WO1996037769A1

WO1996037769A1 - Material analysis device and method, in particular for concentration profiles in layered compositions

Info

Publication number: WO1996037769A1
Application number: PCT/SE1996/000666
Authority: WO
Inventors: Bengt Anderberg; William A. Lanford; Björgvin HJÖRVARSSON
Original assignee: Anderberg Och Modéer Accelarator Ab
Priority date: 1995-05-22
Filing date: 1996-05-22
Publication date: 1996-11-28
Also published as: SE9501906D0; SE509644C2; SE9501906L

Abstract

The invention relates to an apparatus for high resolution analysis of concentration profiles in layered compositions. It comprises a particle beam source (1), e.g. a van de Graaff accelerator. There is also provided a sample or target manipulating device (3), a magnetic spectrometer (4) arranged to detect ions scattered from a target exposed to a particle beam from said particle beam source. Said manipulating device and said magnetic spectrometer are enclosed in an ultra-high vacuum housing (2).

Description

MATERIAL ANALYSIS DEVICE AND METHOD, IN PARΗCULAR FOR CONCENTRATION PROFILES IN LAYEREDCOMPOSITIONS

The present invention relates generally to materials analysis, and in particular to high resolution analysis of concentration profiles in layered compositions.

A. Background of the Invention

In many industrial applications, particularly in the computer and semiconductor industry, there is a growing demand for means and methods for analysing surface and sub-surface structure and compositions of materials.

E.g. in the manufacture of hard discs for computers one utilizes polycrystalline multilayers of Fe-Cr, where it is desirable to very accurately determine the concentration profiles close to the layer boundaries.

It is also desirable to determine the thickness and thickness variations of extremely thin surface coatings in the semiconductor industry.

The best devices known today have a depth resolution in surface thickness analysis of 5-10 nm, which is far from sufficient when surface layers to be analyzed comprise but a few atomic layers.

Technical Field of the Invention Ion Beam Analysis (IBA)

Rutherford Back Scattering (RBS) and Elastic Recoil Detection (ERD) of charged particles are standard techniques for determining composition, depth distribution of elements and position of impurities in a crystal matrix. In these experiments, the energy of the incoming particles is normally lower than the Coulomb barrier of the target nuclei. These techniques have been applied in almost all fields of research in materials science.

When the energy of the incoming ions is higher than the Coulomb barrier of the target nuclei, nuclear reactions are induced by the ion beam. The resulting products, particles and (or) gamma^rrays, are detected by appropriate detectors. This is the basis for the Nuclear Reaction Analysis (NRA) . Depth information is obtained by changing the beam energy or by measuring the energy distribution of the outgoing charged particles. A special branch of NRA is the

) Resonant Profiling method.

Multilayered materials, with layer thicknesses of a few atomic layers, are now well established in the technology of semiconduc¬ tors, where they have provided the basis for completely new applications (in optical cable systems, etc.) as well as giving possibilities to study important fundamental systems in physics, such as isolated quantum wells. Metallic multilayers and their more organised forms, the metallic superlattices, are at present considerably less developed, although they may offer essential new insight into the behaviour of electronic and magnetic structure of metals in general, as well as providing possibilities for several different applications. Mostly discussed at present are applications of their magnetic properties in compact memory devices and in sensors based on magnetoresistivity, but also uses of multilayers as X-ray and neutron mirrors for various^'-kinds of spectroscopy are possible.

Today, there are no techniques available to analyse the chemical composition and profiles of these advanced materials with the required resolution.

The use of MeV ion beams to probe the composition and structure of materials has grown rapidly over the past 2 decades. From its beginning with a few pioneering researches in the early 1970's, there are now hundreds of MeV accelerators around the world being used for ion beam analysis. Generically, MeV ion beam analysis consists of those techniques in which a sample is bombarded with MeV ions and particles which come from the sample are detected and used to deduce information about the composition of the sample. The most important subset of this group of techniques is that in which the bombarding ion causes charged particles to be emitted from the target. This includes Rutherford backscattering spectrometry (RBS), some nuclear reaction analysis (NRA) techniques, and energy recoil detection (ERD) . RBS has become one of the most widely practised techniques ) for quantitative measurement of heavy element concentration profiles. NRA provides a unique probe for profiles of light elements, including hydrogen, carbon, nitrogen and oxygen in samples. ERD is a newer method which holds promise for high sensitivity measurement of light elements, including hydrogen in any target. The analytic capability of each of these methods can be greatly improved with the use of a magnetic spectrometer to detect the outgoing particles, in place of the usual surface barrier detector.

