R 718 Philips Res. Repts 25, 87-94,1970 APPLICATION OF A CHANNEL PLATE IN FIELD-ION MICROSCOPY by A. van OOSTROM Abstract Further progress has been made in applying a channel-plate image intensifier in field-ion microscopy. In particular the usefulness of argon as the image gas is demonstrated. 1. Introduetion It has been shown by many investigators during the last decade that the field-ion microscope is a unique instrument for the study of surfaces on an atomic scale 1). The phenomenon of field evaporation, permitting the inspection of a small volume of the original specimen just below the surface, is a feature that has made the instrument particularly attractive for metallurgical applications. Unfortunately, the low brightness of the image is a severe limitation of the field-ion microscope. This is true even if a gas with a low atomic number, e.g. helium is used for imaging at a pressure of 10-3 Torr. Since helium gas has a high ionization potential (24 5 ev), the electric field at the surface needed for ionization is correspondingly large. At the best-image field of helium most materials will field-evaporate, which limits the application of field-ion microscopy considerably. Recently the use of a channelled image intensifier in the field-ion microscope was described 2). We have since then succeeded in operating the microscope at much lower pressures (10-6 Torr) and with a number of other gases, e.g. neon, argon, xenon and methane. These gases have much lower ionization potentials than helium and their best-image fields are therefore also much lower. In this paper we shalllimit ourselves to argon, which seems to be one. of the most promising gases available. 2. Experimental procedure The field-ion microscope equipped with channel plate is shown schematically in fig. 1. Proximity focussing was used as it was found to be much simpler than magnetic focussing of the electron image as applied with converter mesh-grid image intensifiers. The plate-to-screen distance was chosen at 1 mm. The tip-to-plate distance was varied between 15 and 50 mm, but 20 mm was found to be adequate in most cases.
88 A. van OOSTROM Liquid nitrogen +r-to pump V, =10kV V 2 =OV V3 =lkv ti =5kV Channel plate Fluorescent screen Tin-oxide layer Fig. 1. Schematic diagram of a field-ion microscope with a channel plate, using proximity focussing. Some typical operating voltages are: VI = 10 kv, V 2 = 0 V, V 3 = I kv, V 4 = 5 kv. The plate, obtained from Mullard Research Laboratories, Salfords, England, consists of an array of 40!Lmdiameter channels. lts thickness is about 2 0 mm, its useful diameter 27 mm and the open area 62 %. Further details regarding the performance of these plates have been published by Guest, Holmshaw and Manley 3). The post-acceleration voltage between plate and screen is limited by field-electron emission from the plate-support ring. In normal operation up to 4 kv could be applied between plate and screen. The channel plate itself has been operated successfully up to 1350volts. A brightness gain of about 105 was observed at this plate voltage using He+ions of about 10 kev. The choice ofthe phosphor is not critical as compared with an ordinary field-ion microscope, where ions directly impinge on the screen. Both willemite and ZnCdS have been used successfully. The phosphor was deposited on a conductive tin-oxide layer directly upon the glass wall of the vacuum chamber. The specimen was cooled with liquid nitrogen. A glass uhv system pumped with mercury-diffusion pumps was used for these experiments. Gas could be admitted to the field-ion microscope in two different ways. Various gases could be admitted from a bakeable manifold with Granville-Phillips metal valves via a Vacuum Generators leak valve, while helium could also be admitted through the walls of a heated quartz
APPLICATION OF A CHANNEL PLATE IN FffiLD-ION MrCROSCOPY 89 tube. The latter method permitted the production of a controlled mixture of helium with any other gas. Since this was done with the pumps operating and no valve shut between pump and microscope, the impurity level in the helium could be kept Iow. Field-ion microscopy using other gases than helium or neon for imaging requires ultra-high-vacuum techniques. This is necessary because at the lower fields needed for ionizing gases like argon or methane, impurities present in these gases can adsorb on the surface during imaging. In the present case the residual gas pressure in the field-ion microscope was below 10-10 Torr and 10-8 Torr