A new electron spectrometer design:II

Journal of Efectran Spectroscopy and Related Phenomena, 8 (1976) 395410 @ Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

A NEW

ELECTRON

J. D. ALLEN, Departments

SPECTROMETER

JR.,* J. D. DURHAM**, of Chemistry

and Physics,

DESIGN:

GEORGE University

I1

K. SCHWEITZER

of’ Tennessee,

Knoxville,

and W. E. DEEDS Tennessee

37916 (U.S.A.)

ABSTRACT

A novel charged particle energy analyzer of simple geometry is described. Expressions for the potential distribution and electric field components defined by the geometry are given. A raytrace program using these components is discussed and results of its application are presented in graphical and tabular form. Also presented are experimental results for two versions of the analyzer in the form of representative spectra of well known species (Ar+, H2+, Of). Finally, a comparison of the new instrument with the well known cylindrical mirror analyzer is outlined which suggests the superiority of the new instrument under the specific conditions given. INTRODUCTION

In a previous note’ we reported some preliminary results concerning the predicted behavior of an electrostatic charged particle energy analyzer of novel design, the conception of which was motivated by our need for an ~instrument of reasonably high sensitivity and moderate resolving power (EjAE - 100) suitable for use as a photoelectron spectrometer. Besides the requirements on the resolving power and sensitivity, several additional criteria were to be satisfied by the analyzer, not the least of which was that of extreme mechanical simplicity so that it might readily be constructed by inexperienced personnel. This requirement at once limited the range of electrode configurations to be considered to those which could be machined to the necessary tolerances with only the most modest equipment. Concomitantly, the instrument was to be as insensitive as possible to excursions from the design parameters. There were if possible to be no difficult-to-correct fringe fields such as those produced at the electrode gaps of most of the analyzers currently popular for PES and ESCA work2. As a result of this requirement the analyzed particles would * Author to whom correspondence should be addressed. ** Present address: Hinds Junior College, Raymond, Mississippi 39154.

396 and exit the analyzer through slits in surfaces which were automatically equipotential surfaces. Finally, no geometry was to be considered if the potential distribution in the region of: particle trajectories could not be expressed in a way. which would lend itself to rapid numerical integration of the equations of particle motion. In this article we describe our method of treatment of the analyzer’s behavior and present some characteristic results. Finally we illustrate the observed performance of the first and second versions of the instrument with several representative spectra.

enter

ANALYZER

GEOMETRY

A geometry which satisfies all of the requirements outlined in the Introduction is depicted in Fig. 1. Its physical realization could be described as an oversize pill box, although for obvious reasons it has been dubbed the Bessel Box. It is characterized by two identical circular end plates of radius A and an intervening cyIindrica1 wall of radius A and length C. The end plates are maintained at the same potential (for convenience, ground), the cylindrical wall for reflecting electrode) being held at a potential V,,. Annular entrance and exit slits coaxial with the instrument’s symmetry axis are carried by the lower and upper end plates, respectively. These slits are characterized by diameters D, and D, and widths S, and S,. Three more parameters

s

_l Figure 1.

Geometry of the BesselBox Analyzer.

397 define the sample volume location and its shape. S plate to the center of the sample volume. W is the diameter. This restriction of sample shape to one severe limitation and is in any event appropriate phase photoelectron spectroscopic work for which ANALYZER

FIELD

is the distance from the lower end sample volume length and D, its of circular cross-section is not a for the requirements of the gas the analyzer was intended.

DISTRIBUTION

The raytrace program which we shall describe below involves expressions for the two components of the electric field within the analyzer, E,, and E,. As a result of the cylindrical symmetry of the instrument there is no azimuthal fieId component. The required components are obtained from the potential distribution @(r, z), and @ is in turn determined by the Dirichlet boundary conditions @(A, z) = Yo, s 5 z I

c + s

@(Y, S) = @(v, C) = 0,o 5 r < A and is given by3 @(r, Z) = (4V,j7~) E ((l/k) sin (kn(z oii

-

LUdCj s)/(3)) -~,C~WCj

(1)

