The Electron Probe: An Added Dimension in Chemical Analysis

REPORT FOR ANALYTICAL

CHEMISTS

The Electron Probe: An Added Dimension in Chemical Analysis By L. S. Birks U. S. Naval Research Laboratory, Washington 25, D. C.

Using t h e electron probe, q u a n t i ­ tative analyses m a y now be per­ formed on less t h a n 1 μ/xg. of m a t e ­ rial in a n y selected 1 square micron area on a specimen surface. T h e limit of detectability is about 0 . 1 % and all elements above atomic n u m ­ ber 11 are suitable for measurement. The technique depends on a focused beam of 10 t o 50-k.e.v. electrons t o excite the characteristic x-ray spec­ tra in the 1-micron area a n d on standard x-ray optics to measure the spectral intensity. Applications include segregations in alloys and minerals, diffusion or corrosion zones, thin surface layers, and others.

/"ΛΝΕ of the things t h a t m a k e sci­ entific research such a fascinat­ ing subject is t h a t it often advances by sudden spurts in unexpected di­ rections rather t h a n by mere plod­ ding extension of accuracy by one more decimal place. Such a sudden spurt in x-ray spectrochemical analysis was represented by the de­ scription of t h e electron probe m i croanalyzer concept by Castaing in 1 9 5 1 ( 1 4 ) . T h e concept is simple: A beam of 10 to 50 k.e.v. electrons is focused to a 1-micron diameter spot on the surface of a solid specimen. The high voltage electrons generate the characteristic x-ray spectral lines from the elements present in the 1-micron spot and t o a depth of about 1-micron below t h e surface. This corresponds t o about 1 μμg. of material contributing t o t h e a n a l y ­ sis. B y measuring t h e relative in­ tensity of t h e emitted x-ray lines, the q u a n t i t a t i v e composition is o b ­ tained j u s t as in fluorescent x - r a y spectroscopy. Fluorescent x - r a y spectroscopy could never hope t o

achieve such selective composition, however, because physically, t h e p r i m a r y beam cannot be masked down much below 0.1 mm. (100 m i ­ crons) and of course the penetra­ tion is the same order of magni­ tude . Being able to analyze such min­ ute areas is i m p o r t a n t only if there is a need for such analysis, and in­ deed there is such a need. F o r in­ stance, the properties of m a n y of the new, sophisticated materials such as high t e m p e r a t u r e alloys de­ pend not on the average chemical composition, b u t on the particular composition of segregation and nonhomogeneous distribution of other elements. After the initial work of Castaing there was a lag of almost five years before really satisfactory experi­ mental equipment was constructed and dependable, reproducible re­ sults began to appear. T h e number of electron probes in existence grew from three or four in 1956 t o per­ haps 20 or more at the end of 1959 (this does not include an undeter­ mined number of instruments in R u s s i a ) . W i t h t h e recent release of commercial instruments in both t h e United States and Europe, t h e n u m ­ ber of instruments should begin to increase more rapidly and in time

should perhaps equal the number of electron microscopes. Although the number of instru­ ments has been severely limited, popular interest in the subject has grown rapidly because of the new dimension in analysis offered by the electron probe. I t seems worth­ while at this time, when commercial instruments are just coming on the m a r k e t , to stop and look a t the present state of the a r t and to point out some of t h e possibilities.

THE INSTRUMENT AND ITS OPERATION Specific descriptions of almost all of the individual electron probes have appeared in the literature' 5 · 1 1 , 13,16,19,24,32) ; g 0 o n l y t h e funda­ mentals of operation are given here. The schematic diagram (Figure 1) shows generalized components. The source of electrons at the top of Figure 1 is a hot tungsten fila­ ment type of electron gun operated at 10 to 50 k.e.v. [field emission guns which have been used success­ fully in field electron or field ion microscopes' 2 2 · 2 3 ' do not have suffi­ ciently good long-time stability for q u a n t i t a t i v e analysis]. T w o elec­ tromagnetic lenses below the elec­ tron gun reduce the diameter of the

This article, prepared by Dr. Birks at the request of the edi­ tors, does not describe any commercially available instruments. As an indication of activity in this field, the editors have con­ tacted all instrument manufacturers known to be interested in this type of instrument. Summaries of information received to­ gether with photographs or drawings are set forth. Price of the instruments depends in part on the number of at­ tachments, freight, installation, import duties, etc. The general price range, however, is from $60,000 to $100,000.