While the details vary, each of these methods is able to determine the depth distribution of the element being analysed based on a measurement of the energy of the detected particle. The detector almost universally used in such studies is a surface barrier detector which has an energy resolution of about 15 keV (full width at half maximum) for few MeV He particles and much -worse energy resolution for heavy ions such as carbon or oxygen. The magnet spectrometer proposed here will have resolution of 1 keV for 4 MeV He ions and an equally good resolution for 4 MeV heavy ions. This improvement in resolution will result in a dramatic improvement in the depth resolution of these methods. In some cases, the depth resolution will approach atomic dimensions providing new and important probes.

While there are several other important advantages to be gained by 5 the use of a magnetic spectrometer rather than a surface barrier detector, the improvement in depth resolution is one of the most important features of using a magnet. Hence, it is worth describing these improvements in some detail. All these methods derive their depth information based on the fact that charged par¬ ticles loose energy as they penetrate materials. It is the energy loss of particles incident on and emitted from a target that is used to determine the depth in the sample being probed. For^__all these methods, the depth resolution can be expressed (approximately) as: _σ - ΔJff ₍₁₎ dE/dx

where σ is the depth resolution. ΔE is the energy spread of the detected particles coming from a specific depth in the target, and dE/dx is the energy loss rate including energy loss of both the incident and detected particles.

The ΔE term includes contributions from the detector resolution, the spread in the energy of the incident beam, spread in energy of the detected particle due to the finite size of the detector (kinematic spread), and from the straggle in the energy loss of particles penetrating through solids. Modern accelerators can usually be operated so that the beam energy spread contribution can be neglected. Near the surface of the sample, the energy straggle can be neglected. Magnets can be operated to correct for kinematic spread.

Hence, improving the detector resolution from the 15 keV typical for a surface barrier detector, to 1 keV with a magnet, will improve the near surface depth resolution by a factor of 15. This means that a conventional RBS measurement that had a near surface resolution of 20 nm when measured with a surface barrier detector will have a depth resolution of 1.3 nm when measured with a high resolution magnet. An NRA measurement of nitrogen profile at the surface of a sample using the ¹⁴N(d,α)^lzC reaction which might have a depth resolution of 50 nm when measured with a surface barrier detector will have a near surface depth resolution of 3.3 nm with a magnetic spectrometer. While measurement with a high resolution magnet always gives better depth resolution than those made with a surface barrier detector, at some depth in a target straggle eventually becomes the largest contributor to the depth resolution. In this situation, the advantages of a high resolution magnet becoπre less important. However, even in the case of making an RBS measurement with a He beam (a rather unfavourable case), one has to be probing at depths of about 300 nm before straggle becomes comparable with 15 keV detector resolution. At this depth, the depth resolution with a magnet is still approximately 40% better than that measured with a surface barrier detector.

In summary, the use of a high resolution magnetic spectrometer to detect outgoing charged particles rather than a surface barrier detector improves the depth resolution of IBA profiling technique by an order of magnitude near the surface of the sample being analysed and this improvement persists several hundred nm into the sample.

As important as the improvement outlined above, another major improvement is possible with a magnetic spectrometer. When RBS is being used to analyse a sample, the depth resolution can be improved by decreasing the ΔE term in equation (1), as outlined above, or by increasing the dE/dx term in this equation. For example, going from He to 0 beams (all at 0.4 MeV/amu in Si) increases dE/dx by almost a factor of 10. This then improves the near surface depth resolution by an order of magnitude if the detector resolution does not deteriorate. Unfortunately, surface barrier detectors loose energy resolution even more rapidly than the increase in dE/dx, and, hence, there is no advantage in using 0 beams instead of He to do RBS with such detectors. However, a magnetic spectrometer Has the same energy resolution when detecting 0 or He and, hence, provides the opportunity to utilise the improvement in depth resolution intrinsically available with heavy ions.

A second advantage that comes with heavy ion backscattering is an improvement in mass resolution. The energy of an ion elastically scattering from the surface of a sample is given by conservation of energy and momentum in a two-body collision. At 180 degrees, this is simply:

^E ^ m. * ^ »***^{> (2)}

where E is the energy of the scattered particle, xa₁ and m₂ are the masses of the projectile and target masses respectively, and E_beam is the beam energy. Hence, by measuring E and knowing the projectile mass and the beam energy, the target mass may be determined. A difficulty with conventional He RBS is that all

"heavy" target atoms give nearly the same scattering energy. For example, the difference in energy E given by the above formula for 2 MeV He ion scattered from target masses of 100 amu vs 101 amu is only 2 keV. Hence, this difference is unmeasurable when a 15 keV surface barrier detector is used. However, if 8 MeV O is used for the RBS, the difference in E for mass 100 vs 101 increases to 27 keV which would be easy to measure with a magnetic spectrometer.

