Proc. Nat. Acad. Sci. USAVol. 72, No. 5, pp. 1826-1828, May 1975

Scanning Transmission Ion Microscope with a Field Ion Source(ion optics/field ionization/microradiography)

W. H. ESCOVITZ, T. R. FOX, AND R. LEVI-SETTI

The Enrico Fermi Institute and Department of Physics, The University of Chicago, Chicago, Illinois 60637

Communicated by Albert V. Crewe, February 24, 1976

ABSTRACT Experiments with a low-resolution scan-ning transmission ion microscope, using hydrogen ionsfrom a field ionization source, indicate that it will befeasible by this approach to aim at high-resolution ionmicroscopy. Micrographs of unstained biological speci-mens have been obtained by critical range absorption of a55 keV hydrogen ion beam at a resolution of 2000 A.

As a first step toward the development of a high-resolutionscanning transmission ion microscope, we have constructedand operated a 65 kV prototype ion-gun scanning transmis-sion ion microscope which accelerates and focuses hydrogenions from a field ionization (1) source. Our aim is to emulateand complement the achievements of scanning transmissionelectron microscopy (2), ultimately at comparable resolution,by taking advantage of the interaction properties of fast ionswith matter (3). The latter include processes such as chargeexchange and molecular ion dissociation which should resultin contrast mechanisms unavailable to the electron micro-scope.We present here experimental data that are relevant to the

feasibility of a high-resolution scanning transmission ionmicroscope and show the first low-resolution images of bio-logical specimens obtained with the prototype instrument.

Previous attempts at using ions, and in particular protons,to make images of small objects with conventional electronmicroscope optics have been reviewed by Grivet (4) and morerecently by Levi-Setti (3). The aim was to obtain higherresolution than with electrons. This was believed to be possiblebecause of the ion's shorter De Broglie wavelength. Thepotential of proton charge exchange as a possible new sourceof contrast was recognized by Chanson and Magnan (5).

Several factors, however, conspired to abort this develop-ment of the proton microscope as a high-resolution and viableinstrument. In our view, the principal limitations originatedin the approach of conventional microscopy itself, where thelarge rate of proton energy loss (10-30 eV/A in traversingorganic materials at 100 keV) causes irreparable chromaticaberration. For this reason, we have chosen the approach ofscanning transmission microscopy, where no optics is affectedby the beam-specimen interaction so that the latter can beused to advantage in producing contrast.The considerations leading to the optimum parameters and

the resulting performance in the scanning transmission elec-tron microscope have been recently reviewed by A. V.Crewe (6) and E. Zeitler (7). We have extended the analysisof these authors to the case of a proton probe (8). The basicproblem consists of optimizing the design parameters toobtain the maximum probe current in the smallest beam spot.How closely the theoretical microscope resolution can be ap-

proached in practice depends essentially on the specificbrightness of the ion (or electron) source, since the probe cur-rent must exceed that required to obtain an intelligiblepicture in a practicable scan time. We consider a microscopeconsisting of a source of radius 6 and an imaging lens of overallmagnification M, spherical aberration coefficient C8, andchromatic aberration coefficient Co. We also assume that thesource yields ions with energy spread eAV and that itsspecific brightness j3 is

,B = I/r2a262V, [1]where I is the primary current, a the half-angle of emission,and V1 the source voltage. As discussed in ref. 6, the probecurrent at the focused spot, at overall accelerating voltage Vand semi-angle of incidence aj, is (nonrelativistically)

Jo = 07r2a,262AP2V [2]

From [2] we obtain an expression for the radius of theGaussian image of the source in terms of probe current andspecific brightness:

rG = 1118 = (Io/3V)i/'(1/irac) [3]

This contribution to the final spot size must be combined withthose arising from diffraction and from spherical and chro-matic aberration. As often done (9), we add these contribu-tions in quadrature and seek the optimum value a0p, of asthat reduces the overall probe radius r to its minimum valueropt. It is known that this approach overestimates the finalspot size; for our present purposes of establishing the feasi-bility of a high-resolution scanning transmission ion micro-scope, however, our result will be in error in a conservativedirection.The resulting expressions take a simple form in limiting

situations:(a) When the effect of chromatic aberration can be neglectedor corrected, we obtain

rop = ro8(1 + E2)3/8 [4]

with

[5]

and

2 10o Io2O3V(0.61X)2 0.30503 [6]

where for protons, X = 0.29/V/V (A), and with Io in protonsper see and 13 in protons sec'-A-2sr-'V-'. The parameter e

here, as in ref. 6, equals the ratio of the radius of the Gaussian

1826

ro,, = 0.63X 3/4C.51/4

Dow

nloa

ded

by g

uest

on

June

14,

202

0

Scanning Transmission Ion Microscope 1827

image Mb to that of the Airy disk (to the first zero) from apoint source. As can be appreciated from [4] and [6], thetheoretical resolution r0, [5 ], can only be approached for verylarge fl.(b) When the chromatic aberration is dominant in the probeformation, as it is the case in a scanning transmission ionmicroscope that uses a field ionization ion source (where eAVcan amount to several eV) (10), the expression for rpgreduces to

rops= roe (1 + E2)'/' [7]

where

[8]

