Time-of-Flight Direct Recoil Ion Scattering Spectrometer

There are a number of analysis techniques which are able to characterize the surface properties of thin films and bulk materials. This capability is particularly important during the growth of thin films in which there are a plurality of components. Factors such as relative deposition rates of the various species, migration of materials at the surface, differences between surface and sub-surface composition, compositional and thickness uniformity, and nucleation of growth sites are of key importance in determining film properties. For multi-component films, and particularly for multicomponent films which are grown in an atmosphere of e.g. oxygen or nitrogen, precise control of film properties depends on the ability to monitor the growth process as it occurs. The high temperature superconducting oxides, such as YBa.sub.2 Cu.sub.3 O.sub.7-x, and wear-resistant materials, such as TiN and BN, are examples of such materials.

In order to characterize the process occurring at the surface of a growing film, it is necessary to probe the first few atomic layers, and in principle to identify the uppermost monolayer where the growth occurs. Most surface analysis methods however, are unsuitable as in-situ monitors of thin film deposition processes because they either (1) require ultra-high vacuum (<10.sup.-8 Torr) to operate, (3) physically obstruct the deposition process, (3) take too long to acquire data, or (4) cause significant damage to the film they are trying to characterize. There are several analysis techniques which utilize relatively collimated beams and therefore do not interfere with the deposition process. Reflection High Energy Electron Diffraction (RHEED) utilizes elastically scattered electrons with a kinetic energy in the range of 20 keV. This energy is high enough to provide a reasonably long electron mean free path. RHEED is therefore widely used in molecular beam epitaxy (MBE) systems where the ambient pressure during deposition is relatively low. RHEED provides a measure of the lattice spacing in the direction normal to the substrate, but provides no chemical identification and no information on short-range phenomena such as pinhole formation.

Low energy (5-15 keV) pulsed beam Ion Scattering Spectroscopy (ISS) and Direct Recoil Spectroscopy (DRS) are surface analytical tools which possess the ability to provide a remarkably wide range of information directly relevant to the growth of multi-component semiconductor, metal and metal oxide thin films and layered structures. Ion beam methods have not been widely used for monitoring thin film growth processes however, because the designs of existing commercial instrumentation are unsuitable for the purpose. The inventors have determined that pulsed ion beam surface analysis techniques, if properly implemented, may be used to circumvent these and other problems for application to thin film deposition as well as other processes such as gas phase-surface reactions. This invention disclosure describes an implementation of pulsed ion beam surface analysis which circumvents problems associated with the use of other forms of surface analysis techniques for in-situ analysis purposes, as well as with other implementations of ion beam surface analysis. Additionally, the instrument described provides unique capabilities aside from the application as a monitor of thin film growth and gas phase-surface reaction processes.

The method of ion beam surface analysis generally consists of directing an ion beam of mass M.sub.1, kinetic energy E.sub.o at the surface, which is comprised of atoms with mass M.sub.2, and detecting either the backscattered primary particles at energy E.sub.1 (ISS), or the direct recoil-sputtered surface atoms (DRS) with energy E.sub.2. For primary ions in the approximate range 1-100 keV, the primary ion-target atom collisions are adequately described by two-body classical collision dynamics. The kinetic E.sub.1 energy of the scattered primary is then given by:

provided M2>M1. The kinetic energy E.sub.2 of the recoil- sputtered surface atom is

where a=M2/M1 and q.sub.1 and q.sub.2 are the scattering and recoil angles, respectively.

For in-situ thin film analysis, it is necessary to obtain the data over periods of time which are short compared with the time required for the thin film deposition process, using ion beam doses which result in negligible sputtering or other modification of the surface being studied. Since there are roughly 10.sup.15 atoms/cm.sup.2 at the surface of a typical solid, a nondamaging dose is approximately 10.sup.13 ions/cm.sup.2. Low beam dose and rapid data acquisition are key requirements for real-time, low damage in-situ analysis of thin film growth.

Existing ion scattering spectrometers fall into one of two broad design types: (1) electrostatic energy analyzers (ESA) using DC ion beam currents and detection of the scattered ions only, and (2) time-of-flight (ToF) analyzers using pulsed ion beams with detection of either scattered ions or neutrals. There are no commercially available DRS analyzers.

