- Email: [email protected]
A mini-ALE attachment to UHV surface analysis equipment
applied surface science ELSEVIER
Applied Surface Science 107 (1996) 255-259
A mini-ALE attachment to UHV surface analysis equipment R.G. van Welzenis *, R.A.M. Bink, H.H. Brongersma Faculty of Physics, Eindhoven University of Technology, P.O. Box 513, NL-5600 MB Eindhoven, The Netherlands
Received 12 October 1995; accepted 23 December 1995
Abstract An atomic layer epitaxy (ALE) module has been developed that can be attached to a port of UHV surface analysis equipment. In our case this would, for instance, enable analysis by low energy ion scattering (LEIS), X-ray photoelectron spectroscopy (XPS), and atomic force microscopy (AFM). This so-called mini-ALE will be used to study and optimize the growth mechanisms of mono-atomic and snb-monatomic metal layers and to prepare model samples for studies of catalysts and small metal clusters. The mini-ALE basically consists of a small ( = 1 cm 3) growth chamber suspended in an UHV environment, with valved inlets for the reaction and purge gases. The sample can be transferred to the analysis chamber via a load-lock. The surface temperature of the sample can be controlled from room temperature to approximately 500°C. The system was tested by attaching it to one of our LEIS set-ups and growing CuO on A1203, using Cu(acac) 2 and artificial air as reactants. Cu growth was observed, covering about 3% of the surface. Resuks on growth as a function of surface temperature are given.
1. Introduction W e were interested in the A L E technique for the production of sub-mono-atomic ' l a y e r s ' of metals on ceramic substrates and small metal clusters on oxides. Previously low energy ion scattering (LEIS) measurements had been done in the N O D U S set-up [1] in Eindhoven on N i / A 1 2 0 3 produced by A L E at Microchemistry in Finland [2]. The combination of A L E and LEIS seems ideal, because A L E can produce the desired surface modifications [3], while LEIS is capable of quantizing the atomic composi-
* Corresponding author. Tel.: +31-40-2474157; fax: +31-402453587; e-mail: [email protected].
tion on the surface of highly dispersed insulating substrates. There was one disadvantage in the procedure, however. During transport from Finland to the Netherlands the sample surfaces were contaminated, which complicates the LEIS measurements that have to be done in UHV. To overcome this problem it was decided to build our own small A L E machine as an attachment to the surface analysis equipment. This attachment needed to meet the following requirements: • it should be easily attachable to the input ports of the load-locks of surface analysis set-ups, • sample transfer between the surface analysis equipment and the growth chamber should be possible without exposing the sample to a contaminating environment,
0169-4332/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. Pll S0169-4332(96)00496-5
256
R.G. van Welzenis et al./Applied Surface Science 107 (1996) 255-259
it should be as compact as possible, • it is aimed at research purposes and not at production, exchange of reactants should be simple, the surface temperature of the substrate should be measurable and controllable during growth. The design of the mini-ALE according to these requirements is depicted in Fig. l, while Fig. 2 is a photo of the attachment. The main components of the system are: two container holders (2), the main body (5), the exchange chamber, the retractable wobble stick (6), and the piston sample holder (9). The container holders have a bore in which a small container with the metal organic compound can be placed after unscrewing the top. These holders are suspended
from the main body by thin Ti tubes. Four additional Ti tubes are connected to an outside gas supply system. The bottom of the main body is conically shaped and houses a rotatable ceramic disc with strategically placed holes that serves as a valve for the reactant and flush gases. It should be stressed that this system operates differently from most known ALE systems. Usually, the reactants are transported through the growth chamber by means of a carrier gas (e.g. N2), but in this case the metal-organic compound is simply valved into the reaction chamber, driven by its own vapor pressure in the container. The main body, the container holders, and the piston sample holder are each equipped with a heater controlled by a thermocouple. They are thermally
(9
®
lm
I I
Fig. 1. Drawing of the mini ALE attachment. 1. removable lid; 2. containerholder; 3. water cooling; 4. UHV valves (V1 and V2); 5. main body; 6. retractable wobble stick; 7. rotatable linear motion transport rod; 8. substrate; 9. piston sample holder; 10. heaters; 11. ceramic valve; 12. growth chamber; 13. clamp spring with thermocouple.
