Ion sources for accelerators in materials research

Nuclear Instruments and Methods in Physics Research B73 (1993) 221-288 North-Holland

Ion sources for accelerators

NIOMI B

Beam Interactions with Materials 8 Atoms

in materials research

G.D. Alton Oak Ridge National Laboratory *, Oak Ridge, TN 37831, USA

Received 22 April 1992

Reviews are presented of the principal sources that are presently being utilized or can be potentially used for materials science applications, with particular emphasis placed on recent improvements to existing sources and new source developments that show promise for such applications. Specifically, status reports will be given on a number of state-of-the-art positive and negative ion sources routinely used for many materials research applications, including implantation, ion beam deposition, isotope separation, modification of surface properties, ion beam lithography, secondary ion mass spectrometry (SIMS), accelerator mass spectrometry (AMS), Rutherford backscattering spectroscopy (RBS), and proton-induced X-ray emission (PIXE).

1. Introduction During the past decade, remarkable progress has been made in the advancement of the technologies of both positive and negative ion sources, as well as in an

enhanced understanding of the mechanisms underlying ion formation and factors which limit ion production in such sources. As has been the case historically, ion source development continues to be driven by the needs of the basic and applied research communities and in more recent years, by the needs of the industrial communities for sources with improved performance attributes, including operational reliability, lifetime, beam quality (emittance), and intensity for an ever-increasing number of applications. The pervasiveness of the usage of ion beams exemplifies their importance to modern science and technology. These applications include basic and applied research, isotope separation, mass spectroscopy, fusion energy, inertial confinement, radiation therapy, as well as a growing and diverse number of industrial applications such as ion beam lithography, semiconducting material doping, ion beam deposition, modification of material surfaces (e.g., conductivity, wear and corrosive resistance, etc.), and as probe beams for the important analytical fields of secondary ion mass spectrometry (SIMS), accelerator mass spectrometry (AMS), Rutherford backscattering spectroscopy (RBS), proton induced X-ray emission (PIXE), nuclear reaction analysis (NRA), and elastic Correspondence to: G.D. Alton, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, TN 37831, USA. * Research sponsored by the U.S. Department of Energy under contract no. DE-ACOS-840R21400 with Martin Marietta Energy Systems, Inc.

0168-583X/93/$06.00

recoil detection analysis (ERDA). All of these applications are dependent on the use of ion beams and, consequently, on the technologies for their production. Such sources of ions constitute an important set of technologies, each of which may vary widely in complexity, depending on the method required to produce the desired ion species and beam intensity: Due to the many types of sources and variations within a particular source type, a comprehensive review of all sources which have been described in the literature will not be possible. Since most applications involve low-charge-state sources, multiply-charged sources, such as the electron cyclotron resonance (ECR) and electron bombardment ion source (EBIS), will not be emphasized. However, since the ECR technology can be used effectively to produce low-chargestate ion beams, this technology will be briefly reviewed. The reader is encouraged to consult recent references contained in previously published review articles, books, conference proceedings and workshops which deal with ion sources and their applications for more detailed information. Review articles which deal with the chemistry, physics and technology of high-intensity, positive ion sources are included in ref. [l], while ref. [2] emphasizes the physics, chemistry and technology of both positive and negative ion sources based on a wide range of ionization phenomena. The properties of high-intensity sources have been previously reviewed by Green [3]. In recent years, manuscripts have been written which deal with the generation and propagation of high-intensity ion beams [4] and the physics and technology of ion sources [5]. Since the performance of a particular source type can best be measured by the results obtained from its use

0 1993 - Elsevier Science Publishers B.V. All rights reserved

222

G.D. Alton / Ion sources for materials research

in a particular field, reference is made to conference proceedings which deal with specific applications or deal with particular source types or ion sources. 1.1. Ion source selection considerations Several physical and physiochemical mechanisms can be employed to effect positive or negative ionization. A wide variety of ion sources have been developed that utilize one or sometimes a combination of mechanisms for producing ions. The probability for ion formation depends on the atomic structure properties of the particular atom or molecule. The mechanism utilized for positive ion formation may be rather general in scope in terms of species capability, as is the case for ionization through electron impact, or rather restrictive, as is the case for surface ionization. While electron-impact ionization offers a species-independent method for positive ion formation, no such universal method exists for producing negative ions. As a consequence, specific methods must be employed for negative ion formation that depend on the species of interest. However, techniques exist which can be employed to produce negative ion beams containing most of the elements in either atomic or molecular form. The ion species capability of a source may, as well, be design-limited. For example, to provide a wide range of ion species, the source must be equipped with a means of providing an adequate vapor supply of the material in question; thus the source must be equipped with either a vaporization oven or a high-voltage sputter probe, as well as a gas-inlet-metering valve system. Many source concepts and designs are not commensurate with ovens or probes and, therefore, are specieslimited because they must operate with a restricted number of high-vapor-pressure liquid or gaseous-feed materials. The selection of an ion source for a particular application should be made with due consideration of factors such as species and intensity capability, beam quality (emittance and brightness), ionization efficiency, material utilization efficiency, reliability, ease of operation, and maintenance, duty cycle, and source lifetime. These factors obviously are not of equal importance and the ion source selection process should be based on the most desirable combination of characteristics for the particular application. No truly universal source exists that will meet all application requirements. 1.2 Ion source and transport system figures of merit, emittance (brightness and acceptance) Liouville’s theorem states that the motion of a group of particles under the action of conservative force fields is such that the local number density in the

six-dimensional phase space hypervolume xyzp,p,p, is a conserved quantity, where x, y, and z are position coordinates with respective components of momenta px, p,, and pz. If the transverse components of motion of a group of particles are mutually independent in configuration space, they are also independent in the orthogonal phase space planes (x, p,), (y, p,>, and (z,p,) and the corresponding phase space areas are separately conserved. The transverse phase space areas are proportional to the emittances of the beam which are, in turn, also conserved. For a beam moving along the z direction at a constant, nonrelativistic energy E,, we use the following definition for the normalized emittance E,: -T En, =

and E

“y

=

IT

x’ and y’ are angular coordinates in the small angle approximation and the integrations are performed to include specific fractions of the total ion beam. The units of emittance, as defined by eq. (1) are given in units of rr mm mrad (MeV)“‘. Another figure of merit often used for evaluating ion beams is the brightness B. Brightness is defined in terms of the ion current dZ per unit area dS per unit solid angle dR or BE=

d21

dSd0’ Brightness can be shown to be equivalent to 1,421

The term acceptance A refers to the beam transport system and is complementary to the transverse phase space or emittance E. The acceptance of a device is defined as the phase space area containing all particles that can be transmitted through the device without impediment. In order to transmit an ion beam through a system of optical components (lenses, magnets, electrostatic analyzers, and drift spaces, etc.) without loss, the beam must be transformed into the acceptance phase space A, of each individual element from the source to the target. The two-dimensional acceptance phase space for the respective orthogonal x and y directions can be expressed through the following relations: A “X =Ti

and Any=~

(3)

223

G.D. Alton / Ion sourcesfor materialsresearch

where the units of normalized acceptance are again measured in r mmmrad(MeV)‘/2. If the acceptance of the device is less than the emittance of the beam, then only that portion of the beam will be transported for which en1 I A,,; E,,~ I A,,,,. In general, in order to transport a beam through a system, it may be necessary to incorporate properly positioned lenses, steerers, magnets, etc., to transform the beam emittance to properly match the acceptance A,, of the system.

iI

+-

C -0

GlaS -

:*31< IONS

2. Positive ion sources This description of positive ion sources will be limited to those basic types commonly used in material research applications and new developments which show promise for these applications. Only typical examples will be reviewed. The predominant method for producing a highly ionized medium from which intense beams of positive ions can be extracted is by creation of a magnetically directed or confined electrical (plasma) discharge supported wholly or in part by a gaseous vapor containing the material of interest. This method is universal in that any species can be ionized and therefore is the method most often used in the design of versatile heavy ion sources for the production of low-charge-state intense ion beams. In simple plasma sources, the discharge is initiated and sustained by accelerating electrons that are thermally or secondarily emitted from a directly (hot) or indirectly (cold) heated cathode to energies above the threshold for ionization of the gaseous vapor. The discharges are most generally directed by a magnetic field - usually oriented parallel to the direction of electron acceleration. Ions may be extracted in a direction parallel to the discharge axis or perpendicular to the direction of the discharge. The former method is often referred to as an axial or end extraction geometry, and extraction perpendicular to the discharge axis is usually referred to as a side or radial extraction geometry. The end extraction geometry is compatible with circular extraction apertures, whereas the side extraction geometry permits extraction from the length of the plasma column using slittype apertures. Fig. 1 illustrates the two commonly used discharge configurations. 2.1. Elementary physical processes in plasma discharges The ensemble of complex physical and chemical processes, which are in dynamic equilibrium in stable discharges containing chemically active elements, make detailed analyses and understanding of such media difficult. Most investigations involve noble gases, and therefore avoid the additional complexities present in more typical discharges, which may involve simultane-

I-

+-

0

II^^

s -

I +

Fig. 1. Commonly used ion source discharge configurations for low-charge-state, positive ion production. Upper: side- or radial-extraction geometry; lower: end-extraction geometry.

ously several different chemically active elements.

Regardless of the complexity of the situation, it is instructive to consider some of the basic physical processes which are operative in plasma discharges. 2.1.1. The hot cathode discharge

The discharge in hot filament sources is initiated and sustained primarily by thermally emitted electrons from the heated filament which are accelerated to energies sufficiently high to produce ionization of the medium. The emission current density is given by the well-known Dushman-Richardson equation j(A/cm2)

=A(1

- r)T2

e-@‘lkT,

where A is a constant with value A = the reflection coefficient at the metal absolute temperature, 4 is the work metal and k is Boltzmann’s constant.

120 A/cm2, r is surface, T is the function of the From this equa-

224

G.D. Alton / Ion sources .for materials research

tion, we can readily see that for equivalent current densities, high-work-function metals must be heated to higher temperatures than those with lower work functions. In practice, the emission current may not reach the saturation value predicted by eq. (4) because of space-charge effects. However, theoretical values can be achieved if the cathode-to-anode potential is sufficiently high so that electrons are extracted as fast as they are emitted. In the case of a plasma discharge situation, the theoretical value can also be approached when positive ions arriving at the filament reduce space-charge effects. Initiation of an arc discharge is accomplished by introducing a small amount of gas or vapor into the discharge chamber at flow rates high enough to establish an adequate pressure in the anode-cathode region and by accelerating the electrons to an energy sufficient to ionize a fraction of the vapor. In this way, a medium with plasma-like properties is produced. 2.1.2. The Penning ionization cold-cathode discharge The Penning ionization cold-cathode discharge [6] or reflex discharge has been used for many years in ion source applications. Initiation of the discharge is accomplished by introducing a small amount of gas or vapor into the discharge chamber at flow rates high enough to establish an adequate pressure in the anode-cathode region while the cathode potential is adjusted to values typically between 600 and 2000 V. After initiation, the discharge is aided by secondary electrons emitted as a result of positive ion bombardment of the cathode when the source is operated in the low-power mode. In high-power sources, the cathodes reach thermionic emission temperatures, and thus are self-heated; the discharge, then, is sustained both by thermally emitted, as well as secondarily emitted, electrons. The discharge mechanism is very complex and involves avalanche or cascade ionization processes. Electrons born within the electrode system are accelerated toward the anode, but cannot strike it because their radial motion is constrained by the magnetic field. They, therefore, execute oscillatory motion between the cathodes and anode, making collisions with the residual gas, which in turn liberates other electrons, which are also accelerated. The characteristics of the discharge have been studied by several investigators, including those reported by Backus [7] and have been the subjects of extensive review articles such as those by Hooper [8] and by Bennett 191for multiply charged ion generation. Several types of discharges may occur in the coldcathode source, depending on the pressure during operation. For most ion source applications, the discharge is characterized by modes of operation involving relatively low voltages and high pressures (10H4-

10W3 Torr), which produces predominantly lowcharge-state ions. For multiply charged ion production, the pressure of operation must be significantly lower because charge exchange at typical discharge pressures depletes higher-charge-state components produced in the ionization processes. 2.1.3. Sheath thickness Under space-charge-limited flow conditions with no plasma boundary present, the electron current between the cathode and anode (in units of amperes per square meter) would approximately follow the familiar Child-Langmuir space-charge-limited flow formula for a planar diode [lo] (5) where e is the electronic charge, m, is the electronic mass, V is the potential drop between cathode and anode, and d is the electrode spacing. In the presence of plasma, the sheath thickness x is greater than it would be for a sheath containing only positive ions of current density j, and charge q [ll]. This is due to the fact that the electrons reduce the space charge in the sheath so that a greater distance is required to adsorb the potential drop. The actual sheath thickness is given by

The sheath thickness x is increased by the factor /3 over that for a sheath containing only positive ions of mass M. Eq. (6) assumes that the field terminates abruptly at the plasma boundary. However, if we assume, as in the case for electromagnetic waves interacting with a plasma (e.g., see ref. [12]), that static fields penetrate into the plasma dropping off exponentially to a depth Splasmagiven by 6 plasma

E c/w,

9

(7)

where wp = (4rn,e2/m,)‘/2 is the plasma frequency, n, is the plasma density, m, is the electronic mass, and e is the electronic charge. For laboratory-observed plasma densities corresponding to n, = 1012-10’4/cm3, the field penetration depth will be of the order of 5 to 0.5 mm. Because of the penetration of the field into the plasma, positive ions at depth aplasma are accelerated toward the cathode. According to Bohm [ll], ions must reach the plasma sheath with a kinetic energy of at least half of the plasma electron temperature Ti 2 :T,,

which is of the order of a few eV for many sources.

(8)

225

G.D. Alton / Ion sources for materials research

2.1.4. Stable discharge criteria and ion formation

In order to maintain a stable discharge, the relationship between electron, j,, and an ion current density, j+, flowing across the sheath must satisfy the following inequality:

For a neutral density of n,, the rate of ion generation in terms of current density j, can be expressed as

j+“4 yP(fli,

j,, 1, EO).

(11)

2.1.6. Minimum flow rate, pressure and neutral density requirements

where M and m, are the masses of the positive ions and electrons, respectively, and y is the correction factor which varies with the state of the cathode and has values between i and 3. Eq. (9) is referred to as the Langmuir criterion for a stable discharge [13]. The Langmuir criterion provides an upper limit for the electron current density emitted by the filament. When the ratio of positive ions arriving to electrons leaving the filament falls below the limit imposed by eq. (9), the discharge will become unstable and eventually go out. The rate of generation of positive ions is directly proportional to pressure and, therefore, there is a critical pressure below which the condition imposed by eq. (9) cannot be satisfied. 2.1.5. Ionization The probability of creating a positive particle in a discharge of arc current density, j,, is given by P = 1 - exp( -ui j,Z/quO),

(10)

where gi is the cross section for ionization, 1 is the average distance traveled by a neutral atom of mean velocity fiO in the discharge and q is the ionic charge.

In the following derivation, reference is made to fig. 2. After steady-state conditions have been reached, the rate of ion generation is just balanced by the rate at which ions leave the plasma column. For a density n,, the ion current density arriving at the sheath edge is j+=qn+(kT,/M)“‘.

(12)

If we assume that the field penetrates to a distance 6 into the plasma and losses occur with equal probability at each end of the plasma column, then the fractional loss at the ends of the column j,,, (loss) is given by (13) In plasmas directed along a magnetic field, the rate of diffusion across the field (radial direction) is given by u,=DIVn+/n.,

(14)

where n, is the plasma density and D I is the diffusion coefficient across the field, which is approximately given by 1 kT, DFzz

SHEATH REGION’

Fig. 2. Schematic diagram of a plasma discharge column of length Z and diameter p with field penetration 6 at the ends.

(15)

226

G.D. Alton / Ion sourcesfor materials research

for a turbulent plasma, according to Bohm et al. [14]. In the expression for D I, k is Boltzmann’s constant, T, is the electron temperature of the plasma, and B is the magnetic flux density. This solution to the diffusion equation results in an exponential decrease in the ion density given by n+=no+

e

-r/p

,

PLASMA ZERO

SHEATH

I

EOUIPOTENTI

(16) EOUIPOTENTIALS

where no+ is the value of the plasma density at the center of the plasma column and p is a characteristic distance in which the density drops to l/e of its value at the edge of the column. The fractional rate of loss of ions across the magnetic field through diffusion processes is given by l/2

jil(10SS)-4~~+$(~)

.

