Surface Science 50 (1975) 515-526 0 North-Holland Publishing Company
J.M. DICKEY Queens Collfge of The City University of New York, Flushing, New York 11367, U.S.A. Received 10 September 1974; revised manuscript received 28 January 1975
The interaction of nitrogen with niobium was investigated using LEED and Auger spectroscopy. The rate of adsorption of nitrogen was measured at different temperatures and an ordered 5 X 5 diffraction pattern was observed.
1. Introduction Niobium is an important metal since it and its compounds have amongst the highest occurring superconducting transition temperatLlres, in particular NbN has a T, of 17.3 R. Thin layers of the high T, compounds Nb,Sn [1,2] and Nb,Al  on Nb have been produced by depositing a layer of metal on the Nb and heating to an appropriate temperature, so it is of interest to determine whether a thin skin of NbN  could be produced by heating the Nb in N2. An extensive study of the surface properties of Nb has been made since for many superconducting properties the state of the surface is crucial. In earlier papers , hereinafter referred to as I, the emphasis was on the interaction with oxygen, while in this paper we report the details of a study concerning nitrogen, using Auger and LEED techniques. Other LEED studies of Nb, which have previously been reported, were concerned with the clean surface [5,6] and also the interaction with the gases CO , 0, , and H,
The bulk interaction between N and Nb has been extensively studied [9-141 and the equilibrium phase diagram shows the existence of several phases of different composition. At high temperatures Nb contains much N in solid solution, and it has been shown by metaIlograp~c analysis  that as the Nb cools, the excess N is precipitated out as Nb,N, as inclusions in the crystafs as well as at grain boundaries. The kinetics of the bulk interaction between Nb and N have been studied at low pressures by Pasternak and Gibson [ 131 hereinafter referred to as PG, who measured tlte rate of adsorption and the sticking coefficient over a range of temperatures. In this paper we use Auger techniques to monitor the amount of N on the surface and * Research sponsored by the AFOSR under grant No. 72-2292.
of‘ nitrogen with niobium
study adsorption over a range of temperatures and pressures. We find values of the sticking coefficient and of the number of atoms in a monolayer which are in agreement with PG. In parallel with the Auger studies the surface structures were examined by LEED.
The apparatus used was a Varian post acceleration LEED system in which the base pressure was less than about 2 X lo-lo torr. For the Auger work a glancing incidence gun was the source of electrons of about 2000 eV, and the screen and grids of the LEED optics were used as an energy analyzer of the inelastically scattered electrons. The Auger peak was measured in the usual manner by taking electronically the second derivative of the total current collected on the screen. The samples were either single crystals of Marz grade [ 151 niobium (99.999%) or thin foil. The crystals were cut to within 1” of the (100) face and polished. Some of the crystals in the polycrystalline foil were about 4 mm wide and LEED patterns characteristic of the (100)) (110) and (111) faces could be seen. The samples were etched in a 1 : 1 mixture of HF and HNO, and mounted on a niobium sheet holder to prevent contamination by diffusion at high temperatures. The Nb was readily cleaned, as reported earlier [S] , by heating to above 2OOO’C. The principal contaminants were nitrogen, oxygen and carbon, and the principal metallic contaminant was tantalum, the concentration of which was less than 300 parts per million [ 151 . Niobium reacts strongly with 0, and the presence of oxygen on the surface inhibits the adsorption of nitrogen  . Consequently the sample was always heated to a high temperature to remove oxygen before each adsorption run. Using the results published earlier (I) we would estimate from the magnitude of the Auger trace that less than 5% of a monolayer of oxygen was present. The temperature of the sample was measured using a pyrometer. The emissivity, E = 0.35 [lo] , was used to correct the pyrometer readings. In addition a W-Rh thermocouple was mounted close to the sample and was used to measure temperatures below the visible range. The nitrogen used was research grade (99.999%).
3. Auger results The results obtained for Nb foil were similar to those for the single crystal, so the detailed results for the Nb (100) face are reported in the following pages. lncident electrons of 2000 eV were used for excitation, since for this energy, at glancing incidence, the Auger yield from Nb and the major impurities carbon, nitrogen and oxygen is close to a maximum [ 161. The total electron current was usually between 50 and 100 PA. The results were independent of the magnitude of the incident electron beam and there was no evidence of electron induced desorption or
of nitrogen with niobium
Fig. 1, Sample points to show the variation of the Nb Auger peak with the surface N on Nb (100) after different treatments. Both peaks in arbitrary units.