In summary, use of a magnetic spectrometer in place of a surface barrier detector improves the near surface depth resolution by an order of magnitude for all those IBA techniques in which depth profiling depends on measurement of the energy of the detected particle. In addition, another dramatic improvement is possible when heavy ion backscattering is used because of the larger dE/dx of heavy ions compared to He. Heavy ion backscattering with a magnetic spectrometer would also provide a near surface mass resolution of better than 1 amu for all target elements.

In addition to the class of improvements briefly outlined above that follow from the better energy resolution of a magnetic spectrometer, there is second and almost equally important property of magnetic spectrometers when used in IBA. Namely, magnetic spectrometers allow the measurement of low yield nuclear reactions and scattering events in the presence of intense elastic scattering events. Most NRA analysis methods based on the detection of charged particles require some method of eliminating elastically scattered particles from the detector. Without such a system, the detector spends virtually all its time counting elastic scattering events. The usual method of eliminating -these events is to place an absorber foil in front of the detector. This foil is designed to be just thick enough to stop all elastic scattered particles but thin enough to pass the nuclear reaction particles needed for the NRA analysis. The use of such absorber foil has two fundamental problems. One is that the presence of such a foil results in straggle of the NRA particles and the resultant loss of dept resolution. The second is that it eliminates most nuclear reactions as possible NRA probes. Only those nuclear reactions yielding particles with ranges much greater than the range of the elastically scattered particles can be used.

A magnetic spectrometer provides a natural way to avoid these problems. The dispersion of the particles gives the high momentum resolution needed for the analysis, while the solid state detector can discriminate between the large energy difference between different charge states, and thereby uniquely determine the energy of the incoming particles. Hence, it will be possible to measure even rather weak nuclear reaction particles in the presence of intense elastic scattering. Further, these NRA particles -will be detected with much higher energy resolution leading to much better depth resolution.

This ability of a magnet to physically separate elastic particles from others also has very important applications in energy recoil detection (ERD) where light elements are profiled by bombarding the target at an angle and measuring the light ion recoiled out of the target by these incident ions. This technique is commonly used for hydrogen profiling using an absorber foil in front of the detector to eliminate the intense elastic scattering events.

Choosing an appropriate magnet field or placing a solid block in front of the focal plane detector to stop elastic scattered particles, will allow ERD measurements with much better depth resolution and sensitivity than is currently possible in systems utilising surface barrier detectors. In good cases, this approach should provide orders of magnitude improvement in the sensitivity of hydrogen analysis over any other currently available technique.

B. Summary of the Invention

The invention sets out to solve the problems of insufficient resolution that the prior art devices and methods suffer from. The problem is solved in a first aspect by an improved method as defined in claim 1, and in a second aspect by a device for high resolution materials analysis as defined in claim 15.

In a third aspect of the invention there is provided an apparatus for enabling the handling of samples in the ultra-high vacuum required inside the sample chamber of the device of the invention. In this aspect of the invention, as defined in claim 4, a manipulating system is provided which allows manipulation of a sample or target under ultra-high vacuum of the order of 10"¹⁰ mbar. It allows fast loading of many samples. Prior art systems have required extended periods of outgassing by pumping to achive the desired vacuum level, before measurements could be performed, thereby enabling only just a few measurements per day. With the invention it is also possible to scan the samples to allow lateral mapping thereof.

Further scope of applicability of the present invention will become apparent from the detailed description given hereinafter. However, it should be understood that the detailed description and specific examples, while indicating preferred embodiments of the invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become apparent to those skilled in the art from this detailed description.

The present invention will become more fully understood from the detailed description given hereinbelow and the accompanying drawings which are given by way of illustration only, and thus not limitative of the present invention, and wherein

Fig. 1 is a schematic view of a system according to the invention;

Fig. 2 is top view of the magnet spectrometer of the invention with the top half lifted off;

Fig. 3 is a cross section along the line A-A in Fig 2;

Fig. 4 is an overview partly in section of the sample manipulating system;

Fig. 5 illustrates a sample holder;

Fig. 6 illustrates the sample holder and transfer rod assembly;

Fig. 7 shows a cooled version of the sample holder; and

Fig. 8 is a graph demonstrating the improvment in depth resolution achieved by the invention.

