Quantitative electron probe microanalysis of boron in binary borides

Quantitative electron probe microanalysis of boron in binary

borides

Citation for published version (APA):

Bastin, G. F., & Heijligers, H. J. M. (1986). Quantitative electron probe microanalysis of boron in binary borides. (1st ed. ed.) Technische Universiteit Eindhoven.

Document status and date: Published: 01/01/1986

Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

providing details and we will investigate your claim.

Labaratory for Physlcal Chemistry

...

Universlty of Technology Eindhoven

'

'-rn'IIVAtilty of Technology Eindhoven Labaratory for Phys\cal Chemlatry P.O.

Box 513.

~MB

eJndho_veo

The

NethérlandS

ISBN 9().861EHJ06-7 CIP

'

M077724

Quantnative Electron probe

Microanalysis of Baron

in Binary Borides

Cr.fr:

S.F.

Bastin

and Ir. H.J.M. Hetjllgers ,

BN/8

418*

•

CA

t-32K

24K

§

16K

0

u

8192

1,1.

181.

191.

211.

211.

""

QUANTITATIVE ELECTRON PROBE MICROANALYSIS OP BORON

IN BINARY BORIDES

Dr. Ir. G.F. Bastin and Ir. H.J.M. Heijllgers Labaratory for Physical Chemistry

University of Techno1ogy P.O. Box 513, 5600MB EINDHOVEN

The Netherlands

Work performed in the period: March 1984 - J anuary 1985

CIP-DATA KONINKLIJKE BIBLIOTHEEK, DEN HAAG

Bastin, G.F.

Quantitative electron probe microanalysis of boron in binary borides : work performed in the period: March 1984-January 1985

I

G.F. Bastin and H.J.M. Heijligers. - Eindhoven : University of Technology. - Ill.

Publ. of the Labaratory for Physical Chemistry. With ref.

ISBN 90-6819-006-7 SISO 542 UDC 541.1

Subject heading: Electron Probe Microanalysis of Boron

©

University of Technology, Eindhoven Labaratory for Physical Chemistry P.O.Box 513, 5600MB EINDHOVEN The Netherlands

I.

SUMMARl'·

INTRODUCTION

CONTBNTS

11. PRACTICAL PROBLBMS in the ANALYSIS of VERY LIGHT BLBMBNTS 11.1 General Problems

11.2 Specific Problems in the analysis of Boron

lil. BXPBRIMENTAL PROCEDURES

Ill.l III.2 III.3

m.4

Ill.5

Preparation and characterization of binary borides Mounting, pollshing and cleaning procedures Some details on the equipment used

Measurements of Area/Peak Factors <APP's)

Measurements of Peak k-ratios between 4 and 15 {30) keV IV. RBSULTS

IV.l IV.2 IV.3

· Boron-K emission spectra a.

Area/Peak Factors for B-K relative toelemental Boron a.

lntensity ratios for metals and boron

V. DATA REDUCTION AND COMPARISON OP CORRECTION PROGRAMS

V.l V.2

Description of the <1>(pz) approach

New parameterization for the <t><pz) approach V.3 Resuits on previous data

V.3.1 Medlum-to-heavy element analyses V.3.2 Metal analyses in Carbides

V.3.3 Carbon analyses in binary Carbides V.4 Resuits from the present work

V.4.1 Metal analyses in binary Borides V.4.2 Boron analyses page 1 3

4

12 18 23 23 24 25 27 37 41 50 54 56 58 60 62 65 REPBRENCES 75 APPENDIX

A. 1-3 Area/Peak factors as a function of detected peak position for a number of

borldes. 77

B. 1-2 7 Numerical details for the intensity measurements for metals and Boron. 80

C. 1-3 Relative B-K intensities for the binary boridesas a function of accelerating a.

voltage. 107

-1-SUMMARY

Quat:i.titative electron probe microanalysis has been performed in 27 binary borides in the range of 4-30 keV, both for the metals as well as for Boron.

The procedures along which accurate intensity measurements for B-K must be a.

carried out are discussed in detail. It is shown that, similar to the analysis of Carbon, intensity measurements can only be performed in an integral fashion. Apart from peak shifts and peak shape alterations, however, there is an additional problem in the analysis of Boron and that concerns the fact that peak positions and peak shapes may be dependent on the crystallographic orientation of the specimen. Approximately 50% of the borides were found to exhibit these effects. This implies that the approach proposed earlier, i.e. the use of an Area/Peak Factor to deal with peak shape

alterations, can be applied only in 50% of the cases in a straight-forward manner. In the remaining cases the APF has to be known as a function of peak position and this has been measured in the appropriate cases.

The present work has resulted in a total of 196 k-ratios with respect to elemental standards for the metal X-ray lines and 180 k-ratios for B-K relative toelemental

a. Boron.

These data have been used to arrive at an improved parameterization for the <t>Cpz) approach in matrix correction. The resulting new program CBAS861) was :Qext

compared to 5 other current correction programs. At the same occasion the available set of mass absorption coefficients for Boron was tested on lts consistency and better values suggested where necessary. Finally it is shown that our modified version of the Gaussfan <t>Cpz) approach (BAS861 program) is highly successful in the matrix

correction for B-Ka.: a relative root-mean-square value of 6.30% was obtained. This demonstrates that even foranultra-light element like Boron an expertmental

accuracy can be achieved that hitherto was considered normal for the analysis of much heavier elements. This is of course only possible if due care is excercised in the measurements and lf the correct procedures are followed.

-3-I. INTRODUCTION

In a number of publications l-3 we have reported on our systematic workon quantitative electron probe microanalysis of Carbon in binary carbides. It has been demonstrated that quantitative analysis of even such a light element like Carbon is indeed possible with unexpected high accuracy, provided that proper care is exercised in the measurements and that a suitable matrix correction program is used in

conjunction with a consistent set of mass absorption coefficients (mac's).

The workon Carbon was, in fact, part of a much wider research effort aimed at exploring the possibilities for quantitative analysis of elements like Baron, Carbon, Nitrogen and Oxygen; elements which enjoy a growing interest in fields like high temperature materials and special ceramics. It must be expected therefore, that there wi1l be a rapidly growing demand for quantitative analysis of these elements in future, especially for local microanalysis for which the electron probe microanalyzer still appears to be the most powerful tool.

The object of this long-term investigation, which was initiated in

J

anuary 1983, is not only to explore the feasibility of light-element analysis but also to supply (for the flrst time) large data bases of accurate measurements for each of these elements, on which present and future matrix correction programs can be tested. As far as the latter purpose is concernedit is our firm conviction that it is absolutely no use to test programs on a few isolated analyses, as has been done frequently in the past; only consistent sets of measurements taken over widely varying conditlans and systems can provide the necessary information on both the performance of programs as well as on the consistency of publisbed sets of mac's. It must be emphasized that

3

the latter are of ten. of equal importance as the programs themselves . These considerations are the reason for the sametimes extreme conditlans (4-30 keV accelerating voltage) under which we perfarm our measurements and the large number of systems Cl0-30) on which they are performed.

The phllosophy on which our workis basedis perhaps best described as: "Correction programs may come and go in the future but accurate measurements wi1l retain their value almast indefinitely".

It is the object of the present work to report on our efforts concerning the quantitative analysis of Baron in 27 binary borides. As far as we knowit is for the first time that a systematic investlgation on this subject is carried out.