adsorption. The peak to peak height of the 380 eV N transition was measured. If the principal Nb peak at 110 eV was measured simultaneously it was found that the magnitude of this peak decreased linearly to about 3/4 of its value for the clean surface as the amount of N on the surface of the Nb increased, as is shown in fig. 1. The points lie on the same curve independent of the prior treatment, that is whether the N had been adsorbed on a hot or cold sample or an initial layer reduced by heating, so it is assumed that the magnitude of the N Auger peak is proportional to the number of N atoms per unit area of the surface, u. If we assume that the maximum Auger signal corresponds to a monolayer of nitrogen then we could detect about 2% of a monolayer above the noise level. In strongly interacting systems appreciable chemical shifts in the Auger spectrum have been observed, for example, on the V-O system shifts of about 2 eV occur when the oxygen coverage is high. We did not detect any large chemical shifts in either Nb or the N peaks within the limited accuracy, about OS eV, of the present measurements.
4. Adsorption The build up of nitrogen on the surface was measured at different temperatures. The sample was held at a constant temperature and nitrogen gas was admitted to the chamber at constant pressure. The magnitude of the N Auger peak was followed with time. In addition some measurements were made by exposing the sample to N, for a short period of time and pumping down quickly. Within the accuracy of the measurement the Auger peak did not change during the time it took for the
of nitrogen with niobium
Pt ( lO-7
Fig. 2. Nitrogen Auger peak, in arbitrary units, versus exposure to nitrogen of Nb (100). Full curve pressure = 2.5 X 10V8 torr, dashed curve pressure = lo-’ torr. For 61 “C the curves coin-
tide . sample to cool down from the initial temperature to room temperature. The Auger signal does not depend on the temperature of the sample at the time the measurement is being made but only on the exposure to N2 and the temperature of the sample during that exposure. If the sample was at about room temperature, the surface coverage depended only on the product (Pt), for pressures less than about lop5 torr. It does not matter what the rate of incidence of gas atoms is but just the total number of incident N2 molecules. Fig. 2 shows the height of the Auger peak versus exposure to N2 at 3 different temperatures. At 61 “C the same curve is obtained for any nitrogen pressure whereas at higher temperatures only the initial stages of adsorption on a nearly clean surface are pressure independent. Of those N atoms hitting the sample and adhering, some remain on the surface and others diffuse into the bulk, this latter process occurring more readily the higher the temperature. If we assume that the probability of an incident molecules adhering to the surface depends only on u then the difference at high temperatures is due to leakage of N atoms from surface to bulk. More atoms leak away in the longer time taken for a given number of atoms to arrive at the surface at low pressures, so the resultant surface coverage as measured by the Auger peak is less.
of nitrogen with niobium
61 "C 159°C 275°C
Pt Fig. 3. Nitrogen Auger peak, in arbitrary pressure of Nz of 2.5 X 10T8 torr.
exposure of Nb (100) to Nz at a constant
At room temperature u increases rapidly until Pt = 100 X 10e7 torr set, which we can assume corresponds roughly to one complete monolayer coverage. On further exposure u increases more slowly. It does not tend asymptotically to a constant value as reported earlier  but a small continual increase is still detectable. After 5 min at P = lop7 torr the Auger peak rises by 8% in the next 10 min and after 5 min at P= 10M6 torr the Auger peak rises by 6% in the next 10 min. Fig. 3 contrasts the build up of N on the surface at different temperatures, and a constant pressure of 2.5 X 1O-7 torr. The curves are similar to that at room temperature but the rate of initial rise and the coverage both decrease with increasing temperature. This is attributable to both a decrease in the sticking coefficient and to an increase in diffusion from the surface as the temperature increases. At higher temperatures the coverage is less than a complete monolayer and continues to rise slowly as diffusion into the interior must continue unit1 the equilibrium solid solubility is attained. Fig. 4 shows the surface coverage attained after 1 min exposure to N2 at P= 5 X low7 torr, a time long enough so that any subsequent increase in u is slow. Above a temperature of about 900 “C no N could be detected above the noise level. Above 800 “C only a fraction of a monolayer is detectable even with extended exposures. This confirms the suggestion of PC, who found that S was independent of temperature above about 800 ‘C, and suggested that this indicated no nitrogen remained on the surface above this temperature. We can use fig. 3 to deduce the sticking coefficient S. Since the number of atoms
of nitrogen with niobium
Fig. 4. Nitrogen Auger peak, in arbitrary units, after exposure of Nb (100) at a pressure of 5 X lo-’ torr for 1 min with the sample held at a constant temperature.
