The quadrupole mass filter as a commercial isotope separator

Nuclear Instruments and Methods 195 (1982) 447-456 North-Holland Publishing Company

447

THE QUADRUPOLE MASS FILTER AS A COMMERCIAL ISOTOPE SEPARATOR M.F. FINLAN, R.F. SUNDERLAND Cyclotron Group, The Radiochemical Centre Ltd., White Lion Road, Amersham, Bucks, HP7 9LL, England and J.F.J. TODD Chemical Laboratory, University of Kent, Canterbury, Kent, CT2 7NH, England Received 8 May 1981

The theory of operation, design, construction and evaluation of a prototype isotope separator using a quadrupole mass filter is described. The enrichment of 26Mg has been demonstrated and the performance has been found to be consistent with the specifications required of a commercial unit. The use of a glass envelope for the vacuum system has permitted the direct observation of the characteristics of the ion beam within the mass filter.

1. Introduction Although of very much more recent origin than the magnetic sector mass spectrometer, the quadmpole mass filter is now firmly established as an analytical instrument suited to a wide variety of applications ranging from residual gas analysis [1] to highly sophisticated computer-controlled GC-MS systems [2]. What is perhaps less widely appreciated is that the initial development of the quadmpole was in relation to its use as an isotope separator [3], and since these early reports [4] there seems to have been no systematic practical appraisal of the potential of the mass filter in this sphere. In fact if the requirement is to produce a few grams per year of separated isotopes of elements up to A = 100, then the quadrupole appears to be an attractive proposition compared to equivalent electromagnetic separators. Thus the absence of large, high grade, intense magnetic fields reduces the capital cost considerably and the vacuum requirements of the quadmpole are less stringent. Furthermore, since the mass filter will tolerate a significant kinetic energy spread in the ion beam without degradation in performance, much greater ion source extraction efficiency can be achieved leading to a higher rate of production. This paper therefore describes the construction and performance of a low cost prototype quadmpole isotope separator designed to meet the specifications detailed below. 0029-554X/82/0000-0000/$02.75 © 1982 North-Holland

2. Theory The radiofrequency quadrupole mass filter functions on the principle of the stability of ion trajectories in inhomogeneous oscillating electric fields. The theory of operation has been described in detail elsewhere [5,6] and therefore only a summary will be presented here. The electric field is established along the axis of an array of four rod electrodes coupled so that the potentials applied between adjacent rods are

Y X

RF

i SOURCE

GENERATOR I

SWEEP

DC GENERATOR DCIRF RATIO

Fig. 1. Schematic diagram of the quadrupole mass filter.

448

M.F. Fin lan et al,/Quadntpole mass filter

equal and opposite (fig. 1). Ideally the internal surfaces of the electrodes should be hyperbolic, but provided the values of the radii of the inscribed circle (ro) and of the cylindrical electrodes (R) are set to the correct ratio (R/ro = 1.148, [7]) the shape of the field approximates closely to this ideal. In general terms, if potentials -+1/2~o are applied to the electrodes then the potential ~x,y at a point x , y in a rectilinear co-ordinate system is given by ~ , y = ~o(~X 2 + oy2),

(1)

where X and a are constants. In the absence of any additional potentials the Laplace condition V2~ = 0 applies such that eq. (1) becomes

(~ + o)= o,

(2)

and for the mass filter ;~ = - o = 1/2r2o. Hence eq. (1) becomes CPx,y = ¢o(X 2 - f f )/2r 2.

(3)

The actual form of the potential ¢o may be a combination of a dc component U and a sinusoidally varying component Vo such that ~bo = U - Vo cos f~t,

(4)

where g~ is the angular frequency and Vo is the amplitude of the altenating potential developed between oppositely charged electrodes. Thus from eq. (3) ¢x,y = ( U -

Vo cos ~2t)(x 2 - y2)/2r2.

(S)

Now the absence of cross terms between x and y indicates that the components of ion motion in the two mutually perpendicular directions may be considered indepedently and the force acting upon an ion of mass m and charge e in, say, the x-direction will be given by - e E x = m d 2 x / d t 2,

(6)

which together with eq. (5) gives d2x dt 2

e + - - (U - V0 cos ~2t) x = 0 mrg

(7)

Expression (7) is in fact an equation of the Mathieu type for which the general form is d2u dT2

+ (au - 2qu cos 23') u = 0

(8)

with the transformations u =x, au

4eU m r ~ 2 2 =ax ,

(9) (10)

=

qu

2e V o mr~22 =qx,

i!;

7 = ½f2t

|2~

An analogous treatment of the ion motion in the ydirection gives d2y dt 2

e mr2o (U - V o c o s ~ 2 t ) y = O .