R.G. van Welzenis et al, /Applied Surface Science 107 (1996) 255-259
257
opened and the piston is brought into the growth position (fully upward). After setting and stabilizing the temperatures, the growth chamber is flushed with N 2 gas. Next the Cu(acac) 2 gas is admitted from the small container during 7.5 min followed by a N 2 flush. The temperature of the container was kept at 70°C, which corresponds to a local Cu(acac) 2 pressure of 3.6 × 10 - 3 mbar. The last step, removal of the ligands by oxidation, can be done in two ways. Either artificial air is admitted at 500°C in the growth chamber before retracting the piston and opening the growth chamber, or the substrate is transported to the preparation chamber of the NODUS and oxidized with pure 02 at 400°C.
3. Performance
Fig. 2. Photograph of the ALE attachment. The linear motion rod has been removed as well as the pumping connection at the rear. One clearly sees the substrate suspended in the gripper above the piston. The growth chamber is situated just above the horizontal UHV valve. The piston sample holder can be moved up and down, thanks to the extended bellows (© TUE, Stafgroep Reproductie en Fotografie).
insulated by ceramic supports and the vacuum in the chamber. The system is pumped by a small turbo molecular pump. The conical top of the piston sample holder fits precisely into the bottom of the main body, thus closing off the growth chamber.
The system was tested for vacuum tightness and temperature control. It turned out that the piston-cone seal was not perfect. At a flow of 2 c m 3 / m i n N 2 gas at a pressure of 20 mbar, the vacuum outside the growth chamber, with the piston in the up (closed) condition, could just be maintained at 10 .2 mbar. Without gas flow in the growth chamber this pressure easily comes down below 10 .6 mbar. A simple change in the design of the piston top is expected to solve this problem. Of course the temperatures of the main body and the piston can not be controlled completely independently when they are in contact, but that is not a serious drawback. The temperature of the containers holding the organic compound reactant can be controlled independently to some extend, because of the thermal 'insulation' provided by the Ti connection tubes. In fact the system operated essentially correct almost immediately.
4. Test 2. Operating procedures After moving the sample in the center of the exchange chamber and closing V1, the sample is put on top of the piston sample holder using the retractable wobble stick. Next the UHV valve V2 is
The ALE performance by growing Cu on A1203, a simple system and it previous experiences with used as the metal organic flushing.
of the system was tested because this seemed to be also correlated with our Ni/A1203. Cu(acac) 2 was compound and N 2 gas for
258
R.G. van Welzenis et al. /Applied Surface Science 107 (1996) 255-259
Small pure A1 disks were cleaned and mounted in the NODUS set-up. After sputtering with 3 keV Ar + with a total dose 1 × 1017 i o n s / c m 2, the LEIS spectrum showed a clean surface with only an aluminum peak. The L E I S spectra are obtained with 3 keV 4He+ ions and a scattering angle of 142 °. Next the substrate was transferred to the preparation chamber of the NODUS and calcined at 500°C for 30 min. Although this procedure does not guarantee the A1203 stoichiometry, it is close enough for these preliminary experiments. Then either another LEIS spectrum was taken or the substrate was transferred directly to the ALE attachment to undergo an ALE cycle. Thereafter, the substrate was returned to the NODUS again and another (initial) LEIS spectrum is taken. The ion dose during an initial measurement is 1015 i o n s / c m 2, which means that roughly 1 / 4 of the Cu atoms in the outermost atomic layer of the surface will be sputtered away during analysis. The
500 ~ . - - - 0 AI . "Cu
~-.~o
400
-~" -=e
300
_~ 200
100
i
060
30
Cu
f 20 e
d
i
I
i
I
180 220 Temperature (C)
|
I
260
i
I
300
rain).
c
10
b
.
15100
I
140
results of a typical set of initial measurements are shown in Fig. 3. One can clearly see that Cu has been deposited. From the peak area it is estimated that the Cu surface coverage is approximately 3 at%. This appears to be the saturation coverage after one ALE cycle, independent of the reaction time between 1 and 20 min (standard reaction time used is 7.5
_J-x
1O00
i
Fig. 4. The areas of the Cu, A1 and O peaks of the initial LEIS spectra as a function of the substrate temperature during growth. The lines are drawn to connect the points.
AI
0
0500
I
100
2000
_
,
2500
z
3000
Energy (eV)
Fig. 3. 3 keV 4He+ LEIS spectra (0 = 142 °) at several substrate temperatures: (a) T = 80°C; (b) T = 125°C; (c) T = 150°C; (d) T = 175°C; (e) T = 200°C; (f) T = 250°C; (g) T = 330°C; curve z is for the substrate after sputtering and calcination at 500°C. For clarity, the spectra were shifted vertically with respect to one another. At high energies the scattered ion intensity is zero for all spectra.
