Development of a Nuclear Microprobe and Its Application to Neurobiology

DEVELOPMENT OF A NUCLEAR MICROPROBE AND ITS APPLICATION TO NEUROBIOLOGY

Staffan Tapper

Civilingenjör, Helsingkrona Nation

Akademisk avhandling för avläggande av teknisk doktorsexamen vid tekniska fakulteten vid I' Lunds Tekniska och Naturvetenskapliga Högskola, kommer att offentligen försvaras i fysiska institutionens föreläsningssal B, fredagen 19 maj 1989, kl 14.00. Development of a Nuclear Microprobe and its Application to Neurobiology

Staffan Tapper

Department of Nuclear Physics, Lund, 1989 Contents

Abstract

Development of a Nuclear Microprobe and its application to neurobiology 5

Summary of the papers 12

I Achromatic ion-beam focusing using a low aberration 17 quadrupole with symmetric focusing properties

II A computer-controlled magnetic post-lens scanning system for the Lund proton microprobe 33

III Enhanced beam quality for proton microprobes using ultrathin stripper foils in tandem accelerators 43

IV Characterization of the response function from a Si(Li)-detector using an absorber technique 49

V Calibration of a proton microprobe set-up 65

VI Accumulation of calcium in substantia nigra lesions induced by status epilepticus - A microprobe analysis 69

VII Elemental regional distribution in human brain tumours; PDCE analysis of biopsy and autopsy samples 83

V VIII Trace elements in astrocytomas and surrounding brain; Correlation to malignancy and survival 89

Acknowledgements 107

,. Organization Document name LUND UNIVERSITY DOCTORAL DISSERTATION Department of Nuclear Physics Date of issue Sölvegatan 14 April 2S, 19RQ S-223 62 LUND, Sweden CODEN: LUTFD2/(TFKF-101?)/1-106/(19891 Authors) Sponsoring organization Staffan Tapper

Titte and subtitle Development of a Nuclear Microprobe and its Application to Neurobiology

Abstract

A Nuclear Microprobe has been developed at the Pelletron accelerator in Lund. The design of the achromatic beam focusing system as well as of the beam scanning system is described. The focusing system consists of three magnetic- and three electrostatic quadrupole lenses together forming an achromatic quadnipole triplet with symmetric focusing properties. The beam quality from the accelerator has been improved by use of ultrathin stripper foils.

The Nuclear Microprobe set-up has been applied to investigations of brain tissue specimens. The elemental disorder following epileptic seizures has been studied by micro-PIXE analysis. A combination of macro- and micro-PIXE analysis have been utilized in an investigation of elemental differences between normal human brain and human brain tumours. In the context of the quantification procedure in micro-PIXE analysis, calibration and X-ray detector i response function are discussed. i- 31

eyw Nuclear Microprobe, beam focusing, achromatic focusing, beam scanning, beam quality, micro-PIXE analysis, macro-PIXE analysis, quantitative analysis, hram tissue, brain tumour

Classification system and/or index terms (if any)

Supplementary bibliographical information Language F.n p 1 i Q\\

ISSN and key title ISBN

Recipient's notes Number of pages Price v^^ 106 Security classification

Distribution by (name and address) Department of Nuclear Physics Sölvegatan 14 I, the undersigned, being the copyrighröwnw oi me^iwrac^ÖTTn? above-mentioned dissertation, hereby grant to all reference sources permission to publish and d'swmiiute the abstract of the above-mentioned dissertation.