~13~

so that ax = ay

and

qx =

qy.

(14)

The solutions to the Mathieu equation [8] may be characterised according to whether they are "stable". i.e. the values of u(7) remain finite as 7 ~ oo or "unstable", where u(3') increases without limit, and t%r stability the values of the parameters au and qu must remain within certain limits. We find therefore that the transmission of ions through the quadrupole is only possible when the trajectories in the z- and y-directions are stable simultaneously, and that the combination of the quantities m/e, ro, ~2, U and Vo yields values of ax, y and qx, y which lie within the bounded region as defined by the stability diagram shown in fig. 2. Mass-selectivity may be achieved in a number of ways. In the conventional quadrupole mass filter the values of the rf and dc potentials are scanned so that there is a constant ratio between them. In this way the (a,q) co-ordinates of a series of m/e-values move along a "scan-line" which cuts the apex of the stability diagram, as shown in fig. 2. For theexample illustrated, in which rn~ < r n a < m a represent a sequence of ions each of the same charge e and separated by single mass unit intervals, the gradient of the scan line, as determined by the ratio U/Vo, has been set so that ions m2 are stable whereas ions ml and ms are not. The steeper the gradient the higher the resolution but the lower the ion transmission through the quadrupole, since progressively fewer of the ion trajectories satisfy the stability criteria. For an isotope separator, where the resolution required may be fairly low, this mode of operation may not y i d d the necessary rate of collection of separated product. As the dc potential bias is reduced, for a given rf potential, the gradient of the scan line is reduced such that the working points for m b rn~ and m3 all tie within the stability envelope (fig. 2) and the mass pass-band of the quadrupole is increased. For zero dc bias the scan line is now co-incident with the q-axis and for a given rf level the pass-band extends from a

449

M.F. Finlan et al./Quadrupole mass filter O. 0.3

• < 0"2

rn

SCAN

LINE

I

4.

0"1

j

,

0'2

,

v

0'4

,

0"6

v

0-8

1'0

~

CI¢

Fig. 2. Stability diagram for the quadrupole mass filter, plotted in a,q space, showing a "scan line" for mass-selectiveoperation.

value of m equivalent to q = 0.908 (the stability limit) in eq. (11) to infinity. This corresponds to the so-called "total pressure" mode of operation of the quadrupole. At first sight it might be thought that mass-selectivity under these conditions would be impossible, however several modes of operation may be considered. One, first suggested by Paul and co-workers [3], employs auxilliary rf fields applied between geometrically opposite electrodes tuned to the frequency of the secular motion of the ionic species which it is desired to remove. In this case resonance absorption occurs resulting in instability of the trajectories. Other methods, making use of the (a,q) stability limit (0, 0.908) as a working point, have been proposed. Thus Holme et al. [9], following a proposal by Brinkmann [10], showed that mass selection could be achieved by applying a retarding field between the analyser and the detector such that when the rf potential was scanned, only those ions which were on the point of becoming unstable had sufficient kinetic energy to be recorded. Weaver and Mathers [ 1 l] adopted a different approach in that as the rf drive potential was swept an audiofrequency modulation was applied such that peaks appeared as first derivatives of the inflexions in the total ion intensity plots observed as the (a,q) co-ordinates for

the constituent ions passed from the stable to the unstable regions. A third method, employed by us in part of this research and termed the "falling edge" mode, may be applied in instances where it is desired to eliminate only the lighter components from an isotopic mixture. Thus, referring again to fig. 2, the rf potential may be adjusted so that m2 and m3 remain stable whilst ions corresponding to ml possess unstable trajectories and are not transmitted by the quadmpole.