It must be stressed that with LEIS one only observes the outermost atomic layer, therefore these results should not be interpreted as a 0.02 Cu monolayer coverage. When subsequent LEIS experiments are done the Cu signal only slightly decreases. This suggests that each spot is made up of several layers of Cu atoms. After a series of 14 LEIS runs, which means that the equivalent of 3 to 4 full monolayers of Cu should have been removed, the Cu signal has decreased by a factor of 5, leaving a Cu coverage of still approximately 0.4 at%. On two samples (250°C and 380°C growth temperature) the Cu areal density was measured with RBS. Although the signals were rather weak, for both samples an average areal density of approxi-
R.G. van Welzenis et al. / A p p l i e d Surface Science 107 (1996) 255-259
mately 1014 a t / c m 2 could be determined. The surface areal density of Cu as derived from the LEIS measurements is about 3 × 1013 surface a t / c m 2. The factor 3 difference can be understood if one realizes that LEIS will not detect Cu atoms that are buried in the deeper pores of the A1203, whereas during growth the Cu(acac) 2 will penetrate the pores. A plot of the measured Cu, A1 and O peak areas as a function of the substrate temperature is given in Fig. 4. The Cu peak area is almost constant over the whole temperature range, which means that the A L E window must be wider than the investigated temperature range. It is believed that A L E growth is observed, the most probable reaction being the exchange of acac between Cu and c.u.s. AI 3+ sites [4]. In conclusion it appears that the m i n i - A L E attachment works and allows one to directly study the growth processes in detail.
259
Acknowledgements Without the expert support of the staff of our central technical workshops, in particular Meindert Jansen, Theo Maas and Peer Brinkgreve we never would have succeeded. W e thank L e o van IJzend o o m for performing the RBS measurements.
References [1] H.H. Brongersma, N. Hazewindus, J.M. van Nieuwland, A.M.M. Otten and A.J. Smets, Rev. Sci. Instr. 49 (1978) 707. [2] J.-P. Jacobs, L.P. Lindfors, J.G. Reintjes, O. Jylh~iand H.H. Brongersma. Catal. Lett. 25 (1994) 315. [3] E.-L. Laakomaa, Abstracts ACSI-2, Joensuu, Finland (1993) p. 58. [4] J.A.R. van Veen, G. Jonkers and W.H. Hesselink, J. Chem. Soc. Faraday Trans. I, 85 (1989) 389.
Applied Surface Science 107 (1996) 255-259
A mini-ALE attachment to UHV surface analysis equipment R.G. van Welzenis *, R.A.M. Bink, H.H. Brongersma Faculty of Physics, Eindhoven University of Technology, P.O. Box 513, NL-5600 MB Eindhoven, The Netherlands
Received 12 October 1995; accepted 23 December 1995
Abstract An atomic layer epitaxy (ALE) module has been developed that can be attached to a port of UHV surface analysis equipment. In our case this would, for instance, enable analysis by low energy ion scattering (LEIS), X-ray photoelectron spectroscopy (XPS), and atomic force microscopy (AFM). This so-called mini-ALE will be used to study and optimize the growth mechanisms of mono-atomic and snb-monatomic metal layers and to prepare model samples for studies of catalysts and small metal clusters. The mini-ALE basically consists of a small ( = 1 cm 3) growth chamber suspended in an UHV environment, with valved inlets for the reaction and purge gases. The sample can be transferred to the analysis chamber via a load-lock. The surface temperature of the sample can be controlled from room temperature to approximately 500°C. The system was tested by attaching it to one of our LEIS set-ups and growing CuO on A1203, using Cu(acac) 2 and artificial air as reactants. Cu growth was observed, covering about 3% of the surface. Resuks on growth as a function of surface temperature are given.
1. Introduction W e were interested in the A L E technique for the production of sub-mono-atomic ' l a y e r s ' of metals on ceramic substrates and small metal clusters on oxides. Previously low energy ion scattering (LEIS) measurements had been done in the N O D U S set-up [1] in Eindhoven on N i / A 1 2 0 3 produced by A L E at Microchemistry in Finland [2]. The combination of A L E and LEIS seems ideal, because A L E can produce the desired surface modifications [3], while LEIS is capable of quantizing the atomic composi-
* Corresponding author. Tel.: +31-40-2474157; fax: +31-402453587; e-mail: [email protected].