Signature *•' '-— \ Date April 25, 1989 Development of a Nuclear Microprobe and its Application to Neurobiology

1. Introduction

During the last decades, several micro analytical techniques. The micro- analytical methods have been developed beam set-up is often referred to as a based upon the high energy ion-beams Nuclear Microprobe (NM). from electrostatic accelerators. When a The principles of the NM are similar heavy-ion beam of a few MeV/u irradiates to that o^ the electron microscope (EM) a target, atomic and nuclear reactions are which has made important contributions to induced and the emitted radiation from the modern science mainly by its excellent reactions can be detected by different types capabilities to imaging small structures in of detectors. By measuring the the energies samples. The electron microscope can also of the emitted photons or particles, be used for elemental analysis by detecting information about the constituents in the the characteristic X-rays produced in the target can be obtained. If the ion-beam is target atoms by the electron beam. The compressed to micrometre dimensions, detection limits reached by this analysing these analytical methods can be used on method are of the order of mg/g. The rea- the micro-scale at the target. son for the high detection limit is that the The problem of obtaining the small- incident electrons produce an intense con- est possible ion-beam focus was initiated tinuous bremsstrahlung which obscure by Cookson in 1972 (1) and has since then weak-intensity peaks of characteristic X- attracted increasing interest among scien- rays. The bremsstrahlung decreases with tist working in ion- beam analysis. Micro- increasing mass of the impinging particle, analysis can add information into various which make heavy ions a straightforward fields of science e.g. investigation of way to lower the attainable limits of detec- elemental composition of biological cells tion. with the possibility to reveal trace element Compared to the EM, the proton distributions within single cells. The in- microscope, i.e. NM, does not produce creasing demand for characterization of such high quality images of the sample produced materials in material science is surface, equivalent to these of the EM but another application of microanalysis which the detection limits for X-ray analysis are will become more important. significantly better, typically by two orders The fast progress in this scientific of magnitude. In addition, heavy ions can area has partly been facilitated by the cause nuclear reactions among the target decreasing interest for small accelerators atoms and these nuclear reactions can give from scientists in the field of pure nuclear additional information as to the compo- physics. After years of declining import- nents of the sample. In its ability to detect ance these laboratories now experience a and quantify elements with high sensitivity febrile activity from scientists working on the micro-scale, NM has virtually no with application of science, e.g. competitor. PART I: The Nuclear Microprobe by a second larger collimator which limits Instrument the beam divergence from the object. This beam angle aperture limits aberrations in focusing due to vanishing with angle. The lens system then transfers the object to the 2. The Nuclear Microprobe set-up target plane, demagnified by a factor of 5- 100 depending on the particular lens syst- A schematic drawing of the NM set-up in em used. After the lenses are situated the Lund is shown in Fig 1. beam scanning system, which enables movement of the beam focus relative to the target. Different detectors are situated in the target chamber to detect the radiation produced by the ion beam. Focusing and deflection of the beam in the NM is a considerable problem due to the rigidity of high-energy heavy-ions. For ACCELERATOR focusing, quadrupole lenses have mainly been used. They yield field gradients high enough to bend the beam but their team optical properties are rather compliciited. However, the behaviour of charged pj nic- ies moving in electromagnetic fields is collimators well-known from Maxwell's laws, making it possible to investigate focusing properties by using computer programs to simulate the behaviour of the ion-beam. Magne- tic fields are more powerful in bending high energy beams, but electrostatic fields may sometimes be advantageous, e.g. for focusing the reasons discussed in publications no. 1 lenses and 2 in this thesis. With current high-resolution nuclear target microprobes, the spatial resolution attain- chamber able is limited by aberrations in the focusing of the object collimator. These are usually divided into two categories intrin- Fig 1. Schematic drawing of the Lund NM sic aberrations and parasitic aberrations. Intrinsic aberrations consist of spherical and chromatic aberrations, whereas parasitic aberrations are focusing distortions The NM set-up is placed at an acce- caused by lens