Commercially available ISS analyzers are all of the electrostatic type which acquire data corresponding only to the scattered ions which constitute 10.sup.-1 to 10.sup.-3 of the total scattered flux. Furthermore, since the electrostatic analyzer only transmits a small portion of the spectrum at any one time, the data acquisition rate is very low and a very high ion dose is required to produce a spectrum. Consequently, the surface undergoes substantial bombardment leading to ion beam damage. The ion beams are operated in the DC mode and are readily focusable. However, it is not desirable to focus down to a small spot size in order to obtain spatially resolved data for two reasons: (1) the required ion dose per unit area would be so high that the surface would have been destroyed before its composition could be measured; and (2) spatially-resolved elemental mapping would take so long that it would not be practical.

The ToF-ISS method consists of directing a pulsed beam of energetic ions onto the surface of the sample, and measuring the arrival times of the scattered primary particles, most of which are scattered as neutral atoms. By pulsing the beam repeatedly, a spectrum containing information on all masses is collected. The data acquisition procedure is not limited to one surface species at a time, and spectra can be accumulated with a much smaller ion dose and consequently smaller surface damage than with the ESA. Since the ToF scheme, independently detects both ions and neutrals, whereas the ESA detects only the ion fraction of the scattered/recoil sputtered atoms, an overall 3-4 order of magnitude increase in sensitivity (or an equivalent 3-4 order of magnitude reduction in beam damage and data acquisition time) is achieved compared with electrostatic analyzers.

If the scattered neutrals are detected, the ToF-ISS method has a depth sensitivity of three to five atomic layers. However, since the neutralization probability for an ion penetrating more than one monolayer into the solid is nearly unity, the detection of the ion species provides a sensitivity only to the uppermost atomic layer of the solid. This capability is unique among surface analytical techniques. Furthermore, by varying the angle of incidence of the primary beam so that atoms in the second and third layers are "shadowed" by atoms in the first layer, it is possible to determine the distance and bond angle between atoms in the first few atomic layers. Therefore, ToF-ISS is not only capable of measuring the average lattice spacing, but unlike the more conventionally used MEED, it is element specific, and can thus map out the detailed crystal structure of the first several atomic layers.

The shape of the scattering peak for any given element, and the variation of that shape with primary beam energy and angle of incidence is related to both the lattice defect density, and the concentration variation of that species with depth over the first several atomic layers. Varying the angle of ion beam incidence from grazing to normal shows pronounced intensity oscillations in the signal corresponding to subsurface atoms, resulting in peaks which may be identified with specific lattice sites. However, atomic species which reside solely on the surface display no intensity variation with angle of incidence. It is therefore possible to distinguish between atomic species which truly reside on the surface from species for which the concentration varies with depth. By varying both the azimuthal and polar angle of incidence, it has been demonstrated by others that it is possible to generate atomic images similar in information content to those produced by the Scanning Tunneling Microscope.

By placing a detector in the forward scattering direction, surface atoms ejected by direct recoil sputtering are seen in addition to the scattered primary beam. Detection of these recoil-sputtered atoms constitutes the basis of the closely related Direct Recoil Spectroscopy (DRS) analysis technique. ISS provides no signal for ions lighter than the probe beam, but DRS is one of the few surface analytical techniques which is sensitive to helium and hydrogen, and is able to distinguish between H and D at levels down to about 1%.

Because of the long source-detector distances associated with the ToF detection scheme, the analysis method does not interfere with the equipment required for the thin film deposition process. The scattering mean free path of low keV ions is much longer than that of the sub-keV electrons, and ToF-DR/ISS is therefore much more tolerant of high background pressures than analytical methods based on the emission of electrons from the surface. It is also possible to provide differential pumping of the incoming and outgoing ion beam paths to extend the pressure limits of the ToF ion beam techniques even further. This capability has been demonstrated by others for diamond-like films during growth by Low Pressure Chemical Vapor Deposition (LPCVD) at pressures up to 1 Torr, and hydrogen adsorption has been measured on (100) diamond at ambient pressures up to 330 mTorr with very little loss of resolution. This represents six to eight orders of magnitude increase in the permissible operating pressure compared with other surface analytical methods.