VOL. 32, NO. 9, AUGUST 1960

·

19 A

REFORT FOR ANALYTICAL CHEMISTS

Figure 1. Schematic of electron probe microanalyzer s h o w i n g only the major c o m p o n e n t s . The electron gun is at t h e t o p with the two e l e c t r o m a g n e t i c lenses below it f o c u s i n g the b e a m onto t h e s p e c i m e n . At the right is a microscope used to observe the s p e c i m e n . At the left, two curved analyzing crystals and their respective x-ray detectors are set to measure the radiation f r o m two e l e m e n t s in the s p e c i m e n . As m a n y as six ele­ m e n t s may be measured s i m u l t a n e o u s l y

electron beam to about 1 micron and focus it on the specimen sur­ face. Characteristic x-rays from the specimen elements are gener­ ated in a region only slightly larger than the electron beam. In most of the electron probes, a microscope is used to observe the specimen and position the exact area of interest under the electron beam. T h e posi­ tioning m a y be done either by trans­ lating the specimen or by deflecting the electron beam ; the specimen can be translated as much as 2 or 3 cm., but the beam can be deflected only a few tenths of a millimeter. Some of the electron probes sweep the electron beam back and forth to cover a square area on the specimen

and feed the x-ray detector output (set for a particular element) to the brightness grid on an ordinary cath­ ode ray tube which is sweeping in synchronism with the electron probe. In this way, a "picture" of the specimen in terms of the par­ ticular element is obtained. Two precautions should be men­ tioned in connection with specimen preparation: First, the surface should be polished smooth as is done for metallurgical or mineralogical microscopy, because a rough surface gives a nonuniform path length for emergent x-rays. Sec­ ond, the surface should not be etched, because this often leads to selective extraction a n d / o r redeposition of some of the etch products. Two fixed crystals and detectors are shown in Figure 1, each one set to record the x-ray intensity of a chosen element. With the N a v a l Research Laboratory ( N R L ) in­ strument as many as six elements may be measured simultaneously with fixed x-ray optics or the whole spectrum may be examined with a scanning spectrometer. Because the x-ray source is essentially a point source, curved analyzing crystals are required in contrast to fluorescent x-ray spectroscopy where either flat or curved crystals m a y be used with similar results. Geiger counters, proportional, or scintillation detectors are used as in x-ray fluorescence. In some cases, it is possible to use pulse amplitude discrimination circuitry and thus eliminate the analyzing crystal. After measuring, the x-ray intens­ ity must be related to per cent com­ position in the specimen. The rela­

Figure 2. Niobium-zinc d i f f u s i o n zone s h o w i n g i n t e r m e d i a t e phases. Specimen h e a t e d 48 hours at 900° C. 20 A

·

ANALYTICAL CHEMISTRY

tion between intensity and composi­ tion is more nearly linear t h a n in x-ray fluorescence but often not linear enough for satisfactory quan­ titative analysis. For instance, se­ lective excitation of chromium by FeTva in an iron matrix may in­ crease the CrKa intensity by 20 or 3 0 % of the amount present over the proper value. On the other hand, the absorption of Ύ\Κα radiation in a zirconium matrix m a y reduce the ΎΊΚα intensity to one half its proper value. The most common procedure for quantitative analysis has been to prepare calibration curves from a set of chemically an­ alyzed standards just as is done in x-ray fluorescence; however, the re­ quirement of homogeneity on a 1-micron scale has made the prepa­ ration of such standards extremely difficult or impossible for many types of specimens. At the present time, a technique is being developed a t N R L t h a t will allow calculation of expected x-ray intensities for any combination of elements and thus eliminate the need for calibration standards. Preliminary tests on some 50 specimens indicate a pre­ cision of a few per cent of the amount present when absorption co­ efficients are known. Results will be ready for publication shortly.

