due to collective excitation of the conduction electrons [8]. The energy of the elementary plasmon is given by [5] :
262
where W, is the plasma oscillation frequency and n is the density of free electrons (electrons per unit volume) in the material. Powell and Swan calculated h W, for both aluminium (assuming three free electrons per atom) and magnesium (assuming two free electrons per atom). Their experimental and calculated values together with the results of the present study are given in Table 1. The agreement between our results for the pure metals and those of Powell and Swan (both experimental and theoretical) is good. TABLE 1 Experimental
and calculated
Sample
energies of the elementary
Plasmon energy (eV) This work
Al Mg Mg,,,, Al,.,, W’f Mg,,,,Al,.,, (AR)
bulk plasmons
Ref. 5 Experimental
Calculated
15.6
15.5
11.0 14.5 13.7
10.6
15.8 10.9 -
The delisity of Mg,,,,A10.45 is 2.125 g cm -3 [9]. By using this value together with the value of n derived from eqn. (l), we obtain the unexpected result that there should be three free electrons per formula unit (Mg,,,,Al,,,& If we assume that magnesium donates two free electrons to the conductivity band and aluminium donates three, there should be effectively 2.45 free electrons per formula unit. This result is not fully understood. The application of eqn. (l), which is derived from simplified plasmon theory, is probably not justified for the intermetallic Mg-Al compound. The decrease in the plasmon energy in the AR state relative to the SP state is in accordance with the presence of an oxide layer. MgO and Al,O, are known to be insulators. However, the fact that the change in the plasmon energy is small shows that the oxide layer is thin and that the surface still has a metallic character. This is further supported by the fact that no charging effects were detected in the X-ray photoelectron and Auger spectra of the AR state. 4.3. Oxidation b~ha~iou~ Both constituents of the Mg-A1 alloys oxidized during their exposure to oxygen (Sections 3.2 and 3.3), whereas in the MgO.,,In,,,, alloy only the magnesium was oxidized and the indium remained in the metallic state (Section 3.1). The oxygen affinities of magnesium and aluminium in the alloys appear to be similar to those of the pure unalloyed metals. However, the oxidation resistance of indium in Mg-In alloys is greater than that of pure metallic indium as shown by the oxidized AR state of pure indium. The higher oxygen affinities of magnesium and al~ini~ relative to that of indium are consistent with the greater stability of their oxides [8] as measured by the enthalpy AHof formation
263
of the oxides. The relevant values are [lo] AH,,, = 287 kcal (mol 0,)-l, and AH,,,o, = 148kcal (mol 0,)-l. AHAM,~ = 267 kcal (mol0,))’ An additional explanation for the enhanced oxidation resistance of indium in Mg,,,,In,~,, relative to that of pure indium is probably the strength of the electrochemical interactions between magnesium and indium as shown by the presence of electron charge transfer (Section 3.1). It has been reported that the addition of small amounts (less than about 1 at.%) of group IIIa metals to magnesium increases its rate of reaction with gaseous hydrogen. This catalytic activity was similar for various group IIIa additives investigated. The results of the present study indicate that the addition of aluminium and indium to magnesium produces different surface characteristics. It is thus reasonable to conclude that the catalytic effect of the various group IIIa metals is not due to gas-solid interactions at the surface. It had been suggested that bulk diffusion properties are responsible for this catalytic effect [2].
Acknowledgment The authors samples.
would
like to thank
Mr. H. Mamman
for preparing
the
References 1 2 3
4
M. H. Mintz, Z. Gavra and Z. Hadari, J. Inorg. NucE. Chem., 40 (1978) 765. M. H. Mintz, S. Malkiely, Z. Gavra and Z. Hadari, J. Inorg. Nucl. Chem., 40 (1978) 1949. L. E. Davis, N. C. MacDonald, P. W. Palmberg, G. E. Riach and R. E. Weber, Handbook of Auger Electron Spectroscopy, Physical Electronics Inc., Eden Prairie, MN, 1976. C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E. Muilenberg (eds.), Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corporation (Phys-
ical Electronics Division), Eden Prairie, MN, 1979. 5 C. J. Powell and J. B. Swan, Phys. Rev., 115(1959) 869; I16(1959) 81. 6 A. R. Miedema, J. Less-Common Met., 32(1973) 117. 7 8
9 10
G. V. Raynor, The Physical Metallurgy of Magnesium and its Alloys, Pergamon, Oxford, 1959. D. Pines and D. Bohm, Phys. Rev., 85 (1952) 338. D. Pines, Rev. Mod. Phys., 28(1956) 184. P. Nozieres and D. Pines, Phys. Rev., 109 (1958) 1062. E. S. Makarov, Dokl. Akad. Nauk S.S.S.R., 74(1950)935. 0. Kubachevski and B. F. Hopkins, Oxidation of Metals and Alloys, Butterworths, London, 1962.