(17)

By equating the rates of ion loss for a plasma column, such as shown in fig. 2, to the rate of ion generation and utilizing the Langmuir stability criterion, expressions for the minimum flow rate f, the minimum neutral density no, and the minimum pressure Pmin necessary to maintain a stable discharge can be derived. These expressions are given by

(18)

where Co = [(8/a)(kT,,/M)]‘/2 has been used for the average thermal velocity of the neutral atoms of temperature To and the arc current density j, has been equated to j,. The previous expressions, although not general because of the specific geometry considered, illustrate the dependencies of the quantities f, n,, and Pmin on certain physical or source operational parameters. Minimum pressures for stable discharge generally lie between 1O-4 and lo-’ Torr for most ion sources. However, the high frequency (rf) source generally requires pressures between lo-’ and 10-l Torr. 2.1.7. Ion extraction from a plasma Extraction of ions from a plasma boundary is accomplished by applying a potential V,, between the exit aperture and an extraction electrode that is negative with respect to the plasma potential VP. Because of the difference in the mobilities of electrons and positive ions, the plasma potential VP is slightly positive with respect to the walls of the ionization chamber of the source and usually is of the order of S to 20 V. At the exit boundary, a sheath is formed between the

--ION

ION

BEA

CREATION

Fig. 3. Schematic diagram of ion extraction from a source operated in modes: (a) low-plasma-density and (b) highplasma-density, illustrating how ion energy spreads can be induced by the mode of source operation or design of the ion generation extraction regions of the source.

plasma surface area and the extraction electrode which is made of ions extracted from the plasma. The ion generation and extraction system may contribute considerably to the final beam quality and thus careful consideration should be given to their selection and design. For example, the ion production and extraction system may introduce energy spreads associated with the fact that the ions are generated at differing equipotentials within the ionization volume; the effect is illustrated in fig. 3. For a penetrating field extraction system and low plasma densities, ions may be generated over a considerable distance relative to the direction of the extraction field and as a result, an energy spread is superposed on the intrinsic energy spread of the ions. The latter is related to the method used for ion production (fig. 3b). The problem is reduced by increasing the plasma density so that the external field is terminated more abruptly (fig. 3a). Thus, ions may be generated from sources with ion densities that vary from the very low to the fluid surfaces of high density plasma discharge sources or from solid surfaces, as is the case for many surface ionization sources.

221

G.D. Alton / Ion sources for materials research

The ion extraction boundary (meniscus) may assume a variety of shapes, depending on the plasma density ne, variations in plasma density, and the potential Ynx. For fixed ion source parameters and low extraction voltage VEX, the plasma may assume a convex shape. As V,, is gradually increased, more positive ions are extracted and the boundary recedes until it becomes flat and finally becomes concave at appropriately high values of V,,. These effects are schemati~alIy illustrated in fig. 4. Since the shape of the extraction plasma boundary varies with extraction potential V,, and source parameters (arc current, arc voltage, vapor flow rate, and magnetic field), it acts as the first and most important lens in the system, and thus plays an important roll in the final angular divergence of the ion beam as well as beam quality. The ion current density which can be extracted from a plasma boundary generally exhibits a I&” dependence given by j.+= C(q/M)‘~2V&2/d2,

(21)

where the constant C depends on the source geometry, q is the charge on the ion, M is the mass of the ion

and d is the spacing of the extraction electrode from the plasma boundary. Several experimental verifications of space-charge-limited current density versus extraction voltage have been reported, including those of refs. 115-171. It should be pointed out that many species of ions may be generated in the plasma with different masses and different charge states and, therefore, the space charge limited current may also be comprised of many charge states and ion masses. The presence of other charge states and masses can substantially reduce the beam intensity of the desired species. This is especially true when the ion is generated from a complex molecular feed material or when an auxiliary discharge support gas is used. Because of the capability of adjusting the source parameters for optimum production of the desired

r

(4

PLASMA

component, some marginal species differentiation is possible. As a rule, the source discharge potential should be maintained at appro~mately 51j, where Ii is the first ionization potential of the desired species. 2.1.8. Computational simulation of ion extraction from a plasma Considerable advances have been made toward appropriate computer modeling of the plasma-va~um interface ion-extraction problem. Such programs have been developed by a number of groups, including the programs described in refs. [l&19]. Basic mathematical formulation of the problem involves the solution of the Poisson-Vlasov equations for extraction of ions from a ~llisionless plasma: 0’4 = /fi

dui - e-# z>za,

(22)

V,fi = 0

u. V,f - ;v,d.

1 where b; and f are the potential and ion dist~bution functions within the plasma. The boundary conditions are taken to be 4 = do;

f = f,,

(Dirichlet),

and V,,#

=o;

V,, If = 0

(Neumann).

The Dirichlet boundary conditions are easily specified and a solution to the extraction problem may be obtained wherever a solution to the source boundary value problem is achieved. The source plasma boundary is traditionally related through the one-dimensional sheath potential distribution described by the following equation: a2 d”4(r)

2 --=

dz2

z g(y) dy / 0 &(*)-#J(y)]

-ecb(r)Y

ZlZ,.

VIRTUAL r FOCUS

@f

Fig. 4. Illustration of possible plasma boundary curvatures during ion extraction that may be effected by changes in plasma density or ion extraction parameters: (a) concave meniscus, (b) flat meniscus, (c) convex meniscus, and (d) complex curvature that leads to beam aberrations.

228

G.D. Alton / Ion sourcesfor materialsresearch

In this equation, a = /2(h,/A) where A, is the Debye length and A is the mean free path for ionization usually given in units of Debye lengths for the plasma, g is a source function for ionization, and z is an arbitrary axial distance from the center of the plasma in Debye lengths. The Debye length A, is related to the plasma density it, and temperature T through the following relationship A, = {,,kT,/4an,e where k is Boltzmann’s constant. The equations are appropriate for descriptions of the ion extraction problem as long as the Debye length is short in relation to the electrode dimensions. The previous equations and their adequacy for property representing the extraction problem have been critiqued by Whealton [20]. Numerical solutions to the ion extraction problem often do not converge because of the difficulties associated with boundary stabilization. Whealton et al., as well as others, have developed algorithms and numerical techniques that minimize such problems. Fig. 5 illustrates stable solutions for ion extraction from cylindrical and rectangular geometry electrode configurations [18]. Such programs are very useful in estimating the emittance for ion sources as well as for determining optimum perveances for source operation. Beam transport is affected by angular divergence of the beam during extraction from a plasma boundary. Therefore, optimum perveance operational conditions must be met in order to maximize beam transport through the beam transport system. Computational simulation of the effect is illustrated in fig. 6, which displays the angular divergence erms as a function of ion current density j, extracted from both slot and circular aperture sources [ES]. 2.1.9. Computational simulation of ion emission from solid surfaces Computer programs have been developed for the precise solution of Poisson’s equation and calculation of ion trajectories through the resulting electric fields in particular electrode systems. An example of such a program is described in ref. [21]. Poisson’s equation for the potential distribution $J in the presence of space charge density p is v24 = -p/E”.

(24)

Fig. 7 illustrates the numerical simulation of ion extraction from a cone-geometry, cesium sputter; negative ion source [2] by use of the program described in ref. [21]. The negative ion beam is produced by sputtering the sample surface with a positive cesium ion beam. We note that the beam is highly aberrated because of the geometry of the negative ion generation surface. 2.1.10. Ionization efficiency The ionization efficiency, n, of an ion source is defined as the ratio of the flux of charged particles j,

(4

(b)

Fig. 5. Computational analyses of the ion extraction region near a high-density plasma boundary for (a) slot and (b) circular aperture extraction electrode systems. (From ref. [18].)

to the total flux of charged particles j, and neutral particles j, flowing across the ion extraction boundary or

j++ja

(25)

’

The positive ion current density is given by the formula of Bohm et al. [14] according to kT, 7

j+=0.4qn+ (

I”

,

1

(26)

where k is Boltzmann’s constant and T, is the electron temperature of the plasma. The equivalent current density of neutrals flowing across the sheath is given by the well-known formula from the kinetic theory of gases j,=q-= nofio 4

kTo “’ qno ( 23rM i ’

(27)

where T,, is the temperature of the neutrals of mass M. Substituting eqs. (26) and (27) into eq. (23, we obtain

n_ll+o.73~)“2]~‘.

(28)

Ion source performance measurements are very often made immediately after extraction and, therefore, include all ion species generated by the discharge. In such measurements, the natural tendency is to adjust all ion source parameters to achieve a maximum extracted ion current without regard to beam quality, space charge effects, or the spectrum of charges or masses that make up the extracted beam. Therefore, measurements made under these conditions are of limited value. The most important component in the spectrum of ion beams generated by the ion source is the desired ion beam, which must be optimally transmitted through the beam-handling equipment. The ion source parameters and other optical components should

G.D. Alton / Ion sourcesfor materialsresearch

229

be adjusted so as to maximize the current through an aperture at some point after mass analysis. The material efficiency measured in this way may be expressed as n=4_=

Z 9W,A

I

2aM r’z Z qn,A ’ i kT0 I

(29)

where Z is the analyzed current of the desired species of charge q and mass M, and A is the area of the ion source extraction aperture. 2.2. Types of arc plasma discharge sources

WltGOW

WITHOUT

INTERCEPTION

b-CYLINDER

,--I

+-LOT

0

+

I

I

I

I

I

0.1

0.2

0.3

0.4

0.5

j+,

I

(A/cm21

Fig. 6. Computationally determined ion beam angular divergence e,, versus ion extraction current j, for slot and cylindrical apertures. (From ref. [18].)

2.2.1. The duoplasmatron The duoplasmatron source has been the subject of further development since its introduction by Von Ardenne [22]. The source and variations are widely used at various research installations, primarily for the production of high-energy ion beams from noncorrosive gaseous materials. The source exhibits high gas efficiency, high beam intensities of singly charged ions, and is being used as well for the production of multiply charged positive ions and negative ion beams. The duoplasmatron is developed from a low pressure discharge maintained between a cathode and an-

VOLTAGE : 30 kV

Fig. 7. Numerical

computational

analysis

of the negative ion generation and extraction negative ion source. (From ref. [2].)

region

of a Middleton-Adams

sputter-type

G.D. Alton / Ion sources for materials research

230

MAGNETIC MIRROR

,r-

REGION

i-

Fig. 8. Illustration

MAGNETIC

ANODE

FIELD

of the ion extraction matron ion source.

LINES

region

of a duoplasFig. 10. Illustration of a long lifetime duoplasmatron which utilizes a LaB, filament [35].

ode. The cathode may be a hot tungsten or tantalum filament, hollow cathode or low-work-function emitter. A strong axial inhomogeneous magnetic field (3-10 kG) maintained between an intermediate ferromagnetic electrode and the anode concentrates the discharge near the extraction aperture in the anode region by the action of the field. As a consequence, the plasma discharge is characterized by two distinct regions: a high pressure region between the cathode and intermediate electrode and a lower pressure region between the intermediate electrode and the anode. A schematic of the extraction region of the source is shown in fig. 8. Fig. 9 displays the potential distribution within the plasma discharge of the source. The plasma in the anode region attains densities of the order of n, = 1 X 1014/cm” from which ion beams are extracted. The source is characterized by high efficiency (greater than 80% for hydrogen) and high current densities.

200 ANODE

r

Y 2 F 400 -,f, F

2

c

I

DOUBLE SHEATHINTERMEDIATE ELECTRODE -

SHEATH--Y

b-

,

CATHODE

CATHODE

ANODE DISTANCE

Fig. 9. Potential

distribution within the plasma duoplasmatron ion source.

discharge

of a

source

A number of descriptions of the emissive properties of various geometry sources can be found in the literature (see e.g., refs. [23-371). Because of the high plasma densities present near the anode, many sources are equipped with a plasma expansion cup which allows the plasma to expand and thereby cool [31]. This technique increases the area over which the beam can be extracted and transmitted through the system. The effect of the expansion cup on beam optics has been investigated by a number of experimental groups, including the studies described in refs. [31-331. The characteristics of the source plasma have been studied by a number of research groups, including the work reported in refs. [26-281. Fig. 10 displays a long lifetime duoplasmatron which utilizes a LaB, filament [35]. The incorporation of the LaB, filament for H+ (8 mA) and He+ (10 mA) beam generation extends the lifetime from 150 h to several months over that for a tantalum filament. A single-aperture source has been designed for use as a high-current ( 2 200 mA) deuteron source for the generation of high-energy neutron beams for use in cancer therapy [36]. The source can produce intense beams from gaseous feed materials covering a range of intensities l-100 mA in cw mode and has been used to produce peak beam intensities in pulsed mode up to _ 20 A. The source is used widely in SIMS microprobe applications where high beam densities must be imaged to spot sizes in the range from less than 1 up to 100 pm. Although the source is used primarily for noncorrosive gaseous materials, versions have been constructed that can be used with low vapor pressure and corrosive materials [29,30]. In the source described in ref. [29], the required feed vapor is introduced in the expansion cup while the primary discharge is sustained with he-

G.D. Alton / Ion sources for materials research

lium or argon. The source described in ref. [30] has been designed for multiply-charged ion beams and is fitted with a ring-shaped vaporization oven between the intermediate electrode and the anode for processing solid materials. 2.2,2. The duopigatron source The duopigatron source is a duoplasmatron source to which has been added a reflex discharge section between the intermediate electrode and anode structure. The concept of the source is illustrated in fig. 11. The modification, first proposed by Demirkhanov et al. [38], improves the ionization efficiencies of the source and permits the use of multiple-aperture extraction electrode systems because of the large area plasma region which can be formed by expansion of the plasma flowing from the intermediate structure into the reflex discharge region of the source [38,40]. Ion beam intensities of several A have been realized with this source geometry. Because of the high intensity and broad beam capabilities of the duopigatron, the source has been developed by a number of groups for use in plasma heating experiments, including those cited in refs. 139-411. A single-aperture source has been developed for the generation of high beam intensities (2 200 rnA) of mixtures of hydrogen and deuterium gases for use in neutron generators [42]. Coaxial forms of the source have also been developed, which are designed to pro-

duce high-intensity, uniformly distributed H+ beams 1431. A single-aperture source has been modified to produce metal ions [44]. Beams are formed of the metal of interest by placing inserts on the extraction aperture/ electron reflection electrode. The material from the electrode is sputtered into the discharge where it is ionized and extracted. Beam intensities of N 50 FA of singly charged ions are typical with useful beam intensities of low-charge-state, multiply charged ion beams also present in the spectrum. This technique overcomes some of the species limitations of the source which has otherwise been restricted to gaseous or high-vapor-pressure feed materials. A single-aperture form of the source has also been modified for the generation of Li+ ion beams at densities exceeding 15 mA/cm’ [45]. The duopigatron source geometry allows independent control of the primary and reflex discharges. For example, an inert gas such as Ar can be introduced to support the cathode discharge while a very corrosive material such as oxygen can be introduced into the reflex discharge region for the generation of O+ ion beams. The argon serves as a buffer gas which helps to reduce back diffusion of the 0, into the cathode region where it can severely reduce the lifetime of the hot filament. By utilizing Re filaments and argon support gas in the cathode region, the lifetime of this source has been extended from two or three hours,

COPPER 7

INTERMEDIATE/ ELECTRODE

CATHODE (COPPER)

Fig. 11. Illustration

231

of the duopigatron

source.

G.D. Alton / Ion sources for materials research

232

which are typical lifetimes for chemically active filaments such as tantalum, to greater than 25 hours. This concept has been incorporated by Shubaly et al. to produce total current densities of 160 rnA/cm’ and O+ ion beams exceeding 140 mA from a multiple-aperture source [46]. This source has also been incorporated into an ion implanter for ion implantation into silicon (separation by ion implantation into silicon

(SIMOX)) 1471. Fig. 12 displays the oxygen source of ref. [46]. This source type, equipped with an oven which feeds vaporous materials into the reflex discharge region of the source, has also been recently developed for solid materials [48]. The source oven, which can reach temperatures up to llOO”C, greatly extends the range of species that can be processed in this source type. Ion beam densities of 16, 24, 27, and 8

v

PRIMARY GAS INLET (ARGON)

II

I

_

CATHODE FLANGE INSULATOR

Ill

-COMPRESSOR

COIL

SECONDARY GAS INLET (OXYGEN)

RHENIUM FILAME INTERMEDIATE ELECTRODE

PLASMA APERTURE 1

INSULATOR

0-c WI b 0

1

k_=imCCEL ACCEL ELECTRODE

GAP INSULATOR

ACCEL ELECTRODE

Fig. 12. A high-current

duopigatron

for Of

generation

[46].

G.D. Alton / Ion sourcesfor materialsresearch

233

OVEN HEATER FEED THROUGH

Fig. 13. Schematic drawing of a hollow cathode similar to the source described in ref. [49].

mA/cm2, respectively, are typical of Li+, P+, Ca+, and Bi+ extracted from a linear array of 3, 5-mm diameter circular apertures. 2.2.3. Hollow-cathode ion sources A hollow-cathode ion source, developed by Sidenius for use in electromagnetic isotope separation processes [49], is shown schematically in fig. 13. The hollowcathode ion source relies on emission of electrons from a thin wall tube to maintain the discharge. The oven temperature may be varied between 200 and 2000°C by regulating its position in the innermost tube of the source. Analyzed beams of 400 FA of argon and 150 IJ.A of lead have been observed using a OS-mm-diameter exit aperture. The magnetic field produced by the cathode heating filament is canceled by the windings of the auxiliary magnet coil. The ionization chamber of the source is illustrated in fig. 14. 2.2.4. Penning discharge ion sources The reflex discharge, originated by Penning [6] and often referred to as a PIG (Penning ionization gauge) source, has been used for many years for generating singly- and low-multiply-charged ion beams. Fig. 15 displays a cold cathode Penning source. The Penning discharge ion source consists of two cathodes placed at the ends of a cylindrical hollow anode. Electrons, produced in the discharge, oscillate between the cathodes and are constrained from moving to the anode cylinder by means of a magnetic field directed parallel to the axis of the anode. Ions are extracted through an aperture in the end of the cathode (end extraction) or through an aperture in the anode (side or radial extraction). The latter geometry is compatible with slit apertures. Some sources employ hot cathodes that emit electrons thermally to initiate the discharge while others use cold cathodes where electrons are emitted

principally by secondary processes. The cathode in the hot cathode version may be directly or indirectly heated. A low-power source (30 W) has recently been developed which utilizes single-crystal LaB, cathodes [50]. Using N, feed material, the source produces a total beam current of 1 mA, of which u 30% is N+.