The field components within the analyzer are thus given by E, (r,

z) = -

(iW (r, z)‘dr)

= - (4V&)

E (sin (kn (z k odd

s)/C):;(k&~; 0

and

IO and -I, are the modified Bessel functions of order zero and one, respectively. The expression for E, follows from the relation, 1, (v> = Cl (9 The field components outside the analyzing volume are of no interest except in the regions occupied by the electron source (or sample) and the detector where they are forced to be zero by shielding. THE

RAYTRACE

PROGRAM

A raytrace program suitable for the investigation of charged particle analyzers must be capable of computing particle trajectories for any set of initial particle coordinates and velocities, and should, in addition, display the trajectory end point

398 parameters in such a way that the instrumental line shape, focusing characteristics, resolving power, and any other related aspects of the instrument’s behavior may be readily discerned. The essential aspects of a program which we have written to satisfy these requirements are discussed in this section. We make no claim for the originality of the approach, but describe it in some detail because of its broad range of applicability. The raytrace calculations for the pill box analyzer can conveniently be divided into three sections, the first and the last involving the rectilinear motion of a particle in the field-free sample and detector spaces, and the second, the motion in the potential distribution described by eqn. (1). It will prove useful for the subsequent description of particle motion in the anaIyzing region to represent the trajectory of the analyzed particle (hereafter electron) in the sample space by a vector vwhose components are referred to a right hand coordinate system (x, y, z) centered at the midpoint of the cylindrical sample volume. The utility of this description of the electron’s initial field-free motion is the facility it brings to the definition of the initial velocity components for the middle non-zero field section of the trajectory, The coordinate system and the pertinent quantities associated with it are depicted in Fig. 2, where

Entrance

Slit

Sample Volume

Figure

!

fL/?.

0 Y

2. Sample volume-entrancelslit

geometry.

399 q. is the azimuthal angle of the electron’s point of origin in the sampIe space, ‘pl is the azimuthal angle of the electron’s point of entry into the potential distribution (taken to be zero with no loss in generality), y. is the radial coordinate of the electron’s point of origin (0 < r. 5 0,/2), r1 is the radial coordinate of the electron’s point of entry into the potential distribution (0,/2 - S,/2) I r1 I (0,/2 + S,/2), zU is the z coordinate of the electron’s point of origin (-H/2) s z. 5 (H/2), and z1 is the z coordinate of the electron’s point of entry into the potential distribution (zr E S). The electron’s initial velocity (i.e., its velocity as it traverses the distance between its point of origin and its point of entry into the potential distribution) can be expressed as V=

vd

where V is the magnitude of the initial velocity and h is a unit vector in the direction of the initial velocity given by (2= ~

d

PI

with d = (Xr - X,)i: + (JJl - yo)j The components

+ (zr - .&

not previously defined are given by

Xl

=

l-1 coscp,

x0

=

Yg cosipo,

Yl

=

rl

y.

= r. sincp.

sinGPI and

(2,j, I$) are the usual unit vectors in the (x, y, z) system. In order to make the transition from motion in the field-free region to motion in the analyzer interior it is necessary to relate the initial velocity in the sample volume to a new set of initial velocity components along axes appropriate to the cylindrical symmetry of the analyzer. This is conveniently done by defining a set of unit vectors (PI, If;, fI) with common origin at the electron’s point of entry into the analyzer interior (rr, ql, 2,). 3, is directed radially at the entry point, PI tangentially at that point, and 5, along a line parallel to the instrument’s axis. The required unit vectors can be expressed in terms of the previously defined Cartesian units vectors as 3, = cosqo,^l + sinqo,j L, =

-ssincp,i + coscp,j and

c z1 = rt

400 The new initial velocity components are obtained by performing the dot product of the field-free velocity v with each of these unit vectors in turn. Thus, i, = 3. P, = (V/D)((x, r,&,

