- Email: [email protected]
AES analysis of reed switch contact surfaces and gas chromatographic analysis of filling gases
Vacuum/volume37/numbers 1/2/pages 165 to 167/1987 Printed in Great Britain
0042-207X/8753.00 +.00 Pergamon Journals Ltd
AES analysis of reed switch contact surfaces and gas chromatographic analysis of filling gases M M u r k o - J e z o v ~ , e k and F B r e c e l j , Institut za Elektroniko in Vakuumsko Tehniko, Teslova 30, 61000
Ljubljana, Yugoslavia
Contact surfaces of miniature reed switches were examined with a scanning Auger microprobe. After life-time tests under low load conditions, one or more dark spots on the contact surface were visible, consisting mostly of carbon or carbonaceous compounds. Further, some attempts were made to establish the possible origin of this carbon contamination. The N2-H 2 reed switch filling gas was analyzed with gas chromatography and no impurity gases were found in it. Therefore, gases released from glass and metal or formed during the sealing operation were suspected. A special device was constructed to enable high frequency induction sealing of the kovar (Fe-Ni-Co) wire to the glass in purified hydrogen. With gas chromatography the following impurities were detected. H20, CO, CO 2, CH 4 and minor quantities of C2Hn(n = 2-6).
1. Introduction With miniaturisation of electric contact elements, the cleanliness of the contact surface becomes more and more important. Contaminants can cause an unacceptable increase in contact resistance due to formation of isolation or semiconducting films on contact surfaces. Ogren 1 has studied the formation of a highly resistive deposit in reed switches working in conditions of low level loads. He found that for no-load conditions a deposit was formed only on rhodium but, if the open circuit voltage was increased, so much that a molten metal bridge could be developed between the contacts, i.e. above 1 V, a coating was formed also on hard gold. If the load voltage is increased above the one needed to form an arc, carbon deposit is still formed but now has to compete with a transfer of contact material. The deposit, which, besides the base material, contained carbon and probably hydrogen, was assumed to be formed from gaseous organic impurity, trapped in the reed switch atmosphere, though such gas impurity has not been detected. Earlier 2 we found carbon contamination on contact surfaces in reed switches (CC Type, see Figure 5), produced at IEVT. After the life-time test at low load levels one or more dark spots were
?IT
?iii
.......... i
__5 .~_6
~'I L4
3
2
Figure 1. Device for high frequency induction sealing of the kovar wire to the glass: 1, 2--valves for filling the device with dry hydrogen and for introducing it to the gas chromatograph; 3--kovar wire ~b 1.2 mm, l= 80 mm (0.71 g); 4--four glass tubes, ~bo,t=r= 2.4 mm, ~bi.... = 1.55 mm (0.4 g); 5--support of the kovar wire; 6--glass envelope of the whole device; 7--high frequency induction heating coil.
Figure 2. Carbon Auger image C (272 eV) of contact having 0.08 D contact resistance after life-time test (50 million operations, magnification x 500.
visible on the contact area (see Figures 2 and 3). Analyses performed on the scanning Auger microprobe PHI, model SAM 545-A, showed that these were contamination spots:' 3 consisting mostly of carbon or carbonaceous compounds (since it is not possible to detect hydrogen with AES). After life-time tests the amounts of accumulated carbon were much greater than those found after various stages in sample production. The aim of this work was to find: (1) the relation between the contact resistance of reed switch contacts and carbon concentration in the contamination spots; (2) the thickness and the area of these spots and (3) to find the origin of this carbon contamination.
165
M Murko-Jezov.~ek and F Brecelj. AES analysis
support and 4 glass tubes were put on it (see Figure 1). After filling the device with purified hydrogen to a pressure of about 3 x 105 Pa, the kovar wire was heated inductively with high frequency to a temperature of about 1000°C for 5 s so that the glass tubes were sealed to it. The device was then connected to the entrance valve of the gas chromatograph (Perkin Elmer model F 17), which permits the introduction of reproducible volumes {e.g. 0.5 ml) of gas sample into the selected gas chromatographic column.