Detailed Description of Preferred Embodiments

Referring now to Fig. 1 there is shown in block form the basic elements of a system according to the invention for high resolution materials analysis.

The system comprises a particle beam source 1, a vacuum chamber 2. Located inside said vacuum chamber there are a sample (or target) manipulating system 3, a magnet 4, and a focal plane detector 5. Each component will now be described separately in detail.

The particle beam source may be of various kinds already mentioned in the discussion of background of the invention, and does not form part of the invention per se. An illustrative example ^"thereof is a van de Graaff accelerator.

Any position sensitive detector having sufficient spacial resolution (better than 0,1 mm) known and commercially available may be utilized in the invention, and thus the detector does not form part of the invention per se.

A key factor for the success of the present invention is the sample manipulating system generally designated with reference numeral 3. Previously known systems have been operated such that samples have been loaded in the vacuum chamber under atmospheric pressure, and thereafter vacuum has been created. This has several drawbacks. Primarily it becomes substantially more time consuming to reduce pressure to the desired level, i.e ultra-high vacuum of the order of 10^-1^ mbar, if at all possible. This is i.a. a consequence of moisture adhering to the surface of the vacuum chamber.

Thus, the manipulating system of the invention makes it possible 1) to load a plurality of samples under dry nitrogen gas,- thereby reducing pumping time, and 2) to maximize the pressure exposure of the detectors to 10 mbar maximum.

The manipulating system 3, an overview of which is shown in Fig. 4, comprises three main parts: a loading chamber 6, an ultra-high vacuum valve 7 connected to said chamber 6, and an analysis chamber 8, wherein the samples are subjected to ion beams.

The loading chamber 6 may be placed in a gas purge chamber for enabling purging with dry nitrogen from a suitable source.

The sample holder 9 (Fig. 5), provided with a plurality of samples 10 is placed inside the loading chamber 6. The sample holder comprises a sample locating strip 11, made of e.g. steel or copper. The latter material is used if cooling is desired. Along the sides of said strip 11 there are spring support rods 12 attached to the strip 11 by screwing. The support rods 12 support attachment springs 13, by means of which the samples are held in place on said locating strip 11. The strip 11 is provided with a laterally extended top portion 14. Said top portion 14 is mounted inside a cylindrical bayonet member 15. Said bayonet member 15 has a bottom opening the diameter of which corresponds to the width of the strip 11, and is narrower than the lateral extension of said top portion 14. A spring element 16 is provided between the bottom of the bayonet and the top portion 14, thus providing a springing function of the assembly.

The bayonet member 15 is provided with two diametrically arranged, essentially L-shaped recesses 17. These recesses are matched by two pins 18 provided on a manipulator member 19, for manipulating the samples in the analysis chamber, i.a to optimize the position thereof (the manipulator will be described further belo ) .

Loading of samples is carried out as follows.

The sample holder 9 including the bayonet member 15 is placed inside the loading chamber 6. A fork shaped transfer rod^'-20 (Fig. 6) , extending inside the chamber through the bottom thereof via a vacuum seal, and having a pusher member 20' that can be actuated from the outside, is pushed upwards until it engages recesses 21 in said bayonet member 15. Each leg 22 of the fork 20 is provided with a latch means 23. Said latch means 23 comprises a protrusion 24 from the top end of each leg. Said protrusions 24 are essentially T-shaped, thereby forming hooks 25, 26. When the fork (or transfer rod) has been pushed as far as possible into said bayonet 15, the fork is turned slightly, such that the hooks 25, 26 extend over the inner bottom surface 27 of the bayonet member 15. Thereby a locking function is achieved. When the sample has been properly located inside the loading chamber 6 it is evacuated. When the desired pressure has been arrived at, the valve 7 is opened. The transfer rod and fork 20 is pushed upwards thus bringing the sample holder assembly with it until the sample holder is positioned in the ultra-high vacuum region UHV. In this position the recesses 17 of the bayonet member 15 match the pins 18 of the manipulator 19. By turning the sample holder assembly by means of the transfer rod 20 the pins 18 will be locked in the L-shaped recesses 17. When this operation is completed it is a simple matter to release the T-shaped hook members 25, 26 from their engagement in the bayonet member 15, by turning them to a position where they can be completely withdrawn from the assembly and down into the loading chamber, wherefter the UHV-valve 7 may be closed again.