in the gas-inlet system. Fortunately, the channel plate permits the field-ion microscope to operate at pressures below 10-4 Torr. Therefore, it is sufficient to reduce the impurity level to about 1 ppm. In the case of argon this was achieved by purifying the image gas over a titanium getter at a pressure of 1 Torr in the manifold. Another serious effect encountered was gas desorption from the channel plate during operation,.' which may contaminate the specimen when image gases with a low ionization potential are used. However, this gas production can be considerably reduced by baking the plate at 300 oe and subsequent electron bombardment. It was found that after several bake-out cycles at 300 oe the field-ion microscope could be operated in argon without any visible signs in the field-ion pattern of contamination of the imaged surface area over periods as long as half an hour. 3. Experimental results Although the available brightness gain of the order of 10 5 can be used to obtain images suitable for cine recording, it is also feasible to operate the microscope at a much lower pressure level. There are two advantages of using the available gain in this way. Firstly, operating the field-ion microscope at a reduced pressure will correspondingly reduce the electron current in the specimen and therefore will make the imaging of a less conductive material possible. Secondly, the influence of image-gas adsorption will be reduced. Recently, it has been observed in the atom probe that helium gas is desorbed in a field-evaporating pulse, indicating that helium is adsorbed on a tungsten specimen at 5.10-4 Torr in the high field needed for imaging at 78 "K 4.5). At present it is not clear to what extent this observation will influence the interpretation of previous results on e.g. binding-energy determinations of single metal atoms on a field-ion-microscope tip. However, it is clear that operating the field-ion microscope at a reduced pressure will lower the amount of helium,adsorbed... Figure 2 shows the field-ion pattern of a tungsten specimen at 78 "K imaged in helium at a pressure of 2.10-6 Torr. At such a pressure about 100 He+ions/s are formed above each surface atom imaged. Statistical fluctuations be-
90 A. van OOSTROM Fig. 2. Clean-tungsten pattern. Image gas: helium: pressure: 2.10-6 Torr; best-image voltage: 10 4 kv. come important at this level and therefore this pressure sets the lower limit for visual observation. Figure 2 was obtained after field evaporation at 11 5 kv. The best-image voltage was 10 4 kv and 1 30 kv the channel-plate voltage. A unique feature of the channel plate is that it makes it possible to examine the field evaporation in the presence of the image gas over the pressure range 10-6 to 10-3 Torr. For a tungsten tip with a radius of curvature of about 500 A the field-evaporation voltage decreased by 1 % for every order of magnitude increase in pressure. From the electron-emission data in ultra-high vacuum we determined the geometrical factor and arrived in this way at a field strength for field evaporation in 10-4 Torr helium of 5 5 v/a at 78 OK. The final and probably most important aspect of the channel-plate image intensifier is the possiblity of using other image gases which produce patterns
APPLICATION OF A CHANNEL PLATE IN FffiLD-ION MICROSCOPY 91 at considerably reduced electric-field strengths compared with the field needed to ionize helium at the surface of a specimen. The lower electric fields permit the study of other materials which would field-evaporate in helium. Argon has turned out to be one of the most promising gases so far. The best image field is below 2 0 v/a, which makes it suitable for imaging many materials, for studying gas adsorption and biological applications. The brightness of the argon-ion pattterns was found to be lower than that of the corresponding helium ones. This is probably due to a less efficient ion-to-electron conversion at the entrance of the channel plate. Also a blurring effect occurred at about 10-3 Torr, mainly visible in the central part of the picture..at lower pressures the blurring disappeared and the resolution improved considerably. Presumably, a large amount of argon is adsorbed on the tungsten surface kept at 78 "K and at 10-3 Torr. It is likely that the larger size of the argon atom in comparison with helium is at least partly responsible for the blurring effect. It is an interesting,observation that the channel plate permits gas-surface interactions to be