One of the most obvious uses of an in-situ thin film surface analysis technique is to monitor the composition of the growing film. By simultaneously collecting signal corresponding to both the scattered neutrals and scattered ions, compositional information may be obtained for both the immediate surface and subsurface species. In either case, however, the data is more surface-specific than that produced by other "surface" analysis techniques such as Auger electron spectroscopy. This situation is evident in the 10 keV Ne.sup.+ NSS spectrum obtained with the instant invention (FIG. 10) for sulfur which has segregated to the surface of a stainless steel 304 sample. Although S is the dominant species in this spectrum, an Auger spectrum taken on the same sample shows a 6:1 Fe:S ratio. The ISS data is more surface-specific than the Auger data, which samples the iron over an e-folding length of approximately 10 .ANG.. The large S signal in the ISS spectrum therefore confirms the theoretical expectation that segregated S is localized at the uppermost atomic layer.

There are, however, a number of shortcomings associated with existing TOF designs:

1. A number of different species may be used for the primary ion beam. The "best" primary ion is a trade-off between sensitivity and mass resolution, and is also determined by the lowest mass which must be detected. Ideally, the pulse length and pulse repetition rate should be optimized for each primary ion mass and energy. In practice, because of the way in which the pulses are generated in conventional ToF instruments, it is necessary to preset the apertures in the beam line to give the desired time resolution, and to set the pulse repetition rate to match the slowest (i.e., heaviest and lowest kinetic energy) ion species which will be used by the system. The only control available to the operator is the frequency of the oscillatory waveform which drives the beam pulsing. The use of a fixed waveshape with variable frequency as the beam deflection voltage results in a compromise which limits the pulse repetition rate, number of ions per pulse and the attainable mass resolution;

2. In order to offset the resulting limitations on the number of ions per pulse, a rectangular beam spot, which may cover the full length of the sample, is often used. In most existing configurations, this beam spot also sweeps across the width of the sample as the pulse is generated. It is therefore not possible to generate a spatially resolved image of the elemental distribution on the sample or even to analyze a specific area of the sample; and

3. Since the mass resolution of the signal depends on the fact that all detected ions must lie within a relatively well-defined solid angle, good mass resolution is obtainable only at the expense of very low signal collection efficiency. The detector in a typical TOF instrument collects 10.sup.-4 -10.sup.-5 of the total scattered particles. Various features of the preferred embodiment are illustrated by way of comparison with other prior art systems in Table I.

It is, therefore, an object of the invention to provide an improved time of flight direct recoil/ion scattering spectrometer.

It is another object of the invention to provide a novel pulsed time of flight spectrometer.

It is yet a further object of the invention to provide an improved time of flight spectrometer having improved data collection and analysis features.

It is still another object of the invention to provide a novel ion scattering spectrometer and/or microscope.

It is an additional object of the invention to provide an improved time of flight direct recoil/ion scattering spectrometer with very high data acquisition rate and high mass resolution.

It is yet a further object of the invention to provide novel spectrometer for reducing the ion beam damage caused the analysis beam.

It is another object of the invention to provide a novel pulsed time of flight spectrometer with flexible operating parameters to maximize the usefulness of various modes operation.

It is yet a further object of the invention to provide improved time of flight spectrometer having improved data collection and analysis features.

It is an additional object of the invention to provide a novel method and apparatus for performing practical means surface analysis for samples immersed in an atmosphere at an ambient pressure which would make other surface analytical procedures impossible.

It is still another object of the invention to provide a novel ion scattering spectrometer and/or microscope.

It is a final object of the invention to do all of the above in a non-intrusive manner, being therefore compatible with fabrication and processing procedures such as thin film deposition or gas phase-surface reactions.

Other advantages and objects of the invention will become apparent from the detailed description of the method and apparatus and the drawings below, wherein like features have like numerals throughout the several drawings.

FIG. 1A illustrates a dual deflection pulsed ion beam line; and FIG. 1B shows the number of ions transmitted per 10 ns pulse to the sample versus aperture diameter;

FIG. 2A shows the ion beam path corresponding to the illustrated plate deflection voltage pulses and FIG. 2B shows details of pulse timing in FIG. 2A;

FIG. 3 illustrates a schematic functional block diagram of one form of the data acquisition electronics utilized for a small area, single anode detector;

FIGS. 4A-B show a segmented eight channel 4 cm diameter channel plate detector;

FIG. 5 is a schematic functional block diagram of the system for the data acquisition electronics corresponding to the large area, segmented anode detector of FIG. 4;

FIG. 6 illustrates beam profiles at the sample taken from scans across a sharp edge;