APPLICATIONS In general, applications of the electron probe are limited to SOlids i 2 . 0 .". 9 .! 0 . 1 ^,18,20,21,25-28,30,31)

foe_

cause liquids would not furnish a suitable surface for electron bom­ bardment and would be difficult to handle in the evacuated electron optics system. The specimen sur-

Figure 3. Electron micrograph of n i o b i u m - c h r o m i u m diffusion zone, heated 167 hours at 1100° C , shows i n t e r m e d i a t e phases

REPORT FOR A N A L Y T I C A L

fare must be conducting or be made conducting by evaporating 10 to 50 A. of metal over a nonconducting surface. W i t h these slight restric­ tions, there is an unlimited number of things the electron probe will do. Stoichometric Composition. In m a n y corrosion and diffusion p r o b ­ lems, particular stoichometric com­ pounds are formed. Once the ele­ ments present have been deter­ mined, the precision of the electron probe is such t h a t one can distin­ guish, for instance, between N b Z n and NbaZg, or between C u 2 0 and CuO. Figure 2 shows a specimen of niobium t h a t was coated with zinc and then heated 48 hours at 900° C. to bond the two metals. Three dis­ tinct layers are observed and one of them contains fine precipitates. The intermediate layers had proved much too narrow for analysis by other techniques but were readily measured in the electron probe. Table I shows the composition com­ puted using x-ray intensities, the assigned chemical formula, and the true weight per cent for t h a t for­ mula. The determination is unam­ biguous and the average error is 3.6% of the amount present.

Nonstoichometric Compositions. I n m a n y of the minerals and alloys, the segregations or precipitates are not stoichometric compounds, and the composition often varies from one individual segregation to an­ other. For instance, precipitates in Inconel-X alloy ( 7 % F e ; 7 2 % N i ; 1 5 % C r ; 2.6% T i ; 0 . 8 6 % N b plus minor constituents) showed a v a r i a ­ tion of from 50 to 7 0 % Ti and from 5 to 5 5 % N b in a single speci­ m e n ( 1 2 ) . I n other alloys the com­ position of precipitates varies with heat t r e a t m e n t and with over-all specimen composition. T h e elec­ tron probe is also useful in identi­ fying precipitates even if they are smaller t h a n the electron beam. A case in point is the chi and sigma phases t h a t are formed in stainless steels ( 8 ) . T h e y m a y be indistin­ guishable by microscopy, and since both phases often occur in the same specimen, x-ray diffraction or wet chemical analysis of extracted resi­ dues gives only the average compo­ sition of both phases. I n the elec­ tron probe, individual precipitates of 0.5-micron size m a y be identified in situ as chi or sigma from the C r / M o x-ray intensity ratio. For

!00

τι

Λ

I300°C 9 Hours

50

c

«

100

1

cent

Zr

CL c c

:I

(ft Ο CL

I

\

Ι

j

ι

Ti

I530°C 9 Hours

50 Zr

Ε

δ

/

1

0

1

^"""*

too Ti 50

Zr

0 C)

1485 " C 4 9 Hours

___ 1 1

1 2

1 3

I 4

I 5

Distance in Millimeters

Figure 4. Titanium-zirconium composition curves for three diffusion conditions. Courtesy of Interscience Publishers, Inc., New York, taken from "X-Ray Spectrochemical Analysis," by L. S. Birks, copyright 1960.

CHEMISTS

Table I. Composition of Niobium-Zinc Intermediate Phases (Specimen formed by diffusion at 9 0 0 ° C. for 4 8 hours) Measured Composition" Zn~ Nb 57.5 42.5 50 50 39 61 33 67 α Independent normalized by 100%.

Assigned True Weight % Comfor Compound pound Assigned Zn Nb 44 56 NbZn 51.5 48.5 Nb 2 Zn 3 41 59 NbZn 2 31 NbZn 3 69 estimates of Nb and Zn making the sum equal to

L. S. Birks, 4 1 , is head of the X-Ray Optics Branch, U. S. Naval Re­ search Laboratory, Washington, D. C. He has a B.S. from the Uni­ versity of Illinois (1942) and an M.S. from the University of Mary­ land ( 1 9 5 1 ) , both in physics. He has also done further graduate work in this field. In 1942 he entered the U. S. Navy and was assigned to the Naval Research Laboratory where he helped develop x-ray spectrochemical apparatus. He has been with the Naval Research Labora­ tory since that time. He has published over 50 pa­ pers on x-ray and electron optics, has patents on x-ray spectro­ graphs, and is author of a recent book "X-Ray Spectrochemical Anal­ y s i s " (Interscience, New York). He is a member of the Washing­ ton Academy of Sciences, American Physical Society, Research Society of America, Electron Microscope Society of America, member of several ASTM committees, and is a Naval Reserve Officer.