-4-Apart from lts intrinsic relevanee the analysis of Boron is very interesting because, being almost the lightest element that can be measured in an electron probe

mlcroanalyzer, it must be anticipated that it will provide the most testing conditions for any matrix correction program.

ll. PRACTICAL PROBLEMS in the ANALYSIS of VERY LIGHT ELEMENTS

ll. 1. General problems

The practical problems whlch have to be faced during the analysis of very light 1-3

elements llke B, C, Nor 0, have already been discussed in detail and will only briefly be recalled here.

1. Low vield of X-rays (especially for N-K X-rays), coupled with a relatively

_a

_.

inefficient detection system. Por B-K X-rays the situation is not particularly

a

bad because in general the count rates which can be obtained are quite acceptable and PeaJc: .. to-Background ratlos rather favourable.

2. Dead-time problems for the metallines if these are to be measured

simultaneously with the light element and pulse shift problems for the light element itself. These problems are a direct consequence of the use of very high beam currents whlch are frequently necessary and of the sometimes very high dUferences in count rates between standard (e.g. elemental Boron) and specimen. That the latter problem is not merely of academie interest is demonstrated in Fig. ll.l where the shape of the Boron-K pulseis shown u a function of the beam

a

current. It is evident that an increase of a factor of 3 in the beam current produces a shift of -0.1 Volt in the location of the pulse maximum. Considering the fact that differences in count rates between (elementall standards and

specimens may well exceed two orders of magnitude it is obvious that pulse shifts of several tenths of a Volt may occur. Thls may, in turn; lead to appreclable errors in the lntensity measurements if a narrow window (sharp discrimination) is applied in the Pulse-Height-Analyzer.

-5-eruiT AAlt (cPS)

leP

lo3

10 0 2 lj

₆

vUJ

Fig. II.1.---Shift of the

Bo~on~K

puLse with count

~ate

at 10 keV.

a

JEOL ?33; Lead-stearate crystaL;Gas-[Low counter (Ar/10%

methane);High tension 1?00 V;Gain

64x8;0~tec

counting

eLect~onics;Window

0.2 V. Variation in beam current between

10

and

300

nA.

3. Interference of higher-order metallmes with the light-element line can be quite

a nuisance, mainly in the analysis of Carbon, Nitrogen and Oxygen (see e.g. Ref

1). Fortunately, for the present case of Baron these problems are much less

pronounced due tothefact that in this wavelength range<- 67 Angstroms) the higher orders of metallines have gradually faded away. However, insome cases

new problems canturn up, e.g. in combinatlans with the elements Nb and Mo

where the M-( lines of these metals show up very close to the B-K line in

a.

comparable intensities. Examples of these effects and solutions to these problems will be presented la ter on.

-6-4. Background and contamination problems, which played a very important role in the analysis of Carbon, are notprominent in the analysis of Boron, with an exception for the cases mentioned under 3.

However, due to the extremely shallow X-ray emission for B-K X-rays it is still a.

advisable to use some kind of anti-contamination device (preferably an air-jet1) during the measurements, especially during the time-consuming integral

1-3

measurements (see also 8).

5. The choice of a standard, which was a real problem in Carbon ana1ysis and, to a lesser extent, also in the analysis of Nitrogen and Oxygen, is not much of a

problem in the analysis of Boron because an unambiguous form of elemental Boron is readily avallable in the form of crystalline Boron.

In fact, for reasans whlch will be discussed in II.2 we are almost forced to use thls standard, whlch, by the way, makes the problems mentioned under 2. all the more re al.

6. The knowledge of the mass absorption coefficients is an extremely vitalissue. It 1-3

has been pointed out that as a rule of thumb an uncertainty of 1% in these values produces an uncertainty of 1% in the calculated concentrations. Now, it seems hlghly unllkely that the publisbed sets of rnac's for very light elements are anywhere near withln 1 %; in fa ct, publisbed values for C -K X-rays may dlffer by

1 a.

as much as 100 %!

As one of the results of our workon Carbon we proposed a new and consistent set of rnac's for Carbon whlch were found to produce significant improvements in all recent (since 1980) correction programs3•4•5. Considering the fact that the coefficlents for Carbon range from 2,500 to 40,000 and those for Boron 6 from 3,300 to 84,000 (see Table II.1) it must be anticipated that the Wlcertainties in the lattervalues play an even more drastic ro1e than in the case of Carbon. There appears, however, to be no obvious reason to expect hlgher accuracies in the values for B-K . The opposite is probably true. Apart from thls only two

a.

independent sets of rnac's are avallable6'7<Table II.l).

Thls must be attributed to the fact that, contrary to Carbon, nobody bas ever done any systematlc workon Boron.

-7-Table II.l

Mass Absorption Coefficients for Boron-K X-rays according to various sources. a.

Henke (74) 7 Henke (82) 8 Present Work

Absorber B 3353 3350 3350

c

6456 6350 6350 N 10570 11200 11200 Al 65170 64000 65000 Si 74180 84000 85000 Ti 15280 15300 15300 V 16710 16700 18300 Cr 20670 20700 20700 Fe 25780 27600 27600 Co 28340 30900 30900 Ni 33090 35700 35700 Zr 38410* 8270 4100 Nb 4417 6560 4300 Mo 4717 5610 4200 La 3826 3730 2600 Ta 20820 20800 19500

w

19660 19700 19000

u

2247 9020 7500

-8-It will be clear, though, that the doubts expressed here will have a large impact on the reswts calcwated by any matrix correction program.

As a consequence, it is our firm belief that an independent judgement of a

particwar correction program is impossible as long as themac's are not available with an accuracy better than 1% relative. Until then we shall have to satisfy ourselves with statements made in conjunction with the particwar (and consistent) set of mac's used in the test.

7. The choice of a matrix correction program.

For many years it has been common practice to discuss the matrix correction procedure, which is necessary in order to convert the measured intensity ratio between specimen and standard (k) into concentration, in terms of Z, A and F factors.

The Z-factor stands for the ratio between the amount of X-rays generated in standard and specimen, respectively. The emitted intensity, which is of course the vital issue, is obtained by applylng an absorption correction, which is usually

expressed by F(x.), with x. = 1!:.. cosec 1j1 ( _{p p} 1!:.= mac; 1j1 = X-ray take-off angle). The quantity P(x.) is simply the fraction of generated X-rays, which is actually emitted from standard or specimen.

The P -factor, finally, accounts for the fa ct that sametimes additional,

non-electron beam generated, X-rays can be produced by primary X-radiations of other elementsin the specimen. The concentratien C is now related to the

intensity ratio k by the following relationships:

C=k.Z.A.P I or:C= ~ 1 stand.

z

stand.

z

spec. P(x.) ttand. p P( x.) . spec.

Due to the straight-forward proportionallty of the concentratien to all three factors Z, A and F it is obvious that in principle they are of equal importance. However, fluorescence effects are usually small and for very light elements they can be neglected; this leaves only the Z and A factors to be considered.

The atomie number correction (Z) for medium-ta-heavy elements (Z> 11) is usually less than 30% from unity; the absorption correction, on the other hand, can be much larger: A-factors of more than 10 are not uncommon in unfavourable cases9. As aresult the absorption correction has rightfully received the most attention in the past.

-9-In our apinion it is doubtful if this is still appropriate for cases of very light elements. Our experiences with a number of current correction programs for the analysis of Boron (or Carbon) show that the diffetences in atomie number

correctlans produced by these programs (of the order of 20%) might, in

themselves, already account for the discrepancies between calculated and nominal concentrations. What it all comes down to is that the same product CZ times A) ciln be produced by a variety of Z and A factors and that in cases of very light·

elements nobody is at present able to establish what the correct distribution over Z and A factors should be.