3. Objectives, specification and design details The ultimate aim of the project was the development of a system capable of producing gram quantities per year of separated isotopes of elements up to A = 100. For this it was considered desirable to be able to operate with a 5 mA ion beam for period of ca. 120 h per week with the system having automated control and capability of running unattended. The design was to require relatively low engineering tolerances and the components to be made from everyday materials. In particular it was desirable to have low capital and running costs. In drawing up the design details for any mass filter where the aim is to achieve a particular resolution at

M.F. Finlan et al./Quadrupole mass filter

450

a given mass, the interrelated parameters which must be specified are the ion kinetic energy, the length of the quadrupole rods, l, the dimension ro and the values of ~2, Vo and U. Thus the mass range is given by eq. (11) involving f2, 11o and ro, whereas the practical resolution is a function not only of the ratio U/Vo, but depends upon the number of radiofrequency cycles experienced by the ion in passing through the quadmpole, i.e. upon g2, l and the ion kinetic energy. This last-mentioned aspect has been considered in some detail by Austin et al. [12], and experimental data presented by these authors suggests that to achieve a maximum resolution of re~Am = 100 (Am measured at 10% peak height) then ions should spend ca. 60 cycles in the rf field. However, for the present application a further vital requirement is the maximum beam current which may be confined within the mass Filter, a factor which is of little consequence in small-scale analytical devices. A model for estimating the effects of space charge along the axis of a mass filter has already been reported briefly by us elsewhere [13] and is essentially based upon the model of pseudo-potential wells developed for the quadrupole ion store (Quistor), as follows. It has been shown [14] that at low values of a and q the one-dimensional motion of an ion in a radiofrequency quadrupole electric field may be approximated to that of a particle undergoing simple harmonic motion within a parabolic pseudo-potential well of depth D which, for the mass filter in, say, the x-direction, is given by [15]

eVg Dx - 4mr~122 .

(15)

In the absence of a superimposed dc bias (a = 0), Dx = Dy. Ignoring certain weaknesses in the model recently noted elsewhere [16], the two=dimensional pseudo= potential acting in the x,y plane may be written as

~x,y= bx r_~o(X2 +y2).

(16)

Under conditions of space charge limitation ions will occupy the well to a density Pmax such that the electrostatic potential ~i and the pseudo=potential will follow the Poisson relation, --V2q~i = 7 2 ¢ ---- 47rpmax ,

(17)

so that with eq. (16) we have O~

(18)

Substitution by means of eq. (1 l ) and ( 15 ) gives

qVo / ) m a x - 8 ~ r 02 .

(19)

The beam envelope wilt be rectangular in cross-section with dimensions which are a function of the phase-angle of the drive potential [ 17] ; if"we approximate the mean area of this rectangle (averaged over a cycle) to r~ then the ions within the mass filter will be contained within a rectangular parallelpiped of volume equal to lr~. Thus the maximum number (Nmax) of ions of charge e confined within the mass filter will be Nma x -

qVo 8Tre

l

.

(20)

But ifn ions per second leave the source with velocity then the number of ions within the mass Filter wilt be nl/~, and equating this with Nmax gives

_ qVoi n - 87r-----e-

(21)

Thus when operating at a given value of q the space charge limited ion current is determined by the value of Vo and the injection velocity 2. Iio and q are of course related by eq. (11) so that ifq is specified then for a given mass, Vo is restricted by the values of ro and ~2. In the present series of experiments the radio= frequency supply had an output limit of Vo equal to ca. 6 kV ( p - p ) at 2 MHz and rods fabricated from brass tubing of nominal diameter t.25" (R = 0.612") were employed in an array such that ro = 1.354 × 10-: m. For a singly charged ion with m = I00 this arrangement would provide a maximum value of q equal to 0.399, and with an ion injection energy of 5 kV (2 = 9.79 X l0 s ms -I) eq. (21) suggests that the maximum beam current should be ca. 1.04 mA. Thus a system based upon this set of design and operating parameters should at least be capable of transmitting beam currents of the correct order of magnitude. With this particular power supply the maximum value of role which would be transmitted at the cut-off point of q = 0.91 is approximately 44, and for some of the experiments 4°Ca and 4°Ar were used as sampies. Thus with m = 40 and q = 0.399 (a value where the transmission appears to be maximal, see later) one would operate with 1Io = 2400 V ( p - p ) and expect to achieve a beam current of ca. 0.66 mA. It remains to specify the length of the mass filter; this was selected as l = 3 m since an ion of m/e IO0

M.F. Finlan et aL /Quadrupole mass filter Table 1 Design parameters for quadrupole mass f'tlter acting as isotope separator Quantity

Symbol

Units

Value

Inscribed radius Quadrupole length Drive frequency Maximum amplitude of rf drive Ion injection energy

ro l

m m MHz

1.354 X 10 -2 3.0 2.0

V(p-p) eV

6000 5000

s2/2n Vo

451

ANALYSINGOUADSYSTEM- - ~ RETARDING LENSES~

~

=

FARADAYCAGE

~ ~ ~

EXIT COLLIMATOR~ _ . . ~ .