(a)
3. Results Figures 2 and 3 show the typical form of the contamination spots on the contact area of contact reeds. Figure 2 shows the carbon Auger image C (272 eV) on the contact having low contact resistance (0.08 fl) after life-time test (50 million operations). Figure 3 shows the carbon Auger image C (272 eV) (Figure 3a) and gold Auger image Au (69 eV) (Figure 3b) on the contact having high contact resistance (2 f~) after life-time test (5 million operations). Figures 4 and 5 represent the AES sputter depth profiles of samples shown in Figures 3 and 2. Comparing both carbon Auger images (Figures 2 and 3a) and depth profiles (Figures 4 and 5), we can see that the carbon contaminated area of the sample with high contact resistance was much smaller than that of the sample with low contact resistance, though the thickness of carbon spots was much greater. The majority of carbon on the high resistance contact is accumulated
(b)
10or
a
~00
b
_I00~ oc ~e • hl z~co
~o
~c
"
< 20 .
|
25pro
C
~50 cJ -L.~
~v Lx
¢"01
~
20
-
-
~
, 0 25 50 75 100
25 50 75 100 time [ rain ]
0 25 50 75 100
Sputtering
Figure 3. Carbon Auger image C (272 eV) (Figure 3a) and gold Auger image Au (69 eV) (Figure 3b) of the contact having 2 fl contact resistance after life-time test (5 million operations), magnification × 1000.
2. Experimental F r o m a certain quantity of reed switches which were subjected to life-time test at 15 V DC, 0.2 mA resistive load, and 5 × l0 * operations, 20 samples with low contact resistance (mean value 0.08 fl) and 20 samples with high contact resistance (mean value 18 l)) were taken. The switches were broken just before mounting both contact reeds onto the sample holder of the scanning Auger microprobe. AE spectra, on the basis of which all depth profiles were calculated, were taken with the primary electron beam of 5 keV. The analyzed area on the sample surface was in turn approximately 200, 1000 and 6000/~m 2. We supposed that gases released at the glass-to-metal sealing operations could be the origin of the carbon contamination on contact surfaces. Therefore, a special device (Figure l) was constructed to enable high frequency induction sealing of the kovar wire (Vacon 10, Vacuumschmelze, Hanau) to the matching borosilicate glass (No 8250, Schott, Mainz), in a closed space of volume 6.8 ml. The kovar wire (~b 1.2 mm) was previously annealed in a wet hydrogen atmosphere then mounted on the 1 66
Figure 4. The AES sputter depth profile of the contact having 2 f~ contact resistance after life-time test (5 million operations). Depth profile in Figure 4a shows the composition of the area of approx. 200 um 2, depth profile in Figure 4b shows the composition of the area of approximately 1000 ,um2 and depth profile in Figure 4c shows the composition of the area of approximately 6000 um 2. Sputtering rates: approx. 10 nm min 1. To make diagrams more clear, points for N and O, which are present in very low concentrations are omitted. a
.-
50 o
,7 "eed ~,~ltF~IE', T C~ type
0
b
10
20
30
Z,0
~ o ~
~.
c
~, ~0ntoct e.cec
0 10 20 30 40 Sputferinc] time [mtn ]
o u
0
10
20
30
1,0
Figure 5. The AES sputter depth profile of the contact having 0.08 contact resistance after fife-time test (50 million operations). The analysed areas in Figure 5a, 5b, 5c and sputtering rates are the same as in Figure 4a, 4b, 4c.