tion on the surface of highly dispersed insulating substrates. There was one disadvantage in the procedure, however. During transport from Finland to the Netherlands the sample surfaces were contaminated, which complicates the LEIS measurements that have to be done in UHV. To overcome this problem it was decided to build our own small A L E machine as an attachment to the surface analysis equipment. This attachment needed to meet the following requirements: • it should be easily attachable to the input ports of the load-locks of surface analysis set-ups, • sample transfer between the surface analysis equipment and the growth chamber should be possible without exposing the sample to a contaminating environment,
0169-4332/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved. Pll S0169-4332(96)00496-5
256
R.G. van Welzenis et al./Applied Surface Science 107 (1996) 255-259
it should be as compact as possible, • it is aimed at research purposes and not at production, exchange of reactants should be simple, the surface temperature of the substrate should be measurable and controllable during growth. The design of the mini-ALE according to these requirements is depicted in Fig. l, while Fig. 2 is a photo of the attachment. The main components of the system are: two container holders (2), the main body (5), the exchange chamber, the retractable wobble stick (6), and the piston sample holder (9). The container holders have a bore in which a small container with the metal organic compound can be placed after unscrewing the top. These holders are suspended
from the main body by thin Ti tubes. Four additional Ti tubes are connected to an outside gas supply system. The bottom of the main body is conically shaped and houses a rotatable ceramic disc with strategically placed holes that serves as a valve for the reactant and flush gases. It should be stressed that this system operates differently from most known ALE systems. Usually, the reactants are transported through the growth chamber by means of a carrier gas (e.g. N2), but in this case the metal-organic compound is simply valved into the reaction chamber, driven by its own vapor pressure in the container. The main body, the container holders, and the piston sample holder are each equipped with a heater controlled by a thermocouple. They are thermally
(9
®
lm
I I
Fig. 1. Drawing of the mini ALE attachment. 1. removable lid; 2. containerholder; 3. water cooling; 4. UHV valves (V1 and V2); 5. main body; 6. retractable wobble stick; 7. rotatable linear motion transport rod; 8. substrate; 9. piston sample holder; 10. heaters; 11. ceramic valve; 12. growth chamber; 13. clamp spring with thermocouple.
R.G. van Welzenis et al, /Applied Surface Science 107 (1996) 255-259
257
opened and the piston is brought into the growth position (fully upward). After setting and stabilizing the temperatures, the growth chamber is flushed with N 2 gas. Next the Cu(acac) 2 gas is admitted from the small container during 7.5 min followed by a N 2 flush. The temperature of the container was kept at 70°C, which corresponds to a local Cu(acac) 2 pressure of 3.6 × 10 - 3 mbar. The last step, removal of the ligands by oxidation, can be done in two ways. Either artificial air is admitted at 500°C in the growth chamber before retracting the piston and opening the growth chamber, or the substrate is transported to the preparation chamber of the NODUS and oxidized with pure 02 at 400°C.
3. Performance
Fig. 2. Photograph of the ALE attachment. The linear motion rod has been removed as well as the pumping connection at the rear. One clearly sees the substrate suspended in the gripper above the piston. The growth chamber is situated just above the horizontal UHV valve. The piston sample holder can be moved up and down, thanks to the extended bellows (© TUE, Stafgroep Reproductie en Fotografie).
insulated by ceramic supports and the vacuum in the chamber. The system is pumped by a small turbo molecular pump. The conical top of the piston sample holder fits precisely into the bottom of the main body, thus closing off the growth chamber.
The system was tested for vacuum tightness and temperature control. It turned out that the piston-cone seal was not perfect. At a flow of 2 c m 3 / m i n N 2 gas at a pressure of 20 mbar, the vacuum outside the growth chamber, with the piston in the up (closed) condition, could just be maintained at 10 .2 mbar. Without gas flow in the growth chamber this pressure easily comes down below 10 .6 mbar. A simple change in the design of the piston top is expected to solve this problem. Of course the temperatures of the main body and the piston can not be controlled completely independently when they are in contact, but that is not a serious drawback. The temperature of the containers holding the organic compound reactant can be controlled independently to some extend, because of the thermal 'insulation' provided by the Ti connection tubes. In fact the system operated essentially correct almost immediately.