0-

mm

10

GAS FLOW HOLLOW

CATHODE

XTRACTION

z

ELECTRODE

ION BEAM

Fig. 14. Schematic diagram of the ionization chamber of the hollow-cathode source [49].

G.D. Alton / Ion sources for materials research

234

EXTRACTION

EXTRACTED

GAS +1.5kV EXTRACTION VOLTAGE INPUT

Fig. 15. Schematic representation

of an axial-geometry, cold-cathode, Penning-discharge

Multipurpose Penning sources, such as those used in isotope separation and ion implantation, often use a directly heated filament as one of the cathodes. The operational range of the discharge voltage for this source is typically between 50 and 150 V. Early versions of the source are described in refs. [51-531 and the characteristics of the source are described in ref. [16]. A wide variety of materials may be processed in the ion source by external feed of gaseous materials, by direct evaporation of solids in either of two ovens having a temperature range of 600-12OO”C, or by halogenation of heated oxides or elemental materials. Typically, ion currents of a few to several hundred microampkres are produced in sources of this type. The hot cathode source is principally used for low-chargestate ion production because of the limited lifetimes at higher power discharge operation. The discharge chamber of the Nielsen source is illustrated in fig. 16. An improved version of the Nielsen-type Penning source has recently been described in the literature [54]. The modifications include reducing the diameter of the helical coil of the filament and reducing the discharge volume of the source. A comparison of the performance of standard and improved sources is summarized in table 1. A high-intensity, universal version of the hotcathode, Penning discharge has been developed for ion implantation applications [55]. The source is shown in fig. 17. The operating parameters and performance of the source for generating ion beams of the group IIIB and VIB elements are displayed in table 2. The source is equipped with an oven with a controllable temperature range from - 300 to - 1200°C and is used princi-

source.

pally for processing group IIIB and VB elements for ion implantation doping of semiconducting materials. The source is used in conjunction with a tandem electrostatic accelerator for high-energy ion implantation applications and, therefore, initially positive ion beams must be converted to negative ion beams through charge-exchange processes. An axial-geometry source, equipped with a strontium oxide-platinum cathode has been designed for generation of several tens of mA of O+ and C+ 1561. The lifetime of the cathode in the source exceeds 65 hours. Because of the relatively high voltages required to maintain the discharge compared to the hot cathode version, the cold cathode Penning source is often used for production of multiply charged heavy ion beams.

FILAMENT

ANODE CHAMBER

EXTRACTION ELECTRODE

Fig. 16. Structure of a Nielsen-type, hot-cathode, Penning-discharge ion source.

G.D. Alton / Ion sourcesfor materialsresearch

235

Table 1 Comparison of performances of Nielsen and improved Nielsen ion sources 1541 Nielsen Ion species Source gas Discharge volume (cm3) Extraction area (cm’) Arc voltage (VI Arc current (A) Filament/solenoid magnetic direction Extraction voltage (kV) Current density (mA/cm’) Accelerated beam (mA) Lifetime (h) Brightness (uA/rr2mm2 mrad’ MeV) lnx (a mmmrad(MeV))‘/’ 6,” (P mmmrad(MeV))1/2

Improved Nielsen

Ar+ Ar 6.6 0.031 60 0.7

A?+ Ar 6.6 0.031 60 0.7

_ 200 10 0.17 50 61.7 1.96 1.96

200 10 0.02 50 _

The radial extraction geometry is widely used for multiply charged heavy ion beam production in cyclotron applications and for heavy ion synchrotron applications [57-621. Higher charge states presumably result from longer containment times than those in the axial extraction geometry source. The cold cathode Penning source is used primarily to produce ions from gaseous materials, but its range of capability may be extended by direct vaporization and/or internal chemical synthesis [S&60], or the use of the sputtering technique

Ar+ Ar 1.7 0.031 60 0.7

Ar2f Ar 1.7 0.031 60 0.7

antiparallel 200 40 0.7 50 364 1.96 1.96

antiparallel 200 40 0.08 50

[57,60,62]. An axial-geometry source designed for low power consumption and multiply charged ion productions [63] is shown in fig. lg. 2.2.5. Freeman sources The Freeman source [64] operates according to the magnetron principle, but because of its wide use and popularity as a medium intensity source, a separate distinction is made. The source, shown in fig. 19, was developed for use with an electromagnetic isotope sep-

/

FILAMENT POWER CONNECTOR

VAPORIZER

QUICK-RELEASE FILAMENT CLAMP

ION BEAM EXTRACTION APERTURE

\ ANODE

_

“nr”“lLcn OVEN

Fig. 17. Schematic drawing of the high-intensity, hot-cathode, Penning-discharge

source of ref. [55].

236

G.D. Alton / Ion sources for materials research

Table 2 Typical ion source operating conditions for the high-intensity hot-cathode PIG source 1551 Filament current (A) Arc current (A) Arc voltage (V) Solenoid current (A) Extraction voltage (kV) Extraction current (mA) Emission area (cm’) Current density (mA/cm’) Lifetime (h) 3lP+ (mA) “B+ (mA) Positive to negative ion conversion efficiencies (%)

arator by Freeman and, therefore, has the capability of producing a wide variety of ion beams. The most important characteristic of the Freeman source is the quiescent, hash-free characteristic of ion beams extracted from the source which are highly desirable for high-resolution isotope separation applications. The plasma from which the ion beam is extracted must be free of high-frequency oscillations or “hash” in order to accomplish this objective. Such oscillations modulate the ion beam; the consequent changes in space charge compensation cause the quality of the focus at the image position of the isotope separator to be degraded. The temperature range of the source oven lies between 5.50 and 1100°C. The source oven is also equipped with an inlet for introducing halogenating agents such as Ccl, or ClF,. Ion beams of the platinum or palladium

170 2.5 to 8 60 3 35 20 to 30 0.30 65-100 80-100 > 10 >4 20-25 for “B+ > 50 for 13P+ or 75Asf

l-l

/--EXTRACTION

ELECT RODE

--w

COOLING

ANODE

SYSTEM

CYLINDER

ANTICATHODE

x Fig. 18. Schematic drawing of a low-power-consumption,

INSULATOR

cold-cathode, Penning-discharge

ion source 1631.

G.D. Alton / ion sources for materials research

237

REFLECTOR

Fig. 19. Schematic drawing of the Freeman (slit-ape~ure) ion source. metric ion density increase causes an asymmetric beam profile along the extraction slit. As a consequence of this effect, the source usually exhibits a very nonuniform density distribution along the direction of the slit aperture. The source described in ref. [67] has been modified to eliminate this problem.

groups of metals may also be produced by mounting a sputtering probe made of the material in the rear of the discharge chamber. The cathode is usually a tantalum or tungsten rod, often with a flat emitting surface, mounted 3 mm behind a slit aperture (1.5 X 42 mm’>. An auxiliary magnetic field of O-150 G is maintained parallel to the axis of the cathode. The source produces analyzed currents up to a few milliamperes of most elemental materials. This particular source type is widely employed as the principal source used in medium to high intensity ion implantation systems [65,66]. One of the major problems inherent in the Freeman ion source, however, is caused by the axial electron E x B drift to the positive side of the filament. (The electric field E is perpendicular to the axial filament and the magnetic field B results mainly from the current used to heat the filament.) The electron drift results in higher plasma densities at the positive filament end which results in nonsymmetric filament erosion. Nonsymmetric filament erosion leads to decreased ion source service life and the nonsym-

i

2.2.6. Magnetron ion sources The magnetron source (fig. 20) developed for use in mass spectrometry 1681 is similar in principle to the Freeman source. The primary difference is that the magnetron source filament is coaxial with the discharge chamber and auxiliary magnetic field. The magnetron discharge exhibits an abrupt cutoff at magnetic field strengths of a few hundred gauss. Ion beams of metals may be produced by biasing a metal probe, made of the material of interest, negative with respect to the discharge chamber, or by coating the cathode with the material to be vaporized. The source described in ref. 1691has been developed to produce Sb+ ion beams. In this source, the Sb feed material is

I

HOT CATHODE ANODE i--

CHARGE MATERIAL1

, hGA3$ALENT

AMALYZER

Fig. 20. Schematic diagrams of two magnetron (slit-aper~re)

ion source discharge chambers. (From refs. [68,69].)

G.D. Alton / Ion sources for materials research

238

G

-FILAMENT

ITIVE NS

S

I

I

Fig. 21. A schematic drawing of a Calutron, slit-aperture, high-intensity ion source. netic field volume of the large-scale calutron isotope separators [70]. Several smaller versions of the source have been designed and constructed for use outside the main magnetic field and, therefore, utilize small auxiliary magnets to direct the discharge. A schematic drawing of the source is shown in fig. 21. The discharge is effected by a hot filament located at one end of the discharge chamber. This arrangement permits use of a nonreactive primary discharge support gas such as Ar which acts as a differential buffer against corrosive

placed in the discharge chamber and vaporized by filament and plasma discharge power. The source with typical slit extraction apertures of 2 X 10 mm2 produces ion currents from gaseous materials up to a few milliamperes. 2.2.7. Calutron-type sources The Calutron source is a side extraction, hot cathode source with a slit extraction aperture and was originally designed to operate within the main mag160 -

I

I

I

I

I

I

I

FEED MATERIAL:

146 --

I

I

I

0 ELEMENTAL V COMPOUND

-

120 -

THEORETICAL SPACE CHARGE LIMITED CURRENT DENSITY (LANGMUIR-CHILDS LAW)

60

f

j(makm2) = .$ co

t\

0

20

Fig. 22. Magnetically analyzed ion current

40

60 80 100 120 140 160 ION ATOMIC MASS (amu)

density data observed

as a function for the various

of atomic species.

mass.

180

200

Ion currents

represent

average

values

239

G.D. Alton / Ion sources for materials research

Fig. 23. A schematic drawing of a Calutron/Bernas/Nier,

feed gases which are fed independently into the ionization chamber of the source, thus extending the lifetime of the source. The source is versatile and may be used with gases, elemental and molecular solids, or internal chemical synthesis. The controllable temperature range of the source is N 300-1200°C. Ion beams of up to 200 mA have been obtained from a source with extraction aperture of 4.8 X 127 mm’. The material utilization efficiency from the large-scale source typically ranges from 5% to more than 30%. Average current densities versus atomic mass which have been extracted from the source are shown in fig. 22. The source described in

slit-aperture, high-intensity ion source.

ref. [71] and developed for use in ion implantation is in essence a Calutron source. The source generates current densities of N 20 mA/cm2 and has a lifetime of greater than 90 hours. The Calutron/ Bernas/ Nier configuration [72] of the source is almost identical to the Calutron source, except that in this version the filament is mounted within the discharge instead of outside as generally used in Calutron sources. Thus, the hot cathode is exposed to the corrosive lifetimes of all components in the discharge. The lifetime of the cathode, in principle, should be less than the lifetime of the calutron cathFIELD LINES

(a)

’ DIRECTION OF MAGNETIZATION

lb)

IRON

Fig. 24. Illustrations of possible multi-cusp-field configurations formed with permanent magnets for plasma confinement.

240

G.D. Alton / Ion sourcesfor materialsresearch

ode. This source is also used in ion implantation applications [66,73,74]. This source configuration is illustrated schematically in fig. 23.

LEGEND H : HOT ANODE C : CATHODE SP: SPU-ITER ELECTRODE

2.3. Sources based on the use of multicusp, magnetic-field plasma confinement techniques

A multicusp, magnetic-field provides a simple, but effective, means for reducing electron and ion losses at vacuum chamber walls. Cusp-type magnetic-field boundaries can be formed by a so-called “picket” fence array of parallel conductors with currents flowing in opposite directions [75]. The wires may be placed around the circumference of a cylindrical vacuum chamber or in a planar array on a flat vacuum chamber. When the cusp fields are strong enough so that the electron precession radii are small compared to the cusp-to-cusp distance, most of the electrons will be reflected back into the plasma volume; only those electrons which are moving along the loss cone or cusp direction can escape. Arrays of permanent magnets (e.g., SmCo,, NdFeB, AlNiCo) are often used to simulate the picket fence plasma confinement current-carrying wire arrangement. Fig. 24 illustrates a few cuspfield configurations which may be utilized to effect plasma confinement. Various configurations have been utilized effectively to confine the plasmas in volumetype positive and negative ion sources. 2.3.1. The high-intensity, multicusp, magnetic-field source (CHORDIS).

A versatile, high-intensity ion source has been developed at the Gesellschaft fiir Schwerionenforschung (GSI) for use in a variety of applications [76], including the modification of material surfaces. The source referred to by the authors as CHORDIS (cold and hot reflex discharge ion source) is a modular source which is designed to produce mA beam intensities from a wide variety of metal, gaseous, and volatile compound feed materials [76,77]. The source can be operated either in dc or pulsed mode. The discharge is effected by hot tantalum cathodes maintained at discharge potential. The source is equipped with an oven for processing elemental or compound materials which have vapor pressures between 10e4 and lo-* Torr at temperatures between - 300 and 1200°C and a sputter cathode for metal elements. A source equipped with a sputter cathode is shown in fig. 25. Ions are usually extracted from the source with a three-electrode, multiple-aperture extraction system; the source may be operated with single-slit or single circular-aperture extraction system, as well. The source is flexible and versatile as indicated in table 3, which lists the beam intensities and source operational parameters for a few elements that have been provided with the source.

Fig. 25. Schematic drawing of the cold or hot reflex discharge ion source equipped with provisions for sputtering the cathode material [76].

2.3.2. Multicusp, magnetic-field N t sources Nitrogen ion implantation is used industrially to increase the surface hardness and wear resistance of metals. The implantation process also has the effect of producing much smoother surfaces than is possible with untreated material, resulting in less friction for contacting surfaces such as ball bearings. Nitrogen ion implantation is usually carried out at energies of lo-400 keV and dose levels of 1O’6-1O’s ions/cm*. A pure beam of N+ ions is desirable since the usual mass separation process needed to remove the molecular ions (NC) is unnecessary. Experimental results have shown that better wear and corrosion resistance are achieved when implantation is done with only N+ ions rather than a mixture of both N+ and NC ions. A small, single-aperture, multicusp ion source has been developed which is capable of generating nearly pure atomic nitrogen ion (N+) beams (98.5% N+, 1.5% N:) [78]. A magnetic filter is used to achieve high-purity N+ ion beams. The magnetic filter serves to elimiTable 3 Performance of the CHORDIS ion source [76] Species

H He Li N Ne Al Ar Kr Xe I Bi

Ion beam intensity (m.4) 80

120 59 52 53 2.4 42 86 71 28 31

I’, (kV)

Number of apertures

Emission area (cm*)

25 47 30 25 40 20 50 45 50 31 30

13 7 7 7 7 1 7 1 7 7 7

1.0 0.5 2.0 2.7 0.9 0.8 1.3 2.7 2.0 1.4 2.0

241

G.D. Alton / Ion sourcesfor materialsresearch

nate high energy electrons from the extraction region. Without the filter, electron impact ionization of N, near the extraction aperture results in a substantial increase in the fraction of N: impurity ions extracted. This source type can also be used for producing ion beams of many other species from gaseous, molecular compound feed materials. 2.4. High-frequency (@ plasma discharge sources In the case of a continuous discharge, electrons are accelerated by a steady state electric field. At sufficiently low pressures, the electrons will pass through a gas without undergoing ionizing collisions - provided that the mean free path A is greater than the spacing between the electrodes d. In this case, the discharge can only be sustained by secondary electron emission followed by acceleration. At sufficiently high pressures, the mean free path h will be small compared to the spacing of the electrodes d, and the primary electrons will make numerous ionizing collisions and liberate secondary electrons that also contribute to the maintenance of the discharge. The latter process is a much more effective means of producing ions and stabilizing the discharge. Ionization effected by an alternating electric field (ac) differs in many respects from that produced by a steady state (dc) field. During the periodic variation of the field, the charge carriers are not swept out of the discharge region toward the walls or the acceleration electrodes where they can be captured. The reduction in charge loss makes possible a slow increase in ionization with rather weak fields, which can lead to a self-sustained discharge. The secondary processes that occur at the electrode boundaries no longer are important for sustaining the discharge unless the phase of the field is correct for acceleration. As a consequence, the dramatic electrode wear, attributable to sputtering associated with steady state discharges, is not present and high frequency discharge ion sources have much longer lifetimes than those utilizing a steady state plasma discharge. Some of the im~rtant parameters that characterize a high frequency discharge, oscillating under the influence of an electric field E = EOeior are given below (see, e.g., standard reference books such as ref. [79]): (1) The pressure p and consequently the mean free path A and collision frequency v = u/h of the electron where u is the velocity of the electron; (2) the frequency of the alternating field f = w/27~ and the wavelength of the alternating field, h,; and (3) the electrode spacing d and radius of the discharge volume r. Types of discharges. If A > d and r, the electrons strike the walls more often than they collide with the gaseous vapors, and secondary wall effects dominate.