FIG. 7 shows use of the beam pulsing system for an ISS microscope;

FIG. 8 illustrates an exemplary output spectrum for surface composition analysis of a given sample spot;

FIGS. 9A-B show a prior art DR analysis of diamond samples in vacuum versus 330 mtorr;

FIG. 10 illustrates a 10 keV Ne.sup.+ NSS spectrum of sulfur on stainless steel;

FIG. 11A shows a spectrum from a short ion beam pulse and FIG. 11B illustrates a spectrum from a long ion beam pulse;

FIG. 12 illustrates a spectrum taken with a 50 nsec pulse at an ion dose of 10.sup.10 ions;

FIG. 13 shows a double pulse edge transmission (DPET) beam line;

FIG. 14 illustrates a beam line with a "sniffer tube" component; and

FIG. 15 shows relational distance parameters among sample, apertures and detectors.

FIG. 16 illustrates the current practice for differentially pumped pulsed ion beam analysis.

FIGS. 17A-B show a differentially pumped detector assembly and alignment filament geometry.

Because of the manner in which data is collected and the way the beam is pulsed in practicing the invention, we are able to effect significant improvements in all of the three problem areas described herein before. An ion beam line 10 (see FIG. 1A) consists of an Atomika Corp. telefocus ion source 12 injecting a 5-12 keV mass-analyzed ion beam into a transfer tube which includes two pulsed deflection regions 14 and 16, separated by a drift space 18 containing a small set of DC x-y correction plates. Each deflection region is terminated by a pair of apertures selected from among a choice of six sizes ranging from 250 microns to 4 mm in diameter on a movable aperture plate 20.

A final optics stage consisting of an ion lens and deflection plate assembly 22 is added for demagnifying the image of the final aperture and rastering beam 23 to produce a uniform current density over the analyzed area 24. The ion pulse length and interval between pulses are determined by adjusting the timing between the voltage pulses applied to the two sets of deflection plates. The timing sequence is complex, but is under computer control to provide a transparent interface so that the user may specify whatever pulse length he desires in order to optimize the data collection needs of the experiment. This flexibility provides great control over the type of information which can be obtained from the data. An attached Appendix sets forth computer software used to carry out the timing sequence described.

In the preferred embodiment, the interval between successive beam pulses is dynamically set for each combination of ion beam mass and energy to provide the maximum repetition rate consistent with clean separation of the data arising from consecutive ion beam pulses. Insofar as is practical, the deflection plates 14 and 16 run the entire length of the deflection region. This design feature ensures that fringe field effects do not introduce longitudinal velocity components to the beam 23 which would result in degraded temporal and spatial resolution of the beam pulse. The use of an ion source which produces a paraxial beam, coupled with the absence of fringing field effects makes it possible to use relatively simple ion optics to achieve a well-defined beam spot diameter ranging from the selected aperture size (250 microns to 4 mm) to less than 1 micron in diameter. As discussed below, the usable spot size is limited by the damage threshold of the sample and by relatively minor "tail" effects associated with the beam pulsing. Consequently we have been able to design a ToF Ion Scattering Microscope with an anticipated spatial resolution of a few microns.

One form of a data acquisition system 26 consists of two 8 channel, 100 MHz scalers and five detectors (see FIGS. 4 and 5), of which four are differentially pumped small area detectors of channeltron multipliers 28 with an entrance cone 1 cm in diameter. These detectors 28 are positioned 50 cm from sample 30, and have a geometric collection efficiency of 5.times.10.sup.-5. Two of the small area detectors 28 are in the backscattering direction, one off-axis and the other in line of sight with the sample 38, permitting simultaneous detection of both ion and neutral backscattering signals. It is useful to refer to the latter signal as the "Neutral Scattering Spectroscopy" (NSS) signal. The other two small area detectors 28 are in the forward direction, where they function as Direct Recoil (DR) detectors.

Conventional ISS instruments using large area detectors are able to gain data acquisition speed only at the expense of reduced angular (and consequently mass) resolution. The fifth detector of the instant invention is a 40 mm dia. channel plate 32 with a segmented anode collector consisting of 8 concentric rings, each connected to a separate preamplifier and discriminator. The most preferred sample-detector distance is 1/2 that of the small area detectors, providing 4.times. the count rate/cm.sup.2 of detector area.