M Murko-Jezovgek and F Brecelj: AES analysis
on a relatively small area (Figure 4a). Concentration of carbon in this depth profile is much higher than on the other depth profiles (Figure 4b, 4c) and the C signal diminishes more slowly with the sputtering time. Simultaneously, concentrations of gold and other elements composing the contact alloy (Fe, Co, Ni) increase. The depth profile in Figure 4c is for all the elements similar to those we have taken outside the contact area. Essentially different are the depth profiles of the low resistance contact (Figure 5). Carbon concentration on the surface is lower than on the corresponding depth profiles, presented in Figure 4. Carbon spots are evidently spread over a larger area (see Figure 2). They are thinner and, therefore, do not cause an increase of the contact resistance. Analyzing also other contacts with high and low contact resistance we have come to similar conclusions. To find the origin of treated carbon contamination we have analyzed the contact surfaces after different phases of sample production, i.e. different cleaning procedures, gold plating, thermal treatments and final ultrasonic cleaning in freon TF (Du Pont). By completing all these phases very carefully the amount of carbon contamination was not higher than is usually found on similar metal surfaces exposed to the laboratory atmosphere. But in spite of such preparatory procedures the dark spots on contact areas after the life-time tests of reed switches with electrical loading, were again found. Another possible origin of this carbon contamination could be the N2-H 2 (10% H2) gas mixture with which the reed switches are filled. However, analysis by gas chromatography, revealed no impurity gases. Further, we supposed that impurity gases could be released during the sealing of the contact reeds in glass tubes in the presence of N2-H 2 gas mixture. To prove this hypothesis a special device was constructed (Figure 1) enabling high frequency induction sealing of the kovar wire to glass in purified hydrogen. During this sealing, released impurity gases (approximately 25/11) consisted mainly of water (65 vol%), CO (25 vol%), CH 4 (7 vol%) and of minor quantities of CO 2 and CzH n (n = 2 6). These gases derive from dissolved and adsorbed impurities of both partners being released at high temperature during the sealing process or formed in reactions with hydrogen or reactions between glass and the kovar surface. N 2 and 0 2 were also present in small amounts but were analyzed only qualitatively because they cannot be the origin of carbon contamination. By heating only kovar wire previously heated in wet hydrogen (dew point +20°C, at 900°C, 2 h), the amount of released or formed impurity gases was approximately 21/~1. They consisted of water 51.3 vol%), CO (42 vol%) and C H 4 (6.7 vol%).
simulate the first technique, where red-hot surfaces of metal and glass come into contact; the origin of impurity gases are both the kovar and the glass. In both experiments, i.e. by heating of kovar alone and by sealing of glass to kovar in the presence of hydrogen, the released or formed gas impurities contained about 7 vol% of methane. We suspect that the carbon which appears on the contacts is being formed by thermal decomposition of methane during electrical loading4. From the AES sputter depth profiles (Figure 4) and the area covered with carbon we could estimate that the quantity of carbon deposited on the contact was approximately 2 x 10- log. If all deposited carbon on both contacts originated from methane, it would correspond to 5.6 × 10 -4 pl of methane in the reed switch atmosphere. On the basis of the analyses of released gases, we can conclude that the concentration of methane in the reed switch atmosphere is high enough to form the described contamination spots. When sealing reed switches, the atmosphere does not always have equal reducing properties, because in the sealing machines the protective gas could be partly mixed with air. In some series of reed switches, contacts were not contaminated with carbon and we conclude that methane was then not present due to its oxidation.
5. Conclusions (1) After life-time tests of reed switches at low load conditions, the AES analyses of the contacts showed contamination spots consisting mostly of carbon or carbonaceous compounds. (2)From the carbon C (272 eV) Auger images and AES sputter depth profiles it was established that, on the contacts with high contact resistance, the carbon contamination spots were localized on smaller areas and were of higher carbon concentration and thickness compared with the low resistance contacts. (3) The main source of carbon on our reed switch contacts is most probably impurity gases originating from the glass at the sealing of contact reeds in the glass envelope in the presence ofN2-H z gas mixture. They remain partly trapped in reed switches and are thermally decomposed to carbon which deposits on contacts when they operate under electrical load.
Acknowledgement Discussions with Dr E Kansky, Institut za Elektroniko in Vakuumsko Tehniko, Ljubljana, are gratefully acknowledged.
4. Discussion References Based on the results of above-mentioned experiments, we can expect that carbon containing impurity gases release or form during the sealing of contact reeds into glass tubes where they remain trapped. For reed switch production two techniques are used: either the contact reeds are sealed directly into glass tubes in the presence of N2-H 2 gas mixture, or the contact reeds are pre-glazed in air and then sealed in glass tubes in the presence of N z - H z gas mixture. The experiments we have made (by heating kovar and glass)
1 S Ogren, IEEE Trans Components, Hybrids Manuf Technol, CHMT-3, 431 (1980). 2 M Murko-Jezov~ek,Bilten JUVAK, 16, 81 (1976); M Murko-Jezovgek, Thin Solid Films, 32, 366 (1976); M Murko-Jezov~ek, F Brecelj and B Jenko, Proc 7th Int Vac Conor and 3rd Int Conf Solid Surfaces, p 2343, Vienna (1977). 3 M Murko-Jezov~ek and B Jenko, Bilten JUVAK, 17, 443 (1979); M Murko-Jezov~ek and B Jenko, Proc 9th lnt Vac Conor and 5th Int Conf Solid Surfaces, Extended Abstracts, p 174, Madrid (1983). 4 E W Gray, IEEE Trans Parts, Hybrids Paekagin#, PHP-11, 121 (1975).