This is an obvious consequence of a totallack of measured cpCpz)

Cionisation-vs-mass depth) data for very light elements. Hence, it follows that all separate atomie number corrections., which are currently in use, must be regarded as more or less blind extrapolations towards very light elements.

. 3,4,10

The recently mtroduced cp(pz) programs may suffer from the same

uncertainties, of course. The calculated <t>Cpz) curves on which the correctlans are based can only be compared to actually measured <t>Cpz) data for medium-ta-heavy

9

elements, for which they are highly successful . In many respects it is somewhat uncertain whether the equations used in the cp(pz) approaches can be extrapolated to the conditlans common in light element analysis.

The main problem in quantitative electron probe microanalysis is that one is always discussing ratlos between quantities; the intensity (or <t>Cpz) curve) of the standard is compared to that of the specimen. In many respects it is possible that large errors are present in both the numerator and the denominator and that still the ratio is correct. As a consequence it is often very difficult to obtain evidence about the correctness of absolute values.

A further discussion on this subject wil! be given in Chapter V, where the numerical results of the horon-analyses will be evaluated.

8. Peak Shifts and Peak Shape Alterations. 1-3 .

We have shown befare that the intensity measurements for very light elements can be seriously influenced by:

- Peak shifts and

- Peak Shape Alteratiens

which both can occur in the light-element-K emission spectrum. While the first a.

problem can easily bedealt with by retuning the spectrometer when moving from the standard to the specimen the second problem is much more difficult to tackle.

·. ·.

-10-The direct consequence of peak shape alterations is that the net peak intensity, which is usually measured in day-to-day practice, is no langer a good measure for the emitted intensity. In fact, this routine practice is based on the tacit

assumption that the peak intensity is proportional to the integral intensity emitted by the specimen. Fortunately this is true in the vast majority of cases, like all medium-ta-heavy element K-, L- and M-lines.

Due to the sametimes extreme varlaUons in peak shapes in very llght element K -llnes, however, this assumption no langer holds.

a

As a result it is imperative to perform integral measurements which is not very attractive, of course, for a Wavelength-Dispersive-Spectrometer CWDS).

We have shown before1-3 that a considerable reduction in effort can be obtained by the introduetion of so-called Area-Peak Factors (APFs). This APF has been defined as the ratio between the correct integral (Area) k-ratio and the Peak k-ratio. Of course, this factor is only valld fora glven compound with respect to a given standard and for a given spectrometer. Once an APF is known, subsequent measurements can simply be measured on the peak again; subsequent

multiplication by the APF will yield the correct integral k-ratio.

From our work on Carbon in binary carbldes lt was obvlous that the APF was e.ssentially independent of accelerating voltage and that individual values (relative to Fe

3C as a standard) could range from 0.72 (TiC) to 1.05 CB4C). If glassy Carbon would be chosen as a standard then the deviations would be even more extreme: down to 0.5 for TiC; which clearly demonstrates the need for integral

measurements.

Consldering the fact that Boron is even llghter than Carbon we must expect to run into similar problems for the B-K spectrum. That this is indeed the case is

t1

demonstrated in Fig. II.2 where four B-K emission spectra are shown.

a

Apart from the dUferences in peak positions the varlaUons in peak shapes are very- p.romlnent. Quite frequently two or three separate contributions can be distinguished in the spectra, even to the extent tha:t two c:ompletely distlnct satellltes are present in BN. It will be clear that in the latter case the satellite intenstties would go unnoticed in a peak intensity measurement, demonstrating. once more the nece·ssity of integral measurements.

.,; i ::> 0 .... .,;

...

_:a ::> 0 .... -11-Wavelength eR> 60 65 70 75 60 65 70 75 "ZS6r< J"ZBr< B l, ~. (. _LaB 6 l.~.E. 19"21< 961< 1 "ZBr<. 641< 64r< 8 3'Zt< Hr< 161< 8]8( 171.

""

BN 178. 1'11'1 JU·. 191. 181. 1118. .,;

...

_z ::::> 0 .... <IB. "Zll. l, ~. (. 3'Zt< 178. 1111. 191. "ZBB. "ZIJ. 1'11'1 3"Zr<

-

UB4 l. ~.l. '241< 161< 181. 'ZIIl. :2111.

Fig. II.2.---TypicaL Boron-K emission spectra from three binary

a

borides as corrrpared to that from eLementaL Boron.

Note the peak shifts

and

the pronounced peak shape

aLterations. Lead-stearate crystaL;lO keV,300

nA

beam current.

From a chemica! point of view these spectra are also interesting because they must contain a lot of information on the accupation of electron energy levels. In combination with XPS measurements perhaps valuable information on the nature of the chemica! bond could be extracted. Thls is, at the moment, however, beyond

the scope of the present work.

Unfortunately the problems in Baron analysis are not restricted to peak shifts and peak shape alterations as we will see in the next section.

-12-ll.2 SpecHic Problems in the Analysis of Baron.

Our first experiences with integral measurements in borides were made on needle-shaped TiB-crystals (orthorhombic crystal structure), grown in a Ti-matrix. These measurements were carried out relative to elemental Baron as a standard and the results were at first astonishing. From the workon Carbon we were used to APFs with reproducibilities of better than 2%; now we were suddenly facing data which showed varlatlans of up to 30%. The explanation of these peculiar phenomena was eventually found in the observation that the peak position was no langer a constant but showed variations from one TiB-crystal to another.

_ _ TlB 13/18 T.H.E. 12K 1/)

....

z ₈₁₉₂ ::1 0 u 411196 17111. 18111. 19111. 2111111. 21111.

HH

60 65 70 75 wavelength

äö

Fig. II. 3.-The two extremes that can be observed in the shape of the

B

-K

peak emitted

by

various TiB crystaLs or within the same

a.

crystaL upon rotation in a pLane perrpendicuLar

to

the eLectron

-13-Stranger still, the peak shape was found to vary in a completely synchronous way with the peak position: The lowest APF, i.e. the highest peak intensity in combination with the narrowest peak, was always found at the lowest peak position (shortest

wavelength). Reversely, the highest APF was found at the highest peak position Clongest wavelength).

Fig. 11.3 shows the two extremes in the Boron spectra emltted by TiB. Apparently each individual TiB-crystal can emlt lts own specific Boron-K peak, both in shape as

a. well as in peak positlon.

It was further found that each individual crystal could be made to emit peak shapes between the two extremes in FigJI.3, merely by rotating the specimen in a plane perpendicular to the electron beam. This must be taken as an indication that it is not so much the position of the specimen with respect to the electron beam that matters, but rather the position of the specimen with respect to the analyzer crystal. In extreme cases a rotation of 90 deg. was enough to shift the B-Ka. peak over 1 mm (0.358 ,8,) from the maximum to the minimum position. A further rotation of 90 deg. brought it back to the original position and so on. These observations strongly remlnded us of the effects observed when examlnlng a non-cubic specimen under an optical microscope with crossed Nicoll's prisms. A literature research revealed that the observed effects were indeed due to polarizatlon effects in the emitted

X-radiation and that the phenomenon ltself has been known since 196911. It can be expected 12' 13 in all compounds with a crystal symmetry lower than cubic and higher than trlclinic. Polarization takes place (Fig. 11.4) in two mutually perpendicular

planes, which are aligned along the principal crystallographic directlans of the crystal lattice.

Unfortunately for quantitative analysis, the analyzer crystal can assume the function of a polarization filter and lts action as such is optima! when the angle of incidence of the X-rays on the crystal is 45 deg.