misalignment, distorting lerator beam-line after the analysing stray fields, non-perfect lenses etc. While magnet. The beam is collimated by a slit intrinsic aberrations define the ultimate system, the object collimator, with an limits of resolution in NM, in most systems aperture of about 0.02-0.10 mm in both it is the chromatic aberrations which are directions. The function of the accelerator responsible for the dominant contribution. is to accelerate the ions and transfer an It is possible to construct achromatic image of the ion-source to the object colli- lenses and lens systems with non-spherical mator with minimum beam quality degra- focusing properties by using complex dation. The object collimator is followed combinations of magnetic and electrostatic lenses in the former case and by the use of (Nuclear Reaction Analysis). Used to- quadrupole and octupole lenses in the gether, these analytical techniques facili- latter. However, with increasing system tate analysis of essentially the entire complexity, parasitic aberrations producing periodical table simultaneously, making beam distortions larger than the corrected NM a very powerful tool for micro- aberration can be introduced. Corrections analysis. for chromatic aberrations and spherical aberrations in NM have been reported, but l)PIXE-analysis so far no improvements in probe resolution arising from these corrections have been Charged particles ionize atoms by reported. During irradiation, the target interacting with the inner-shell electrons emits radiation which can be detected in producing X-ray photons characteristic for suitable detectors. When an event is detect- the ionized atoms. The X-ray photons pro- ed, the following information has to be duced are detected by a Si(Li)-detector. transferred to the data aquisition system: The PIXE technique yields data on all detector no, energy of particle, and the x- elements heavier than Mg simultaneously and y-coordinates of the target (D,E,X,Y). with detection limits ranging from 0.1-10 A common counting-rate from a detector is ppm in thin targets. Due to limited ion 1000 cps which fills 1Mb memory in 40 currents, PIXE-analysis with NM often seconds, if four detectors are used simul- gives slightly higher detection limits. taneously. To handle this amount of data, the capacity of modern computers are re- 2) PESA quired. If all events are stored consecuti- vely row, then the time is also recorded. Some of the high energy ions pene- The data-acquisition system used at trating the target come close enough to the Lund NM enables on-line elemental atomic nuclei to interact with their Cou- mapping to be presented at a computer lomb fields. If the ions are scattered in the terminal and off-line data-sorting followed backward direction, they can be detected by spectrum evaluation at chosen irradia- by a surface-barrier detector and their tion spots (pixels). A computer controlled energy reveals the mass of the target scanning system is almost a necessity be- nucleus. The detector wil! only be able to cause of the long irradiation-times required resolve the energy for the lighter elements in micro-beam analysis. If beam scanning making this analysing technique suitable is completely software controlled, a scan- for combination with a simultaneous ning pattern exactly matching areas of PIXE-analysis. interest in the sample can be programmed thereby mimimising the exposure-time de- 3) FSA voted to non-interesting structures. Forward scattering is used to detect the hydrogen content in the target. After the interaction between the ion (e.g. a 3. Analytical methods at NM proton) and the target hydrogen nucleus, both can be detected if they are not stopped The analytical methods which have in the target. If both protons are detected been used at the Lund NM are PIXE ana- by use of a coincidence technique, the lysis, Particle Elastic Scattering Analysis background in the spectra can be eliminat- (PESA) and Forw;ird-Scattering Analysis ed. PESA and FSA are mainly used for the (FSA). In addition to these techniques, it is determination of the hydrogen, carbon, possible to use e.g., PIGE (Panicle nitrogen and oxygen contents in the Induced Gamma ray Emission) and NRA sample. 