167
0042-207X/8753.00 +.00 Pergamon Journals Ltd
AES analysis of reed switch contact surfaces and gas chromatographic analysis of filling gases M M u r k o - J e z o v ~ , e k and F B r e c e l j , Institut za Elektroniko in Vakuumsko Tehniko, Teslova 30, 61000
Ljubljana, Yugoslavia
Contact surfaces of miniature reed switches were examined with a scanning Auger microprobe. After life-time tests under low load conditions, one or more dark spots on the contact surface were visible, consisting mostly of carbon or carbonaceous compounds. Further, some attempts were made to establish the possible origin of this carbon contamination. The N2-H 2 reed switch filling gas was analyzed with gas chromatography and no impurity gases were found in it. Therefore, gases released from glass and metal or formed during the sealing operation were suspected. A special device was constructed to enable high frequency induction sealing of the kovar (Fe-Ni-Co) wire to the glass in purified hydrogen. With gas chromatography the following impurities were detected. H20, CO, CO 2, CH 4 and minor quantities of C2Hn(n = 2-6).
1. Introduction With miniaturisation of electric contact elements, the cleanliness of the contact surface becomes more and more important. Contaminants can cause an unacceptable increase in contact resistance due to formation of isolation or semiconducting films on contact surfaces. Ogren 1 has studied the formation of a highly resistive deposit in reed switches working in conditions of low level loads. He found that for no-load conditions a deposit was formed only on rhodium but, if the open circuit voltage was increased, so much that a molten metal bridge could be developed between the contacts, i.e. above 1 V, a coating was formed also on hard gold. If the load voltage is increased above the one needed to form an arc, carbon deposit is still formed but now has to compete with a transfer of contact material. The deposit, which, besides the base material, contained carbon and probably hydrogen, was assumed to be formed from gaseous organic impurity, trapped in the reed switch atmosphere, though such gas impurity has not been detected. Earlier 2 we found carbon contamination on contact surfaces in reed switches (CC Type, see Figure 5), produced at IEVT. After the life-time test at low load levels one or more dark spots were
?IT
?iii
.......... i
__5 .~_6
~'I L4
3
2
Figure 1. Device for high frequency induction sealing of the kovar wire to the glass: 1, 2--valves for filling the device with dry hydrogen and for introducing it to the gas chromatograph; 3--kovar wire ~b 1.2 mm, l= 80 mm (0.71 g); 4--four glass tubes, ~bo,t=r= 2.4 mm, ~bi.... = 1.55 mm (0.4 g); 5--support of the kovar wire; 6--glass envelope of the whole device; 7--high frequency induction heating coil.
Figure 2. Carbon Auger image C (272 eV) of contact having 0.08 D contact resistance after life-time test (50 million operations, magnification x 500.
visible on the contact area (see Figures 2 and 3). Analyses performed on the scanning Auger microprobe PHI, model SAM 545-A, showed that these were contamination spots:' 3 consisting mostly of carbon or carbonaceous compounds (since it is not possible to detect hydrogen with AES). After life-time tests the amounts of accumulated carbon were much greater than those found after various stages in sample production. The aim of this work was to find: (1) the relation between the contact resistance of reed switch contacts and carbon concentration in the contamination spots; (2) the thickness and the area of these spots and (3) to find the origin of this carbon contamination.
165
M Murko-Jezov.~ek and F Brecelj. AES analysis
support and 4 glass tubes were put on it (see Figure 1). After filling the device with purified hydrogen to a pressure of about 3 x 105 Pa, the kovar wire was heated inductively with high frequency to a temperature of about 1000°C for 5 s so that the glass tubes were sealed to it. The device was then connected to the entrance valve of the gas chromatograph (Perkin Elmer model F 17), which permits the introduction of reproducible volumes {e.g. 0.5 ml) of gas sample into the selected gas chromatographic column.