Por our specific crystal spectrometer (JEOL Superprobe 733) with a Pb-stearate crystal C2d=l00.15A) the relatlon between wavelength 0.., A) and linear position CL, mm) of the crystalis given by:

. L

n . ).. = 2d .

2R

= 2d . sin

e

in which 9 is the Bragg angle, n is the order of the reUeetion and R Cmm) is the radius of the Rowland clrcle (140 mm).

-14-1\'W..YlER CRYST/IL

y

x

Fig. II.4.-Schematic drahling showing the interaction of the analyzer

crystal with the polarized components in the emitted

B-K

a

radiation. PI and PII are the main polarization directions

which are aligned along the principal crystallographic

directions in the crystal lattice of the specimen.

The electron beam is considered to hit the specimen in the

origin of the coordinate system X,Y,Z. (After Wiech

13

J.

-15-It follows that at the positlon of the B-K peak (189.0 mm) the corresponding Bragg a.

angle

e

is equal to 42.5 deg. This coincidence explains a great deal of the exciting and interesting phenomena observed.

The most pronounced effects have, so far, been found in hexagonal compounds like TlB

2 and ZrB2, withup to 40% variation in APF (rel. toelemental Boron) and shifts of approx. 1 mm (0.358A). As was to be expected the very few cubic borides lilce LaB

6 did notshow these effect and could thus be considered as potential complex standards. Unfortunately, due to the low mac for B-K in these compounds (see Table

a.

11.1 ), the emitted intensity is already 30-50% of that from pure Boron; so, there is little to be gained. Crystalline Boron doesnotshow these effects, although its rhombohedral crystal structure would perhaps allow this, so we were ulUmately forced to adopt this as a standard.

An immediate impHeation of these polarizatlon phenomena is that anyone who is unaware of these effects could make large errors in the intensity measurements; even if these measurements were performed with the same compound as a standard and if the spectrometer would be retuned for each individual measurement. One could easily come to the conclusion that the Boron content in compounds like TlB

2 and ZrB2 showed varlatlans of 25%, whereas they actually have very limlted homogeneity ranges.

For quantltaUve analysis these effects are, of course, very much of a nuisance because one is forced either to measure all intenslUes in an integral fashlon, or to determine first the varlation of the APF with the observed peak posiUon for each non-cubic compound.

We have done the latter for all 27 binary borides and the general results are that the Area (lntegran intenslUes showed only a relaUvely small varlation (approx.±. 4%) relative to the average values. Apparently there is also a small effect of

-16-Fig. II.5 shows the results of the measurements on ZrB

2, which must be regarded as a typical example of a non-cubic boride. Note the very strong varlation in Peak k-ratio as compared to the sllght varlation in Area k-ratio.

The resulting APF shows almost 40% variation. In Fig. II.6 the B-Ka. emission spectra corresponding to the two extremes are represented.

The detailed results of the APF measurements for the other borides will be discussed later on.

fff,

REL. TO

B

r

~

10

KV

:m

NA

K-PATIO

\

.3

0~0

0 ff/lJ(

•

fiD.

-·----=-··

--1

1.00

.2

~ar--e-.81

;g~

,(i) 1.88

1.89

ffllJ< ~ITH~ U+~>

Fig. II.5.---Variation of Peak k-ratio,Area (integral) k-ratio

and

Area/Peak Faator with deteated peak position in ZrB 2 (hexagonal)

as a typiaal representative of the non-aubie borides.

1(81'< 96K 64K V) r -:7. :::l _3ZI'L 0 lJ 60

----·

1/B. 11M

-17-WAVELENGlli

(/1.)

65 70 75 1. H.E. J 88. HUL ZUL

Fig. II.6.---The two extremes in the peak shapes of the B-Ka spectrum

emitted

by

ZrB

2

. The corresponding values for Peak

k-ratio~

-18-m

EXPERIMENTAL PROCEDURES

Por the purpose of the present work it is, in principle, required to have boride specimens with 100% density and accurately known compositions. Like it was the case for Carbon such specimens can hardly be obtained commercially.

It was, therefore, decided to prepare them in our own laboratory. Only 3 out of 2 7 boddes were obtained from outside our laboratory; these were a specimen of BN (kindly supplied to us by Mr. G. Verspui; C.F.T. Phlllps, Eindhoven, The Netherlands), UB

4 (single crystal supplied by Dr. Menovsky, Univ. of Amsterdam, The Netherlands) and LaB

6 (single crystal supplied by Prof. L. Swanson, Field Electron and Ion Sourees Company, Me Minville, Oregon, U.S.A.).

Ill.l Preparation and characterization of binary borides.

The vast ma)ority of boride specimens (Table Ill.l) were prepared by repeated Argon-Are melting techniques. Mixtures of elemental Boron and metal powders (purity both better than 99.8%) were first pressed into pellets and subsequently melted in an Argon plasma (20 cm Hg) on a water-cooled copper plate. After repeated melting (3-5 times) with intermediate turning-over of the buttons, the specimens were sectioned and polished using conventional polishing techniques for very hard ma terials.

Optica! microscopy was used to get a first impression of the quallty of the specimens, as far as presence of second and/or third phases, presence of unreacted starting matetials, grain size (using polarized light) etc. are concerned.

In some cases (e.g. BN) X-ray diffraction procedures were used to identify the

compounds. Mieroprobe analysis of the metal component was used to check such vital items like homogeneity, identity of second phases and the compositions of the

specimens. Special emphasis was hereby laid on the consistency of the results obtained in cases where 2 or 3 representatives of the same system (e.g. Ni

3B, Ni2B, NiB) were avaUable. This was, by the way, the reason that we preferred to have more than one compound in the same binary system.

During melting is was frequently observed that small pieces of Boron splashed away from the plasma, due to the often very high melting points of many borides. On the other hand it was sometimes observed that evaporation of the metaltook place. As a result it must be expected that the nomina! (real) compositions of the

-19-Table Ill.l

Survey of the binary borides used in the present work, with their nomina! compositions and crysta1la ttices.

(S) indica tes stoichiometrie composition, (L) line compound.

Compound Comp.(wt%B) Structure

B 99.80 Rhombohedra1 B 4C 79.81 Rhombohedra1 BN 43.48 (S) (L) Hexagona1 A1B 2 44.49 (S) (L) Hexagona1 AIB 12 82.78 (S) (L) 1 SiB 3 52.20 Rhombohedral SiB 6 68.57 Orthorhom bic

TiB 16.78 Orthorhom bic

TiB 2 30.07 Hexagonal VB 2 28.40 Hexagonal CrB 16.80 Orthorhom bic CrB 2 27.31 (Lim) Hexagona1 Fe 2B 8.82 (S) (L) Tetragonal

FeB 16.22 (S) (L) Orthorhom bic

Co

2B 8.40 (S) (L) Tetragonal

CoB 15.50 (S) (L) Orthorhom bic

Ni

3B 5.78(S)(L) Orthorhom bic

Ni

2B 8.43 (S) (L) Tetragonal

NiB 14.40 Orthorhom bic

ZrB 2 17.89 Hexagonal NbB 10.42 (S) Orthorhom bic NbB 2 17.15 Hexagonal MoB 10.13 (S) Orthorh.(HT)ITetragonal (LT) LaB 6 31.83 (S) Cu bic

TaB 5.64 (S) Orthorhom bic

TaB

2 9.23 (S) Hexagona1

WB 5.55 (S) Orthor ./Tetragona1

UB

-20-specimens (Table III.l) may differ significantly from the weighed-in compositions. In the majority of cases this led to the production of second and third phases, which, in turn, made an independent chemica! analysis impossible and meaningless.