Such processes generally occur at very low pressures (p I 10e3 Torr). At medium or high pressures (p I 10-3-10-’ Torr) and with low frequency fields such that h
dud

m--e&sin&--mvud, dt

(30)

which can be readily integrated to give V, =

eE, sin(wt - 4) m(v2 + u2)l’*

(31)

’

In the previous expression, E, is the amplitude of the electric field which oscillates at frequency U, 4 is the phase angle, and v is the collision frequency of the electrons in the medium. Integration of eq. (31) results in the average motion of an electron in the medium: x=-

e& cos(ot - I$) mw (P2 + *2)t/2 + constant,

(32)

which occurs about a fixed position in the medium. The electron current density can be expressed as nOe2

j, = nOeud = -Eo m

sin(wt-4) (v’

+

02)

l/2

’

(33)

where no is the number density of neutral atoms or molecules per unit volume. The energy gained by the electrons per unit volume P then can be readily determined from the product P = jE or P=

noe2~~~~s

tir,-

cm ha)

2m( v2 + ~0~)~”

’

(34)

which, when averaged over time, gives the mean value of energy gained by the electrons: p=

nOe2 -E;+ 2m

v +u**

If the electric field is represented by E = EoeiO’, then the medium possesses complex conductivity given by fl=--

1

noez

m (iw + Y)

G.D. Alton / Ion sources for materials research

COOLING FIN (Al)

DISCHARGE TUBE

TO OSCILLATOR

GAS INLET -

EXTRACTING ELECTRODE (Al)

Fig. 26. Illustration

coupled rf ion source.

of a capacitatively

with real part ur reaching a maximum whenever o = v is given by rzoe2

Y

(37)

u r =-v2+w2.

m

2.4.1.

Types of high-frequency

(rf) ion sources

The radio frequency discharge has been investigated by several experimenters. The discharge is effected by an rf oscillator with frequencies in the range

of 20-100 MHz. An axial magnetic field is usually employed to direct the discharge and the ions are extracted through a canal in one end of the chamber. The source generally requires an order of magnitude higher gas pressure (10P2-10-l Torr) than conventional hot cathode discharge sources. The rf power may be capacitatively or inductively coupled to the plasma. A typical capacitatively coupled source is shown in fig. 26. The vessel is made of quartz or glass to reduce recombination at the inside surface of the vessel. Wall recombination includes not only neutralization of ions by electrons, but also recombination of neutral atoms and ions. A large steady potential difference is maintained between the extraction electrode and anode. The degree of ionization is observed to increase monotonically with the high frequency power applied. Typical rf discharges require between 30 and 700 W of power, depending on the source type. In certain sources, a steady magnetic field is applied parallel to the steady state electric field or transverse to it. The greatest increase of ion density is achieved with a transverse field whose magnitude corresponds to the ion cyclotron frequency. The rf power is inductively coupled to the plasma by a coil placed around the discharge vessel. In this case, the gas electrons are accelerated by the magnetically induced rotational electric field. An additional dc magnetic field can be used to constrict the plasma to the central portion of the discharge chamber. High-frequency ion sources have been used extensively for the production of protons for use in accelerators. The extracted ion intensities can be high, but the energy spread of the ions is also high (40-100 eV>. These sources are particularly valuable for the produc-

0

5

EXTRACTION

CATHODE PYREX

ITI)

J GLAZED

CERAMIC

BEAM

CURRENT

TRANSFORMER

Fig. 27. Schematic

representation

of a high-intensity

rf ion source

for protons

[90].

G.D. Alton / Ion sourcesfor materialsresearch

tion of beams of high proton content ( - 70%) in comparison with H: ions. Radio frequency sources are also suitable for use with heavier ions and they can be used for the production of negative ions. The first high frequency source was built by Getting [80] and the technology advanced further by Thoneman et al. [Sl] and Moak et al. [82]. Many sources, including those described in refs. [83-861 have been described in the literature. An early review of high-frequency sources was given in ref. [87], which includes a tabulation of the physical details of several sources. The pressure and intensity characteristics of the rf discharge are described in ref. [88]. Various extraction geometries have been investigated to optimize beam shape, source efficiency, and gas consumption. For microfocused beam applications, the potential of this source appears to be limited due to the high energy spread (30-500 eV) of the extracted ions, which is about one order of magnitude higher than in the duoplasmatron and more than two orders higher than in the surface ionization source

CS91. Proton currents of the order of 100 mA have been obtained in pulsed-mode operation 1901.The source is generally used for simple gases, but versions have been reported that can be used to produce beams from more complex gases and solids. A schematic of the source of Regenstreif for the production of proton beams is shown in fig. 27 [90]. While the rf source is clearly better suited for processing gaseous feed materials, a few sources have been developed, such as the sources described in refs. [91,92], which utilize sputtering of the exit canal made of the metal from which the ion species is produced. However, the useful lifetimes of rf sources which utilize sputtering or gaseous molecular compounds may be reduced from 2 50 h to a few hours by coating of the quartz vacuum envelope with sputter deposits or chemical reaction and dissociation products. The ratio of H+ to H+ and Hl is also known to be seriously affected d; deposits on the quartz tube. In recent years, high-power rf sources have been developed which utilize multicusp, magnetic-field, plasma-confinement technology to generate ion beams from gaseous feed materials. Sources designed for generating negative H-/Dand heavy negative ion beams will be described in section 3. This type of source has the potential for generating high-intensi~ H+ and N’ beams, as weil. 2.5. Electron-cyclotron resonance sources During the past several years, multiply charged ion sources based on the electron-cyclotron resonance (ECR) principle have been developed. The concept was first utilized by Geller et al. for multiply-charged ion production [93,94]. The ECR source technology has advanced steadily over the years, driven by needs for

243

high-charge-state beams for high-energy applications. A major advantage of this source concept is that the bombarding and ionizing electron current is generated from the plasma itself and there are no emissive electrodes which can erode as a result of ion bombardment and thus limit the lifetime for continuous operation. The ECR source, as applied for multiply-charged ion generation, has been reviewed in refs. [95] and [96], respectively, while scaling laws for multiply-charged ion formation have been postulated by Geller [97]. The source is comprised of a multi-mode cavity that serves as the plasma generation and containment cavity. Magnetic mirror coils are situated at the inlet and ion-extraction ends of the cavity to confine the plasma in the axial direction. Multicusp, magnetic fields are usually used to assist in containing the plasma in the radial direction. A high frequency cavity is positioned on the inlet side of the source and high frequency power of several gigahertz is used to produce a plasma from the gaseous feed material that is metered into the source. The operating pressure in this so-called first stage of the source is - lOem Torr. After generation, the plasma drifts down the axial magnetic field gradient into a second cavity of the source where the electrons are resonantly excited by the high frequency field at a frequency w,~ = oE = eB/m,;

(38) in eq. (38) wrf is the excitation frequency, o, is the electron cyclotron frequency, e is the magnitude of the electronic charge, B is the magnetic flux density, and 111, is the mass of the electron. Wherever the resonance condition of eq. (38) is satisfied, the electrons are stochastically heated, a small fraction can actually attain energies of a few kiioelectronvoits and are thus capable of removing tightly bound electrons from heavy ions. The pressure in this so-called second stage of the ECR source is - 10e6 Torr. Ionization occurs principally by single electron removal with contributions from multiple electron removal and inner shell vacancy creation, which may result in Auger electron ejection. Single electron loss processes are believed to dominate so that multiple collisions are necessary to produce highly charged ions. Thus, it is necessary to make the product nyi as large as possible, where n, is the electron density and +ri is the containment time of the ion beam. Power can be coupled into the plasma until the plasma density reaches the critical value ner at which time the plasma excitation wrf, and electron cyclotron w, frequenW,Pj cies are equal, i.e. o,r = w, = Up. The plasma frequency is related to the critical plasma density through the following expression: I 4-i7n,e2 \“’

“p=!yf ;

(39)

G. D. Alton / Ion sources for materials research

244

therefore, n, at&.

(40)

However, for a fixed magnetic field design configuration, o,r and thus n, is limited by the attainable magnetic field B in the multimode cavity of the source. Geller et al. [93] have verified that the plasma density increases according to eq. (401.. To increase the number of highly charged ions, multiple-stage sources (usually two stages) have been developed so that the ion containment time ri can be increased to values compatible with those required for production of higher charge state ions. The source shown in fig. 28 is a multiply-charged, two-stage source which is utilized for injection into a cyclotron [95]. Ion-ion collision recombination processes tend to lower the charge state distribution at higher operating pressures and therefore differential pumping is utilized so that lower pressures can be maintained from stage to stage. Each stage can have its own magnetic mirror configuration and high frequency power generator. Ions diffuse along the axial field gradient and flow from stage to stage until they are extracted. The charge state distributions from a single stage are somewhat better than those obtained from high power radial geometry Penning discharge sources with self-heated cathodes while increasing stages move the charge state closer to that achieved in an electron beam ion source (EBIS). The rf power generator usually has the capability of producing a few hundred to several kilowatts of power, a fraction of which is coupled to the plasma. The

power may be pulsed with typical widths of tens to hundreds of milliseconds at variable repetition rates or be continuous. Electrons at the resonance frequency are excited essentially in a direction perpendicular to the axial magnetic field so that their energy perpendicular to the field is much greater than that parallel to the field, or T, z=-T,, . Because of the direction of excitation and the magnetic mirror geometry, the electrons are very effectively trapped and therefore it is possible to bombard ions continually with several amperes of electron current during their diffusion along the field. Ions arriving at the outlet end of the source diffuse through an aperture into a field region where they are extracted. Total current densities of a few amperes per square centimeter have been extracted from the single-stage source. While the source is perhaps the most efficient and best method for producing cw beams of highly charged ions for injection into high-energy acceleration devices, e.g., the RFQ linear accelerator and cyclotron, low-charge-state, single-stage ECR sources have also recently been developed for use in producing beams of rare isotopic elements [98,99]. According to eq. (391, the maximum plasma densities that can be achieved under ECR conditions are limited by the plasma excitation frequency o,r which, in turn, is related to the magnetic field strength B through w,. Since the critical density scales as o$ and B*, the increase in density requires simultaneous increases in the power source frequency and magnetic field. (The cost of microwave power supplies increases

NY-B

IronYo e f

Sextupole

0.8

0.6

10 Fig. 28. Schematic

drawing

20

30

of a two-stage,

40 14-GHz,

60,

60

70

80

ECR source for multiply-charged

90

100

ion generation

[95].

G.D. Alton / Ion sources for materials research

MICROWAVE

245

POWER

DISCHARGE

CHANBER BORON NIT

ACCELERATION

ELECTR MIDDLE ELECTRODE DECELERATION

(20

ELECTRODE (-2

GROUND ELECTRODE

(0 kV

Fig. 29. Schematic drawing of the high-intensity, slit-aperture, microwave source described in ref. [104].

considerably with w,.) However, the density limitation can be overcome by operation in the so-called “overdense” mode by using right-hand circularly polarized waves for which there is no cutoff limit (see e.g., refs. [lOO,lOl]). This means that a low cost 2.45-GHz power source can be utilized to achieve densities far exceeding the theoretical ECR limit of n, = 7.45 X 10”/cm3. For example, densities II, 2 3 X 1013/cm3 have been achieved using a 2.45-GHz magnetron power source [102]. Circularly polarized waves are produced by transforming the incident polarized TE,, waves generated by the magnetron to TE,, waves by inserting a dielectric plate polarizer in the wave guide. The magnetic field necessary to excite an “overdense” plasma,

Table 4 Comparison of performances of Calutron, Calutron/Bernas Calutron a Ion species Source gas Extraction area (cm*) Arc voltage (V) Arc current (amp) Microwave power(W) Extraction voltage (kV) Current density (mA/cm*) Gas flow (cm’/min) Lifetime (h)

Bf BCl, 3.2 100 4.5 N/A 35 65-100 IO-12 60-80

a Calutron (ORNL). b Calutron/Bernas; Freeman (Eaton/Nova). ’ Sakudo Microwave.

however, must be increased over that required to meet the ECR conditions, e.g., w,/w,~ > 1.

2.5.1.

of

low-charge-state

ECR

(microwave)

Several microwave sources have taken advantage of the ECR phenomenon. Sakudo et al. have developed a multiple-aperture ion source for fusion-reactor-related research, modification of material surfaces and SIMOX applications [103] and a slit-type source [lo41 suitable for use with ion implantation or electromagnetic isotope separator applications. All of these sources use microwave power to produce “overdense” plasmas. A

and Sakudo ion implantation sources Calutron/Bernas B+

BF, 3.2 80 3.0 N/A 35 65-100 l-2 60-100

Types

sources

1.2 100 5.5 N/A 20 44 7 40-50

b

PC PO 1.2 89 3.0 N/A 20 35 1 43

Freeman b B+ BF, 0.8 100 4.5 N/A 35 44 -7 40-50

Sakudo ’

;i 0.8 N/A N/A 300 30 50 0.1-0.5 r 100

246

G.D. Alton / Ion sources for materials research

schematic representation of the slit-type source is shown in fig. 29. The discharge is initiated and sustained by microwave-induced excitation and ionization of the gaseous feed material. The frequency of the microwave generator is 2.45 GHz. Waveguide techniques are used to optimally transmit and couple several hundred watts of microwave power to the discharge volume, which is effective in producing a highly ionized plasma. A quarter-wave dielectric located in the wave guide of the source produces circularly polarized radiation which is used to produce “overdense” plasmas. The dielectric also seals against atmospheric pressure. The discharge volume is designed to be compatible with ion extraction from slit apertures. Analyzed beams of P+ of 10 mA have been observed from an exit aperture of 2 x 40 mm’. This technique has proved to be very successful, especially for gaseous and high-vapor-pressure feed materials. Because of the nature of the plasma generation technique, the source has a very long lifetime unless the exposed dielectric components degrade, because of exposure to discharge deposits and chemically corrosive materials. The principal problem which limits lifetime is due to backstreaming electrons which strike the dielectric polarization plates located in the rear of the source and ultimately cause vacuum leaks. A comparison of the performance of this source with selected Freeman, Calutron, and Calutron/Bernas sources is made in table 4. There has been increasing demand in many fields for high-current ion sources with reasonable lifetimes which can be used to produce ion beams from a variety of chemically active and corrosively reactive gases. In the ion implantation industry, for example, ion implantation into semiconducting materials is being used at an increasing rate for preparation of active regions between buried oxide layers and to provide isolation barriers in devices. Increasingly higher energies are required because of device fabrication with deeper active layers. For example, the revival of CMOS technology and the need for latch-up and single-event immunity have created a need for deep-layer isolation barriers beneath active regions of a particular device. To date, one of the most promising approaches to substrate isolation is separation by implantation of oxygen (SIMOX), which also has the beneficial effects of lower leakage currents and faster switching speeds for devices. High-energy implantation is also being used for the creation of CMOS p-wells which utilize much lower power and offer greater latch-up immunity. The number of other applications for high-energy ion implantation continues to grow, such as the programming of read-only memory chips (ROMS), grid formation for charge collection in dynamic random access memory chips (RAMS), junction field emission transistors (FETs), ion lithography, mask replication,

Fig. 30. Schematic diagram of the high-intensity, O+ microwave source described in ref. [106].

wave guide formation, ion beam mixing and modification of material surfaces. Oxygen is particularly deleterious to hot cathode sources, the use of which may reduce the lifetime to a very few hours due to the chemical reactions between the oxygen feed gas and hot Ta or W filaments traditionally used in such sources. The microwave or ECR ion source has an advantage over most conventional sources in that it lacks a filament, the component which limits the ultimate functional lifetime of the source. Thus, stability of operation with reactive gases can be easily achieved due to their nonfilament structure while simultaneously having the benefit of a long lifetime. A few sources have been successfully developed for corrosive feed materials processing such as oxygen [105-1071. O+ beams are important in the semiconducting materials industry for use in the separation by implantation of oxygen (SIMOX) technology. SIMOX requires the implantation of very high doses (- 10 ” ions/cm’) of oxygen beneath the silicon surface. These sources have the capability of producing O+ currents exceeding 100 mA as required for use in commercial ion implantation devices. Fig. 30 displays the source developed by Torii et al. which produces O+ beams exceeding 138 mA at current densities exceeding 150 mA/cm2 [106]. This source has the following features: (a> a high-current density of 1.50 mA/cm’ and a large extracted current of more than 200 mA, (b) a long lifetime of more than 200 h, and (c) a high O+ ratio of more than 80%. Fig. 31 displays the ion beam intensity versus ion extraction aperture area [106]. A multicusp-field source has been developed which utilizes microwave plasma cathodes to supply electrons for ignition of the primary plasma in the cusp field region of the source [log]. The plasma in the cathode region of the source is ignited by three coaxial rod-type antennae located on the back plate of the source which