The data acquisition package for the described system uses custom-designed electronics to partition a commercial histogram memory 34 into eight sections, each section corresponding to a separate data channel of the 100 MHz scalers. Consequently, each of the eight independent data channels sees only a small range of scattering angles and consequently represents a well-defined mass corresponding to a given time-of-flight. An instrument control computer 36 deconvolutes the signals from the eight data channels to produce a single composite spectrum with the same mass resolution as the small area detector, while the data acquisition rate is increased by a factor of sixty-four, resulting in a geometric detection efficiency of 3.2.times.10.sup.-3.

The aperture plate positions are controlled by micrometer drives 38 for reproducibility (see FIG. 1A). The attainable beam current as a function of aperture size is shown in FIG. 1B. Except for the smallest aperture for which lateral alignment becomes critical, the beam current scales closely as the aperture area. At the start of a deflection cycle, the first deflection region 14 has no applied deflection voltage while the second region 18 does, and therefore does not allow the ion beam to pass. At a certain time, a voltage pulse applied to the first set of deflection plates in order terminate the back end of the beam as shown in FIG. 2. Just before the back end of the beam enters the second deflection region, the voltage on the second deflection stage is removed, thereby allowing a short pulse to be transmitted. The timing of the pulses applied to the two stages is calculated by the instrument control computer program (see Appendix); based on the dimensions of the ion beam line, and the ion beam mass, energy and desired pulse length. Increasing the pulse length increases the count rate, but results in decreased mass resolution. One can vary the pulse length "on the fly" over a usable range of approximately 5.times.10.sup.-9 to 1.5.times.10.sup.-6 seconds. The pulse frequency is independently set by the computer software, based on the ion beam mass and energy; and the mass range anticipated atomic species on the sample surface, to produce the maximum possible pulse rate consistent with clear separation of the data corresponding to each ion pulse 40.

By using a very short pulse with small apertures, excellent time resolution, corresponding to good mass resolution can obtained, at the expense of reduced data acquisition rate shown in FIG. 11A. A short ion beam pulse (50-100 nsec) provides high mass resolution and exhibits a distinctly asymmetric peak as shown in FIGS. 10, 11A and 12. Most existing ToF ISS/NSS instruments have a fixed resolution which is determined by the beam aperture size and the rise time the deflection pulse, and can not be easily varied. The beam line geometry of these instruments is typically fixed at some compromise between resolution and count rate. However, the peak shape contains useful information on the depth distribution of each element. If the interface between film and substrate is not sharp, i.e., if the dominant surface species extends into the substrate, the signal corresponding the backscattered neutral will have an asymmetric lineshape with a long time-of-flight "tail" resulting from scattering from atoms two to three layers below the surface. Since the backscattered ion (ISS) signal is believed to be representative of the composition of the uppermost atomic layer, the formation of a coherent film is therefore indicated by the disappearance of scattered ion signal from the substrate. By comparing the ISS and NSS peak intensities and line shapes, one can detect the formation of islands and pinholes in layers only a few .ANG. thick. Such depth-specific information is of paramount importance in the deposition of HTSCs in conjunction with other materials for device fabrication since the superconducting coherence length may be less than 5 .ANG., and HTSC devices will therefore require abrupt interfaces between smooth and pinhole-free films only a few .ANG. thick.

Conversely, in situations where high mass resolution is not as important as rapid data acquisition for e.g., real-time monitoring of surface processes, longer pulses and larger apertures can be used to increase the count rate as shown in FIG. 11B. In the case of stainless steel, a 500 nsec Ar.sup.+ ion pulse is able to completely resolve Fe and Cr, with a total data acquisition time of ninety-three seconds. No special care was taken to minimize the beam dose for the spectra shown in FIGS. 11A and B which were both taken using a dose of approximately 3.times.10.sup.12 ions. In fact, the signal/noise is much greater than needed for most purposes. FIG. 12 shows a spectrum with a s/n ratio of approximately 20:1, taken using a 50 nsec pulse with a dose of 1.times.10.sup.10 ions. Installation of the large area detector and moderate beam rastering is expected to reduce the required dose and data collection time an additional 2-3 orders of magnitude.