However, precisely the presence of second phases, like traces of Fe at the grain boundaries of Fe

2B, and the observations made during melting, led us to adopt the final nominal compositions in Table III.l. In cases where we observed second phases richer in metal we have assumed that the Boron content of our boride specimen corresponds to the roetal rich boundary according tothelatest phasediagram14. On the rare occasions where second phases richer in Boron were found we assumed that the Boron-rich limit of the homogeneity region was attained.

Fortunately most boddes have rather high Boron-contents and in general the homogeneity regions are rather restricted. The estimated accuracy of the nominal compositions in Table ill.l is therefore better than 5% relative.

In the following we will give a brief description of the preparation of the specimens, together with some remarks relevant to their compositions.

B

4 C Arc-melted; no special remarks

Weighed-in: 79.81 w/o B Nominal: 79.81 w/o B

BN Chemical Vapour Deposited layer of hexagona1 BN.

Extreme1y fine-grained structure (X-ray diffraction).

Line Compound Nominal: 43.48 w/o B

R..F .me1ted; s1owly cooled down overnight into two-phase region AlBlAlB

12

Gold-coloured (strongly polarizing) crystals of AlB 2 (few) + grey AlB

12 crystals in Al-matrix. Line compounds Nominal: 44.49 w/o B

82.78 w/o B

SiB

3 Arc-melted Si (30 w/o B) alloy, rapidly coo1ed. SiB

6 SiB3 crystals (light in backscattered image) + SiB 6 crystals (dark) in Si-matrix. Non-equilibrium situation

Nominal: 52.20 w/o B 68.57 w/o B

-21-TiB Arc-melted Ti (2-10 w/o B> alloys, slowly cooled. Primary Ti-rich TiB-needles in Ti-matrix.

Nominal: 16.78 w/o B.

TiB

2 Arc-melted, very large grains; traces of second phases at grain boundariès and near edges of specimen (evaporation of Ti; m.pt. of TiB

2 3225 °C). Splashing of Baron.

Weighed-in: 31.10 w/o B. Nominal: 30.07 w/o B.

VB

2 Arc-melted; splashing of Baron, resulting in second and third phases

(V -richer).

Weighed-in: 29.80 w/o B. Nominal: 28.40 w/o B.

CrB Arc-melted; splashing of Baron, resulting in presence of fine lamellae of Cr-richer phase.

Weighed-in: 17.21 w/o B. Nominal: 16.80 w/o B.

Arc-melted; splashing of Baron, resulting in traces of Cr-richer phases. Weighed-in: 29.37w/o B. Nominal: 27.31 w/o B.

Arc-melted; presence of eutectic Fe/Fe

2B at grain boundaries. Line compound.

Weighed-in: 8.82 w/o B. Nominal: 8.82 w/o B.

FeB Arc-melted; presence of Fe

2B at grain boundaries. Weighed-in: 16.22 w/o B. Nominal: 16.22 w/o B. Line compound.

Arc-melted; presence of Co at grain boundaries. Weighed-in: 8.40 w/o B. Nominal: 8.40 w/o B. Line compound.

CoB Arc-melted; presence of Co

2B at grain boundaries. Weighed-in: 15.50 w/o B. Nominal: 15.50 w/o B. Line compound.

- NiB

-22-Arc-melted; two-phase mixture of Ni

3B + Ni.

Weighed-in: 5.78 w/o B. Nominal: 5.78 w/o B. Line compound.

Arc-melted; two-phase mixture of Ni

2B + Ni3B. Weighed-in: 8.43 w/o B. Nominal: 8.43 w/o B. Line-compound.

Arc-melted; traces of (unreacted ?) Boron.

Weighed-in: 15.55 w/o B. Nominal: 14.40 w/o B.

ZrB

2 Arc-melted; splashing of Boron + evaporaUon of Zr possible (m.pt of

ZrB

2 3245 °C); very large(» 100 l!ID) grains. Presence of Zr at grain

boundaries Clocally).

- NbB

- MoB

Weighed-in: 19.16 w/o B. Nomina!: 17.89 w/o B.

Arc-melted.

Weighed-in: 10.42 w/o B. Nominal: 10.42 w/o B.

Arc-melted; splashing of Boron; traces of Nb-richer compound very locally.

Weighed-in: 18.88 w/o B. Nomina!: 17.15 w/o B.

Arc-melted, very large(» 100 l!ID) grains;

Single-phased.

Weighed-in: 10.13 w/o B. Nominal: 10.13 w/o B.

LaB

6 Sintered bar, converted into single-crystalline rod (approx. 1.5 mm

diameter) trough triple-are floating zone technique. Nominal: 31.83 w/o B.

TaB Arc-melted, large elongated grains; presence of Ta-richer phase at

grain boundaries.

-23-Arc-melted, very large grains(» 100 '}Uil); presence of Ta-richer phase

at grain boundaries.

Weighed-in: 10.67 w/o B. Nomina!: 9.23 w/o B.

WB Arc-melted, very large grains. Traces of unidentified second phase only at the edges of the specimen.

Weighed-in: 5.55 w/o B. Nomina!: 5.55 w/o B.

UB

4 Arc-melted and drawn from the melt using triple-are technique. Nomina!: 15.37 w/o B.

III.2 Mounting, polishing and cleaning procedures

Each of the boride specimens, lumps of crystalline boron (used as standards) and pieces of the pure metals were mounted and polisbed separately. This was done in order to avoid polishing problems with materials with extreme dUferences in hardness. The specimens were mounted in copper-filled resin and very carefully polished. Diamand abrasive discs were used for the coarse stages <70,30,15 1J.m); final polishing was done on a nylon cloth with diamond (3-1 ')Uil). When a satisfactory polish was obtained the specimens were cut out again in small cubes of mounting material containing the specimen after which all edges were carefully rounded off. Next the specimens were turned upside down to be remounted groupwise (e.g. TiB, TiB

2, B and Ti) with the polisbed sides facing the carefully cleaned bottorn side of the mould. This procedure was found to guarantee perfectly plane and parallel top and bottorn planes of the mount which is extremely important in view of the correctnessof the take-off angle in light element work. Superficial contaminants were removed from the

polisbed faces by a brief pollshing treatment on a soft cloth using 0.05 lUil

y-Alumina. Final cleaning was done ultrasonically using alcohol.

III.3 Some details on the eguipment used.

All measurements were performed on an automated · JEOL 733 Superprobe, equipped with 3 wavelength-dispersive spectrometers (at present 4) and an energy-dispersive system (Tracor Northern, TN2000). The automation system was also supplied by Tracor Northern (TN-1310). The first spectrometer, specifically for light elements, was equipped with a lead-stearate crystal which was used for the boron measurements, and a TAP crystal. The counter was of the Gas-flow

-24-Proportional type, counter gas Argon/10% methane. The other two spectrometers each contained a PET and an LIF crystal while the counters were of the sealed Xenon-type.

On a previous occasion1 we have already described the tests on the eperating conditlans of the mieroprobe (accelerating voltage, beam cw·rent stability, stability of beam position) which were found highiy satisfactory. Besides, our instrument is fitted with an automated beam currect detector and count rates are automatically corrected for any drift that might occur.

1

During all Boron-measurements an air-jet was used to prevent carbon

contamination of the spot to be analyzed. This was necessary in view of the aften very long (up to 3 hrs) measuring times on the same location during integral

m~asurements. It must be expected that without anti-contamination device the B-K count rate would gradually deteriorate due to carbon build-up. This, in turn, must mainly be attributed to the presence of an oil-diffusion pump. At present this has been replaced by a turbo-molecular pump.