G.D. Alton / Ion sourcesfor materialsresearch

MICROWAVE

FREQUENCY:

MICROWAVE

POWER:

2.45 GHz

1 kW

EXTRACTED

0+ ION CURRENT

BEAM DIAMETER: (20mm9) (31 mm$)

0

1

2

3

(35mm+)

4

5

EXTRACTIONAREA (cm2 ) Fig. 31. Ion beam intensity as a function of extraction aperture area [106].

at 2.45 GHz under ECR conditions. Total beams of Ar (exceeding 230 mA) and 0 (exceeding 130 mA) have been extracted from the source with a multiple-aperture emission diameter of 115 mm. The lifetime of the source, which is determined essentially by the lifetime of the antennae used to produce the microwave plasma, exceeds 100 h. A low-emittance, high-intensity ECR (microwave) source operating with a 2.45-GHz magnetron has been developed for producing high-proton-ratio H+ beams for postacceleration with an RFQ injector [109]. Proton fractions up to 90% and total extracted beam densities up to 350 mA/cm’ have been extracted from a 4-mmdiameter single-aperture source. The rms emittance of the source for a total beam intensity of 25 mA is 1.55a mm mrad (MeV)‘/*. A compact ECR source has been developed which utilizes a quartz tube [llO]; the plasma is produced with a 2.45-GHz magnetron power supply. The source produces hydrogen ion densities of 200 mA/cm2 with an 80% proton fraction. are driven

2.6. Vacuum-arc sources A simple and practical method of producing ions from solid materials, often utilized in mass spectrometry applications, can be effected by producing a periodic low- or high-voltage-spark discharge between two conducting or semiconducting electrodes in a vacuum. The theory of the vacuum arc is described in ref. [ill]. The results of early studies of both low and high voltage spark discharges have been reported by Dempster [112] with primary emphasis on the low-voltage type. Additional information has been accumulated as a result of studies of low voltage discharges by Venkatasubramanian and Duckworth [ 1131 and by Wil-

241

son and Jamba [114]. The high voltage spark discharge has also been investigated by Suladze and Plyutto [115]. These authors have also reviewed previous work on the subject. Sources of each type are illustrated schematically in fig. 32. The low-voltage discharge is produced by mechanically separating two current-carrying electrodes initially in contact, one of which contains the material of interest. During the mechanical interrupt, the contact area decreases and consequently the resistance of the interface increases, leading to a localized heating of a small volume of the material. The temperature of the material quickly rises to a high value, which results in vaporization of the electrodes. As the contacts are further separated and the resistance increases, the potential difference increases through the vapor, producing a highly ionized plasma if the potential difference is greater than the first ionization potential of the vaporized material. The discharge may sustain itself for a long period of time if the correct conditions prevail. Such discharges principally produce singly charged ions. The high-voltage discharge is initiated by applying a high voltage across a gap between two electrodes that exceeds the vacuum breakdown voltage. The high voltage spark discharge, usually effected by applying lo-50 kV across the gap, produces a very highly ionized plasma at high densities (10’5-1016 ions/cm31 with very high electron temperatures (T = lo5 K). Such high temperature plasmas produce a spectrum of highly ionized particles with charge states up to +4 or +5. Positive ion current densities of j+= qn, (/cTJM)~/’ of lo-100 A/cm2 can be generated. Such sources have the advantage that ions can be produced from almost any element from which electrodes can be fabricated. Because of the nature of the discharge, electrode lifetime is very short and the discharge duration is also variable, having typically low duty cycles. The ion energy spread in the low voltage source is significantly lower than produced in the high voltage source [112,113]. 2.6.1. Types of vacuum arc sources Wilson and Jamba [114] have utilized a low-voltage, low-frequency-spark source to produce ions of the group I, II, III, V and VI elements. Two electrodes of the desired element are employed and the discharge is initiated by striking the arc between a third central electrode of either the same element or carbon using a 50-60 Hz interrupted dc voltage of 45 V. Their data indicate that high conductivity elements (such as Al, Cu, Ag and Au) produce large C+ components of current when used with carbon electrodes. The unwanted C+ ion current can be eliminated by choosing identical electrodes containing the element of interest. Techniques for large-area (5-7 cm21 emitting surfaces in high voltage sources have also been developed,

G.D. A/ton / Ion sources for materialsresearch

248 ACCELERATING VOLTAGE-j

IONS

:

trigger-to-cathode. The trigger voltage is connected positive with respect to cathode and attains values between 10 to 20 kV and lasts several microseconds. The ion source repetition rate is limited by the charging times for the arc and trigger circuitries and by average power constraints on various components. Typical repetition rates of a few pulses per second up to a maximum repetition rate of - 10 Hz are achieved. The usual method for triggering is from an annular ring which surrounds the cathode. For hard materials such as niobium, > 100000 triggerings can be effected before failure. For softer materials, the number of triggerings possible before failure is considerably lower. The long pulse length (2 250 us), and low repetition rate (a few Hz) of the source make it well suited for synchrotron accelerator basic research applications; the source is also finding uses for implantation and modification to material surfaces. These sources have demonstrated intensities for many metal ions of several hundred mA when accelerated with extraction voltages up to 100 kV. A schematic of a single-electrode, vacuumarc source [120] is shown in fig. 34. Fig. 35 displays Ti ion beam current as a function of extraction voltage [121]. A high-frequency (1 MHz) spark source has been developed which produces C3+ and C4+ beams with an rms variation from pulse to pulse of less than 10% 11221.

kfiE+l

ACCELERATING

SPARK “_/‘GAP

IONS

Fig. 32. Illustrations of spark discharge ionization sources: lower: high-voltage spark discharge arrangement, upper: lowvoltage spark discharge arrangement.

and ion current densities of 5-15 A/cm’ and total ion currents of 50-200 A of H+ and D+ have been achieved in microsecond pulses [115]. The high-voltage-type-source produces multiply charged ions as evidenced by the results of Zwally et al. from a triggered high voltage spark source 11161. Such sources have also been utilized extensively in high resolution secondary ion mass spectrometry [117,118]. In more recent years, a series of high-current-density, metal-vapor, vacuum-arc-type sources have been developed by Brown et al. [119-1211. The arc power supply for this source is a simple, low-impedance (N 0.5 fl) line with a pulse length of _ 300 us. The line is charged to a voltage of several hundred volts with a small dc power supply, and the output is connected to the anode and cathode terminals of the source. A schematic of the trigger circuitry is shown in fig. 33. When the triggering pulse is applied to the trigger electrode, the cathode-anode circuit is closed by the plasma and the arc proceeds throughout the duration of the pulse. The trigger pulse is generated by the discharge of a O.l+F capacitor of a few kV (up to 5 kV) and is switched with a thyratron, through a step-up isolation transformer, the output of which is connected

2.7. Sources based on field ionization and field euaporation 2.7.1. Field ionization Field electron emission refers to the transfer of electrons from the surface of a metal into the vacuum as a result of the action of very high electric fields at the surface (- lo’-10’ V/cm), which lowers the po-

T

T

Fig. 33. Schematic representation of the trigger circuitry for the vacuum-arc source of refs. [119-1211.

G.D. Alton / Ion sources for materials research

I/

Extractor

grids

,,,,“(~~er /’

Electrode ,, Pyrex vacuum chamber

/’

Quartz sheath Magnetic field coil Fig. 34. Schematic diagram of a high-intensity, vacuum-arc source described in ref. [120].

tential barrier so that electrons can leak out of the metal. Ionization of atoms or molecules adsorbed at or near a surface at high potential can also occur whenever the field polarity is reversed, and thus the technique can be utilized to form positive ions. The field ionization process usually requires electric fields of the order of lo8 V/cm. To achieve surface fields of such high values at moderate potential differences, small diameter spherically tipped or thin, hollow, needle-like electrodes are usually employed. The basic mechanisms of the two processes are essentially the same and involve the tunneling of electrons from the metal into the vacuum (field emission) or from the atom or molecule into the surface (field ionization). The tunneling mechanism can only be explained by quantum mechanical theory and has no classical analog. Analytical expressions for the probability of field emission have been derived by Fowler and Nordheim [123] for an abrupt potential step at the metal surface that neglects the image potential term -e2/4x, where x is the distance of the electron from the surface. More rigorous theoretical calculations that include the image term have been made by Nordheim [124]. Summaries of work in field emission have been given by Good and Miiller [125] and a textbook has been written on the subject by Gomer [126]. The problem of field ionization can be treated in an analogous manner, but is slightly more difficult since the image potential term now must include both nu-

clear and electronic fects. An atom near is strongly polarized the field and image

600

I

terms, as well as polarization efa surface in a strong electric field as a result of the superposition of effects. The potential barrier is

I

I

SAMPLE

Ti

10

30

I

I

I

I hlrc

0 0

20

EXTRACTION

40

50

VOLTAGE

60

70

80

(kV)

Fig. 35. Illustration of the dependence of total Ti ion beam intensity

as a function of ion extraction voltage discharge currents [121].

for different

2.50

G. D. Alton / Ion sources for materials research

lowered by the superposition of the strong electric field and the induced image charge of the electron and nucleus vi,,,. The one-dimensional potential energy for the atom is therefore the sum of the applied and induced image potentials or

E

= -eEx

V(x)

+ y,,

(41)

where e is the electronic charge and x is the distance of the atom from the surface. For the transfer process to occur, the electric field must raise the potential energy of the atomic electron at least to the Fermi level, of the r.:etal. The critical distance for which tunneling can occur is given approximately by

x,=(eY-4)/IEl,

(42)

where eVi is the first ionization potential of the atom, and 4 is the work function of the surface. The probability of barrier penetration or tunneling and hence ionization can be estimated by applying the Wentzei-Kramers-Brillouin (WKB) quantum theory approximation, provided realistic potential energy curves V(X) are available. For the case where the image potential in eq. (41) is assumed to be a Coulomb potential of the form V,, = Ze’/x,

where Z is the effective nuclear charge, then Pi becomes Pi =

I

exp -6.8 X 10

,

( eVi)3’2 , E,

1 -

i

. dno z=-p where q is the ionic charge. The rate of arrival dn,/dt is

dno

-

dt

P

= 2ar:

(21rMkT)‘/2(T7

where P is the pressure, it4 is the mass of the ion, k is Boltzmann’s constant, T is the absolute temperature, and

2 V(rJ

a=l_

kT

3

7.6 x 10-4Z1’2-

‘I2

eVi

)

I. (44)

The field ionization current can be calculated in principle by multiplying the arrival rate of atoms at the surface of the ionizer, dn,/dt, and the probability of ionization Pi. However, the mechanism of generation depends in a complex way on the ambient temperature, field strength, and polarizability of the atom or molecule. At very high fields, all particles arriving at

’

Here V is the potential energy of the particle at the surface of the ionizer of radius rt and - I/ is a positive quantity. Thus, the rate of arrival due to polarization forces is increased by the factor (T over that associated with thermal motion alone. The quantity o may have values from 10 to 100. For a hemispherical ionizer tip of area 2rrr:, the positive ion current which can be generated per Torr of pressure at 300 K is 7

i+(p,A/Torr)

El/2 x

the ionizer tip are ionized so the current is limited only by the rate of arrival. Since polarization forces attract atoms in the vicinity of the ionizer, the rate of arrival is faster than that associated with thermal motion. The high field ion current from the tip of a hemispherical ionizer of radius rt and area A = 2rr: in the central field approximation for polarization forces is

z 2x

lo-

urt

2O

(A)

,

m

For an atom such as He with a value (T= 10, i, (t_tA/Torr) = 0.25. The ion currents for low and intermediate fields are more complicated; the reader is referred to the derivations given by Gomer [126] for these cases.

2.7.1.1. Types of field ionization sources The method of field ionization has been applied in mass spectrometry ion sources, in field ion microscopy, and as sources of

r

NEEDLE

CAPILLARY IONIZER

l- IONIZER

ION BEAM

Fig. 36. Illustration

of two types of field ionization

(48)

sources:

(a) needle

ionizer,

(b) capillary

ionizer.

251

G.D. Alton / Ion sourcesfor materialsresearch

intense ion beams for general use. Two source concepts that produce cylindrically symmetric ion beams are illustrated in fig. 36. The hemispherically tipped ionizer (fig. 36a) produces ions from atoms that pass within the critical distance of the tip. At high pressures, the un-ionized particles in the path of the accelerated ion beam may act as scattering centers and thus degrade the beam quality. An attractive method shown in fig. 36b reduces this effect and thus may be more desirable. It consists of a thin capillary tube through which the gaseous or vaporous material to be ionized can be fed. The method has been implemented by Hendricks [127] and in more recent years, by a number of groups for microfocused beam applications (see e.g., refs. [128-1301). Ref. [131] contains a compilation of work related to the performance of several field-ionization sources. 2.7.2. Field-evaporation or liquid-metal ion sources A potential route to increased brightness beams is through the use of field-ionization or liquid-metal (field-evaporation) ion sources. The liquid-metal ion source (LMIS) has the advantage of higher beam intensities and longer lifetimes, and avoids the gas loading and scattering problems associated with the field-ionization source. In addition, the LMIS is simple in structure, easy to operate, has a very long lifetime, and consumes very little power. Microfocused ion beams are now used in a wide range of fields. These include such diverse applications as micro-fabrication, ion immaterials characterization and space plantation, propulsion. The extremely high brightnesses of the central cores of beams extracted from field-ionization and liquid-metal ion sources make them primary choices for such applications. As a consequence, the liquid-metal ion source (LMIS) has been studied by many groups. A recent comprehensive bibliography, which contains N 1100 references to both field-ionization and liquid-metal ion sources and their applications, as well as several review articles on these source types, has been compiled by MacKenzie and Smith [131], while a textbook has been written by Prewett and Mair [132] which deals with the physics and technology of the LMIS with emphasis on the use of microfocused beams for microscopy and analysis, micromachining and deposition, microcircuit lithography and ion implantation. Although the precise ion formation mechanism in the LMIS is still the subject of controversy, generation of ion beams in this source is quite simple. A strong electric field, applied between a capillary or needle wetted with molten metal at anode potential and an extraction aperture, pulls the metal up into a modified Taylor cone (apex angle 98.6”) with a rounded apex [133]. The two principle types of liquid-metal ion sources are shown schematically in fig. 37. If the field

ION BEAM

ION BEAM

Fig. 37. Illustration of two principle types of liquid-metal ion sources: (a) needle ionizer, (b) capillary ionizer.

is sufficiently strong (10’ V/cm) at the surface of the molten cone, ion desorption will occur. The process whereby ions are removed from the tip region of the molten surface is often referred to as field evaporation. It is highly likely that field ionization and field evaporation processes both contribute to the ionization process. Since the field ionization model has been discussed previously, a brief description of the field evaporation model is given here. For further details of models used to describe the process, the reader is referred to the review article by Marriot and Riviere [134] and the more comprehensive field evaporation model described by Forbes [135]. Although the “image-hump” model has certainly not been confirmed, it is usually used to explain the field evaporation ionization process for the high temperature case. The model views the field evaporation process as the thermal escape of an ion over a one-dimensional barrier. The ion current density, j, which can be extracted through a barrier height of AH can be expressed as j=qNA(w)exp

,

-g (

(49)

1

where q is the charge of the ion, N is the number of active sites per unit area, A(w) is a factor which depends on the vibration frequency of the atoms in the potential well, k is Boltzmann’s constant and T is the absolute temperature of the surface. AH can be expressed according to the following simple relationship: AH=H,,+&+Zq-4-

qlEl l’* 4?rFTE0 ’ i 1

(50)

where Had is the heat of adsorption, Vi, is the image potential energy between an ion and image charge