VOL. 32, NO. 9, AUGUST 1960

·

21 A

REPORT FOR ANALYTICAL CHEMISTS

the chi phase, the ratio is about 2:1 and for the sigma phase 3.5:1 or 4:1 for a wide variety of stainless steel compositions. Intermetallic Diffusion. B y far the most powerful advance in technique for studying solid-solid or solid-liquid diffusion is the electron probe. Because of the micron size of the area analyzed, diffusion m a y now be studied with at least 10 times greater detail than before or in zones less t h a n Vio the previous length. Being able to examine very short zones means t h a t in preparing diffusion couples the heating time is reduced from perhaps 1000 hours to 1 or 2 days. Figure 3 is an electron micrograph of a N b - C r diffusion zone of only 8 microns total length after 167 hours at 1100° C, and containing N b precipitates in an N b C r matrix plus a broad N b C r 2 zone plus a narrow (not visible) N b C r 7 zone. The phases are very hard and brittle and could not possibly have been examined by any other analytical tool. Even for ordinary diffusion couples such as T i - Z r ( 4 ) in Figure 4 t h a t extend over a large enough distance to be examined by mechanical sectioning techniques, the electron probe offers savings of more than 9 0 % in time and money over other techniques.

A single source for all your spectro needs.. .Wherever you are, whatever your needs, call on your Beckman dealer for precision spectrophotometers... he can supply any—or all —of the five different instruments that cover the full wavelength range from 170 m/x to 3500 m/i. Page 1A describes each of them briefly; your dealer can do the rest.

Three-dimensional

diffusion

around precipitates or along grain boundaries is another field where only the electron probe allows composition measurements on the size scale required. Figure 5 shows the selective diffusion of zinc along a grain boundary in copper (1>. On the left, careful etching and microscopy (after electron probe examination) show t h a t there is selective diffusion ; on the right, the composition contours obtained with the electron probe indicate much more vividly and accurately the extent of the diffusion. Using the electron probe results it was possible to calculate both grain boundary and lattice diffusion coefficients and show t h a t the grain boundary coefficient is about 1 million times greater. Thin Films of 50 to 5000 A. Films in the 50 to 5000-A. thickness range correspond to evaporated metal layers such as are used for storage tubes in high speed computers and to the thin corrosion layers t h a t form on resistant alloys. Such films are too thin to stop all of the incident electrons and therefore the x-ray intensity generated is not as high as from the bulk material. Figure 6 shows relative intensity vs. thickness curves for evaporated layers of gold, zinc, manganese, and chromium. Thickness was measured independently by interference t e c h n i q u e s ' 2 9 ' . As would be ex-

Beckman dealers, offering sales, application, and service assistance, are located in 69 countries throughout the world. Write for your dealer's name and address.

Beckman® International, Beckman

Instruments

S.A.

Rue des Pierres-du-Nilon Geneva, Switzerland

Manufacturing facilities: Beckman Instruments, Inc. Fullerton, California Beckman Instruments, G.m.b.H. Munich, Germany Beckman Instruments, Ltd. Glenrothes, Scotland Circle No. 112 on Readers' Service Card 22 A

·

ANALYTICAL CHEMISTRY

!'.

Figure 5. Selective diffusion of zinc along a grain boundary in copper. On the left is a micrograph of the region and on the right are the composition contours as determined by the electron probe. Reprinted courtesy of J. Appl. Phys. 30, 1825 (1959) article by M. R. Achter, L. S. Birks, and E. J. Brooks.