We are now in the position to calculate the source brightnessneeded to reach a certain resolution, if we fix the variables Io,X, C., and C,. In order to obtain a discernible picture with aproton probe, where each proton is contributing a contrastsignal, we estimate the need for a minimum probe current of2.5 X 104 protons per see (500 X 500 point elements, 10protons per point element, 100 see exposure). Using the valuesfor C, and C, (1.2 and 0.8 cm, respectively) attainable with asuperconducting lens of 1.2 cm focal length (3) at 100 kV, wethus estimate, that to reach a probe radius in the high resolu-tion range of 5-10 A, values of # in excess of about 104 protonssec-l-2sr-lV-1 will be needed. Good pictures will requireOs in excess of about 10' proton sec-'A-2sr-'V-1. Thesefigures become comparable with those that apply to thescanning transmission electron microscope (6) when multipliedby a factor of order 102 (for thin specimens, this correspondsto the ratio of proton versus electron cross sections for theprocess originating contrast). The above values of fB areseveral orders of magnitude larger. than those provided byconventional ion sources.The development of a sufficiently bright ion source is,

therefore, the determining factor for the feasibility of a high-resolution scanning transmission ion microscope.We have undertaken to experimentally determine whether,

by the choice of a field ionization source, this requirement canbe met. Our apparatus is schematized in Fig. 1. The opticalsystem is based on the design by Crewe et al. (11) for a fieldemission electron gun microscope. The specific brightness ofthe field ion source for our preliminary operating conditions[H2 pressure not greater than 10-4 torr (0.013 Pa), operatinglimit of the electron multiplier detector] was measured bycounting the particles transmitted by the optical system andassuming a virtual source radius of 3.5 A (1). Typical valuesof # are in the range of 103 protons sec-'A-2sr-'V-1. Wehave verified that the tip current (and hence fl) increaseslinearly with the H2 pressure up to at least 2 X 10-' torr (0.26Pa), so that it should be possible, by differential pumpingbetween the ion source and detector regions, to meet the re-quirements outlined above for high-resolution work.

Field ionization tips, made of (111) oriented W wire, haveprovided stable microscope operation with lifetimes up to 40hr. The resolution in the operating conditions of our prototypemicroscope is limited to 1000-2000 A; this is close to the valuecalculated from the known energy spread of field-ionizedhydrogen (10) and the chromatic aberration coefficient of oursimple optical system. Fig. 2 contains our first micrographsof biological specimens taken with hydrogen ions at 55 kV.

DISPLAYTUBE

FIG. 1. Schematic diagram of 63 kV scanning transmission iongun microscope. The field ionization W tip is cooled by thecryostat to 780K. After evacuation of the microscope to about10-9 torr (about 1.3 X 10-7 Pa), hydrogen is admitted by a Pdleak to reach an operating pressure of 10-4 torr (0.013 Pa). Thetip voltage V1 is in the range 5.5-6.5 kV. The ions are acceleratedand focused by a two-electrode Butler lens (11). Two sets of de-flecting plates provide for scanning and the correction of astig-matism. A large continuous dynode electron multiplier detectsthe transmitted beam.

The fragmentary appearance of the images shown is due toinsufficient statistics per point element. The probe current(103-104 ions/sec), limited by the reduced source operatingpressure, is still well below the minimum acceptable leveldiscussed previously. In the transmission mode presentlyavailable, high contrast with unstained specimens is achievedmostly by total and partial absorption of the incident beam.In regions of Fig. 2b, c, and d, however, we approach theconditions for microradiography by critical range absorption(3, 12). This obtains when the specimen thickness is withinthe range straggling distribution of the ion beam for thatmaterial. Then the contrast between areas with areal densitiesabove and below the mean range is strongly enhanced. Sinceboth H+ and H2+ ions are present in our beam, there are infact two range regions where the critical condition occurs.(For soft biological material, the mean range of 55 keV H+ions is about 0.06 mg/cm2, or 0.6 Mrm, that for the protons fromthe breakup of 55 keV H2+ ions is about 0.04 mg/cm2 or0.4 Mm.)