Typically in ToF experiments, the instrumentation is only able to detect one signal event per ion beam pulse. The preferred embodiment of the invention has a beam line, which when operated with a large aperture with a long pulse is capable of producing several signal counts per ion beam pulse. All signal pulses following the first would be lost in a conventional single-hit data acquisition system, both reducing the data rate and possible severely distorting the data. In the instant invention, a special circuit directs consecutive signal events to separate channels of an eight channel scaler, thereby allowing up to eight signal events per ion beam pulse as shown in system 42 of FIG. 3.

Using a 1 cm. diameter electron multiplier 28 at a distance of 50 cm from the sample 30 as the detector, an angular resolution of 1.1 degrees is obtained, with a collection efficiency of 5.times.10.sup.-5, as shown in FIG. 4. However, by using the 4 cm. diameter channel plate 32 with a central hole to allow primary beam 44 to impinge on the sample 30 at normal incidence, an eight section segmented anode can be used at a distance of 25 cm while retaining the same angular resolution per segment as the 1 cm. diameter electron multiplier at a distance of 50 cm. The collection angle is now increased to 9.2 degrees, and the detection efficiency is increased by a factor of sixty-four to a value 3.2.times.10.sup.-3. In this case, each segment of the collector plate 32 is connected to one input channel of eight channel scaler 46, and eight separate spectra are directed via a custom-designed data router 48 into one of eight designated areas of the histogram memory 34 as shown in FIG. 5. By adjusting the time scale of each spectrum to compensate for the slightly different detection angles, a composite spectrum can be generated by the computer 36 which represents a sixty-four fold increase in detection efficiency and data acquisition rate.

The shape of the ion beam is a circular spot whose size is determined primarily by the size of the apertures in the beam line. For a given size aperture, the transmitted ion pulse forms a footprint on the sample surface consisting of a stable central portion with "tails" which typically spread 50-200 microns beyond the central "hot spot". The number of ions in the tails and the degree of lateral spread is determined by the aperture size and the pulse length. For e.g., a 500 micron aperture, pulses shorter than 15 nsec consist entirely of the tail portion. Longer pulses consist primarily of the stationary "hot spot" and provide a high count rate with a high degree of beam stability and focusability, although excessively long pulses also result in a loss of mass resolution. Beam profiles taken by scanning the beam across a sharp edge and measuring the current to the sample for the 500 micron apertures with both DC and pulsed beams are shown in FIG. 6 for collimated operation (unfocused beam). The profiles for both dc and pulsed beams show little beam divergence; i.e. most of the beam is confined within the 500 micron collimation diameter and the beam is therefore expected to be highly focusable. As shown in FIG. 7, the beam pulsing system can also be used as an ISS "microscope" 50, i.e., an ISS instrument with spatially resolved surface analytical capability. Since ion beam 52 is highly collimated, simple ion optics, consisting of a three-tube einzel lens 54 is capable of focusing to a very small spot, in principle, to approximately 100 .ANG. in diameter.

By utilizing the above-described techniques for rapid data acquisition and high detection efficiency, it is possible to focus ion beam 56 down to a spot 5 microns in diameter while obtaining a spectrum of 10.sup.3 -10.sup.4 total counts with a beam dose of 10.sup.13 ions/cm.sup.2. This dose corresponds to approximately one percent of the number of atoms/cm.sup.2 at the surface 24 of the sample 30, and is taken as the criterion for production of insignificant surface damage. Five microns is therefore estimated to be the minimum resolvable diameter which may be analyzed without introducing significant surface damage. If the surface 24 of the sample 30 is not particularly subject to damage, the resolvable area can be reduced still further.

Calibration measurements using the 1 cm diameter single anode detector 28 at a distance of 54 cm from the sample 30 have produced count rates of 80 kHz for a 4.2 nA dc beam. Assuming a calculated value of 0.088 for the ion reflection coefficient, the observed count rate corresponds to a detection efficiency of 3.5.times.10.sup.-5, or 70% of the theoretically obtainable value of 5.times.10.sup.-5 for this detector configuration.

As shown in FIGS. 13-15, the invention further includes a modification of the detector assembly. The modification primarily involves a change in the detector housing to include a small aperture 60 at the entrance to the detector channel. This geometry allows differential pumping of the detector 28, permitting the detector 28 to operate with the sample chamber at a pressure of several hundred millitorr. The beam line is already differentially pumped and requires no further modification for sample chamber pressures in this range.