REPORT FOR ANALYTICAL CHEMISTS pected, the relative intensity, for a given thickness, increases with atomic number and for gold begins to approach the bulk intensity a t a thickness of about 0.5 micron. Two uses m a y be made of data such as Figure 6. Most obvious is the mere estimation of thickness of similar films using the relative x-ray intensity and the calibration curves. A second and probably more valuable application arises

from the fact t h a t Figure 6 m a y just as easily be plotted in terms of grams of each element (density times thickness) instead of in terms of thickness alone. The x-ray intensity is then a measure of the absolute amount of each element present. For alloy or corrosion layers, the total composition is obtained by merely adding up the absolute amounts of each element present. For thin film measurements, the

substrate or support should not contain the elements being measured in the film. FUTURE TRENDS The large majority of the work to date has been done in the metals and alloys field. Perhaps this is because the most pressing problems where large amounts of money can be saved are in those fields or perhaps it is because companies with —; such interests have the personnel capable of building and using such equipment. For the future, there is no reason t h a t the electron probe should not extend to the mineralogieal, biological, ceramic, and other fields. It will perhaps become as commonplace as the electron microscope and with as m a n y instruments in use. The problem of chemical analysis of minute areas would seem to be of equal importance to the physical structure as determined by microscopy. Although the electron probe cannot reach into the Angstrom range to compete in t h a t way with electron microscopy, it seems probable that by proper selection of operating voltage and with reduced beam size, the electron probe will be able to analyze areas of 0.1 micron or less within the next few years. As well as the need to lower limit of probe size possible, there are many where even the 1 micron

push the as far as problems size now

Don't miss the boat! In fact . . . don't miss any product in the complete line of ALUNDUM* Lab Ware. Made of fused white alumina, this rugged ware is inherently immune to the rigors of today's chemical and physical laboratories. Available in a wide range of types and sizes of crucible, capsules, dishes, boats, etc., it effectively meets the requirements of modern melting, filtering, extraction, and incineration procedures. Save time and money . . . add life to your lab ware . . . with ALUNDUM Lab Ware.

Send f o r this " Q u i c k S e l e c t i o n " C a t a l o g on ALUNDUM Lab Ware. Write NORTON COMPANY, 287 New Bond St., Worcester 6, Mass. •Trade-Mark Reg. U. S. Pat. Off. and foreign countries

NORTON REFRACTORIES Engineered...

t\^

... Prescribed

75 years o f . . . M a k i n g b e t t e r pro d u c t s . . . t o m a k e y o u r p r o d u c t s b e t t e r NORTON PRODUCTS: Abrasives · Grinding Wheels · Machine Tools · Refractories · Electro-Chemicals BEHR-MANNING DIVISION: Coated Abrasives · Sharpening Stones • Pressure-Sensitive Tapes Circle No. 70 on Readers' Service Card 24 A

·

ANALYTICAL CHEMISTRY

Figure 6. Relative x-ray intensity vs thickness for evaporated layers. Re printed courtesy of J. Appl. Phys., article in press by W. R. Sweeny, R. E. Seebold, and L. S. Birks.

REPORT

DON'T S T I R WITHOUT KONTES Because tests prove Kontes stirrers superior. • Shaft-bearing clearances measured in ten-thousandths of an inch. This prevents vacuum loss, and makes all Kontes component parts inter­ changeable. • Even after vacuum operation over a period of time, there is no loss of lubricant—further evidence of a tight seal. • Virtually friction-free ground sur­ faces; diamond-honed bearings. • Borosilicate glass used in all parts for added strength and longer life. (Teflon stirrer blades available.) • Convenient lubricant reservoir on bearing top. • Inexpensive adapters permit con­ necting shafts to motors with chuck openings as small as lA".

KONTES GLASS COMPANY Vineland, New Jersey First Choice For Quality Technical Glassware Midwest Distributor: Research Apparatus, Inc., Wauconda, III.

Circle No. BO on Readers' Service Card

GAS A N A L Y S I S

GOW-MAC INSTRUMENTS By the Thermal Conductivity

Method

• HOT WIRE FILAMENTS IN MATCHED PAIRS . ON-STREAM DETECTOR CELLS (AvaMab ' e • CHROMATOGRAPHY DETECTOR CELLS

fr m

°

St Ck)

°

For V4" Columns and Capillary Columns*

• TEMPERATURE REGULATED CELLS TO 300° C+ • PORTABLE ANALYZERS (Gas Masters) • PANEL I N S T R U M E N T S — Indicate — Record — Control • GAS BLENDER (Automatically mixes forming gas and other binaries) • EXPLOSION-PROOF U N I T S

By the Gas Density Method — • DETECTOR CELLS* — No Calibration Required — Linear Across Broad Ranges — Less than 1 % Error