252

G.D. Alton / Ion sources for materialsresearch

given by eq. (431, Z, is the ionization energy required to ionize the atoms to charge state q, C#Iis the work function of the emission surface, 1E ) is the electric field strength at the surface, and E,, is the permittivity of free space. The last term in eq. (50) is attributable to the Schottky effect which lowers the escape barrier. In this way, beam intensities of a few to several tens of p,LAcan be formed. The remarkable feature of this source type is that the area of ion emission is extremely small - having a radius of curvature of a few Hngstrijms [136,137]. Thus, the LMIS is, in principle, the brightest of existing positive ion sources. The central cores of beams extracted from this source are estimated to have brightnesses of N lo5 times those of conventional gaseous discharge ion sources [138,139]. The liquidmetal ion source offers a means for generating beams with central brightnesses of approximately 1 x 10”’ A/srm2 [139-1411. LMIS beams can be focused into submicron images with current densities of a few A/cm2 and thus offer a means for direct writing while ion implantation doping of semiconducting materials, and for a number of ion beam lithography applications, including mask repair. However, in order to utilize the source for submicron imaging applications (e.g., microprobe surface analysis, ion implantation, microfabrication, and microetching), the angular divergence and, consequently, ion beam intensity must be severely limited. This is a consequence of the highly divergent characteristic of ion beams from this source type and the difficulty of accelerating and focusing beams without severely distorting their phase spaces through spherical aberrational effects (such effects increase as 8:, where 13s is the half-angular divergence into the optical device). The half-angular divergence of LMIS beams range from a few degrees at low beam intensities to a few tens of degrees at higher intensities [142,143]. Inherent in LMIS beams are space-charge effects which cause increases in the angular and energy spreads; these, in turn, lead to increases of the phase space of the ion beam through spherical and chromatic aberrations. Energy spread attributable to space-charge influences near the region of ion emission may be caused by collective and individual particle CoulombCoulomb interactions (Boersch effect) [144]. Theoretical treatment of the Coulomb-collision-dominated problem in a high-density laminar-flow electron beam predicts an energy spread AE proportional to the square root of the ion current Z or A E a 1"' [145]. For the case of a high-density, highly divergent beam dominated by Coulomb-Coulomb interactions, an energy spread AE proportional to Z2j3 or AE a Z213 [146] is theoretically predicted. (Energy spreads A E in the LMIS are found to follow, approximately, the Z1/*to-Z213 relation and have widths of several eV [142,146-1481.) The energy spread AE associated with

collective space-charge effects increases linearly with ion current or AE a I. The resultant energy distribution associated with these effects is superposed on other distributions intrinsic to the source which are independent of ion current (e.g., energy spreads attributable to thermal and ion-formation processes). All additively affect the emittance through chromatic aberrational and dispersive effects in the extraction postacceleration, lensing, and momentum-analysis systems used in the experiments. 2.7.2.1. Types of liquid metal ion sources Liquid-metal ion sources usually rely on the wicking of a molten elemental or alloy metal along the outer surface of a metal needle (e.g., refs. [147-153]), the inner surface of a metal capillary tube (e.g., refs. [154,155]), or through a sintered matrix (refs. [156,157]). Many examples of each of these source types have been developed as exemplified in ref. [131]. The source may be composed of single or multiple emitters. The ion currents achievable from the LMIS increase linearly with the number of emitters [158,159]. A schematical representation of a needle-type Li LMIS equipped with provisions for inserting or withdrawing the needle in the molten Li metal is shown in fig. 38 [160]. The currents which can be achieved with this source type range from a few to several hundred kA. The current which can be achieved can be enhanced by use of multiple emitters such as described in refs. [158,159]. Clampitt [155] has reported high brightness beams of Li+, Cs+, Sn+, Gaf and Hg+ from single emitters with beam intensities ranging from 500 to 700 kA. Current densities from these sources reached densities in excess of lo4 A/cm* at extraction voltages of 3-10 kV. The operating characteristics of liquid-metal sources, therefore, are strong functions of flow characteristics of the particular metal or alloy along the surface of the metal (tungsten) needle or capillary tube. Wagner [161] has explained the flow characterization of the LMIS by use of a hydrodynamic model balancing the pressures due to surface tension and electrostatic forces on the end of the liquid cone of the source and the viscous capillary flow along the tip. A number of sources have been developed which are predicated on ion beam generation from elemental metals. The versatility of the technique has been significantly extended by utilizing the low melting (eutectic) property of alloys. By utilizing low melting point alloys, the LMIS can be used to generate beams of most of the metallic elements. The versatility of this source type in terms of species is illustrated in table 5, which references a number of liquid-metal ion sources that have been developed to generate a variety of ion beams from both elemental and eutectic alloy materials [162-1761. A more comprehensive listing of field emission and liquid-metal ion sources is given in ref. [131]. Fig. 39

253

G.D. Alton / Ion sourcesfor materialsresearch

Table 5 A partial listing of metals and alloys for use in liquid metal ion sources Species

Liquid element or alloy

Be

Au-Be, Au-Be, Au-Si-Be, Ga-Si-Be Pt-B, Pd-Ni-B, Pt-Ni-B Al Au-S& Ga-Si-Be Pd-Ni-B cu Zn Ga B-Pt-Au-Ge Pd-Ni-B-As, As-Sn-Pb, As-Pt Rb Pd-Ni-B In Sn, As-Sn-Pb Sb-Pb-Au cs Pt-B Au Pb, Sb-Pb-Au Bi U

B Al Si Ni cu Zn Ga Ge As Rb Pd In Sn Sb cs Pt Au Pb Bi U

J Fig. 38. Schematic drawing of a needle-type Li* LMIS equipped with provisions for remote wetting or rewetting the needle [1601. 50mm

I

References [162,163,164,165] [162,162,166]

H671 [162,166,165]

K'l [la M91 11701 11711 [162,171,172] 11731 [1621 [1701 [168,171] [166,172] D681 [162,168] [I741 [168,171] v751 [I761

potential (eV,> can be ionized by contact with a surface of high work function that is hot enough to thermally evaporate ions. In this process, a valence electron of the adsorbed atom or molecule is lost to the surface upon evaporation or desorption of the positive ion. The most easily ionized materials for which the technique can be effected are ions of the alkali metals (Cs, Rb, K, Na, and Li). The alkaline-earth, rare earth, and transuranic elements can also be formed as positive

displays the measured emittance of a gallium LMIS as a function of percentage of the beam contained within a particular phase space contour as measured by Alton and Read [141,149,150]. 2.8. Surface ionization and aluminosilicate sources Positive ion beams of several elements can be produced by surface ionization. An atom of low ionization 7.9 6.0 -

Species:

“Ga+

;ny = AJ' + A2P2 + ASP3 _ 2 -v

6.o _

g

4.0 -

$ ‘; ,t

A, = l.O7~10-~;A,

=2.86x10-‘;A,

= 1.31~10~

3.0.

2.0 -

10

20

30

40

PERCENT

60

60

70

OF BEAM INTENSITY,

80

90

II

P

Fig. 39. Emittance as a function of percentage of the beam contained within a particular emittance contour [141,149,150].

254

G.D. Alton / Ion sources for materials research

ions through the process, but with lower efficiencies. The elements that are most difficult to ionize are the elements with high heats of vaporization or high ionization potentials. If we supply enough energy to ionize the atom, transferring the electron to the metal in the case of electropositive atom adsorption in the process, then the atomic and ionic potential curves are separated by an amount evi - 4, where evi is the first ionization potential of the electropositive atom and 4 is the work function of the surface. If enough energy is supplied to evaporate the atom or ion from the hot surface, the probability for arrival at a position far from the metal in a given state depends on the magnitude of eK - 4. For thermodynamic equilibrium processes, the ratio of ions to neutrals that leave an ideal surface can be predicted from Langmuir-Saha surface ionization theory. The probability of positive ion formation, Pi, leaving a heated surface is given by

2 z loo : : _

\

IRIDIUM IONIZING SURFACE (+ = 5.27eV) AT 2000°K

&a ‘TI tea \

I

Q’ 3.5

4.0

I 4.5 Vi, IONIZATION

I

I

5.0 5.5 POTENTIAL (eV)

= -

I 6.0

6.5

Fig. 40. Surface ionization efficiencies of the group IA, IIA, and IIIA elements evaporated from hot iridium metal, calculated from eq. (51).

-’

4-eVi

Xexp

( ~

kT

)I

’

(51)

where r+ and rO are the reflection coefficients of the positive and neutral particles at the surface, w+ and wa are statistical weighting factors, T is the absolute temperature, and k is Boltzmann’s constant. Optimum ionization efficiencies are obtained for high work function materials and low ionization potential atomic species. For elements for which eVi > 4, the process is much less efficient. For example, the work function for clean tungsten is about 4.6 eV and the ionization potential for indium is 5.8 eV. Thus, in this case, the exponential term (4 - eV,> in the Langmuir-Saha relation is negative and, therefore, the probability of ionization is low. A particular technique that helps to improve the ionization efficiency is to incorporate an oxygen spray that is directed onto the ionizer surface. This increases the work function of the emitting surface and hence the efficiency of ionization. The work function of a particular material may vary significantly with crystallographic orientation of the surface. A plot of the efficiency of ionization of a number of elements as a function of the ionization potential evi for the high-work-function iridium surface is shown in fig. 40. It is seen that the efficiency is relatively high for some atoms, but is low (lo-’ or 10m3) for atoms of In, Ca, Al, Ga, and Tl. The incident particles which are not ionized are evaporated as neutral atoms. In recent years, surface ionization sources have been utilized widely to generate Cs+ ion beams for sputtering and effecting low-work-function surfaces

from which negative ion yields can be enhanced basic and applied research and SIMS applications. 2.8.1.

for

Types of surface ionization sources

Ion sources based on the surface ionization principle are generally characterized by a high degree of ion beam purity, low energy spread, and limited range of species capability. The ionization efficiency can be very high or low, depending on the ion-substrate combinations as evidenced by the Saha-Langmuir relationship [eq. (5111. However, the energy spread of the ion beam is characteristically very low and is of the order of thermal energies _ 2 kT ( < 1 eV>. Ion current densities of alkali metals ranging from 10e3 to above lo-* and densities 10m6 to 2 lo-‘A/cm2 of the group IIIB elements have been produced by surface ionization. Several types of surface ionization sources have been used and are characterized by the means by which the atomic vapor is fed onto the ionizing surface and the method used to extract the ions. Three examples of this source type are illustrated in fig. 41. Numbers of sources have been developed such as those described in refs. [177-l@]. The most highly developed and in many respects the simplest type of surface ionization source is illustrated in fig. 41a. The ionizer is a porous body made of sintered tungsten. The atomic vapor is fed from the atomic oven through a sealed tube and then the atoms are diffused through the porous tungsten ionizer where they are ionized. The porous tungsten ionizer has a density _ 80% of that of solid tungsten. At the front surface, a substantial frac-

G.D. Alton / Ion sources for materials research ACCEL

ELECTRODE

V = -2DOO TO 7000 VI P”L”B

IONIZER

255

CLIZI

NEUTRALIZER

HEATER

10

-

!g 0.6

-

5 3

ii O6 z 0.4 4 L 0.2 -

V = +25CKl V /al CESIUM-POROUS

TUNGSTEN ION

ION SOURCE

[

0.0

a__,&&f

800

BEAM

650

9x3

950

’

,

,

lDco

Km3

ID0

1150

TEMPERATURE

PCC)

IONIZER

,

,

1200

1

,

1250

ljoo

Fig. 42. Illustration of the critical temperature effect for ionization of cesium with a porous tungsten ionizer [188]. VAPOR FOCUS

GUIDE AND ELECTRODE RIDIUM ONIZER

HEAT

LIQUID

SHIELD

METAL

/b/ CYLINDRICAL

SURFACE

IONIZATION

SOURCE

FOCUS ELECTRODE ACCELERATING

-IRIDIUM

ELECTRODES

ELECTRODE

LIQUID

METAL

OVEN

(cl CURVILINEAR

70% and 80% for various cesium oven temperatures; these data were derived from the source described in ref. [188]. Examples of surface ionization sources which utilize porous tungsten ionizers that have been developed for generation on Li+, Na+, and In+ ions are reviewed with several other types of sources in refs. [178,187]. Ions of Al, Ga, and Tl require high ionizer temperatures and the ionization process generally exhibits low efficiencies. A solid ionizer is desirable for these materials with the neutral vapor directed away from the target. The curvilinear-type source used by Jamba [181]

.-f&j

SURFACE

IONIZATION

3.5

SOURCE

I

Fig. 41. Illustration of (a) porous tungsten, (b) direct cylindrical, and (c) curvilinear cesium surface ionization sources.

SURFACE AREA: 0.32 3.0

EXTRACTION IONIZER

of the metal is ionized and extracted with an acceleration-deceleration type of electrode system. This type of ion source has also been highly developed to form ion propulsion devices [182]. It is especially suitable for the alkali metals (0, Rb, K, Na and Li), which exhibit relatively high vapor pressures and relatively low critical temperatures - the temperature required to evaporate the ion. The properties of a porous ionizer may change at the high temperatures required for ionization of those metals that have high critical temperatures because of recrystallization processes that tend to change the porosity of the ionizer. Fig. 42 illustrates the critical temperature effect for the ionization of cesium in a source of this type. Characterization of this source type for Cs+ ion beam generation is described in ref. [188]. The perveance of the source and consequently cesium ion current is found to depend on the porosity of the tungsten ionizer. Cesium ion current versus extraction voltage are shown in figs. 43 and 44 for ionizers with porosities, respectively, of

I

I

I

I

I CESIUM WEN TEMP

SOURCE: REFOCUS SPUTTER cm2

GAP: 4 cm

POROSITY:

f~ = 0.7 p,,

tion

2.5

2.0 % +H f.5

9.0

0.5

0

0

4

6

(2

46

20

24

26

V”

Fig. 43. Cesium ion beam intensity I+ versus extraction V,, for a porous tungsten cesium surface ionizer type source [188].

256

G.D. Alton / Ion sources for materials research

SOURCEREFOCUS SPUTTER IONIZER POROSITY ,Q= 0.8~~ SURFACE AREA 0.32 cm2 EXTRACTION GAP’ 4 cm CESIUM OVEN TEMP x50c* /

0

4

8

16

rz

20

24

20

V,, (kV)

Fig. 44. Cesium ion beam intensity If versus extraction voltage VEXfor a porous tungsten type cesium surface ionization source [188].

adapted from the original work of Kino [182] is illustrated in fig. 41~. In this source, the atoms to be ionized are directed from an atomic oven onto the ionizer surface, which is resistively heated. The ions are extracted in a curvilinear path by a relatively complex arrangement of electrodes. In this way, the neutral atoms evaporated from the ionizer can be directed toward a cold surface. The source was designed especially to produce a rectangular beam with parallel ion trajectories and with uniform current density. The Tl+ surface ionization source, described in ref. [189], utilizes an electron bombarded Ir ionizer to ionize Tl vapor, which impinges on the surface of the ionizer from an independently heated reservoir. Raiko et al. have used surface ionization for the production of very high intensity beams of potassium and rubidium in an electromagnetic isotope separator with a nickel ionizing surface that has a high work function (5.03 eV) [179]. In the source, the alkali vapor from an oven is passed between a heated nickel surface and the front face of the ionization chamber and extracted through an aperture. A capillary-type cesium ion source has been developed, which is especially well suited for microfocused beam applications such as micro-SIMS [190]. The source, shown in fig. 45, utilizes a heated tantalum tube ( N 1 lOPC> through which cesium vapor is transported from an independently heated reservoir. Ce-

sium vapor emitted from the hot tantalum capillary (0.1 mm) is surface ionized and the resulting Cs+ ions extracted by impression of a high-electric field between the capillary tip and electrode at ground potential. The brightnesses of the source is estimated to be 80 A/cm’sr at a beam energy of 4 kV. This correlates to an ion emission density from the orifice of N 380 mA/cm2. The source produces Cs+ ion beams of 30 p,A within a solid angle of - 0.5 sr for periods of time exceeding six weeks. Metallic ions may also be produced by heating metal oxides or halides to very high temperatures in a highwork-function crucible. A source of this type has been developed which functions by heating a tungsten crucible containing one of the halides of the rare earth metals of _ 2800°C [191]. Thermal dissociation of the halide releases elemental rare earth atoms that make many collisions with the crucible walls before being emitted. The source has been used to generate a wide variety of rare earth species, with material utilization efficiencies greater than 70% recorded for europium and samarium. Because of the high thermal temperatures involved, perhaps both surface and thermal ionization processes may be involved in the production of ions from this source. 2.8.2. Types of aluminosilicate ion sources The sources described earlier are of the conventional type in which surface ionization is effected by a hot metal surface. An alternative method for producing high currents of alkali metal ions is by heating certain chemical compounds. This is one of the oldest methods of producing such ion beams [192-1981. For lithium ions, /3-eucryptite (lithium aluminum silicate) can be used. This is a naturally occurring compound, but for ion source requirements, it is normally synthesized by heating alumina and silica with lithium carbonate at around 145O”C, Li,CO,

+ 2Si0,

+ Al,O,

4 Li,O, co,.

Al,O,,

2Si0,

+

(52)

The compound can be melted onto a high-work-function material such as platinum, which is then used as

Fig. 4.5. Schematic representation of the capillary-type cesium surface ionization source of ref. [190].