The use of differential pumping poses alignment problems which in the past have prevented the practical application of ion beam analysis to samples in an ambient atmosphere in excess of 10.sup.-6 Torr. Another aspect of the instant invention relates to a new beam alignment procedure for use with the differentially pumped detector. This procedure has particular value for analysis of thin films of one or more components which must be deposited in the presence of a chemically active gas such as oxygen or nitrogen. The alignment procedure utilizes a light-emitting filament which casts a bright pinpoint of light on the sample, clearly indicating the sample area being analyzed. A preferred modification includes the filament in a six-way cross in the detector housing with the filament at right angles to two detectors, one detector in a line of sight with the sample, and the other in an off-axis arrangement. The alignment procedure uses a dot of gold or other visually identifiable material at a predetermined location on the sample holder. The sample is positioned so that the spot of light falls on the visual reference sample. The ion beam position and focus are then electrically adjusted to give maximum ISS peak intensity for the reference sample, thereby completing the alignment process without any need for mechanically adjusting ion beam entrance and exit apertures.

The Double Pulse Edge Transmission (DPET) beam line shown in FIGS. 13 and 14 (which we are using as part of the preferred form of the pulsed ion beam spectrometer design) provides differential pumping of the beam line via the four apertures used to provide the pulsed ion beam. These apertures are adjustable from 125 .mu.m to 4 mm and are normally in the range 0.5-1 mm for surface analysis.

In both the ISS and DRS configurations, we can place a "sniffer tube" at a distance A from the sample corresponding to the radius of the inner cylindrical shell (see FIG. 16). We immediately gain the advantage that the work space is considerably opened up. Other equipment such as thin film deposition equipment and other diagnostic instrumentation may reside outside radius A as long as they do not conflict with the sniffers. The aperture at the end of the sniffer defines the scattering/recoil angles and the "illuminated" detecter area. For a 1 mm diameter sniffer aperture, a 1 cm diameter detector will receive full signal if B/A=10 where A is the sample-aperture distance and B is the sample-detector distance. In our system, B=50 cm.

Assuming a 1 mm dia. aperture as shown in FIG. 15 (which provides full illumination of the detector), and a 90 1/sec differential pumping speed (the maximum obtainable through the 23/4" flange fitting we are currently using for the detector housing), the detector differential pumping ratio is

That is, if the sample chamber is at a pressure of one torr, the detector will be at a pressure of 1/813=1.2.times.10.sup.-3 torr. This is the approximate upper limit at which a channel electron multiplier may be safely operated.

It should also be noted that at this pressure, the mean free path is nearly the same as the aperture diameter. There is therefore little to be gained by providing a second stage of differential pumping since most of the gas flux will be associated with molecular rather than viscous flow. The degree of differential pressure drop therefore is comparable with prior art designs.

The only alignment criterion is that the axis defined by the aperture and the center of the detector intersects the sample somewhere on its surface. The sample may be moved within the chamber (using a three axis goniometer mount) if this condition is not immediately met. The ion beam must then be directed to that spot by means of the electrostatic deflection plates. This can be done interactively by maximizing the detector signal as the beam is scanned across the surface. The detector aperture, once installed, does not need to be disturbed when changing either the sample or detectors.

The first step of the alignment process can be simplified even further as follows: We are presently using a 4-way cross for the detector housing. This permits differential pumping and the use of two detectors, one line of sight and the other off-axis for detection of scattered ions which are more surface-specific than the scattered neutrals. By replacing the 4-way cross with a 6-way cross in which a tungsten filament runs at right angles through the intersection of the other two axes, as shown in FIG. 17, several advantages can be gained:

1. Since the filament is equidistant from, and on the axis of both detectors, the detectors may be biased to detect the electrons emitted by the filament (rather than energetic neutral atoms or positive ions). Therefore the performance of the two detectors may be calibrated and amplifier gain and discriminator levels may be unambiguously checked and absolute count rates verified.

2. The filament and aperture will cast a bright pinpoint of light on the sample, clearly indicating the sample area being analyzed.

3. A standard calibration sample (such as a 1 mm dia. dot of gold located a predetermined distance from the center of the sample being analyzed) can be moved into the spot of light, and the ion beam position and focus can then be adjusted to give the maximum goal ISS peak intensity. The position of this peak then unambiguously establishes the ToF calibration for all elements.

While preferred forms of the invention have been described and illustrated, one of ordinary skill in the art will appreciate the invention is not so limited and is defined by the scope of the appended claims and their range of equivalents. ##SPC1##