investigated, while the gas under investigation is used as the image gas. In a particular experiment a clean tungsten tip at 78 "K was operated in the electron-emission mode in an argon p'~essure of 5.10-6 Torr. In all, 7.10 15 argon ions hit the specimen during this bombardment, The applied potential difference in this case was 1400 volts. Numet1cal calculations of the ion bombardment of field-emission tips due to gas ionization indicate that the maximum of the ion energy for a tip in its field-evaporated end form is about 0 03 of the highest possible energy 6). This means that in our case the maximum in the ion-energy distribution would occur at about 42<'eV. The damage caused was visualized by increasing the.pressure to 5.10-5 Torr and reversing the applied voltage. Figure 3 shows the field-ion pattern at 5 4 kv and fig. 4 at 6 5 kv. Many large spots visible in fig. 3 are considerably smaller in fig. 4 or have even disappeared. However, both patterns could be obtained with increasing and decreasing voltage. Field evaporation at 16 0 kv removed the bright spots of fig. 3, but did not remove the damage in the substrate. Since the ratio of this field-evaporation voltage to the best-image voltage of fig. 3 equals the ratio of the field-evaporation field for tungsten to the best-image field in argon, it is evident that the bright spots in fig. 3 are tungsten atoms. It is therefore likely that the bright spots in fig. 3 correspond to protrusions consisting of tungsten atoms on a damaged tungsten substrate. This explanation is supported by the fact that the spot size in fig. 3 is considerably larger than in fig. 4, indicating a larger magnification. Overlapping of spots was also sometimes observed. From the best-image voltages a field-enhancement factor of about 1 2 is found for these protrusions. Field evaporation above 16 0 kvalso removed the damaged substrate layers and produced again the perfect field-evaporated end form.
92 A. van OOSTROM Fig. 3. Tungsten pattern after argon-ion bombardment. Image gas: argon; pressure: 5.10-5 Torr; best-image voltage: 5 4 kv. Fig. 4. Tungsten pattern after argon-ion bombardment. Image gas: argon; pressure: 5.10-5 Torr; best-image voltage: 6 5 kv.
APPLICATION OF A CHANNEL PLATE IN FIELD-ION MICROSCOPY 93 Fig. 5. Clean-tungsten pattern. Image gas: argon; pressure: 10-4 Torr; best-image voltage: 4 3 kv. Contamination may seriously affect results obtained with argon-ion microscopy. However, we succeeded in reducing the impurity level in the argon during operation to a very low level. The tungsten pattern shown in fig. 5 was imaged in argon at about 10-4 Torr after field evaporation at 10 7 kv. The best-image voltage was 4 3 kv. This pattern could be kept clean for about 30 minutes when one or two extra spots appeared. Tip temperature was again 78 ok. Finally, we measured the best-image fields for helium, neon, argon and methane. As given above, the field-evaporation field strength for a tungsten specimen at 78 "K in 10-4 Torr of helium was found to be 5 5 v/a. If field evaporation occurs at the same field strength in other image gases at the same pressure, we arrive at the following values for the best-image field at 10-4 Torr:
94 A. van OOSTROM helium: neon: argon: methane: 4 5 v/a, 3 4 v/a, 1 9 v/a, 1 3 v/a. The observed values are considerably below the ones expected from a simple 13/2 rule, where I represents the value of the ionization potentialof the gas. 4. Conclusions The excellent performance of the channel-plate image intensifier is shown for helium- and argon-ion microscopy. The present experiments demonstrate that field-ion microscopy can be carried out at much lower pressures as well as with argon. In particular adsorption phenomena can now be studied on an atomic scale. The recent development of the atom pro be and the availability of channel plates would seem to herald a breakthrough in field-ion microscopy. Eindhoven, December 1969 REFERENCES 1) E. W. Müller 'and T. T. Tsong, Field ion microscopy, American Elsevier Publishing Company, Inc.; New York 1969. 2) P. J. Turner, P. Cartwright, M. J. Southon, A. van Oostrom and B. W. Manley, J.S.I. (J. of Phys, E) 1969, Series 2, Vol. 2, 731. 3) A. Guest, R. T. Holmshaw and B. W. Manley, Mullardtechn. Comm. 97,210,1969. 4) E. W. Müller, S. B. McLane and J. A. Panitz, Surface Science 17,430, 1969. 5) J. T. McKinney and S. S. Brenner, 16th Field Emission Symposium, Pittsburgh, 1969. 6) H. Vernickel and H. Welter, 16th Field Emission Symposium, Pittsburgh,1969.