(a)
3. Results Figures 2 and 3 show the typical form of the contamination spots on the contact area of contact reeds. Figure 2 shows the carbon Auger image C (272 eV) on the contact having low contact resistance (0.08 fl) after life-time test (50 million operations). Figure 3 shows the carbon Auger image C (272 eV) (Figure 3a) and gold Auger image Au (69 eV) (Figure 3b) on the contact having high contact resistance (2 f~) after life-time test (5 million operations). Figures 4 and 5 represent the AES sputter depth profiles of samples shown in Figures 3 and 2. Comparing both carbon Auger images (Figures 2 and 3a) and depth profiles (Figures 4 and 5), we can see that the carbon contaminated area of the sample with high contact resistance was much smaller than that of the sample with low contact resistance, though the thickness of carbon spots was much greater. The majority of carbon on the high resistance contact is accumulated
(b)
10or
a
~00
b
_I00~ oc ~e • hl z~co
~o
~c
"
< 20 .
|
25pro
C
~50 cJ -L.~
~v Lx
¢"01
~
20
-
-
~
, 0 25 50 75 100
25 50 75 100 time [ rain ]
0 25 50 75 100
Sputtering
Figure 3. Carbon Auger image C (272 eV) (Figure 3a) and gold Auger image Au (69 eV) (Figure 3b) of the contact having 2 fl contact resistance after life-time test (5 million operations), magnification × 1000.
2. Experimental F r o m a certain quantity of reed switches which were subjected to life-time test at 15 V DC, 0.2 mA resistive load, and 5 × l0 * operations, 20 samples with low contact resistance (mean value 0.08 fl) and 20 samples with high contact resistance (mean value 18 l)) were taken. The switches were broken just before mounting both contact reeds onto the sample holder of the scanning Auger microprobe. AE spectra, on the basis of which all depth profiles were calculated, were taken with the primary electron beam of 5 keV. The analyzed area on the sample surface was in turn approximately 200, 1000 and 6000/~m 2. We supposed that gases released at the glass-to-metal sealing operations could be the origin of the carbon contamination on contact surfaces. Therefore, a special device (Figure l) was constructed to enable high frequency induction sealing of the kovar wire (Vacon 10, Vacuumschmelze, Hanau) to the matching borosilicate glass (No 8250, Schott, Mainz), in a closed space of volume 6.8 ml. The kovar wire (~b 1.2 mm) was previously annealed in a wet hydrogen atmosphere then mounted on the 1 66
Figure 4. The AES sputter depth profile of the contact having 2 f~ contact resistance after life-time test (5 million operations). Depth profile in Figure 4a shows the composition of the area of approx. 200 um 2, depth profile in Figure 4b shows the composition of the area of approximately 1000 ,um2 and depth profile in Figure 4c shows the composition of the area of approximately 6000 um 2. Sputtering rates: approx. 10 nm min 1. To make diagrams more clear, points for N and O, which are present in very low concentrations are omitted. a
.-
50 o
,7 "eed ~,~ltF~IE', T C~ type
0
b
10
20
30
Z,0
~ o ~
~.
c
~, ~0ntoct e.cec
0 10 20 30 40 Sputferinc] time [mtn ]
o u
0
10
20
30
1,0
Figure 5. The AES sputter depth profile of the contact having 0.08 contact resistance after fife-time test (50 million operations). The analysed areas in Figure 5a, 5b, 5c and sputtering rates are the same as in Figure 4a, 4b, 4c.
M Murko-Jezovgek and F Brecelj: AES analysis
on a relatively small area (Figure 4a). Concentration of carbon in this depth profile is much higher than on the other depth profiles (Figure 4b, 4c) and the C signal diminishes more slowly with the sputtering time. Simultaneously, concentrations of gold and other elements composing the contact alloy (Fe, Co, Ni) increase. The depth profile in Figure 4c is for all the elements similar to those we have taken outside the contact area. Essentially different are the depth profiles of the low resistance contact (Figure 5). Carbon concentration on the surface is lower than on the corresponding depth profiles, presented in Figure 4. Carbon spots are evidently spread over a larger area (see Figure 2). They are thinner and, therefore, do not cause an increase of the contact resistance. Analyzing also other contacts with high and low contact resistance we have come to similar conclusions. To find the origin of treated carbon contamination we have analyzed the contact surfaces after different phases of sample production, i.e. different cleaning procedures, gold plating, thermal treatments and final ultrasonic cleaning in freon TF (Du Pont). By completing all these phases very carefully the amount of carbon contamination was not higher than is usually found on similar metal surfaces exposed to the laboratory atmosphere. But in spite of such preparatory procedures the dark spots on contact areas after the life-time tests of reed switches with electrical loading, were again found. Another possible origin of this carbon contamination could be the N2-H 2 (10% H2) gas mixture with which the reed switches are filled. However, analysis by gas chromatography, revealed no impurity gases. Further, we supposed that impurity gases could be released during the sealing of the contact reeds in glass tubes in the presence of N2-H 2 gas mixture. To prove this hypothesis a special device was constructed (Figure 1) enabling high frequency induction sealing of the kovar wire to glass in purified hydrogen. During this sealing, released impurity gases (approximately 25/11) consisted mainly of water (65 vol%), CO (25 vol%), CH 4 (7 vol%) and of minor quantities of CO 2 and CzH n (n = 2 6). These gases derive from dissolved and adsorbed impurities of both partners being released at high temperature during the sealing process or formed in reactions with hydrogen or reactions between glass and the kovar surface. N 2 and 0 2 were also present in small amounts but were analyzed only qualitatively because they cannot be the origin of carbon contamination. By heating only kovar wire previously heated in wet hydrogen (dew point +20°C, at 900°C, 2 h), the amount of released or formed impurity gases was approximately 21/~1. They consisted of water 51.3 vol%), CO (42 vol%) and C H 4 (6.7 vol%).