Appropriate to thin specimens (20-50 A) and high resolu-tion, several other contrast mechanisms are available to ascanning transmission ion microscope using hydrogen ions.The process of pickup and loss of electrons, by which a size-able fraction of the beam will emerge after specimen traversalas neutral hydrogen, seems particularly promising for itssensitivity to hydrogen content. The collision dissociation ofthe H2+ component, upon interaction with the specimen, willprovide a unique dark-field signal made out of protons carry-ing half of the H2+ beam energy. In addition, other processescommon to the interaction of ions and electrons with matterwill contribute. Details of these, our plans for implementinga high-resolution scanning transmission ion microscope, and

Proc. Nat. Acad. Sci. USA 72 (1975)

,AV 1/2

rc =- 1.1 xccV

Dow

nloa

ded

by g

uest

on

June

14,

202

0

Proc. Nat. Acad. Sci. USA 72 (1976)

+

C * .-,ld" 9c .' *. .,' . .S'. V.-k 3

FIG. 2. Proton beam micrographs taken at 55 kV, 103-104 ions/sec, 200 lines. (a) Myofibrils from rabbit back muscle. Critical-pointdried specimen, unstained, courtesy of M. Lamvik. The individual fibers, about 1 ,um thick, absorb the beam almost completely. Scalebar = 10 Mm. Exposure = 2 min. (b) Nucleus of human lymphocyte, unstained. Critical-point dried specimen, courtesy Dr. H. M. Golomb.Areas of lower mass concentration at the periphery of the nucleus are partially penetrated by the beam. Scale bar = 5 um. Exposure =1 min. (c) Critical-point dried human chromosomes, unstained. Specimen courtesy of Dr. H. M. Golomb (13). The differential packing ofchromatin fibers is detected here in conditions approaching critical range absorption. Scale bar = 5 Mum. Exposure = 2 min. (d) One ofthe chromosomes in panel c from a higher magnification scan. Scale bar = 1 Mum. Exposure = 4 min.

and a discussion of the limitations from radiation damagehave been presented elsewhere (3). Since, however, the fearof fatal radiation damage due to the high rate of proton inter-action versus electron interaction seems a common objectionto proton microscopy, we wish to summarize our views onthis matter. The factor that determines the cumulative dam-age to any atomic structure in both scanning transmissionelectron microscope and scanning transmission ion microscopewhile taking a picture is not the absolute rate of linear energytransfer, but the ratio of useful to damaging signals, e.g.,elastic to inelastic scattering. This is of the same order ofmagnitude for both electrons and protons. Ultimately, thedamage is determined by the statistics of contrast-yieldingcounts per point element, in both instruments. This impliesthat since proton cross sections are of order 102 larger than forelectrons, comparable quality pictures of thin specimens willrequire scans involving of order 102 fewer protons than elec-trons and that the radiation damage will be comparable inboth cases.

In conclusion, we have experimentally established that afield ionization source can provide the brightness required toreach high resolution in a scanning transmission ion micro-scope. We believe that even at intermediate and low resolutionthis new instrument will find useful application in the bio-logical, medical, and physical sciences.

We thank Prof. A. V. Crewe for his generous assistance in bothvaluable equipment and advice, and Mr. M. Lamvik and Dr.H. M. Golomb for providing us with the biological specimens.This work is supported by the Louis Block Fund at the University

of Chicago, the Alfred P. Sloan Foundation, the National ScienceFoundation, and the National Institutes of Health. T.R.F. wasa Union Oil Co. of California Foundation Fellow during 1973-1974.

1. Muller, E. W. & Tsong, T. T. (1969) Field Ion Microscopy(American Elsevier Publ. Co., New York).

2. Crewe, A. V. & Wall, J. (1970) J. Mol. Biol. 48, 375-393.3. Levi-Setti, R. (1974) "Proton scanning microscopy: Feasi-

bility and promise," in Scanning Electron Microscopy/1974,eds. Johari, 0. & Corvin, I. (lIT Research Institute, Chi-cago, Ill.), pp. 125-134.

4. Grivet, P. (1972) in Electron Optics (Pergamon Press,Oxford), 2nd English ed., pp. 557-559.

5. Chanson, P. & Magnan, C. (1954) C. R. H. Acad. Sci. 238,1797-1799; also (1954) Proc. Int. Conf. Electron Microscopy,London, pp. 294-299.

6. Crewe, A. V. (1974) J. Microsc. (Oxford) 100, 247-260.7. Zeitler, E. (1975) "Scanning transmission electron micros-

copy," Lectures presented at Centre for Scientific Culture,Erice, Sicily (1973), in Electron Microscopy in MaterialsScience, ed. Valdr6, U. (Academic Press, New York), in press.

8. Escovitz, W. H., Fox, T. R. & Levi-Setti, R. (1975) Proc.Third Conf. on Applications of Small Accelerators, NorthTexas State Univ., October, 1974, EFI Report 74-57, inpress.

9. Grivet, P. (1972) in Electron Optics (Pergamon Press,Oxford), 2nd English ed., Sect. 16.2.4.

10. Jason, A. J. (1967) Phys. Rev. 156, 266-285.11. Crewe, A. V., Isaacson, M. & Johnson, D. (1969) Rev. Sci.

Instrum. 40, 241-246.12. Cookson, J. A. (1974) Naturwissenschaften 61, 184-191.13. Golomb, H. M., Bahr, G. F. & Borgaonkar, D. S. (1971)

Genetics 69, 123-128.

1828 Physics: Escovitz et al.

r. I .. I I

.-

z-t

.I

Dow

nloa

ded

by g

uest

on

June

14,

202

0