IIL4 Measurements of Area/Peak Factors CAPFs).

The APFs for B-K radlation were measured by recording the integral emission a.

a.

profile both for Baron (standard) and each of the boride-specimens. The spectrometer was hereby scanned in steps of 0.05 mm (0.01788 A) over the range 164-214 mm (58.66-76.54 A). In each of the 1000 intervals the B-Ka. intensity was measured during 10 seconds after which the counts were transferred into successive channels of the multichannel-analyzer and displayed on the CRT of the TN-2000 system.

After completion of the spectrum the data were automatically stared on floppy disk and the procedure repreated on a different location of the specimen. The beam current was measured at the beginning and the end of each measurement. The measurements were carried out in the following alternating sequence: Baron

standard, 3 times a boride specimen, Baron standard etc. Allintegral measurements were carried out over night and in the weekend, so, full advantage of the automation system was taken.

All tagether approximately 500 spectra have been accuinulated. In a first run the global varlation of APF with peak position was established (See chapter IV). In a second run certain gaps in the graphs (missing crystallographic orientations) were fllled in by rotating the specimen mount with respect to the first run. In principle each location was chosen to represent a different grain in the specimen. In general this could easlly be accomplished using the backscattered image in which differe.nt grains could be distinguished by their differences in gray-tones.

I

I

-25-The vast ma)ority of APF measurements were carried out with the same 1-3

accelerating voltage of 10 kV. We have shown befare that the APFs for Carbon-K were essentially independent of accelerating voltage and there is no

a.

reason to expect that it would be different for the case of Boron-K . The beam a.

current was usually 300 nA.

The Pulse Height Analyzer (Ortec) conditlans were chosen to produce a pulse of 2.0 Volt (Counter High Tension 1700 Volt; Gain 64 x8). The lower level (threshold) was set at 0.6 Voltand the window at 3.5 Volt. These settings (see Fig. 11.1) were selected in order to cut off the continuous noise level and at the same time to accept as much of the pulse as possible, thus allowing for possible shifts.

The stared spectra were processed to obtain the net integral (area) and peak intenstties from standard and specimen by subtracting the linearly interpolated background over the relevant region of interest. In cases of speetral interference like the Nb- and Mo-borides, where a relatively strong M-C peak interferes with the B-K a. peak (see Chapter IV) a special procedure has been used tostrip the M-C. peak from the spectrum. This will be discussed in Chapter IV.

The accuracy of the APFs is, of course, strongly dependent, on the system analyzed: individual measurements in TiB

2 have an estimated accuracy of better than ±. 3% relative. In a heavily absorbing boride like Ni

2B this probably deteriorates to ±. 8%, mainly because of the uncertainty in the peak count rates which drop to- 100 cps.

On the basis of the large number of measurements performed we expect that the final APF's quoted have an accuracy of±. 1.5% for the "easy" specimens and ±. 3% for the "difficult" ones.

III.5 Measurements of Peak k-ratios between 4 and 15 (30 keV)

The APF-concept has been introduced to facilitate accurate integral

measurements with a considerable reduction in time. Onc.e the APF for a certain system is known, or like in the present case the varlation of APF with peak position, the measurements can simply be carried out again on the peak. Subsequent

multiplication of the peak k-ratio with the APF for the appropriate peak position will simply yield the correct integral k-ratio. In order to appreciate the reduction in time it must be realized that when using the conditlans applied by us only 4 boride spectra can be recorded (apart from the standard) during one night. If the range 4-30 keV has to be measured at 9 different accelerating voltages then two weeks would be

necessary and still the statistles would be very poor with only 4 integral k-ratios per point.

Using the APF-concept this full range can be measured in much less than a day and with very much improved statistles as we will see later on.

-26-With a few exceptions (La, U), where no suitable elemental standards are available, the k-ratios of the metals have also been measured over the range 4-30 keV. Where possible more than one X-ray line has been measured (e.g. W-L and M )

a. a.

in order to supply complementary evidence to the boron-measurements. In view of the very high beam currents riecessary for Boron-K , which would cause dead-time

a.

problems for the metallines, it has been decided to measure the latter separately. The procedure for the me tal lines was the following: In a preliminary survey 10 suitable areas for analysis were selected. These locations were stared in computer memory.

In each area 5 measurements were taken and the automation system was instructed to move the specimen over a speelfled dlstance in a speelfled direction. In thls way 50 measurements were performed for each specimen and for each accelerating

voltage. The beam current was usually ad)usted as to ensure a maximum count rate in the order of 3000 cps to avoid significant dead-tlme correctlons. A total number of 196 metal k-ratios were thus collected. Backgrounds were measured on elther slde of the peak and interpolated. The standard deviations in the final k-ratios are usuaUy better than 1%.

Por the Boron measurements a slightly different procedure was followed. Here the airjet was used and the 10 areas were selectedeach to represent a different grain and, consequently, probably a different crystallographlc orlentatlon. Care was taken that all 5 consecutlve measurements were allinside the same grain. Por each new area a very slow peak search procedure was used to ensure tha t the measurements would be performed at the correct peak posltlon. Backgrounds were usually measured at±. 25 mm <±. 8.94 A>, except where spectralinterference occurred <Ta, W, see Fig. IV.l). The peak k-ratlos for each area were next multiplied by the APP approprlate for the observed peak positlon. Inthls way a total number of 180 k-ratlos were obtained. Insome cases, with not too heavy absorption and/or high boron contents CB

4C, AlB12, TiB2 etc.) the measurements were performed in the range 4-30 keV; in heavier cases of absorptlon the measurements had to be braken off at 15 keV. Beyond that value the observed count rates simply became too low.

-27-IV. RESULTS

IV.l Boron-K emission spectra

- - a.

Fig. IV.l gives a survey of the B-Ka. spectra emitted by the various borides. All spectra have been moved from the full memory (1000 steps; 10 secondscounting time) to the flrst half of the memory before platting which means that the stepsize has become 0.1 mm and the counting time 20 seconds per step. The other conditlans are the sa me for all spectra: 10 kV and 300 nA beam current.

Upon inspeetion of these spectra some features become iinmediately obvlous:

In the flrst place there is a very large varlation in peak shapes possible from one

boride to another. Some spectra appear to contain only one single-valued Baron-component CTlB, TiB

2, ZrB2 as far as the short-wavelength extremes are

concerned), while others clearly contain at least 2 components (e.g. LaB

6, CoB,

NiB).

In several cases even three components can be distinguished. Examples are B,

SiB

3, SiB6, AlB12 and B4C. The most extreme example is BN where two separate

small peaks are visible which genuinely belong to the B-K emission spectrum. a.

In the second place it is apparent that shifts in the peak positlon of up to 1 mm (0.358. A) relatlve toelemental Boron can occur; not only from one boride to another, but also within the same boride from one crystallographic orlentatlon to another.

Fig. IV.l.a-h---Survey of the B-K spectra emitted by the 27 binary

a

borides investigated

in the present work as compared to that

of

eZementaZ

Boron. ExperimentaZ

conditions: 10 keV,

300

nA,

Zead-stearate crystaZ. In those

cases

where

onZy

one spectrum

is

given no peak shape aZterations in dependenee on

crystaZZographic orientation

have

been found. In

aZZ

other

Wavelength (~) 60 65 70 75 60 65 70 75 I 2SiiK

""P.

-

B I.H.L . - LaB 6 i. H.E. I~<!KJ S&K l'28K 64K V1 .,; ....

...