-21

t7

injected at 5 kV would then experience ca. 61 cycles of rf power from a source running at 2 MHz, compatible with the requirement to achieve unit resolution at m/e 100, as noted earlier. The design parameters are summarised in table 1. After experimenting with a number of earlier designs the layout shown schematically in fig. 3 was adopted. The potentials indicate the approximate conditions needed to detect a 1 mA beam o f N e ÷ ions with zero dc bias and a drive potential of 3 kV ( p - p ) at 2.00 MHz. The ion source was a hot cathode Penning-type, incorporating a temperature-controlled sample oven, and had an ion exit aperture of diameter 3 mm. It was capable of use with gaseous or metallic materials and the cathode life was ca. 4 0 h . The source was mounted so as to permit full adjustment both in azimuth and attitude and could be floated up to potentials in excess of +6.2 kV. After extraction the ions were transmitted through various cylindrically symmetric focussing lenses and injected into the quadrupole mass filter; the ion detector took the form of a Faraday cage wtih secondary electron suppression (see fig. 3). The whole assembly was mounted vertically within a glass (Quickfit Visible Flow) envelope and evacuated to a pressure of ca. 10 -6 Torr by means of oil vapour diffusion pumps. The mass filter rods were mounted under tension to eliminate "bowing" effects, and the rf and dc power was coupled to the centre of the rod assembly to minimise the extent of phase lag (ca. 3.5 ° ) between the ends and the centre of the rods at 2.0 MHz. Details of the electronic circuits employed are shown in fig. 4, from which it is seen that the rf and dc supplies were decoupled from one another and independently referenced to ground. This allowed one to fix a dc level and then scan the rf potential (manually) in order to explore the transmission characteristics of the ana-

MAINOUADS--

FOCUSSING ELECTRODES --

EXTRACT GUN- -

4

-

ION SOURCE--

I

Fig. 3. Layout and dimensions of the isotope separator system.

O'01~F n

RE DECOUPLING D.C.

L~F Ol.la,O CAPACITANCE /~00pF

--

J

/~

I: i c

J CHANGE-OVER

I j

SWITCH

Fig. 4. Schematic diagram of the rf and dc drive system.

452

M.F. Finlan et al./Quadrupole

lyser. An unexpected advantage of the glass vacuum housing proved to be the fact that the ion beam was clearly visible, presumbly because of the decay of excited ions, and therefore the focussing and ion source position could be readily optimised. To the authors' knowledge this is the first report of the direct observation of an ion beam passing through a quadrupole mass analyser. An earlier version of the system was fitted with a set of quadrupole "pre-filters", to which rf power only was supplied in the manner described by Brubaker [18], but these were seen to be unnecessary. In fact the ion beam appeared as a well-collimated cylinder of ca. I cm diameter which exhibited very little "blow-up" on entry to the mass filter. Evidently fringing field effects were minimal since, unlike the case of a small-scale analytical quadrupole, the longitudinal component of the ion kinetic energy was considerably greater than the transverse fields arising from the dc and rf potentials. The precise orientation of beam within the mass filter was, however, seen to depend upon the amplitude of the drive

rTUlSS

filter

potential and under certain conditions nodes were visible in the beam envelope. This effect will be considered in more detail in a future publication. In the most recent version of this system, in order to prove that mass separation is indeed being achieved, a small Centronic Model AIG50 residual gas analyser (without ion source) has been mounted behind a hole drilled in the ion collector plate. With the whole of this analyser (and its associated control unit) floated up to the same potential as the ion source (ca. 6 kV), the ion energy through the miniature quadrupole is reduced to the normal value for this instrument (ca. 5 eV) and the quality of the resolution is found to be sufficient for on-line monitoring of the performance of the main separating mass filter. In one set of experiments (see later) using magnesium as the sample, the system was operated for a total of ca. 50 h and the enrichment of 26Mg demonstrated directly by analysis of the deposition on the collector using spark source mass spectrometry. A typical deposit of magnesium on the copper target is shown in fig. 5, and is seen to be centred approximately on the small hole (1.5 mm diameter) which leads to the miniature quadrupole.