incident from the gas is proportional to Ptand CJis the number of atoms residing on the surface, S is proportional to d(o)/d(Pt). It should be noted that S is also often defined in terms of the fraction of incident atoms leaving the gas phase so that there will be a difference between these two definitions when diffusion into the bulk occurs. In order to obtain absolute values of the sticking coefficient some assumptions must be made about the relationship between the Auger signal and the number of atoms on the surface. We have assumed that an exposure of Pt= 100X 1O-7torr set corresponds to one full monolayer and from the LEED results that one monolayer contains 7.2 X 1014 atoms. For nitrogen at 20 “C the number of molecules striking a surface is 3.6 X 1014 Pmolecules/cm2 where P is in torr. Fig. 5 shows the variation of S with surface coverage for several different temperatures between 61 and 575 “C. S decreases fairly rapidly from the initial values. Similar trends were found by PC but they found that S remained essentially constant up to about half a monolayer coveiage. Starting from as clean a surface as possible the initial sticking coefficient So was measured and the results are shown in fig. 6 for different temperatures. For different samples at room temperature PC found values of So in the range 0.25-0.57, which is consistent with our value of 0.3 at 60 “C.
of nitrogen with niobium
Fig. 5. Sticking coefficient, on Nb (loo), measured temperature of the sample was, a-f, 61, 159,275,
at a N2 pressure of 2.5 X 10e8 torr. The 355,460, and 575°C.
When the sample is heated the nitrogen originally on the surface may be lost through two mechanisms, diffusion into the bulk or evaporation. Since nitrogen is desorbed as N2 [ 141 the rate of surface diffusion is also important. The sample was exposed to 5 X lop7 torr of nitrogen for one minute at room temperature in order to form a monolayer. The sample was then heated as quickly as possible, and the current was switched off when the temperature reached the required value. The Auger trace was recorded. The amount of N remaining on the surface after this procedure is shown in fig. 7, curve a. If the temperature of the sample were subsequently held constant, u decreased slowly with time, fig. 7, curve b. In both figs. 4 and 7, u falls steeply in the temperature range 400-600 “C where diffusion becomes significant. The base pressure was less than 1O-9 torr. At this pressure, for the times involved, the rate of build-up of N on the surface from the ambient was insignificant.
of nitrogen with niobium
Fig. 6. Initial sticking
on Nb (100) for the sample at different
After the sample had been heated to a given temperature the magnitude of the Auger signal was followed as the sample cooled down. u remained approximately constant and did not increase as the temperature fell. This is in contrast to the results reported for oxygen  on niobium where the surface coverage increased as a sample cooled down, indicating a migration of oxygen atoms to the surface. For nitrogen, on the contrary, there appears to be no preferential segregation at the surface.
6. LEED results The sample was exposed to N2 under similar conditions to those described previously and the LEED pattern formed on the fluorescent screen was observed. If the sample were held either at room temperature or at a high temperature during the exposure a diffuse pattern was seen. On the other hand, if the sample were held at an intermediate temperature during the exposure, a sharp pattern could be produced, indicating a high degree of order. At room temperature, on exposure to [email protected]
torr for 1 min, there was an increase
of nitrogen with niobium
Fig. 7. Desorption of a monolayer deposited on Nb (100) at room temperature. Full curve instantaneous, dashed curve after 10 min at constant temperature. N Auger peak in arbitrary units.
in background intensity and blurred satellite spots were visible around each Nb spot together with streaking between the Nb spots, see fig. 8a. A similar pattern was seen for other exposures at room temperature and was reported in I. A well formed 5 X 5 pattern was seen (figs. 8b, 8c), when the sample was held between 250 and 450 “C with a N2 pressure of 10e6 torr for 1 min. Increasing the time of exposure did not affect the pattern much. From the Auger results this exposure would result in between 2/3 of a monolayer at 450 “C to a complete monolayer at 250 “C. Thus we can conclude that at these temperatures the N forms an ordered structure on the surface. This same exposure at a higher temperature of 700 ‘C, resulted in increased background intensity and faint diffuse satellite spots around the Nb substrate spots. If the N2 pressure was increased to 10P5 torr for 1 min a 5 X 5 pattern was seen once again but with more background than at lower temperatures. This is consistent with the Auger results which show that only a fraction of a monolayer remains on the surface, and increasing the pressure results in an increase in the surface coverage. The N tends to take up the same ordered pattern as at lower temperatures but with more incoherence. At a still higher temperature of 835 “C and a pressure of low6 torr
Fig. 8. LEED pattern (b) and (c) adsorption
of N adsorbed on Nb (100). (a) Adsorption at 350’ C, 30 eV and 127 eV respectively.