=

- x,)coscp,

3. p1 = (V/D){-(xi

2, = p* 2, = (Y/D)(z, D = ((x1 -

+ (yi

- x,)sinqp,

- Yo)sinqD1) = + (vl

V,

- yO)coscp,) = V,

and

where

- zO) = V3,

%)2 + (Y1 - Y0)2 + (21 - z*W2

In addition to the initial velocities and coordinates for each eIectron, the initial linear components of acceleration, (U,, _!?J, at the electron’s point of entry into the potential distribution (rl, cpl, zl) are required for the initiation of the raytrace process. These are calculated by substituting the initial coordinate values in the general expressions for i-’and 2 obtained from Lagrange’s equations for the nonrelativistic motion of a particle in the potential given by eqn. (1): ii = (l/300) (4eVJmC)

Z (sin (kn (z -

z,)/C))

k

I,(km’IC) ~-~ lo( knA/C)

+

r:V; ~ r3

(2)

odd

2 = (l/300) (4eYJmC)

Z (cos (kn

(Z

-

zlj/C))

I°CkZ”‘) I,(kWC)

(3)

oakd

In the above e and 1-22 are the electronic charge and mass. All other terms are previously defined except for the factor (l/300) which permits expressing V, in volts. Units are otherwise c.g.s. Since the electron experiences no torque, angular momentum is conserved and the angular acceleration 4 need not be calculated directly. Raytracing in the angular coordinate thus will involve only the angular velocity, (4) A trajectory in the non-zero field region of the analyzer is generated by straightforward numerical integration of the component expressions of the vector equation, ASi

=

eat

+ (lj2)ia’(At)’

(5)

according to the algorithm, i-1 Saj = ZASai + Sal, i=l

a = (r, 4, z)

(6)

subject to the applicable initial conditions. ASi in eqn. (5) is the vector displacement of the electron from the ith to the (i + 1)st position along the path. AS, is thus the displacement from (rl, cpl, zl) on the lower end plate (defined as the first interior position) to the second interior position.

401 vi and Gi are respectively the velocity and acceleration computed at the ith position. The two velocity components, ii and ii, are obtained from the components of (7) The two acceleration components, pi and z~, are obtained from eqns. (2) and (3). As indicated previously the angular part of the raytrace calculation is treated somewhat differently from the linear parts since no account is taken explicitly of the angular acceleration. Thus only the first term in eqn. (5) is used to compute the angular increment with the angular velocity obtained not from eqn. (7) but from eqn. (4). Note that although the angular acceleration is not explicitly included, it is in fact taken into account through conservation of angular momentum so that eqn. (4) gives the correct angular velocity at the beginning of any increment. The important effect of the angular momentum on the radial acceleration is expressed exactly by the second term in eqn. (2). The axial acceleration given by eqn. (3) does not depend on the angular momentum. The time interval of integration Af is chosen to limit the end point kinetic energy error to l/2 percent or so. Values for At range in practice from 2 x lo-” s to 5 x lo-’ s. Provision was made originally to vary At along the trajectory to maintain point by point limits on the error in the total energy, but it proved time consuming and (fortunately) unnecessary to do so, at least for the class of trajectories so far considered. In the raytrace algorithm eqn. (6) S,j is the value of the subscripted coordinate after (j - 1) steps and S,, is the initial value of that coordinate. AS,i is the ith increment in the corresponding coordinate computed by eqn. (5). Note that there is no sum for j = 1 and that the initial conditions determine the first increments. Raytracing in the potential distribution (i.e., summation according to the three constituents of eqn. (6)) continues for a given ray as long as the following conditions are satisfied: ri < A

(8)

s

(9)

2-i >

Zi
(10)

If for any point, (ri, Zi), eqns. (8) or (9) fail to be satisfied the program terminates the raytrace for the current set of initial conditions. Failure to satisfy eqn. (10) indicates that a normal trajectory has been completed. Interpolation is performed to determine the final values of r, i, 9, and 4 for z = C + S. (A similar interpoIation is performed at ri = 0 for those higher order trajectories which intersect the axis with the additional proviso that Cpiis augmented by 71.) The final phase of the raytrace procedure begins upon the completion of a normal trajectory (i.e., one which terminates on the end plane at z = C + S). For each ray the final coordinates, Ye, cp,, and zr( = C + S), and the final coordinate