::z: 2 ~ !i4K ::. :S<!K 0 0 Li u 11 8

.

l/1. llllt. !91'!. '21!1'!. '2!1'!. 171'!. ]81'!. !91'!. 21!8. <!lB. 1'111 1111 I 3'2Kl

'"l

N

A

())

-

BN i. H. (. - UB4 1\ i. H.l.

'

HKJ 'HK

'"j

_I

_\

!SK

~

·

818'2

.,..

...

₂ ::. 8]9(!

e

0 u I! ll J/1'!. J 81!. 181'!. 21'111. Lilt. 1711. 181'!. !911. 21'11!. () 11. 1111 1'111 Fig. IV.l.a

-29-..;

"'

"'

... N N

'"'

""

i i .... ..; ....

...

...

...

N

"'

0 ...

...

...

"'

"'

"'

...

..; \0

"'

"'

N N ... çQ z:. çQ 1:: .-I ,z:. .-I • 1:: ,.:(

...

,.:(

...

"

"

0

I

\0 ó< ~ >::; >::; r. r. "-r. 1"-1 ~ UI N N UI

...

er

"'

N UI

...

..,

"'

"'

ID ..,.

"'

~

Cl' SI <:

"'

CU .... 'SLHOO.J 'SLHOO.J CU ()) >

"'

·~ 3 si k,

"'

...

... N N

..,

w

1:: i. ....

1-...

...

...

...

N N 0 ... ..; ..;

"'

"'

"' ..; \0

...

"'

..

M 1.0 çQ 1:: çQ z:. · rol ·1:: · rol · 1:: Ul

...

Ul

...

"

"

0 >.!) >::; >::; >:: ""-r. UI N N ID

...

N

...

ID N

...

U)

"'

".. N t1) s

""

"'

'SLNOO.J ' SLNOO::J

Wavelength (~) 60 65 70 75 60 65 70 75 3;!1( _J21ll<

...

TiB l.H.l. - B4C l.tl.E. 24ft 961( 16K

•

,

641<

•

I

,

11'1 I 11'1 2 .... :z ;::,. _{8}92 ;::,} 321< 0 ₀ u _u

"

e J;lil. 188. 1'1!1 198. U8. 21 B. 178. 1111!.. 198. ZV.il. 21 B. I w 32!< _32K 0 1. 11. E. 1. H. (. I TiB 2

_•

VB2

' '

•

441< I

•

24K I

'

'

_'

•

_•

I _I I

'

I _I 16.1<

_•

I

_'

\

16K

•

I

'

•

_'

11'1

•

'

'

11'1 ....

'

.... z

'

z ;::, ₁₁₁₉₂

'

;::, ₈₁₉₂ 0

'

0 u

\

Ll \

,,

...

il ₀ 178. I 1111. 198. 288. 211. l ;lil, 188. 198. 'ZBB. i1B. MM _Mf'1

Fig

.

IV.l.a

Wavelength (~) 60 65 70 75 60 65 70 75 Jli~ 32K 1. H. [. 1. H.E. - e r a CrB 2 -~~· 12K

'

.

'

.

'

24K

•

•

•

_'

•

_'

•

_'

8192

•

I _161<

•

I

•

I

•

•

•

I V>

•

'

.,; ....

_"

...

2

'

2; :::> _41!96

"

:::> ₈₁₉₂ 0 Q u

\

u

'

il il

Pil. !Bil. J9il. 2ilil. 2111. l?il. lei!. 198. 2ilil. 21 il.

MM MM I 41!96 8]92 w 1. H.E. ...

-

Fe 2B FeB I Jen 6!44 2il48 41l96 V> .,;

...

_z

...

_z :::> _{lil24 ::::> 2il48} 0 0 L> L> il B

l/lil. l8il. l9il. 2illl. 21 il. l /il. l8il. l91il. 2ili!J. 2lil.

MM 1111

Wavelength (~) 60 65 70 75 60 65 70 75 8!9<! - 4096

- c o s

l.H. (. _Ni 2B l. t1. (. 6H4 :sen 4896 . 28~8 "' .,; .... t -% z :::> <!848 :::> 101~ 0 0 u u 8 8

DB. l8B. !SI!. 281!. 218. l/il. ]811. JSil. "!0. 2)0.

!1!1 1'1!1 I w 4896 41!96 IV - _Co2B l.tt.E.

-

_Ni l. t1. (. I 3B )872 :sen "ZIH8 21!48 "' .,; i t -z :::> lll"Z~ :::> l8"Z4 0 Q u u ll 8 l/8. l8B. l!!B.

zn.

"ZlB. l 78. 180. JSB. <!BB. 21 B. !1!1 I'IM Fig. IV.l.e

6() (,'j 70 75 NiB 1. H.E. 4il96 ::.. ::> c 21148 l.J 8 L'll. 1811. J91l. 21!8. 218. !'IM 1281<

-

ZrB 2 l.H. E. 9611 6411

"'

~ :7. ::> :32~ t.l l.J 0

178. 1 Bil. JSil. Zilil. 218.

1111 Wavelength eR> JSK .,; ,... z :::> Q l.J 8192 &144 4ilSS "' ,... 2 :::> _Q <?il48 l.J ll 60 MoB 178. 1'11'1 - W B 1.?8. 1'1M

Fig. IV.l.f

65 70 75 T.H.E. J"88. 198. Zilll. 2J0. I w w T .11. (. I 188. 1911. Zilil. "2 JIJ.

Wavelength (~) 60 65 70 75 60 65 70 75 &4k 81.92

Nb-M

1-

T.H. E. NbB TaB

---·

ç qei\J

.

fl

6144

B-K

(l ~ZKJ

...

J I q99G "'

V

.,;

·-7.

5

'21l41l ;::, _16K C> 0 w u 0 ll DIL 181!. 198. 21lll. 21 ll. Pil. 1 81!. 1 .90. 21lll. '1 ll. 1'11'1 _1'11'1 I w

""

I 54K1

0192]

NbB 2 1. H.E. - T a B

"

l.H.f:. ---· 2 'llli'IJ ~ _&144

~

I

I

~21'1 _409&

"'

"' ; .- ~-::J 1&1'1 ::J :ZIH8 0 ₀ u _u

a

..

1!1

111). 1811. 190. 2illl. lll:l. 111!1. 180. l9il. :Zilll.

'

J

_0.

1'11'1 _1'11'1

60 6!> 70 G4~ -NbB 1. t1. L 2

_B-K

CL qal<

_Nb-M

ç )2~ V) 2 :::> 161< 0 u ll 171!. 1 !lil. 1 91!1. 21!11. MM 64~ _NbB 2 eorreeted T.ti.E. 48~ 32~ V) z 161< :::> c u

"

l/8. 1 BB. 198. Zllll. 1'11'1 75 z 111. Zlll. Wavelength l28k ss~ 64~ .,; r-:>: '32~ ::;) c u 1!1 S4tl 48tl 32tl V) r-2. ::;) _16K c u (~) 60 Nb 171!1. Ml'l - N b 1 ?ll. 1'111 Fig. IV.l.h 65 70 75 me tal 1. tf. E'. 191!. I w (JI idealized T.lol.f. I I Bil. 198. .281!. z 11'1.

-36-Where thls has been observed the extremes in the spectra have been represented in Fig. IV.l. (solid vs dotted lines). It must be emphasized that apart from the shifts in peak positions als pronounced peak shape alterations, due to the filtering of polarized components from the B-K emisslon spectrum by the action of the

a 11 12 13

analyzer crystal, take place. (See also 11.2). On theoretica! grounds ' ' the presence of polarized components can only be expected in non-cubic compounds. Tha t explains why the only cubic boride in the present investigation (LaB

6) does notshow these effects. As it turns out, however, 13 out of the 27 borides do not · show the varlatlans in peak positions and shapes, although their crystal structures

would allow this. Among these is elemental Baron ltself and this has been the reasen why we adopted this as the Boron-standard. The only other posslble candldates would be LaB

6 or AlB12, consldering the magnitudes of rnac's in all other cases (Table TI.1 ).