4. Results and discussion

4.1. "Total pressure" curves

Fig. 5. Photograph of the collector plate showing discoloration arising from deposited magnesium, together with a note of the orientation and dc bias of the quadrupole rods. The small hole in the collector plate allowed a fraction of the ion beam to be analysed on a miniquadrupole (not shown).

The simplest possible test for a quadrupole mass filter is to choose a sample which gives a single ionic species and then record the detector output when the amplitude of the rf drive potential is scanned, in the absence of any applied dc level: this yields a "total pressure" curve. Fig. 6 shows such a curve recorded for a 400 IlA beam of 5.5 kV Ca+ ions. The trace appears to exhibit a degree of fine structure, and is similar to a result obtained by Paul and Reinhard [3] for Zn+ ions. As noted by Paul, this behaviour is, in fact, a focussing phenomenon associated with the number of secular oscillations which the ions undergo within the mass filter, and direct visual examination confirmed that the structural minima coincided with the beam being rotated axially and deflected so that it struck the shield of the case rather than passing through to the collector plate (see inset). This effect was subsequently eliminated by widening the entrance aperture of the Faraday cup so that, regardless of the value of the drive potential, all the transmitted ions were collected efficiently. Fig. 7 shows a family

M.F. Finlan et al./Quadrupole massfilter

453

c.O0

300 >-

z2005"SKV Ea ions cz

100<

2 3KV

Vo P-P KV

3

Vo P-P KV

-I-.

Fig. 6. Early "total pressure" curve for 5.5 kV Ca ions. The fine structure arose from axial rotation of the ion beam caushag it to collide with the shield around the collector (see inset).

of curves obtained for Ca+ ions at a succession of dc levels. Essentially these correspond to a series of horizontal scans through the stability diagram in fig. 2, at a series of different a values. With zero dc potential the plot displays a broad maximum, and which is displaced progressively to higher values of V0 (i.e. q) as the dc level is increased. The cut-off points for each curve correspond to the onset of zero transmission of the ions, and co-incide with the boundaries of the stability diagram; from the U, Vo co-ordinates of

th cut-off points it is possible to derive the stability boundaries shown in fig. 8.

4.2. Stability diagram determinations Stability diagrams were determined for the six species 4°Ca +, 24Mg+, 2SMg+, 26Mg+, 2°Ne+ and 22Ne+, and are shwon plotted in U, Vo space in fig. 8; also shown is the theoretical limiting "scan line" which joins the origin with the apex of the theoretical stability diagram. The boundaries for the magnesium species were located by observing the cut-off points of curves scanned in the same manner as described

/

700,

//

500. ~,00. 300. 20O 100 a+

0

0

ols

1',o

R.E KV P-P

Iis

Lo

Zs

5

Fig. 7. "Total pressure" curves for Ca+ ions recorded at the stated values of the dc bias potential.

800,

600.

~

6'KV

310

3,s

sio

sls

Fig. 8. Stability diagrams for 4°Ca +, 24Mg+, 25Mg+, 26Mg+, 20Ne+ and 22Ne+ plotted in U, V0 space.

M.F. Finlan et al./Quadrupole mass filter

454 i

! I

~'ZERO-BLAST'

f

02

i

i

i

01

i

_i VQ P-P

0 I 0

' 01

0-2

03

1 0z~

0'5

06

0.7

08

09

SKY

10

Fig. 9. Stability diagrams for 4°Ca+ ions plotted in a,q space: solid line, theoretical boundaries; points, experimental values (ion beam currents: X, ca. 150/zA; +, ca. 400 ~zA).

above for calcium ions, except that extrapolation was necessary in order to obtain the data relating to each type of ion. For the neon ions, the boundaries were located by monitoring the output o f the respective signals on the AIG-50 mini-quad. The extent to which the experimental diagram obtained for 4°Ca ÷ ions agrees with the theoretical diagram may be seen from fig. 9, where the data are re-plotted in a,q space and the theoretical boundaries shown as solid lines. Two sets of experimental points are included, corresponding to maximum beam currents of ca. 150 and ca. 400 ~zA. The experimental shifts in the boundaries are similar to those observed for the Quistor [16] and may possibly be attributed to the effects of space charge. The studies on trapped ions suggest that greater divergence from the theoretical behaviour occurs with lower mass ions, and this trend is visible in fig. 8, where the apexes of the diagram for 2°Ne* and 22Ne* fall below the limiting scan line. From these stability diagram data we conclude that the quadrupole mass separator is working generally in the predicted manner.