4) PIGE and NRA could perhaps assist in orienting the sample to be analysed. If protons are to be PIGE and NRA are in principle simi- used for imaging, the method requires the lar techniques. In both cases the high spatial resolution at the ion probe to energy ion penetrates the Coulomb barrier increase with decreasing proton beam cur- of the atomic nucleus resulting in different rent. Otherwise, the image produced is types of nuclear reactions. The disturbed similar to mass- density mapping using nucleus can produce either charged partic- backscattered particles. In biological les which can be detected by surface samples, there may not be contrast for the barrier detectors or gamma-ray photons de- STIM technique to be applicable. tectable e.g. by a Ge(Li)-detector. These analytical methods are also suitable for combination with the PIXE technique. Due to the proportionality of the Coulomb PART II: Quantification barrier height to the atomic number, Z, the efficiancy of ion-probes falls off with increasing Z. Hence these analysing methods detect mainly light elements, not detect- 5. Quantitative elemental micro analysis able by the PIXE technique. One of the distinct advantages with micro-PIXE analysis at a NM is the possibility of quantifying the detected elements 4. Sample imaging techniques accurately. The method has a high sensitivity with detection limits in the range 1-10 One particular problem in analysing ppm depending of element to be analysed with a high resolution NM is that the and matrix composition of the sample. In a structures in the sample to be analysed spot analysis, the spectrum evaluation may be smaller than can be resolved opti- procedure is analogous to a macro-PIXE cally, making it difficult for the NM analysis. The X-ray spectra acquired are operator to orientate on the area to be evaluated by a computer program making a analysed. The beam-probe must then itself linear or non-linear fit to the continuous be used to produce an image of the sample background and the characteristic peaks in to be analysed. An obvious way is to use the spectra. If the analysed sample is not elemental mapping of the produced X-rays, thin compared to the range of the ions, or but it is preferable to use a much faster not thin relative to the attenuation of the method. X-rays produced in the irradiated sample, As in electron microscopes, the thick-target corrections are also required. secondary electrons produced by ion bom- The evaluation of a single spectrum bardment of the sample surface can be takes a few minutes, making this procedure detected, resulting in an image of the cumbersome for the large numbers of spec- topography at the sample. However, for tra produced in a scanning microanalysis analysis, thin samples with flat surfaces are of a single sample. It is thus impossible to mainly used and they do not give much use this strategy in calculating entire contrast in a secondary electron image. elemental maps consisting of thousands of Transmission images, e.g. Scanning spectra. In addition, if trace elements are to Transmission Ion Microscopy (STIM), is be detected, it takes several minutes to adequate for imaging thin targets. Imaging acquire a single spectrum. Hence, it is with STIM has to be performed with ion- impossible to produce large elemental currents too small for analysis but if the maps of trace elements in the sample. Even STIM-imaging precedes the analysis, it after several hours of mapping the acquisition-time for each pixel spectrum is very secondary electrons, where Tm represent limited and the spectra contain too poor the maximum energy transferred from the statistics to be evaluated by a computerized impinging particle (e.g. a proton) to a free fitting program. A rule of thumb is that electron. each hour of irradiation of 64x64 pixels at The few characteristic peaks from the sample corresponds to about 1 s the organic matrix in the spectra is acqusition time per spectra. From this beneficial, because they thereby do not figure follows that 16x16 pixels yields 16 s interfere with the peaks from trace ele- per pixel for each hour of irradiation, ments embedded in the matrix. The detec- which is close to an upper limit of numbers tion limits in a PIXE analysis at a proton of pixels if trace elements are to be detect- energy of 2.5 MeV, a common energy in ed. PIXE analysis, are lowest in the region of In microprobe analysis, the damage the spectra where calcium (Z=20) to induced by the ion beam should also be bromine (Z=35) are situated. Among these monitored since the ion-current concen- elements are many of the "essential trace trated in a micrometer spot yields very elements" of biological tissues, which high local current densities. makes the PIXE technique suitable for In conclusion, elemental maps show analysing them. the distribution of the major elements, but If the matrix composition is of im- not quantitatively, and from these maps the portance, e.g. for determination of the irra- operator can zoom in on finer structures to diated mass, it is possible to measure the make a fully quantitative analysis by spot H,C,N and O content by detecting protons analysis, line analysis or small area scans, elastically scattered from the atomic nuclei where the number of spectra acquired of these elements. This also gives the should not be so large as to make spectrum opportunity of modifying the amounts of evaluation impossible. From these spectra, trace elements determined due to matrix the trace-element concentrations can also effects i.e., thick target corrections. When be determined. the irradiated mass is measured, determined elemental amounts can be quoted as concentrations, which is often desirable.