The manipulator 19 may now be operated to position the sample holder accurately in the X-Y-Z directions. It may be driven by an electric motor and/or manually operated. The mechanism is located inside a metal bellows in order to render the device sufficiently leakproof.

Now the magnet 4 will be described in detail.

The magnet 4 shown in Fig. 2 is a split pole spectrograph. Jt is based on a spectrograph designed and described by Harald^'-A. Enge in 1964 (Nuclear Instr. and Meth. 28, 126, 1964). The principal design of the magnet of this invention is the same as said known device, and comprises two poles mounted into a common return yoke 33. Particles scattered from a sample or target 10 are first bent or deflected in a first dipole, Dipole I, and then in a second dipole, Dipole II. The pole geometry is such that all particles with the same momentum are focussed into one point on the focal plane. Particles with different momenta hit different spots, and by detecting the point where a particle hits the focal plane, the momentum is also derivable. An energy spectral range Emax:Emin of 8:1 can be detected simultaneously in this way. Although it is preferred to use two dipoles in order to achieve the desired versatility, it is possible to provide the spectrometer with only one dipole, and thus both single and double dipoles are within the scope of the invention. 5

The pole edges are furthermore designed in terms of their shapes (see the article by Enge, supra), such that vertical focussing is obtained and such that 1st and 2nd order aberrations are nearly eliminated. The magnets are excited by one single pair of coils -vO surrounding the two poles.

At the entrance of dipole I Dl, there is an extra steel edge 36 referred to as a "field clamp". It helps shaping the rather sharply curved field in that area. 15

The two poles of the magnet are mounted into one common return yoke 33. They are also excited by one single pair of coils 28 surrounding the two poles.

0 The poles Dl, D2 and parts of the yoke 33 are machined out of just two steel plates 30, each plate constituting one half of the assembly. In this way the technical requirements for ion beam analysis applications are satisfied, namely 1) the geometric tolerances must be very tight, and 2) the vacuum in the system 5 should preferably be in the order of 10~¹⁰ mbar, although^" the principle of the invention is operable at higher pressures as well, such as 10~° or even 10~" mbar. At higher pressures contamination of targets becomes a problem and accuracy will deteriorate rapidly. The mechanical tolerances will be limited 0 only by the precision offered by the milling machine used. The pole surfaces and the adjoining yoke surfaces will also form the surfaces exposed to vacuum. The steel sufaces are metal plated to give them outgassing properties similar to stainless steel. A preferred form of plating is chemical Nickel plating, which 5 renders the surface substantially non-porous. All vacuum surfaces are outgassed by baking. In Fig. 2 there is shown one half of the the vacuum chamber assembly viewed from above.

As mentioned above each half is machined out of a single steel block. The "topology" of the machined surface is indicated~-by depths given in millimeters, measured from the zero-level, i.e. the surface of contact between two blocks. The surface profile is clearly seen in the cross section in Fig. 3, which is taken along the line A-A in Fig. 2, and shows two halves mounted together to form a complete assembly.

The coil 28, also indicated with an "X", is placed in a recess 29 that is machined in the steel plate 30 from the back side. On the interior side a recess 31, corresponding to the recess 29 is machined in the steel, thereby forming a thin material section 32, joining the dipoles (Dl and D2), and the return yoke 33. This is a key feature of the system, and will be explained below.

Dipoles Dl and D2 are formed by removing material in the steel plate such that the desired shapes are created. The pole geometry as such is known. However, as indicated above, instead of having an air gap between dipoles and return yoke, or making that portion of a non-magnetic material, this portion 40 is made of iron. The material becomes magnetically saturated already at low excitation of the magnet, but it is made so thin (2 - 4 mm, preferably 3 mm) that the extra magnetic flux through this section is easily handled by the return yoke. The flux pattern in the useful magnet gap is practically unaffected.

There are two main advantages with this design. First it becomes possible to manufacture with high precision each half of the spectrometer from single blocks of steel, avoiding extra details of other materials. Secondly, this is beneficial in ultra-high vacuum application, since i.a. it provides an all metal vacuum system.

The detection chamber 34 comprises a focal plane detector 35. A 15 suitable detector must have a spacial resolution better than 0,1 mm. Such detectors are commercially available and it is within the competence of the skilled man to select a suitable detector.

In an embodiment there may be provided an electrostatic deflection device in front of the detector to electrostatically separate differently charged particles. This may simply be a couple of plates arranged on opposite sides of the detection plane.