4.3. Mass spectral scans Confirmation that the system was indeed capable of operating as a mass filter was obtained by scanning the rf and dc levels together so as to generate a mass spectrum. Thus a scan obtained with 2.1 kV neon ions is shown in fig. 10, which also reveals peaks arising from ions in the background gases within the

Fig. 10. Mass spectral scan of 2.1 kV Ne~ ions obtained with the isotope separator. The remaining peaks arose from background gases in the system. The "zero blast" ftgnal is caused by the indiscriminate transmission of ions at zero rf]dc potentials.

vacuum system. The ratio of intensities for 2°Ne*/ 22Ne ÷ is close to the ratio of the abundances of the respective isotopes.

4.4. Enrichment studies Two distinct sets of experiments were performed. In the first neon was used as the sample gas and the intensities at m/e 20 and 22 were monitored on the mini-quad as the de and rf tuning, of the main quadrupole was altered. The results listed in table 2 show the data obtained in two runs, where, for a given setting of the dc level, the rflevel was tuned to optimise first the transmission of 2°Ne +, and then the transmmsion of 22Ne*. The values of the ion current listed in columns (2) and (4) show how the transmission of the beam is attenuated as the dc level is raised, and is a measure of the extent to which the rate o f collection of separated isotope decreases as the degree of enrichment is increased. The figures in columns (3) and (5) indicate how the enrichment changes with dc level. Recalling that the natural abundance ratio 22Ne/2°Ne is ca. 0.097 it is somewhat surprising to see that for a d c level of zero volts the ratio 22]20 is so high. Presumably this must reflect the poor resolution of the miniquad under these conditions, with a considerable degree of "break through" of role 20 contributing to the signal at role 22 when the main beam current is quite high. With the 2°Ne + ion beam optimised [column (3)], it is only very near the cut-offlimit for transmission that the 22/20 ratio falls below that for

455

M.F. Finlan et al./Quadrupole mass filter

Table 2 Relative intensities of 2°Ne+ and 22Ne÷ from 4.5 kV beam of neon ions (1) dc level (V)

0 100 200 260 320 350 370 380 395

20 Ne÷ o ptimised

2 2 Ne ÷ optimised

(2) Max. current on Faraday cage (/~A)

(3) Rel. int. on mini-quad

(5) Rel. int. on mini-quad

Ratio 22/20

(4) Max. current on Faraday Cage (~A)

20 Ne÷

22 Ne÷

2°Ne+

22Ne+

Ratio 22/20

60 60 60 45 22 8 1 0 0

120 120 120 100 42 20 1.4 0 0

60 60 20 20 5 1 0 0 0

0.50 0.50 0.17 0.20 0.12 0.05 0 -

50 45 40 6 3 2 1 0 0

I00 100 80 10 6 1.6 0.2 0 0

50 60 50 20 12 3.8 1.8 1.2 0

0.50 0.60 0.63 2.00 2.00 2.34 9.00

the natural abundances; in other words this method of operation does not appear to be suitable where it is desired to strip out low levels of a heavier isotope. On the other hand, with the 22Ne* ion beam optimised [column (5)] the 2°Ne* intensity is seen to fall rapidly as the dc level is increased so that at 370 V the proportion of 22Ne has been increased to 100fold over the natural ratio, although the main beam current has fallen to 2% of its value in the "total pressure", zero dc level, mode. The fact that it is easier to strip out the lighter isotope, even when this is present in considerable excess, is readily explained in terms of the U, V0 stability diagram plots shown in fig. 8. Thus there is only a very narrow region between the lefthand boundaries for the lighter and heavier species, where the latter should exhibit instability, whereas the right-hand boundaries are more widely separated so that it is much easier to locate a region where the lighter-mass species will be unstable and therefore not transmitted. For this reason the working point m a r k e d X in fig. 8 was chosen for an attempt to enrich 26Mg, as follows. For the second set of experiments designed to test the effectiveness of the quadrupole as an isotope separator, a solid element was selected since this more clearly reflected the eventual application of the system, and afforded the actual collection of the separated species for independent assay. For the investigation a 5.0 kV beam of Mg ÷ ions was employed and the total running time of ca. 5 0 h was spread over a two-week period. In the "total pressure" mode, the main beam current collected in the