The boridesof Nb and Mo (Fig. IV.1) give rise toa special problem due to the presence of the MC lines of the metals in high intensities very close to the B-Ka peak. In the farmer case the Nb-MC line is at the right hand side of the B-K a peak; in the latter case at the left hand side. Several procedures have been tried in order to strip the MC lines from the Baron-spectra.

Efforts to multiply the spectrum from pure Nb (upper right-hand side) by the calculated k-ratio for the Nb-MC line and to stripthls from the NbB

2 spectrum failed, due to small shifts in the peak posltion of the MC line. As a result areas with negative count rates appeared in the resulting spectrum. The following procedure was mo're successful: It was observed that the MC peak in the Nb-standard was hlghly symmetrical and that the rlght-hand slde of the MC peak in the NbB

2 spectrum (upper left-hand corner) is probably not influenced by the B-K peak. Symmetrizing the _{a .} rlght-hand half of the peak (approx. 203-214 mm) and connecting the new left-hand end through linear interpolation to the background at approx. 180 mm yielded the so-called "idealized" Nb-MC spectrum in NbB

2 (lower right-hand corner). Thls was finally stripped from the NbB

2 spectrum to yield the corrected NbB2 -spectrum. The result is shown in the lower left-hand corner of Fig. IV.I.h. Apart from a very small artlfact at 200 mm the strlpped spectrum has a quite reliable appearance, when compared to e.g. WB, UB

4, VB2 or CrB2, which gives us confidence in the correctness of the procedure.

In MoB the left-hand side of the Mo-MC peak was used for symmetrlzing and the right-hand end of the symmetrized MC peak connected to the background at approx. 200 mm.

-37-Considering all the variations in peak shifts and certalnly those in peak shapes lt is clear that lntenslty measurements made on the peak will have to fall for ultralight elements. Errors of 20-30% can easlly be made. Even if the spectrometer is retuned for each measurement in a certain compound, and lf the measurements are performed relative to the same compound as a standard, substantial errors are stllllikely, due to the varia tions in peak shapes.

As we have stated befare these effects can all be accounted for by using the

appropriate APF. The only alternative is to record the full integralemission spectrum for each measurement, whlch is evidently not a very attractlve option for a

wavelength-dispersive spectrometer.

IV.2 Area/Peak Factors for B-K relatlve toelemental Baron. a.

1-3 .

It lies at hand to draw a parallel to the case of Carbon where we have shown that strong shape alteratlons but nodependenee on crystallographlc orientation occur. In addition we have found a strong dependenee of APF on the atomie number of the roetal partner, glving rise to a consplcuous saw-tooth like varlation of APF in the periadie system.

In table IV.l the APF's for B-K are represented numerically, tagether with the a.

observed peak positions. A single value for the APF indicates that no shape alterations withln the same compound have been observed. The results of the indivldual Area/Peak measurements graphlcally displayed for each boride as

a

function of detected peak posltlon are given in Appendix A where an lmpression can be obtained on the accuracy and the reproduclbillty of the measurements.

From Table IV.l it is evident that both substantial narrowing of the Boron-peak with respect to elemental Baron, as wellas broadening can occur. Thls is reflected in the varlation of the APF whlch can assume values as low as 0.665 in ZrB

2 or as hlgh as 1.31 in NiB.

A very peculiar observation was made on BN (Carbon coated). Thls specimen was a hollow cylinder of CVD-deposited BN, cut and polished in a plane perpendlcular to the cylindrical axis. The APF measurements were performed in a sequentia! way, covering approx. % of the circumference. The position of the B-K peak was thereby

a.

found to behave in a most peculiar way: At the starting position it was found to have a minimum value 091.05 mm), 90 deg. further the maximum value (191.60 mm) was found; almast the minimum value 091.20 mm) was found agaln in a positlon opposite to the starting point .etc. Strange enough no peak shape alterations could be detected.

-38-Tab1e IV.1

Area/Peak Factors for B-K radlation re1ative to Boron. The figures in parentheses are a.

the number of measurements on which the data are based. (see a1so Appendix A)

Boride APF Peak.pos.(mm) Wave1ength (,6,)

B 1.000 (by def). 189.00 67.60 B 4C 1.014 (27) 189.00 67.60 BN 1.198 ( 9) 191.05-191.60 68.33-68.53 A1B 2 1.152-1.095 ( 7) 189.50-189.90 67.78-67.92 A1B 12 1.008 ( 9) 189.03 67.61 SiB 3 1.003 ( 5) 188.85 67.55 SiB 6 0.922 ( 5) 189.00 67.60 TiB 0.690-0.835 (12) 188.80-189.45 67.53-67.76 TiB 2 0.799-0.945 (12) 189.35-190.25 6 7. 73-68.05 VB 2 0.950-1.045 ( 9) 189.85-190.70 67.91-68.21 CrB 0.825-1.000 (19) 189.25-190.25 6 7.69-68.05 CrB 2 0.950-1.090 (10) 189.90-190.90 67.92-68.28 Fe 2B 1.242 ( 6) 188.90 67.57 FeB 0.985-1.160 (11) 189.55-190.00 67.80-67.96 Co 2B 1.015 ( 5) 189.10 67.64 CoB 1.186 ( 5) 190.40 68.10 Ni 3B 1.020-0.935 ( 7) 189.40-189.85 67.74-67.91 Ni 2B 1.059 (14) 189.10 67.64 NiB 1.060-1.310 (18) 189.80-190.45 67.89-68.12 ZrB 2 0.665-0.915 (13) 188.90-189.90 67.57-67.92 NbB 0.775-0.855 ( 7) 189.05-189.55 6 7.62-6 7.80 NbB 2 0.810-1.025 ( 9) 189.50-190.55 67.78-68.16 MoB 0.936 ( 4) 189.70 67.85 LaB 6 0.898 ( 6) 189.55 67.80 TaB 0.853-0.881 ( 5) 189.55-189.85 67.80-67.91 TaB 2 0.990-1.120 ( 8) 189.95-190.60 67.94-68.17 WB 0.978 ( 5) 189.80 67.89 UB 4 1.028 ( 6) 189.95 67.94

APF', relative to Boron 1.4 1.2 1.0

•

B 4C .8 .6 5 Fe 2B

•

•

CoB N1B BN AlB2

•

I

'"1

}

~,

r~,

r··,

I SiB 3o Ni 2B

•

_INi;B

•

•

V/8 I AlB12 Co2B

f

~

·

•

•

0 [

C<O

]

LaB 6 NbB 2 • I S1B 6 I TiB2

I

TaB NbB]

d

ZrB 2 I 24 40 41

Atomie number of Metal partner

Fig.

IV

.

2

.---

Area/Pea

k

Factor~

relative toelemental

Boron~

as

a

funetion

of

the

atomie number

of

the m

et

al

partner for 27 binary borides

.

Solid points indiaate that only

one

unambiguous

AP

F

app

l

ies:

i

.

e

.

no dependenee on erystallographia orientation

exist

s.

Bars

indieate the variation in APF with peak position (arystall

.

orie

nta

tio

n)

in

aU other

cases

(See also Appendix A)

.

Ull 4

•

I w 1.0 I