402

are used to determine the field-free detector space motion of the electron after it passes through the exit slit. It has proved convenient to express this motion functionally as the radial separation r(z) between each ray and the conical surface defined by the central ray. The central ray is defined as that ray which originates at the midpoint of the sample volume, and passes through the centers of the entrance and exit slits. Its energy and the associated final coordinates are determined by the central ray search subroutine of the raytrace program which by successive trials and interpolations finds the ray which satisfies the defining criteria. The central ray search requires on average five or six successive ray traces for convergence to within five micrometers of the desired final radial coordinate. In practice it has proved sufficient for comparison purposes to trace no more than 75 (and as few as 15) rays for a given set of instrument defining parameters. A plot is generated which displays the function Y(Z) for each of the rays accepted by the exit slit. Considerable insight into the focal and lineshape characteristics of the analyzer can be acquired simply from a cursory inspection of the plot. In addition to generating the pIot the program lists the initial conditions for each ray, the instrumental parameters pertinent to the run, the final coordinates (I, cp, z) for each ray, the kinetic energy error for each ray, and two parameters which are of importance in choosing the electron multiplier location, the minimum distance of approach to the axis by the electron in the detector space, and the corresponding z coordinate. Finally, the instrumental resolving power (E/AE) and figure of merit (defined for our purposes as the product of entrance solid angle and resolving power) are calculated.

increments

are

saved.

RAYTRACE

RESULTS

These

Perhaps the most difficult aspect of an empirical investigation such as this is the choice of reasonable starting values for the various instrumental parameters. In making our choices we tried to combine pragmatism and intuition. We chose the analyzer diameter such that the completed instrument would fit inside one of the seven inch diameter vacuum lines already serving PES instruments in our laboratory. Next we chose the length so that the sum of the surface areas of the two end plates equaled that of the cylindrical wall (thus, C = A). The initial value of S was chosen principally for convenience in machining and access. Slit widths were adopted which were approximately equal to those currently in use in our other instruments. Initial slit diameters were, on the other hand, quite arbitrarily chosen. However, once the initial S, entrance slit diameter, and entrance slit width were fixed, all subsequent work for the first comparison study was carried out using slitwidth values chosen for each Dl - S combination which would subtend a solid angle at the sample center equal to that subtended by the original slit. The value of V, was quite arbitrarily set at - 10 volts, its absolute value being without significance in the nonrelativistic limit. We chose for the value of D,, a diameter typical of the sample volumes in our other

403 instruments. Since in subsequent experimental studies it would be inconvenient to modify the sample geometry for each combination of S and D, we fixed D, at 0.10 cm for all computer work not specifically devoted to investigating the effect of varying the sample size. As it happened, the parameter values which we chose for the initial computer studies (and for the initial construction as well) yielded very nearly the hoped for theoretical resolving power of 100. These values are indicated in Table 1. In order to proceed as quickly as possible from so fortunate a beginning we adopted the following TABLE INITIAL A C S DO H