PART III: Analysis of Brain Tissue

7. Micro analysis

6. Analysis of biological specimens With micro-PIXE analysis, additional information about the tissue can be ob- For various reasons, the PIXE tained. The focused ion-beam of diameter method is particulary suitable for analysing 1-10 nm is about the same size as cells or biological specimens. Biological material even as organelles. Hence the high- has a matrix consisting of hydrogen (H), resolution probe can monitor individual carbon (C), nitrogen (N) and oxygen (O) cells or cell layers in situ, in thin sections where characteristic X-rays have too low of frozen and freeze dried tissue speci- energy to be able to reach the X-ray de- mens, or cells cultured in vitro. It may tector. Hence the matrix only produces a even be possible to analyse structures with- continuous background in the spectra re- in the cells. However, problems with corded, on which are superimposed heavier sample preparation are encountered, e.g., elements. The background caused by the to produce dry samples with the in vivo matrix in the region 1.0 - Tm keV is intracellular gradients still intact. mainly bremsstrahlung from scattered 10

A micro probe analysis also provides A good analysing strategy would be the opportunity to reveal the distributions to start the analysis with elemental map- of the determined elements, i.e., elemental ping of the major elements, for orientation mapping technique. In Fig 2 is presented on the sample, followed by spot analysis, an elemental map of aluminium and sili- or smaller area mapping, of the most con. The map is from an analysis of brain interesting structures. With this smaller tissue taken during autopsy of a patient numbers of spectra with good statistics with a diagnosis of Alzheimer's disease. traditional evaluation procedure for PIXE- The major difficulties encountered in spectra can be performed leading to fully producing elemental maps is that they are quantitative results, even for elements non-quantitative and it is not possible to present only in small concentrations, i.e., determine the concentrations of trace ele- trace elements. ments over large areas in the sample. The However, in the example in Fig 2, resulting elemental intensity map may pro- the Al detected would not have been duce false results, e.g. interferences in the possible to determine in a macro-PIXE spectra can simulate a non-present ele- analysis of the whole sample. Micro-PIXE ment. analysis can consequently detect local

AL * 2.7B2 SI * 1. Element 1 ElEirent £

496 549 605 G54 707

Fig 2. Elemental distribution of Al and S, relative scale, in a thin section of tissue from an

Alzheimer's patient. The detected "hot spot" is the senile plaque. *.--••?. 11

"hot-spots" of an element which would rum fitting procedures should, if possible, have remaind not detectable in a macro- be avoided due to the problems which PIXE analysis with better detection limits. occur at spectra with large statistical spread. With the possiblity to produce quantitative elemental maps at small structures in the sample the nuclear microprobe 8. The future NM will become an even nore important instrument in modern science. As long as the resolution ultimately attainable using NM continue to increase new fields of application open up and there Reference will probably be continuous effort towards higher resolution. The indications are that 1) J.A. Cookson and F.D. Pilling the best resolution achieved today is main- J. Radioanal. Chem. 12(1972)39-52 ly limited by chromatic aberrations in the focusing of the ion beam delivered by the electrostatic accelerator. There are two ways to deal with this problem: one is to increase the stability of the terminal volt- age of the accelerator and the other to develop high quality lens systems with achromatic focusing properties. There is also a third way which is to develop higher brightness ion-sources, enabling further collimation of the ion- beam in the aperture collimator and hence suppression of the chromatic aberration due to its dependence on the angle of the beam diverging from the object collimator. The achromatic focusing system developed in Lund requires higher brightness ion- sources to transmit ion currents sufficient for analytical purposes in a scheduled 100 nm ion-probe. To be able to use a NM with a probe resolution below 1 micron, imaging techniques have to be further developed. Seve- ral processes induced by ion-beams in samples which may permit information about sample structures to be obtained have not yet been investigated. Above has been mentioned that analysis of 16x16 pixels is close to the upper limit for detection of the trace elements in the sample. However, for quantification from such amounts of spectra, from a single sample, the evaluation procedure has to be simplified. The evaluation has to done by computer programs without any interactive part. Spect- 12