The sample is positioned in the circular chamber 37. The manipulating system is located such that it extends into said chamber.

There is a channel 38 for the ion beam from e.g. an accelerator, directed towards the sample chamber. A further channel 39 is provided for inspection purposes, such that the target may be monitored by television.

On the outside of the assembly of the two halves there is welded a stainless wraparound thin wall to complete the vacuum chamber.

In Fig. 8 there is shown a graph illustrating improvements in depth resoulution achievable by using the invention.

The invention being thus described, it will be obvious that the same may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the invention, and all such modifications as would be obvious to one skilled in the art are intended to be included within the scope of the following claims.

Claims

CLAIMS :

1. A method of determining material properties by ion beam scattering analysis, comprising

i) subjecting a target sample to a particle beam;

ii) collecting scattered particles from said target sample on a detector;

c h a r a c t e r i z e d b y

iii) letting backscattered particles from said target pass a magnetic field to variably deflect said particles depending on their momentum,

iv) detecting said deflected particles by means of a position sensitive focal plane detector, and

v) wherein said magnetic field is applied in such a way that ions having the same momentum are focussed on the same point on the detector.

2. The method as claimed in claim 1, wherein only light ions recoiled from said target by incident ions are detected, and elastically scattered particles are not collected on the detector.

3. The method as claimed in claim 1 or 2, wherein the analysis operation is performed under ultra-high vacuum of at least 10^~8 mbar, preferably 10^~10 mbar.

4. A sample manipulator for use in ultra-high vacuum material analysis and designed to be connected to or integrated with an analysis chamber (8) of an analysis apparatus, comprising 17 a) a loading chamber (6) for loading samples (10) at atmospheric pressure;

b) an ultra-high vacuum valve (7) connecting said loading chamber (6) and said analysis chamber

(8);

c) a transfer means (20, 20' ) for moving a sample holder assembly (11, 12, 13, 15), having a plurality of samples (10) placed thereon, from said loading chamber (6) into said ultra-high vacuum analysis chamber (8).

5. A sample manipulator as claimed in claim 4, said sample holder assembly comprising a sample locating means (11, 13), a bayonet member (15) resiliently connected (16) to the upper end (14) of said sample locating means (11), a transfer means (20, 20') comprising means (23, 24, 25) for engaging said bayonet member (15, 21, 27), said transfer means (20, 20') extending through the bottom of said loading chamber (6) via a vacuum seal.

6. A sample manipulator as claimed in claim 5, said transfer means (20, 20') having the shape of a fork with two legs (22), each having a said engaging means (23, 24, 25), said legs (.22) extending in the longitudinal direction of said sample locating means (11).

7. A sample manipulator as claimed in claim 5 or 6, wherein said sample locating means (11) is made of a material with good heat conductivity such as copper for cooling purposes.

8. A sample manipulator as claimed in any of claims 4-7, comprising means for providing a blanket of dry inert gas during loading of samples.

9. A magnetic spectrometer device having a magneto-optical system for focussing particle beams onto a position sensitive focal plane detector; said system comprising at least one dipole (Dl, D2), energized by a coil means (28) arranged around said dipole(s), and a return yoke (33), c h a r a c t e r i z e d i n t h a t said dipoles (Dl, D2) and said return yoke (33) are joined by a sufficiently thin metal bridge (40) such that said bridge becomes magnetically saturated already at low excitation of the magnet.

10. The magnetic spectrometer device of claim 9, comprising a J position sensitive detector (5) .

11. The magnetic spectrometer device of claim 10, wherein said dipole(s) (Dl, D2) are such that ions having the same momentum are focussed onto the same spot on the detector (5 ) .

12. The magnetic spectrometer device of any of claims 9-11, comprising two halves, each half being machined from one piece of solid steel, and wherein the vacuum surfaces of said metal are metal plated.

13. The magnetic spectrometer device of claim 12, wherein said plating is a chemical Nickel plating.

14. The magnetic spectrometer device of any preceding claim, wherein there is an electrostatic deflecting device in front of the detector for electrostatically separating differently charged particles.

15. An apparatus for high resolution analysis of concentration profiles in layered compositions, comprising

a) a particle beam source (1 ) ;

b) a sample or target manipulating device as defined in claim 7;

c) a magnetic spectrometer as defined in claim 12 arranged to detect ions scattered from a target (10) exposed to a particle beam from said particle beam source (1);

said manipulating device and said magnetic spectrometer being enclosed in an ultra-high vacuum housing.