= = = = =

1 PARAMETERS

7.50 cm 7.50 cm Z.OOcm 0.10 cm 0.00

AND

RESULTS

D1 = 0.64 cm S1 = 0.038 cm

DZ = 0.64 cm SZ = 0.038 cm vo = -10.00 volts

(Eo/AE)ca~c. = 95.65 {Eo/AE)obs -

85

scheme for subsequent work. Keeping the initial values of A and C fixed at 7.50 cm we selected four values of S (2.00, 2.50, 3.00, 3.50 cm) which would characterize four subgroups of raytrace studies. Within each subgroup focusing behavior would be sought for each of several values of D1, the entrance slit diameter. Thus, for each D, we would determine the D, yielding the highest resolving power. Finally, after a reasonably extensive examination of the instrument’s behavior for each of the four values of S we would make modest variations (&- 20 %) in the ratio of C to A for several of the previously optimized parameter sets in order to gain an initial feel for the effects of alterations of the field defining electrodes. One further parametric restriction was imposed to limit the expenditure of computer time. This involved using as sample volume points-of-origin only those lying on the circumference of a circle defined by the sample volume diameter (D,) and centered at the midpoint of the sample volume (z = 0). It follows from symmetry that only the half-circle from qpo = 0 to 180” need be considered. Since for our work with gaseous samples the sample volume would be defined by a pinhole at (z = 0) such a restriction is entirely justified. In addition to the parametric restrictions discussed above a limitation was imposed on the time to be spent investigating the higher order modes of operation. As we pointed out in our previous note the potential distribution produced by the geometry of this analyzer can support a number of trajectory classes (or modes) with common initial and terminal radii on the end planes, each characterized by a different energy and a different number of axis crossings (we define mode number as the number of axis crossings for the central ray). These higher order modes exhibit extremely high dispersion but, unlike the fundamental mode, do not exhibit focusing

to any useful degree. Some thought had been given to the elimination of the higher order modes should they prove troublesome and two approaches were at once evident. The first, and simplest, involved baffling corresponding 1SOo segments of the entrance and exit slits. A more elegant solution would have depended for its operation on the fact that the fundamental mode is characterized by the highest kinetic energy. Thus, a simple retarding element placed before the detector would have isolated the desired signal without the loss in sensitivity which would have resulted from the baffling. We have found in our computer studies and have confirmed experimentally that for the proper choice of instrument defining parameters the exit angles for the various modes characterized by a common exit radius differ sufficiently to permit ntlode separation by the insertion of a simple pinhole baffle between the annular exit slit and the electron detector. Thus, only for the first version of the analyzer was it necessary to employ the 180 ’ baffles. The results of our initial raytrace studies are summarized in Fig. 3 and Table 2. In Fig. 3 the figure of merit is plotted against a normalizing parameter (DJS) for the four S values. Note that since the acceptance solid angle was maintained constant for the pertinent raytrace calculations, the figure of merit can be interpreted as a relative measure of the resolving power. For reference, a figure of merit of I.OQ corresponds here to a resolving power of 53.82. Table 2 displays the results of three sets of calculations for common A, S, De, D,, and S,. For each of the three characteristic values of C {OXA, I.OA, 1.2A)

5

l

*

2.50 3.00

l

3.50

n

.

e

2.00

5 )

+

. n

.

Figure 3. Figure o>fMerit (F.M.) as a function of Ill/S.

405 TABLE 2 COMPARISON

OF BEHAVIOR OF THREE RATIOS OF A TO C

Qptimized for

Optimized for

C=A

C=1.2OA

Optimized C=0_80A

for

= 7.50 cm S =3_OOcm DO = 0.10 cm Dl = 1.35 cm Sl = 0.043 cm

A

D E

= =

Ec,/lJE

=

F.M.

=

1.35 cm 3.031 ev

1.60 cm 4.509 eV

128.37

1.10 cm 1.610 eV

149.14

2.416

105.27

2.806

1.981

the D, was determined for which the resolving power was a maximum. The superiority of the longer instrument is expected since the analyzed electrons not only spend a longer time in the analyzing field but make larger radial excursions as well. This latter fact is reflected in the larger E,, value which obtains for the longer instrument. EXPERIMENTAL

RESULTS

Figures 4 and 5 exhibit several characteristics of the analyzer in its first, unoptimized form (the version described by Table 1). Figure 4 is a spectrum of Ar+ excited by He 584 with a discharge lamp current of about 20 mA and a sample chamber pressure of approximately 15 millitorr. The importance of this spectrum lies not only in the fairly high sensitivity which it implies, but in the reasonably close agreement between the calculated resolving power (-96) and an observed value (- 85). Machining errors affecting the slitwidths probably accounted for most of the discrepancy. The slit diameters were, on the other hand, held well within design tolerances as is implied by the resulting close agreement between the calcuIated

Si’8

5.46

ev

Figure 4. Argon spectrum obtained with original version of the analyzer (see Table 1). Breadth of peak indicates excessiveexit slitwidth.