G.D. Alton / Ion sources for materials research

257

Fig. 46. Alumina-silicate ion source for the generation of Li+, NaC, K+, Rb+, and Cs+ ion beams [200].

the ion-emitting surface. P-eucryptite Li+ currents of several milliamperes have been achieved using a low extraction potential (9 kW 11971. 120

-I

l-

110 ExtractIon

Voltage:

18 kV

100

60

80

h 9 + 3 IL)

70 60

60

Other alkali metals can be ionized in a similar way using analogous and other aluminosilicate compounds. A source has been developed which produces 100~uA cesium beams for up to 50 hours from a cesium oxidealumina-ferric oxide mixture [ 1951. Crystalline aluminosilicate compounds containing group IA elements (Li, Na, K, Rb and Cs) will emit singly-charged ions of these elements when heated to _ 1000°C (see e.g., refs. [198,199]). Sources based on this principle can be very simple as evidenced by the Li+ source shown in fig. 46 [200]. The ion beam intensity versus heater current is shown in fig. 47. For low beam intensities (low operational temperatures), sources of this type, formed from thin layers of the aluminosilicate material coated on the emission surface, can run 100 to 200 hours before depletion of the particular group IA element. The lifetimes of Cs+ sources have been recently extended to greater than 2000 h by forming the aluminosilicate in a pellet and then drifting the cesium ions forward to the emission surface by placing a bias voltage across the pellet [201]. This technique can be applied as a means of increasing the lifetime of aluminosilicate pellets containing other members of the group IA elements, as well. A schematic of the Cs+ ion source based on this principle is shown in fig. 48.

o18

20

22

24

26

3. Negative ion sources Ionizer Heater Current (A)

Fig. 47. Li+ ion beam intensity versus ionizer heater current [200].

The existence of the negative ion state has provided us with an additional means for producing singly

G.D. Alton / Ion sources for materials research

258 GLASS

ELECTRODES

L

PELLET

Fig. 48. Schematic illustration of a long-lifetime alumino-silicate cesium ion source with provisions for drifting the Cs+ to the extraction surface of the source [201].

charged negative ion beams containing most of the naturally occurring chemically active elements. Negative ion beams have been used for many years in high-energy, accelerator-based atomic, nuclear, and applied research, as well as low-energy fundamental atomic physics and controlled thermonuclear research programs. Negative ions are also being utilized in a variety of applied research applications, including tandem accelerator-based high-energy ion implantation, radiation damage, high-energy sputtering, and in surface analysis based on Rutherford backscattering spectrometry (RBS), secondary ion mass spectrometry (SIMS), tandem accelerator mass spectrometry (TAMS), X-ray fluorescence, and hydrogen depth profiling. Negative ion beams of H- also play an important role in Lamb shift, optically pumped, and atomic beam ion sources for the production of intense, nuclear-spin-polarized proton beams for use in high-energy spin physics experiments. These and other negative ion applications have provided the impetus for the development of sources capable of producing negative ion beams containing almost every element in the periodic chart. In this portion of the review, brief descriptions will be given of the negative ion formation processes of radiative capture, dielectronic attachment, polar dissociation, dissociative attachment, charge transfer, and thermodynamic and nonthermodynamic equilibrium surface ionization. Particular emphasis will be placed on sources for generating negative ion beams by the more universal and high probability processes of dissociative attachment, charge transfer, and secondary ion formation through non-thermodynamic surface ionization. In addition, the electron affinities of the elements are reviewed and a limited amount of electron affinity data are provided. For more detailed information concerning mechanisms of negative ion formation, the

reader is referred to refs. [202-2041 and to other references cited within the present report. Negative ions are fragile entities because of their low electronic binding energies (electron affinities) and thus can be readily collisionally detached to form neutral beams. Because of the high equilibrium neutral beam fractions that can be formed in collisional detachment processes compared to those formed in positive ion collisional attachment processes, they have been used as sources of neutral beams for injection into plasma fusion devices. Therefore, considerable efforts have been put forth toward the development of high intensity negative H- ion sources for this purpose. Negative ions are valuable, as well, in a fast-developing mass analysis technique that employs the tandem accelerator (tandem accelerator mass spectrometry). The fact that several elements do not form stably bound ground state negative ions permits the elimination of possible isobar contaminants in the mass spectra through the use of this species discrimination technique. A classic example of this technique and its experimental utilization is exemplified by the elimination of 14N isobaric contamination in 14C sample dating tandem mass spectrometry experiments. The use of negative ions in this application, coupled with the mass per unit charge (M/q) and energy E selectivity of the tandem accelerator permits analyses of geological and biological samples to degrees of precision unattainable with other techniques. Negative ions have also been used extensively in basic research activities designed to determine the negative ion atomic structures and their interaction properties with photons, electrons, positrons, positive and negative ions, and with neutral atoms and molecules. These and other negative ion applications have provided the impetus for the development of sources capable of producing negative ion beams containing almost every element in the periodic chart with the exception of many of the noble gas elements. Several versions of negative ion sources that utilize processes such as polar dissociation and dissociative electron attachment, charge exchange, and surface ionization have been described in the literature. The species capability of a particular source may be quite limited or rather universal depending on the mechanism employed. The negative ion intensity capability of a particular source also depends strongly on the mechanism of generation utilized as well as the electron affinity of the species involved. Reviews exclusively devoted to the subject of negative ion sources have not been previously given. However, brief inclusions of the subject have been contained in general ion source review articles such as those of refs. [2,3]. The principal mechanisms of negative ion formation are discussed in refs. [202-2041.

G.D. Alton / Ion sources for materials research

259

ity EA is defined as the difference between the ground state neutral E, and negative Ei energies or

3.1. The negative ion state 3.1.1. The electron affinity The processes involved in the attachment of electrons to neutral atoms and molecules are exothermic in contrast to the endothermic processes required for positive ion formation. The binding energy or electron affinity EA of the negative ion is a measure of the stability and ease of ion formation. The electron affin-

E,=E,-Ei

(53)

and must be positive for stable or metastable ion formation. The probability of negative ion formation depends significantly on the value of EA as well as on the method used in the formation process. Approximately

Table 6 Ionization potentials and electron affinities of the elements IONIZATION IA

ELECTRON

1H 13.595 0.7542 3 Li 5.39 0.620

POTENTIAL VIII A

AFFINITY I

I

III A

IVA

VA

4 Be 9.32 0.19

58 8.30 0.28

6C 11.26 1.268

7N 14.54
80 13.61 1.462

9F 17.42 3.399

10 Ne 21.56
13 Al 5.98 0.46

14 Si 8.15 1.385

15P 10.55 0.743

16s 10.36 2.0772

17 Cl 13.01 3.615

18Ar 15.76
31 Ga 6.00 0.3

32 Ge 7.88 1.2

33 As 9.81 0.80

34 Se 9.75 2.0206

35 6r 11.84 3.364

36 Kr

49 In 5.76 0.3

50 Sn 7.34 1.25

51 Sb 8.64 1.05

52 Te 9.01 1.9706

53 I 10.45 3.061

54 Xe 12.13
81 TI 6.11 0.3

82 Pb 7.41 0.36

838i 7.29 1.1

84 PO 8.43 1.9

85 At 9.5 2.8

86 Rn 10.74
l

11 Na 5.14 0.548

12Mg 7.64
19 K 4.34 0.5012

20 Ca 6.11

+

0.043

37 Rb 4.18 0.4860

38 Sr

55 cs 3.89 0.4715

56 Ba

5.69 >o

5.21 >o

VII A

0.078

IIA

VI A

l

14.00
VIII 6

*

IV 6

21 SC 6.56
22 Ti 6.83 0.2

25 Mn 7.43
26 Fe 7.90 0.25

27 Co 7.86 0.7

28 Ni 7.63 1.15

29 cu 7.72 1.226

30 Zn 9.39
39 Y 6.5 =o

40 Zr 6.95 0.5

43 Tc 7.28 0.7

44 flu 7.36 1.1

45 Rh 7.46 1.2

46 Pd 8.33 0.6

47 Ag 7.57 1.303

48 Cd a.99 -co

57 La 5.61 0.5

72 Hf 7.
75 Re 7.87 0.15

76 OS 8.7 1.1

77 Ir 9. 1.6

78 Pt 8.98 2.128

79 Au 9.22 2.3086

80 Hg 10.43
lMETASTABLE

V.6

VI B

I6

Ill 6

\

VII 6

1

II 6

260

G.D.Alton / Ion sourcesfor materialsresearch

78% of the naturally

occurring elements have positive electron affinities. Table 6 displays experimentally determined atomic electron affinities for many of the naturally occurring elements - many of which were taken from the compilations of ref. [205]. The electron affinities of the elements have values ranging from E, I 0 to EA = 3.618 eV for Cl. We note that groups IIA, IIB, and VIIIA, as well as the elements N, Mn, SC, Y, and Hf, all have negative electron affinities and thus do not form stably bound ground state negative ions. The group IIA elements span the complete spectrum of possible electron affinities. Be has a negative ground state electron affinity, but is bound metastably in the ls’2~2p*(~P) state; the ion is metastable against direct Coulomb autodetachment and electric dipole radiative decay processes, has a lifetime of a few ks, and thus lives long enough for practical use [206]. Mg, on the other hand, forms neither a bound ground state nor metastably bound state and thus cannot be produced as an atomic negative ion. On the other hand, the heavier members of the group (Ca, Sr and Ba) all form or are believed to form stably bound ground state negative ions. For example, Ca- is bound in the [Ar]s2p(*P) electronic configuration by 43 meV [207]. While the electron affinities of Sr and Ba have not been measured to date, they are expected to be stably bound in the [Kr]s’p(‘P) and [Xe]s2p(2P) states, respectively, similar to Ca- [208]. In addition to negative atomic species, many molecular negative ions have been observed. In many cases, molecular negative ions containing the atom of interest have much higher electron affinities than the atom itself and, therefore, can be formed with higher probability than can the atomic species. In some cases, molecular ions offer the only alternative for producing beams containing elements which do not form stably bound negative ion states. For tandem accelerator applications, the unwanted species can be easily rejected by collisional dissociation in the positive ion formation process (stripping) followed by magnetic (M/q) analysis. Ref. [203] provides a partial list of electron affinities of molecular species containing elements which have negative or low positive electron affinities.

angle is small and the product ion is scattered nearly perpendicular to the impact momentum vector. From the standpoint of beam quality degradation, lowmomentum transfer interactions are clearly desirable. Such transfer processes can be cast into two distinct categories: (1) symmetrical (resonance) processes, and (2) asymmetrical or nonresonance processes. In the first category, the projectile and target are the same species, while in the latter category, which is generally of more practical importance as a means of producing negative ion beams, the projectile and target are different.

3.2. Negative ion formation through charge exchange

Obviously, the energy defect can be minimized by selection of a charge transfer medium which has a low first ionization energy E,(Y). Cesium is an excellent choice because it has the lowest ionization energy of the elements. However, cesium is heavy and will scatter the projectile to larger angle than, e.g., lithium.

The single-charge transfer process is often referred to as “charge exchange” or “electron capture.” The charge transfer process can be represented by the following projectile X-target Y interaction, where A E is referred to as the energy defect: X++Y+X+Y++AE.

(54) In general, the collision takes place at relatively large impact parameters, for which the projectile scattering

3.2.1. Symmetrical (resonance) charge transfer The symmetrical charge transfer process in which the projectile and target are identical can be described by the following reaction: X++X+X+X++AE.

(55) The energy defect is equal to the difference between the ionization energies, i.e., AE = E,(X) -E,(X) and is thus zero. Hence, the cross sections can be very large. Negative ion formation requires the following sequential projectile-target reactions: X++X+X+X++AE

(AE=O),

X+X+X-+X++AE

(A E= am

-E,(X)).

(56) As noted, the energy defect for the second reaction is equal to the difference between the electron affinity of the first projectile atom EA and the ionization energy Ei of the stationary target atom. Thus, symmetrical processes are not necessarily commensurate with efficient negative ion formation because of the possibility of a high ionization energy Ei for the donor species. However, consideration may be given to a two-step process in which the first process is symmetrical, for which the cross section is large for neutral energetic atom formation, followed by an asymmetrical nonresonance process in which the energy defect is much smaller, e.g., X++X+X+X++AE, where AE = 0, X+Y+X-+Y++AE, where AE = EA(X) -E,(Y).

(57)

3.2.2. Asymmetrical (nonresonant) charge transfer The most general application of the charge exchange formation technique is for interactions between unlike ions. These processes occur with high probabil-

G.D. Alton / Ion sources for materials research

ity between projectiles interacting with exchange vapors which possess a low energy defect. The probability for charge transfer from the electron donor (exchange vapor) and the projectile ion is sensitively dependent on the velocity of the projectile. The atomic charge transfer process occurs between unlike ions and differs from the symmetric resonance process in that it involves an electronic transition which requires a change in internal energy of the system referred to as the energy defect A E. The energy defect AE of the process X++Y+X+Y++AE

(58) can be equated to the difference between the ionization energies Ei of the interacting atoms or molecules as

261

The criterion has been applied by Hasted to many charge transfer cross-section reactions to estimate the at which the cross sections projectile energy, Em, have maximum values [210]. The assumption is made that the criterion holds as an equality at this energy. The energy, EM, of a projectile of mass M, interacting with a stationary target then is given by E

(64)

The parameter a, obtained from many different reaction processesd was found to have a value of approximately a = 7 A. The “adiabatic maximum rule” is thus a convenient method of estimating the energy of a cross-section function maximum, apparently valid over a large range of impact energies.

AE=E(X+)-E(X)+E(Y)-E(Y+) = E,(X) - E,(Y).

(59)

For negative ion formation, an additional charge transfer collision must take place, e.g., X+Y=X-+Y++AE,

(60) where A E = EA(X) - E,(Y). Two-electron capture during a single collision is much less likely due to the very high energy defect as seen from the following reactions: X++ Y + X-+ Y*++ AE

(61)

where A E = E,(X) + E,(X; - Ei(Y) - E,(Y ‘). 3.2.3. The Massey adiabatic criterion The adiabatic criterion proposed by Massey [209] is of practical importance in charge transfer collisions. At low projectile energies, where the relative motion of the atoms is slow enough so that electronic motion can adiabatically adjust to small changes in the internuclear distance, the electron transfer process becomes unlikely. However, if the impact energy falls outside this “adiabatic region” and the electronic transition time is comparable to the collision time, the probability for electron transfer can be very high. The time of collision is taken as a/v, where v is the impact velocity and a is known as the “adiabatic parameter”, the parameter a is of the same order as atomic dimensions within which the charge transfer transition becomes likely. The characteristic time for the electronic transition is given by h/AE, where AE is the energy defect. Thus, the condition v <
(62) the adiabatic velocity region and

(63) characterizes the velocity region for which the maximum in the charge transfer process occurs.

3.2.4. Charge exchange production efficiencies The electron capture and loss processes which take place during collisions between energetic ions and atoms or molecules in a gaseous vapor result in a distribution of charge states of the emergent beam. Equilibrium between the competing capture and loss processes takes place whenever the beam passes through a target of sufficient thickness so that a balance between production and loss occurs. A particular emergent charge state is in equilibrium whenever the fraction no longer depends on further increases in target density. In order to efficiently generate negative ions by the charge transfer technique, knowledge of the dependence of the negative ion fraction F? on cell temperature and ion energy is essential. For the charge-exchange process, negative ion production efficiencies depend primarily on ion energy, the electron affinity of the element under consideration, as well as the electron binding energy of the donor material. Experimental schemes for measuring such processes have been described by several experimental groups, including the devices described in refs. [211,212]. Equilibrium fractions F” for H in several group IA, and group IIA metal vapors are shown in fig. 49 [213]; analogous equilibrium fractions for D in various group IA have also been measured in these vapors by this group. A review of charge transfer processes related to polarized ion sources is given in ref. [214]. Many other investigations have been made of the probabilities and energy dependencies of charge transfer negative ion formation, including the following: H- in Na [215] and Cs vapors [216] and He- in Li, Na and Mg [217], K [218], Rb [219] and Cs [220] vapors. Charge transfer, using both sodium and magnesium vapor, has also been utilized for the efficient production of many negative heavy ions [221,222]. These investigations show production efficiencies ranging from N 0.5 to > 90% for the elements considered. Equilib-

G.D. Alton / Ion sources for materials research

262

I

I

I

I

400 -

Mg TARGET

=

50 -

20

-

to-

2 8 2

-0.1

0.5

0.2

1.0 2 E (keV)

5

10

5-

20

Fig. 49. H negative ion charge exchange equilibrium fraction FE as a function of H+ energy E in group IA and group IIA vapors [213].

rium fractions as a function of ion energy for several ions in the exchange vapors of Na and Mg are illustrated in figs. 50 and 51. These studies show that Mg is a more effective electron donor for high-electron-affinity elements, while Na vapor is desirable for the production of beams from low-affinity materials. Further evidence of the efficiency and universal character of

2-

Be(O.24)

(-

0

20

40

60 E

60

IkeV)

Fig. 51. Negative ion equilibrium charge exchange versus ion energy for several ions in magnesium vapor [221].

fractions exchange

the charge transfer process is given in figs. 52-55 for groups IA, IIIA, IVA, and VA projectiles in cesium vapor [223].

mo

I

I

I

I

I

I

I

I=

No TARGET 50

l-•-.r

1 -

20 10

‘ii a9 T I&

5 Be( 0.24)

2

Li (0.62) -*-.L.l-

-

4

= i-

,/’ .

7

0.5

0.2 0 0

Fig. 50. Negative versus ion energy

20

40 E (keV)

60

60

ion equilibrium charge exchange fractions for several ions in sodium exchange vapor [2221.

E (keV)

Fig. 52. Negative ion equilibrium charge exchange fraction FY versus projectile energy E for various group IA projectiles in Cs vapor [223].