Faraday cage was 6 0 0 - 8 0 0 pA, and when the conditions were optimised for transmission of 26Mg+ (according to readings on the mini-quad) this fell to 6 0 70/aA. The voltage levels employed were U = 380 V, and V0 = 3.1 kV ( p - p ) at 2.0 MHz. The weight of magnesium collected on the target during the separation was determined by carefully etching off the deposit with dilute nitric acid, and found to equal ca. 20 rag. 20 mg. This was considerably higher than the expected increase, ca. 3.4 mg, based upon the measured current and the duration of the run, suggesting therefore that the current readings were rather on the low side. Independent analysis using spark source mass spectrometry revealed the isotope ratios of the deposited magnesium to be 24Mg, 21.4%; 2SMg, 32.2%; 26Mg, 46.4%, representing a better than fourfold enrichment relative to the natural abundances 24Mg, 78.6%; 2SMg, 10.1%; 26Mg, 11.3%.

5. Conclusions Using a quadrupole and associated circuitry fabricated from relatively cheap components, the practical enrichment of milligram quantities of 26Mg has been demonstrated, and the prototype system appears to fulfil the criteria for a commercial isotope separator noted in the paper. The use of glass tubing for the vacuum envelope has permitted the first reported direct observation of an ion beam traversing a quadrupole mass filter, and the appearance of this beam has provided a valuable guide towards optimising the

456

M.F. Fintan et al./Quadrupole mass filter

operation of the system as well as providing a basis for future comparison with the theory of quadrupole devices. The authors wish to thank Mr. D_P. Bunker for his contribution to the experimental work, including the design and construction of much of the circuitry.

References [1] A.E. Holme, W.J. Thatcher and J.H. Leck, Vacuum 24 (1974) 7. [2] See for example W.H. McFadden, Techniques of combined gas chromatography/mass spectrometry: applications in organic analysis (Wiley-Interscience, New York, 1973). [ 31 W. Paul and H.P. Reinhard, in Forschungsberichte des Wirtschafts-und Verkehrsministeriums Nordrhein-Westfalen, No. 450 (Westdeutscher Verlag, KiSln1958). [4] W. Paul and M. Raether, Z. Physik 140 (1955) 262. [51 J.F.J. Todd and G. Lawson, in International reviews of science, physical chemistry series 2, vol. 5, ed., A. Maccoil (Butterworths, London, 1973) p. 289.

[6] P.H. Dawson (ed.), Quadrupole mass spectrometry and its spplications (Elsevier, Amsterdam, 1976). [7] D.R. Denison, J, Vac. Sci. Technol. 8 (197l) 266. [81 For a detailed account see N.W. McLachlan, Theory and application of Mathieu functions {Ctarendon Press, Oxford, 1951). [9] A.E. Holme, S. Sayyid and J.H. Leck, Int. J. Mass Spectrom. Ion Phys. 26 (1978) 191. [10] U. Brinkmann, Int. J. Mass Spectrom. Ion Phys. 9 (1972) 161. [11] H.E. Weaver and G.E. Mathers, in Dynamic mass spectrometry, vol. 5, eds., D. Price and J.FJ. Todd (Heyden, London, 1978) p. 41. [12] W.E. Austin, A.E. Holme and LH. Leck, in Quadrupole mass spectrometry and its applications, ed. P.H. Dawson (Elsevier, Amsterdam, 1976) p. 121. [13] J.F.J. Todd, G. Lawson and R.F. Bonnet, ibid, p. 181. [14] H.G. Dehmelt, Advan. At. Mol. Phys. 3 (1967) 53. [15] J.F.J. Todd, R.M. Waldren and R.E. Mather, Int. J. Mass Spectrom. Ion Phys. 34 (1980) 325. [16] J.F.J. Todd, D.A. Freer and R.M. Waldren, Int. J. Mass Spectrom. Ion Phys. 36 (1980) 185. [17] J.F.J. Todd, R.M. Waldren, D.A. Freer and R.B. Turner, Int. J. Mass Spectrom. Ion Phys. 35 (1980) 107: [181 W.M. Brubaker, in Advances in mass spectrometry, vol. 4, ed., E. Kendrick (Inst. Petroleum, London, 1968) p. 293.