406

1.00

eV

Figure 5. Fourth band of OZ+ obtained with originai version of the analyzer.The structure near zero eV is due to oxygen impurity in the lamp.

and observed central ray energy values. An error of less than 0.2 oA in the measured position of the Ar 2P,,, line resulted when the calculated value of the corresponding reflecting potential (V,) was applied. Figure 5 is a spectrum of the fourth band of 0: taken under the conditions just described. No preacceleration of the electron beam prior to analysis was employed. The sensitivity for low energy electrons is apparent as is the relative absence of the background from scattered higher energy electrons. No baffles other than the 180” segment masks required in this first version for higher order mode elimination were employed. The annular slits and the pinholes were thus the only limiting apertures in the system. Following our initial success we made a simple modification which, it was hoped, would improve the resolving power, eliminate the higher order modes exhibited by the first version, and yet require only the most minor refabrication. The exit slit diameter was made as large as possible within the constraints imposed by the supporting end plate (which we did not wish to remake). Then the diameter of the entrance slit which would yield focusing for the new exit slit diameter was determined with the raytrace program. S was not altered. For the new slit dimensions (Table 3) the raytrace results indicated that elimination of the higher order modes would be (marginally) p ossible. In practice this was found to be the case; only the first order higher mode could be observed and its intensity was approximately 0.274 of that of the fundamental mode. The accuracy with which the new slits were fabricated is TABLE 3 MODIFIED A = C = S = DO = N =

PARAMETERS

7.50 cm 7.50 cm 2.OOcm 0.10 cm 0.00 cm

AND RESULTS

D1 = 1.00 S1 = 0.01 D~=1.66

cm cm cm Sz = 0.015 cm vo = -10.00 volts

Eo talc. = 3.027 Eo obs. = 3.030 CEo/AE)talc. = 296 (Eo/AE) obs. N 303

407 reflected in the close agreement between the theoretically predicted and experimentally observed values of both the central ray energy and the resolving power (0.1 ok error for the former and 2.4 oA for the latter). Characteristic spectra obtained with this version (again without preretardation) are presented as Figures 6 and 7. Figure 6 is a spectrum of Arf taken under the previously mentioned conditions. Figure 7 is a partial spectrum of Hl taken at room temperature under similar operating conditions. The J” = 3 c J’ = 3 transitions are clearly seen. After our more careful initial studies we decided to try to determine, at least crudely, the characteristics of the analyzer under the application of preretardation. To this end we hastily fabricated (from brass shim stock and copper tubing) a preretarding lens-sample chamber of approximately (!) cylindrical symmetry and inserted it in the analyzer. Our first spectrum of argon (what else) using this system exhibited a line width of 10 meV at a pass energy of 1 volt (see Fig. 8). The reduction in counting rate from the non-preretarded case in excess of that physically required by the retardation process is due to geometrical limitations of the very crude pre15000

:-_ . .

c/s

.. .-‘; * -. . -

..

-

528

.

5.46

rv

Figure 6. Argon spectrum .obtained with modified version of the analyzer (see Table 3).

16.50

lEu.35

BE

Figure 7. Partial spectrum of Hz+ obtained with modified version of the analyzer (see Table 3). Note the pronounced rotational shoulders.

4.28

4.4 6

v,

Figure 8. Argon spectrum obtained with preretardation applied to the modified analyzer (see Table 3 and text). C’, is the retarding potential. The anomalously low count rate is primarily the result of geometrica limitations in the very crude preretarding lens.

retarder. of 0:

Even

thus limited

and H20+

ground

which

of a future indication electron

the instrument

characterized

we have

paper.

not seen equaled

These

first efforts,

of the potential

which

WITH

In order we have figuration.

the new

to place

very

analyzer

beautiful will

be the subject

executed,

has for

spectra

and signal-to-back-

These

imperfectly

its performance

with The

of equivalent

high

served

resolution

as an photo-

dimensions = 6.lr,

ments (L,,

identical

is found

to

the choice

where

is the common

=

L,

S + above

result

It is interesting

should Box

of the Bessel

base resolving analyzers

again,

unoptimized)

order

focusing

con-

an acceptance

solid

3) imply

of approximately Bishop’s4

of inner

radius

axial source-to-focus

300. For a CMA

and source

disc diameter,

eqn.