4. Test 2. Operating procedures After moving the sample in the center of the exchange chamber and closing V1, the sample is put on top of the piston sample holder using the retractable wobble stick. Next the UHV valve V2 is
The ALE performance by growing Cu on A1203, a simple system and it previous experiences with used as the metal organic flushing.
of the system was tested because this seemed to be also correlated with our Ni/A1203. Cu(acac) 2 was compound and N 2 gas for
258
R.G. van Welzenis et al. /Applied Surface Science 107 (1996) 255-259
Small pure A1 disks were cleaned and mounted in the NODUS set-up. After sputtering with 3 keV Ar + with a total dose 1 × 1017 i o n s / c m 2, the LEIS spectrum showed a clean surface with only an aluminum peak. The L E I S spectra are obtained with 3 keV 4He+ ions and a scattering angle of 142 °. Next the substrate was transferred to the preparation chamber of the NODUS and calcined at 500°C for 30 min. Although this procedure does not guarantee the A1203 stoichiometry, it is close enough for these preliminary experiments. Then either another LEIS spectrum was taken or the substrate was transferred directly to the ALE attachment to undergo an ALE cycle. Thereafter, the substrate was returned to the NODUS again and another (initial) LEIS spectrum is taken. The ion dose during an initial measurement is 1015 i o n s / c m 2, which means that roughly 1 / 4 of the Cu atoms in the outermost atomic layer of the surface will be sputtered away during analysis. The
500 ~ . - - - 0 AI . "Cu
~-.~o
400
-~" -=e
300
_~ 200
100
i
060
30
Cu
f 20 e
d
i
I
i
I
180 220 Temperature (C)
|
I
260
i
I
300
rain).
c
10
b
.
15100
I
140
results of a typical set of initial measurements are shown in Fig. 3. One can clearly see that Cu has been deposited. From the peak area it is estimated that the Cu surface coverage is approximately 3 at%. This appears to be the saturation coverage after one ALE cycle, independent of the reaction time between 1 and 20 min (standard reaction time used is 7.5
_J-x
1O00
i
Fig. 4. The areas of the Cu, A1 and O peaks of the initial LEIS spectra as a function of the substrate temperature during growth. The lines are drawn to connect the points.
AI
0
0500
I
100
2000
_
,
2500
z
3000
Energy (eV)
Fig. 3. 3 keV 4He+ LEIS spectra (0 = 142 °) at several substrate temperatures: (a) T = 80°C; (b) T = 125°C; (c) T = 150°C; (d) T = 175°C; (e) T = 200°C; (f) T = 250°C; (g) T = 330°C; curve z is for the substrate after sputtering and calcination at 500°C. For clarity, the spectra were shifted vertically with respect to one another. At high energies the scattered ion intensity is zero for all spectra.
It must be stressed that with LEIS one only observes the outermost atomic layer, therefore these results should not be interpreted as a 0.02 Cu monolayer coverage. When subsequent LEIS experiments are done the Cu signal only slightly decreases. This suggests that each spot is made up of several layers of Cu atoms. After a series of 14 LEIS runs, which means that the equivalent of 3 to 4 full monolayers of Cu should have been removed, the Cu signal has decreased by a factor of 5, leaving a Cu coverage of still approximately 0.4 at%. On two samples (250°C and 380°C growth temperature) the Cu areal density was measured with RBS. Although the signals were rather weak, for both samples an average areal density of approxi-
R.G. van Welzenis et al. / A p p l i e d Surface Science 107 (1996) 255-259
mately 1014 a t / c m 2 could be determined. The surface areal density of Cu as derived from the LEIS measurements is about 3 × 1013 surface a t / c m 2. The factor 3 difference can be understood if one realizes that LEIS will not detect Cu atoms that are buried in the deeper pores of the A1203, whereas during growth the Cu(acac) 2 will penetrate the pores. A plot of the measured Cu, A1 and O peak areas as a function of the substrate temperature is given in Fig. 4. The Cu peak area is almost constant over the whole temperature range, which means that the A L E window must be wider than the investigated temperature range. It is believed that A L E growth is observed, the most probable reaction being the exchange of acac between Cu and c.u.s. AI 3+ sites [4]. In conclusion it appears that the m i n i - A L E attachment works and allows one to directly study the growth processes in detail.
259
Acknowledgements Without the expert support of the staff of our central technical workshops, in particular Meindert Jansen, Theo Maas and Peer Brinkgreve we never would have succeeded. W e thank L e o van IJzend o o m for performing the RBS measurements.
References [1] H.H. Brongersma, N. Hazewindus, J.M. van Nieuwland, A.M.M. Otten and A.J. Smets, Rev. Sci. Instr. 49 (1978) 707. [2] J.-P. Jacobs, L.P. Lindfors, J.G. Reintjes, O. Jylh~iand H.H. Brongersma. Catal. Lett. 25 (1994) 315. [3] E.-L. Laakomaa, Abstracts ACSI-2, Joensuu, Finland (1993) p. 58. [4] J.A.R. van Veen, G. Jonkers and W.H. Hesselink, J. Chem. Soc. Faraday Trans. I, 85 (1989) 389.