Summary of the publications The beam-scanning system was designed to be inserted between the focusing lenses and the target chamber. Compared to pre-lens deflection which influences the focusing properties, this position is to be prefercd. However, with Achromatic ion-beam focusing using a low the short distances between a post-lens aberration quadrupole triplet with scanning system and the target plane, symmetric focusing properties strong electromagnetic fields are required. The Lund system use ferrite-cored A vital part in a Nuclear Microprobe scanning coils which permits large areas to is the lens system which focuses the ion- be scanned (2x2 mm2) with high scanning beam onto the target. This paper describes frequencies (residence time/pixel=100 |is). an achromatic quadrupole triplet system. The lens system consists of six quadrupole lenses, three magnetic and three electrostatic. If excited properly, a combination of Paper 3 an electrostatic and a magnetic lens can act as a quadrupole lens insensitive to the Enhanced beam quality for proton energy spread in the ion beam i.e. chroma- microprobes using ultra-thin stripper foils tic aberrations are cancelled out. in tandem accelerators To reach higher resolution in the ion- probe the lenses must have pure quadru- The Lund Nuclear Microprobe set-up pole fields with also minor distortions of has been developed at the tandem higher order components. The quadrupole accelerator in Lund. As microprobes, the field quality has been investigated using major drawback of tandem accelerators is the grid-shadow method and the results beam degradation in the stripping stage at will be presented in another paper. The the terminal. We have investigated the achromatic focusing properties were also beam degradation caused by ultra-thin demonstrated by the grid-shadow method. carbon foils (0.5 fig/cm^) and compared This system may be a step towards higher the results with thicker foils and gas resolution in Nuclear Microprobes. stripping. The results show that ultra-thin carbon foils are preferable even when compared to gas stripping, at least for accelerators without terminal pumping. Paper 2

A computer-controlled magnetic post-lens scanning system for the Lund proton Paper 4 microprobe Characterization of the response function A beam scanning system is a from a Si(Li)-detector using an absorber prerequisite at a Nuclear Microprobe. technique Beam scanning enables imaging of the sample and prevents thermal equilibrium A new X-ray spectrum evaluation in the irradiation spots from being reached. program is under development at Lund. If Computer-control of the scanning system high precision and accuracy in elemental makts possible any desired scanning determination are to be reached, the pattern, which is very important for characteristics of the detector have to be minimizing the time of analysis. known. To investigate the response 13

function of a Si(Li)-detector the spectra Paper 7 should be free from interfering background. We have evaluated a procedure for Elemental regional distribution in human extracting the tail from continuous brain tumours - PIXE analysis of biopsy background in the spectra. The method and autopsy samples utilizes the different response of the tail compared to the background when an Since malignant human brain absorber is placed between the radiation tumours are almost always lethal, new source and the detector. The results are in therapeutic treatments are been looked for. agreement with measurements obtained Differences in trace element concentrations using monochromatic X-rays. between normai and malignant tissue may give clues for new treatment approaches. We investigated the differences between normal and pathological human brain. Paper 5 Macro-PIXE analysis revealed elemental differences for different tissues. Iron, Calibration of a proton microprobe set-up calcium, zinc and selenium concentrations were higher for tumour tissue. Micro-PIXE The homogeneity of commercially analysis in the border zone between the available standards, at the micron scale, edge of the tumour front and normal brain were investigated. The standard selected tissue showed decreases in zinc and copper for the microprobe calibration showed to in normal tissue just outside the edge of be homogeneous when analysed with a the tumour front. probe size of 10 x 20 \ivnP. The determined amounts were in agreement with a macro- PIXE analysis. Paper 8

Trace elements in astrocytomas and sur- Paper 6 rounding brain: Correlation to malignancy and survival Accumulation of calcium in substantia nigra lesions induced bv status epilepttcus. This work describes an extended in- A microprobe analysis vestigation of the hurran brain tissue described above. The less malignant astro- The calcium and potassium cytomas grade II were included in the alterations in various pan of rat brain material and the results from all analyses regions following epileptic seizures have were evaluated using multivariate statis- been investigated with the Nuclear tics. This analysis showed that different Microprobe. A significant calcium increase types of brain tissue could be separated could be detected in regions which are significantly in multivariate space. By known to show cell death after seizures, using the multivariate statistic method while the calcium concentration remained PLS, it was possible to predict the survival unchanged in regions not affected by the time of the patient from the elemental seizures. concentrations pooled with dry/wet weight ratio and age of the patient. Several elements, e.g. potassium, were found to be correlated with long-term survival. This may be of importance in the search for new treatments of malignant brain tumours.