By

(13).

defined

distance

equivalent

by the relation4 of the two instru-

C).

of the Bessel

characteristics

of a CMA

we note

solid angle,

158 from

in perspective

in the second

of Table

power

acceptance be

(and

CMA

resolving

we mean

The superiority

in a specific

data for the Bessel Box (those

power

ANALYZER

of the Bessel Box analyzer

that of the well known

dimensions,

the resolving

MIRROR

the characteristics

angle of 0.007 sr for the observed

better

in the literature.

although

THE CYLINDRICAL

compared

configuration

both

some

of linewidth

spectroscopy.

COMPARISON

L,

produced

by a combination

exhibit

power

over Box

by

no

means

the CMA.

seem to indicate

than the current

theoretical

that the CMA

be taken

Our brief

values

version

that

to

indicate

studies the CMA

yields

absolute somewhat

of the new instrument,

in excess of 10’ for the quoted

seems to suffer more

the

of the point-source

from

although

parameters.

the use of sources

of finite

409 size than does the Bessel Box (again, for the quoted parameters). Perhaps more significant is the fact that a carefully optimized Bessel Box of approximately twice the length of the original version can in theory yield an order of magnitude better resolving power than the modified version of the original (i.e., E,/AE - 3000) at essentially the same acceptance solid angle. Parameters leading to such performance form the basis of the instrument discussed briefly below. WORK

IN PROGRESS

We are currently pursuing two projects involving analyzers of the Bessel Box type. Of the greater importance is the construction of an entirely new Bessel Box for the high resolution study of gases. Design parameters have been conservatively chosen to yield linewidths of something less than 1 meV for a 1 V pass energy at counting rates which will equal or exceed those exhibited by the original instrument. Source width and kinetic broadening will be limiting factors in the obtainable line width in most cases. When necessary a molecular beam will be used for the reduction of the latter width. With attention to operating parameters the lamp width can probably be reduced to slightly less than 1 meV. During construclion of the new Bessel Box we will continue to use the present instrument in a newly modified form. A properly designed preretarding lens has been installed_ In addition, the cylinder has been lengthened and reduced in diameter so that (with properly chosen exit slit parameters) a resolving power of five or six hundred should be obtainable at counting rates only slightly lower than those exhibited by the original instrument (for equivalent conditions of lamp current and sample pressure). This instrument will be fully operational shortly and if its performance meets our expectations we will present a brief note containing a complete list of parameter values for those who may wish to duplicate the device. CONCLUSION

We have described the theoretical treatment and experimental testing of a mechanically simple charged particle analyzer and have presented characteristic results. We have found the agreement between the calculated and observed results close enough to give us some confidence in the predictive ability of the raytrace program used for the numerical studies and have proceeded with the design and construction of a higher resolution version of the analyzer. Using our early data we have attempted a rough comparison of the CMA and the Bessel Box and find that for specified conditions the Bessel Box is substantially superior. It is noted that recent computer results (not included here) imply the possibility of a lo-fold enhancement in the Bessel Box performance.

410 ACKNOWLEDGEMENTS

We gladly acknowledge the financial support of both the Research Corporation and the National Science Foundation_ The services of the University of Tennessee Computing Center were indispensable. Allen, Deeds, and Schweitzer thank Durham for his exemplary efforts in the fabrication of the original instrument. REFERENCES 1 2 3 4

J. D. Allen, Jr., J. Preston Wolfe and Geo. K. Schweitzer, Znr. J. Mass Spectrom. Zon Phys., 8 (1972) 81. Kenneth D. Sevier, Low Energy Electron Spectrometry, Wiley-Interscience, New York, 1972, Chap. 2. W. R. Smythe, Static and Dynamic Eiectricity, 2nd ed., McGraw-Hill, New York, 1950, p_ 196. H. E. IBishop, J. P. Coad and J. C. Riviere, J. Electron Spectrosc., 1 (1972/73) 389.