simulate the first technique, where red-hot surfaces of metal and glass come into contact; the origin of impurity gases are both the kovar and the glass. In both experiments, i.e. by heating of kovar alone and by sealing of glass to kovar in the presence of hydrogen, the released or formed gas impurities contained about 7 vol% of methane. We suspect that the carbon which appears on the contacts is being formed by thermal decomposition of methane during electrical loading4. From the AES sputter depth profiles (Figure 4) and the area covered with carbon we could estimate that the quantity of carbon deposited on the contact was approximately 2 x 10- log. If all deposited carbon on both contacts originated from methane, it would correspond to 5.6 × 10 -4 pl of methane in the reed switch atmosphere. On the basis of the analyses of released gases, we can conclude that the concentration of methane in the reed switch atmosphere is high enough to form the described contamination spots. When sealing reed switches, the atmosphere does not always have equal reducing properties, because in the sealing machines the protective gas could be partly mixed with air. In some series of reed switches, contacts were not contaminated with carbon and we conclude that methane was then not present due to its oxidation.
5. Conclusions (1) After life-time tests of reed switches at low load conditions, the AES analyses of the contacts showed contamination spots consisting mostly of carbon or carbonaceous compounds. (2)From the carbon C (272 eV) Auger images and AES sputter depth profiles it was established that, on the contacts with high contact resistance, the carbon contamination spots were localized on smaller areas and were of higher carbon concentration and thickness compared with the low resistance contacts. (3) The main source of carbon on our reed switch contacts is most probably impurity gases originating from the glass at the sealing of contact reeds in the glass envelope in the presence ofN2-H z gas mixture. They remain partly trapped in reed switches and are thermally decomposed to carbon which deposits on contacts when they operate under electrical load.
Acknowledgement Discussions with Dr E Kansky, Institut za Elektroniko in Vakuumsko Tehniko, Ljubljana, are gratefully acknowledged.
4. Discussion References Based on the results of above-mentioned experiments, we can expect that carbon containing impurity gases release or form during the sealing of contact reeds into glass tubes where they remain trapped. For reed switch production two techniques are used: either the contact reeds are sealed directly into glass tubes in the presence of N2-H 2 gas mixture, or the contact reeds are pre-glazed in air and then sealed in glass tubes in the presence of N z - H z gas mixture. The experiments we have made (by heating kovar and glass)
1 S Ogren, IEEE Trans Components, Hybrids Manuf Technol, CHMT-3, 431 (1980). 2 M Murko-Jezov~ek,Bilten JUVAK, 16, 81 (1976); M Murko-Jezovgek, Thin Solid Films, 32, 366 (1976); M Murko-Jezov~ek, F Brecelj and B Jenko, Proc 7th Int Vac Conor and 3rd Int Conf Solid Surfaces, p 2343, Vienna (1977). 3 M Murko-Jezov~ek and B Jenko, Bilten JUVAK, 17, 443 (1979); M Murko-Jezov~ek and B Jenko, Proc 9th lnt Vac Conor and 5th Int Conf Solid Surfaces, Extended Abstracts, p 174, Madrid (1983). 4 E W Gray, IEEE Trans Parts, Hybrids Paekagin#, PHP-11, 121 (1975).
167