at room temperature,
of nitrogen with niobium
diffuse extra satellite spots were still seen showing a hint of the 5 X 5 ordered pattern; however these spots were more diffuse. At high temperatures the N dissolves in the sample so much higher pressures are required to see evidence of surface structures. This is also borne out in I where at a temperature of 1400 “C traces of what appear to be the same surface structure could be seen using pressures of 1O-4 torr. A highly ordered surface structure was formed by heating the Nb at 250 “C in N,. The changes in the diffraction pattern were observed as the sample was heated to progressively higher temperatures, in the ambient pressure of less than 10eg torr. At about 450 “C the pattern started to degrade, most of the extra spots disappeared but those adjacent to the Nb spots became diffuse. This confirms the Auger results that desorption of a monolayer becomes rapid above about 400 “C. The patterns were similar to those observed when a disordered monolayer formed at room temperature was heated  . The simplicity of the structures formed by N on Nb is in marked contrast to the results found for 0 . Depending on the conditions, 0 can form several different surface structures  , including a faceted structure. The appearance of a high index pattern is usually indicative [ 181 of a mismatch between the surface layer and substrate. A complete interpretation of the LEED patterns would require an analysis of the intensities of the diffracted beams, as several different surface structures could be consistent with the same pattern of extra spots. In order to estimate absolute values for the number of atoms in a monolayer and the sticking coefficient, we shall make the simplest assumption [ 18 ] , that in the 5 X 5 pattern 4 units cells of the surface layer correspond to 5 of the substrate to produce a coincidence lattice. The lattice constant of Nb is 3.3 A [ 191 . Hence
of nitrogen with niobium
the number of N atoms in a monolayer is 7.2 X 1014 atoms cm-2. This is consistent with the valeus obtained by PC in the range 4-8 X 1014 atoms cmU2. In earlier studies , Nb foil was heated in N2 and it was found that the superconducting transition temperature T, was depressed and the residual resistance was increased. It was suggested that N diffuses into the Nb to form a bulk impurity rather than forming a thin skin of the high T, compound NbN. Our present results reinforce this interpretation. Some diffusion into the bulk occurs even at relatively low temperatures and this results in the degradation of the superconducting properties. In addition the sticking coefficient becomes small as the surface coverage increases so uptake of N2 is slow. 0 prefers to segregate at the surface of Nb and a faceted oxide structure grows readily on the surface . In contrast a nitride does not grow on the surface of the Nb without contamination of the bulk. A 5 X 5 structure was usually formed when the surface coverage was high. Traces of this structure could also be discerned even at partial coverages or when some disorder was present.
Acknowledgements I should like to thank R. Arvanitis, E. Kuhner and C. Spiteri for valuable assistance and 0 .F. Kammerer for help in preparing the samples.
References [l]       [S]  [lo] [ll]        
A.G. Jackson and M.P. Hooker, in: The Structure and Chemistry of Solid Surfaces, Ed. G.A. Somorjai (Wiley, New York, 1969). J.M. Dickey, M. Strongin and O.F. Kemmerer, J. Appl. Phys. 42 (1971) 5808. J.M. Dickey, N. Garcia and M. Strongin, J. Appl. Phys. 44 (1973) 1919. J.M. Dickey, H.H. Farrell, O.F. Kammerer and M. Strongin, Phys. Letters 32A (1970) 483; H.H. Farrell and M. Strongin, Surface Sci. 38 (1973) 18, 38. T.W. Haas, Surface Sci. 5 (1966) 345. D. Tabor and J. Wilson, Surface Sci. 20 (1970) 203. T.W. Haas, A.G. Jackson and M.P. Hooker, J. Chem. Phys. 46 (1967) 3025. T.W. Haas, J. Appl. Phys. 39 (1968) 5854. R.P. Elliot and S. Komjathy, in: Columbium Metallurgy, Eds. D.L. Douglas and F.W. Kunz (Interscience, New York, 1961). J.R. Cost and C.A. Wert, Acta Met. 11 (1963) 231. V.E. Gebhardt, E. Fromm and D. Jakob, Z. Metallk. 55 (1964) 423. V.G. Horz and E. Gebhardt, Z. Metallk. 57 (1966) 737. R.A. Pasternak and R. Gibson, Acta Met. 13 (1965) 1031. R.A. Pasternak, B. Evans and B. Bergsnov-Hansen, J. Electrochem. Sot. 7 (1966) 731. Materials Research Corporation, Orangeburg, New York. R. Arvanitis, paper presented at the Eastern Colleges Science Conf., Worcester,.1974. F.J. Szalkowski and G.A. Somorjai, J. Chem. Phys. 56 (1972) 6097. G.A. Somorjai, Surface Sci. 34 (1973) 156. C. Kittel, Introduction to Solid State Physics (Wiley, New York, 1953).