Nitrogen Eluting Gas Non-destructive to Sample * Indicates new product

VISIT BOOTH 107, A.C.S., Ν. Υ. SECT. CHEMICAL EXPOSITION HOTEL STATLER-HILTON, SEPT. 13th THROUGH 15th

GOW-MAC

INSTRUMENT COMPANY

100 KINGS ROAD, MADISON, NEW JERSEY, U.S.A. Telephone FRontier 7-3450

Θ Α 5 A N A L Y S I S BV l-HBJgiylAL CONDUCTIVITY ί jSiNJSË 1 9 3 5 Circle No. 143 on Readers' Service Card 26 A

·

ANALYTICAL CHEMISTRY

available is smaller than necessary. T h e 5 to 100-micron size range would be sufficient for countless mineralogical and metallurgical problems t h a t cannot now be reached by fluorescent x-ray spectroscopy. Both the electron optics and the viewing microscope would be greatly simplified for a 5 to 100micron instrument and the cost might easily be reduced to compete with ordinary electron microscopes. T h e electron probe is making it possible to study phenomena heretofore not remotely possible and the results indicate the crucial need for further work in other fields. F o r instance,, the mathematics of the solid-solid diffusion process is largely undeveloped for the commonplace situation where intermediate phases of fixed composition form in the diffusion zone. Experimental data can be turned out so rapidly with the electron probe t h a t there should be sufficient measured values to aid in perfecting and testing any mathematical method. In another area, the composition of individual segregations in alloys m a y be related to the position with respect to grain boundaries, dislocations, etc. This means t h a t the metallurgical theory of alloy p r o p erties must become much more sophisticated in order to catch u p to empirical facts. Still another area is corrosion, where the interactions t h a t occur at surfaces m a y now be analyzed in situ and so rapidly t h a t the chemistry of corrosion will have to develop to keep pace with the flow of experimental data. Shortcoming. One shortcoming of the electron probe in common with fluorescent x-ray spectroscopy is the present inability to measure elements below atomic number 11 (Na). T h e limitation arises from the difficulty of generating the long wave lengths (10 to 100 A.) from lower atomic numbers in anything but the cleanest vacuum system and from the difficulty of dispersing and detecting such x-rays even if generated. Metal-ion pumps seem to offer possibilities for cleaner vacuum systems for generating the radiation, and ruled diffraction gratings have already been used for dispersion under laboratory conditions. Energy discrimination and unfold-

REPORT FOR ANALYTICAL CHEMISTS ing techniques for s e p a r a t i n g overlapping energy p e a k s will p a r t l y eliminate t h e need for dispers i o n ( 1 7 ) . Open-window detectors or photomultipliers of B e - C u or other suitable m a t e r i a l s will increase detection efficiency. T h u s it a p p e a r s t h a t the t e m p o r a r y limit a t a b o u t element 11 ( N a ) m a y be overcome in t h e next five y e a r s , a t least on a l a b o r a t o r y basis.

CONCLUSION B o t h the concept and the practice of t h e electron probe are relatively very new. As more and better ins t r u m e n t s become available and a wider v a r i e t y of researchers begins to m a k e use of the results, only then will t h e full potential of the technique be realized. I t can, however, be said t h a t m a n y more things can be done t h a n have now been done

CLEANEST CLEANEST CLEANEST CLEANEST CLEANEST CLfcANEST CLEAr * ^ N E S T CLEANEST CLEANEST CLEANEST C IES"L )1ST CLEANEST CLEANEST CLEANEST CLE ;T Si CLEANEST CLEANEST CLEANEST CLEAI CLE) ÎT CLEANËSlTu.fcANfS' £ l £ â M E S 2 - £ i £ A M f S l CLEANEST CLEANEST CLE J CLEANEST CLEAf CLEANEST CLPξ$Ί CLEANEST C JEST CLEAN CLEANEST CLE ÎT CLEAMESI CLEANEST C ί ϊ CLEANEST CLEAr iT CLEANEST CLEANESTXnr ^TBCjmESrCLEANEST CLEANEST Clfc EANEST C iEST CLEANEST CLEANEST CLEA J CLEAf CLEANEST CLEANEST C /€ST CLE >T CLEANEST CLEANS