G.D. Alton / Ion sourcesfor materialsresearch

102

L

,

I

1

GROUP

_

Cs TARGET

I

IU

I

I

I

I

402

5

‘A

263 I

I

I

I

I

I

I

Y

PROJECTILES

: b,o\b,5Aj .3’P 10’ 2

2

= 1

GROUP P

1

Cs TARGET

PROJECTILES

8

8

7 II.

h (0°

do-’

100

r

1

0

I

I

I

I

I

I

I

10

20

30

40

50

60

70

lo-’ 00

E (keV)

Fig. 53. Negative ion equilibrium charge exchange fraction FE’ versus projectile energy E for various group IIIA projec-

There is an important class of negative ions which can be formed in excited states of the parent atom and live long enough to be of use experimentally. Such long-lived states are categorized as being metastable.

402

b

I

I

I

I

I

I

I

El

8 GROUP

Ip

400

I

I

I

I

I

I

I

10

20

30

40

50

60

70

00

Fig. 55. Negative ion equilibrium charge exchange fraction IT versus projectile energy E for various group VA projectiles in Cs vapor [223].

In order to induce electron attachment and thus form such states, it is necessary, first of all, to form a particular excited electronic state of the neutral atom to which an electron can be attached. The charge exchange process is particularly well suited for metastable ion formation. The subsequently formed negative ion may live for extended periods of time, if decay of the excited compound state is forbidden. Classic examples of metastable negative ions which can only be formed through attachment to excited states of the parent atoms are He- [224,225] and Be[226]. He- and Be- form with high probabilities in the 4P states through sequential charge exchange with a low-ionization-potential charge-exchange vapor, M, such as a member of the group IA elements through the following spin conserving interactions: He’(*S)

PROJECTILES

Cs TARGET

0

E (keV)

tiles in Cs vapor [223].

3.2.5. Metastable negative ion formation

-

+ M(*S) -+ He(3S) + M+(S),

He(3S) + M(*S) + He-(4P)

+ M+(‘S),

(65)

Be+(*S) + M(*S) + Be(3P) + M+(S), Be(3P) + M(*S) + Be-(4P)

46

1

0

I

I

I

t0

20

30

I

I

I

I

40

50

60

70

00

E (k&l

Fig. 54. Negative ion equilibrium charge exchange fraction versus projectile energy E for various group IVA projectiles in Cs vapor [223].

i?

+ M+(S).

(66)

The attachment energy for He- relative to the He (3S) state is 78 meV. He- is thus metastable against direct Coulomb autodetachment and electric dipole radiative decay processes [224,225]. The lifetime of the negative ion depends on the particular magnetic substate; the P,,2,3,2 states have lifetimes of N 10 and 16 us, respectively, and the Ps,* substate has a much longer

G.D. Alton / Ion sources for materials research

264

ON ELECTRODE ESSION

ment of He; into the He,(X’Z$:) + e repulsive continuum. The electron affinity for the 4He,(a3Z:) state is found to be 0.175 eV, while that for the analogous 3He,(a3Z:) state is 0.178 eV.

ELECTRODE

HEATING

ELEMENT

3.2.6. Types of charge exchange sources Several exchange-type negative ion sources have been described in the literature. However, in more recent years, more universal positive ion sources (e.g., the sources described in refs. [49,53-551) have been utilized which greatly increase the range of negative ion species that can be generated by this technique. The source consists of a positive ion source (usually a duoplasmatron or radiofrequency source) and a charge-exchange canal where the exchange interactions take place (see e.g., refs. [229-2351). The exchange canal is usually a tube - 0.75 cm in diameter and - 10 cm in length to which is attached a gas line or an oven for introducing the gaseous or solid exchange material. A scheme utilized for negative ion production by charge exchange using a duoplasmatron positive ion source is illustrated in fig. 56. In the case of the use of solid materials, the temperature of the oven and canal can be controlled rather precisely. The canal region can be biased negative with respect to the source at ground potential or the source can be biased positive with respect to the canal at ground potential (bias potentials of 10 to 40 keV are usually employed). The former configuration permits extraction of negative ions which are generated by the positive ion beam from the exchange vapor itself [229]. In this particular mode of operation, the formation processes are not through exchange but polar dissociation, dissociative attachment, and ternary collisional processes. An einzel lens is usually placed between the positive ion source and

Co-CONTAINER ELECTROOE

T

2

F

Fig. 56.

Schematic diagram of a duoplasmatron charge exchange negative ion source utilizing calcium as the exchange vapor.

lifetime (- 500 JLS)[227]. Be- is also formed in the 4P state which lives for a few JLSbefore autodetaching. The He; molecular ion is a very interesting example of a spin-aligned, complex negative ion state. Two distinct autodetachment channels are observed in the electron energy spectra of He; formed by sequential charge transfer of energetic He: in lithium vapor [228]. A single narrow peak is observed which is attributed to vibrational autodetachment, e.g., He;(411,)DZ,

+ He~(a3ZJ),=,

+ e.

(67)

The second peak is broad and results from autodetach-

HIGH VOLTAGE FEE0 THROUGH

NEGATIVE BEAM OUT

TO OIFFUS!ON.PUMP

Fig. 57. Schematic

drawing

of a universal

charge-exchange

source [231].

G.D. Alton / Ion sources for materials research

the canal for focusing the positive ion beam through the small-diameter canal. Considerable efforts have been expended toward development of high-intensity negative ion sources for the production of H- and He- for use in neutral particle injectors for fusion energy research. The charge-exchange technique has been utilized by Hooper et al. [230] in the production of > 70 mA of Hethrough exchange with Na vapor. The positive ion beam was generated in a large multiple aperture source which was operated in pulsed mode. Gaseous-feed-type positive sources, usually utilized (duoplasmatrons or radiofrequency-type sources), restrict the number of species that can be generated in this type of negative ion source. Incorporation of a universal type positive ion source permits generation of negative ions from any element which has a positive electron affinity as well as a number of metastable ions. This universal source concept is illustrated schematically in fig. 57 [231]. This source employs a hollow-cathode-type positive ion source and a recirculating-type lithium charge exchange canal developed for use in high-intensity He- ion sources [232]. Among the ion beams that such sources have been used to generate are the following: 10 uA H-; 12 JLA He-; 1.5 FA Li-, 2 p,A C-; 15 JLA O-; 25 u.A F-; 0.4 FA Na-; 15 p,A Cl-; 1 FA S-; 0.2 uA K-; 10 pA II; and 10 FA Br-. This list accounts for only a small fraction of the beams which can be generated by this technique. Useful negative ion beams can be formed by charge exchange from almost every element which has a positive electron affinity. 3.3. Principal negative ion formation processes inoolving electron impact

265

whenever third-body energy transfer processes are not involved. The process, however, is a low probability process. 3.3.2. Dielectronic attachment Another attachment mechanism is possible involving radiative stabilization processes. Attachment is possible if the incident electron has energy such that the energy of the atom plus electron is within the level width of a doubly excited state of the atom. Thus, it is possible to capture the electron in certain allowed states without emission of radiation. Once formed, the ion may revert to the ground state by radiative emission or revert to the neutral by ejecting the electron back into the continuum. Dielectronic attachment is a low probability process and, therefore, not important in high-intensity negative ion sources. The process for capture can be represented by the following interaction: e+X+X-+hv.

(70)

3.3.3. Polar dissociative attachment Attachment may occur as a result of photon or electron, as well as other heavy particle interactions with molecular neutrals in which sufficient energy is imparted to the molecule to excite the molecule to an unstable state that dissociates spontaneously into positive and negative ion fragments. Polar photodissociative attachment is a process whereby a molecule, XY, absorbs a photon of sufficient energy hu to cause spontaneous fragmentation according to the reaction XY+hv+X-+Y+.

(71)

Negative ions may be formed by means of several physical or physiochemical processes, including radiative capture, dielectronic, polar dissociation, dissociative attachment, charge exchange, and surface ionization.

Negative ion formation may also occur by this mechanism, whenever an electron of sufficient energy for molecular excitation to an unstable state interacts with a molecule XY, producing the following reaction:

3.3.1. Radiative capture The simplest way in which negative ions can be formed is by direct capture of a free electron by a neutral atom according to the following interaction: (68) For such processes to take place, the initial kinetic energy of the electron T, plus the electron affinity EA must be released in the process through radiation or transfer to a third body in order to effect stable attachment. Such mechanisms are referred to as radiative capture processes and are characterized by a continuum emission of radiation with longest wavelength given by

We note that the electron is not itself captured, but only serves as the means by which the molecular excitation occurs. In polar dissociation processes, the Franck-Condon principle [236,237] can be invoked to estimate the relative kinetic energy of the X+ and Y- ions after the collision. If the minimum energy transfer necessary for molecular fragmentation to occur is given by Emin, and the dissociation energy, D,,, is known along with the ionization energies of the positive and negative ion fragments, E, and EA, respectively, the minimum relative kinetic energy T of the fragments can be calculated from the following expression:

h = he/E,,

T=E,in+EA-Ei-D,,.

e+X+X-+hv.

(69)

e+XY-+X++Y-+e.

(72)

(73)

266

G.D. Alton / Ion sourcesfor materialsresearch

The process can be visualized by referring to the potential energy curves as described in ref. [202]. Since low ion energy spreads are desirable for ion source applications, a knowledge of the respective potential energy curves is useful in predicting energy spreads in ion sources that utilize this principle. 3.3.4. Dissociative attachment Electrons may be stably attached to atoms during their interactions with molecular neutrals according to the following reactions: e+XY-+X-+Y e+XY-+X+Y-.

(74)

The process may be viewed as a three-body process where the excess energy released during the reaction can be adsorbed by transfer to the relative motion of the atomic nuclei or fragments, and thus the state can be readily stabilized. Such processes are characterized by curve crossing of the respective molecular-neutral and negative-neutral potential energy curves as given by Massey, for example [202]. 3.3.5. Types of electron-impact attachment sources Sources that employ electron-impact ionization methods from which negative ions can be extracted are often referred to as direct extraction sources. Almost all sources of this type produce ions by means of electron attachment in a plasma. Many sources have been described in the literature among which the following serve as examples: 3.3.5.1. Plasmatron sources Moak et al. [238] first discovered that useful intensities of negative ions could be directly extracted from a duoplasmatron operated with reversed extraction polarity. Later it was discovered that the negative ions were more abundant in the periphery of the plasma and the yield of negative ions could be increased by offsetting the extraction electrode with respect to the plasma center. Because of the high on-axis plasma densities present in conventional hot cathode duoplasmatron sources, negative ions generated in the source are found in the peripheral regions of the discharge. The hollow cathode plasma densities are lower on axis and this type of source is commensurate with on-axis negative ion extraction. Hollow-cathode duoplasmatron sources have been developed specifically for the generation of negative ion beams. Beam intensities of 10 mA (peak) H- using H, feed gas and 60 mA (peak) using a mixture of H, and Cs have been reported with a hollow-cathode source operated in the pulsed-beam mode [239]. A schematic representation of the hollow-cathode duoplasmatron is shown in fig. 58. The hollow cathode thus eliminates the necessity of having to offset the anode aperture relative to the axis of the plasma column within the

SmCo MAGNETS r

Ni HOLLOW CATHODE ANODE

/

EXTRACTION ELECTRODE

Fig. 58. Schematic drawing of a simple, low-piker, hollowcathode, direct-extract, duoplasmatron, negative ion source.

structure as required in more conventional duoplasmatrons [240]. Other direct-extraction versions of the duoplasmatron that are being used to produce negative ion beams from gases are those of the duodecatron [241] and triplasmatron [242]. The duodecatron version has produced currents of over 100 uA H-; and 10 uA or greater of O-, F-, and Cl-. Currents of O- in excess of 50 mA have been extracted from the triplasmatron ion source. 3.3.5.2. Penning discharge sources Although the Penning discharge source has traditionally been used to produce positive ion beams, the source can De used to efficiently produce H- ion beams. A hot cathode, radial-extraction-geometry version of the PIG source was first developed by Ehlers for generation of steadystate ion beams with current densities of 40 mA/cm’ for cyclotron applications [243]. The discovery that the addition of cesium to the discharge can greatly enhance the negative ion beam intensities has led to the development of high-intensity H- sources (2 3 A/cm*) [244,245]. A pulsed-mode, radial-geometry Penning source equipped with LaB, cathodes has demonstrated H- beam densities of 2 350 mA/cm2 [246]. Many other direct extraction sources have been developed for production of negative ion beams. The diode source of Bastide et al. [247] and the cold cathode Penning source of Heinicke et al. [248] are noteworthy examples. The diode source has been used to generate several negative ion species. Among those reported are 600 uA H-, 20 uA BO-, 0.5 P/IA C, 10 uA CN-; 4 p.A O-, 50 PA F-, 4 )IA P-, and 4 ALAS. The radial extraction geometry Penning source of Heinicke et al. [248] can be used to generate a wide variety of negative ion species. The material to be ionized may be introduced into the discharge as a gas, directly vaporized from an oven, or sublimed from a solid rod of the material. A variety of negative ion species have been extracted from the source including 1.2 ~.LALi-, 60 ~.LAH-, 0.2 )LA BeH-, 1.0 FA MgH-, 100 PA F-, 10 uA BP, 50 )*A S, and 50 p,A Cl-.

G.D. ALton / Ion sourcesfor materialsresearch

Fig. 59. Schematic illustration of the surface plasma source described in ref. 12521.

Another direct extraction radial geometry Penning ion source has been described by Smith and Richards [249]. The source utilizes sputtering from the cathodes to generate neutral vapor of the desired species while introducing cesium vapor into the discharge chamber. Beam intensities of 2.7 PA C-, 6.5 p.A Cu-, and 4 WA Ni- have been reported. Discovery of the enhanced HP production on cesium covered surfaces has led to the modification of original Penning sources [244,245]. These sources are frequently referred to as surface plasma sources (SPS). This source type has been previously reviewed [250,251]. Fig. 59 illustrates schematically the SPS described in ref. [252]. The gap between the two cathodes is decreased to less than 1 cm to reduce the loss of negative ions while moving through the plasma. Negative ions are produced on molybdenum cathodes by sputtering and backscattering and accelerated back into the plasma by the cathode voltage fall. Dimov et al. [245]. have obtained up to 150 mA of H- ions from their source with a density in the slit of 3 A/cm’, in pulses of 0.2 ms duration and 2% duty factor. A number of other sources based on this principle have been developed at other laboratories, including the sources described in refs. [253,254]. The emittances of these sources are usually asymmetric due to the slit geometries utilized. The emittances have been measured by Allison [253] and Alessi et al. [254]. The normalized emittances of the source described in ref. [253] in the respective planes are 3.1~ mmmrad (MeV)‘/’ and 2.3~ mm mrad (MeV)‘/2 for a 36-mA H- beam. 3.3.5.3. Multicusp, magnetic-field, volume-type H -/D sources The generation of H-and D- negative ions in volume-type, multicusp field sources has been reviewed previously by Bacal [2.55,256]. H- and D- sources of this type operate

with pure hydrogen

or deuterium

gas

267

or mixtures of the two. Negative ions are formed by dissociative-attachment processes which take place within the plasma volume in a two-step process. The first process involves the vibrational excitation of the molecule by energetic electrons, followed by a second collision whereby a low-energy electron is attached. Separation of high- and low-energy electrons is usually effected by use of a magnetic filter field arrangement [257]. The energetic electrons which are formed in the primary discharge region of the source are deflected by filter field magnets and thus do not enter into the extraction region of the source which contains mainly low-energy electrons. Low-energy molecular collisions in this region of the source result in negative ion formation. A number of volume-type H-/Dsources have been described in the literature, including those of refs. [258-2651. Small H- and D- volume-type sources equipped with multi-cusp plasma containment magnetic fields have been developed by Leung et al. [260,261] and McAdams et al. [263]. An optimized version of the source described by Leung et al. [261], which is equipped with a permanent magnetic filter field and either LaB, or W filaments, is shown in fig. 60. Highquality, pulsed-mode beams of H- and D- with current densities in excess of 2.50 mA/cm2 have been achieved when the source is operated with pure H, or D, gas. When equipped with a cesium oven [261], current densities in excess of 1 A/cm* can be realized. The effect of adding cesium to the discharge is clearly shown in fig. 61. An rf-driven version of the source with a glass-coated rf antenna can produce ms Hbeam pulses at a repetition rate of 150 Hz at current densities in excess of 200 mA/cm2 from pure H, and D, gas f262]. The multicusp-field source, developed by McAdams et al. utilizes a magnetic filter field and a conventional Ta filament [263]. The source is designed

EXTRACTION

WATER JACKET MAGNETIC

FILTER 0,2

cm

Fig. 60. Schematic drawing of the multi-cusp, magnetic-field, volume-type source optimized for H-/Dgeneration [261].

268

G.D. Alton / Ion sourcesfor materialsresearch

for cw operation. H- current densities of 15 mA/cm* can be achieved in this mode. The normalized emittance of beams extracted from the source increases with extraction current density. For a current density of 7 mA/cm’, and source emission aperture of 8 mm, the normalized emittance approaches 22~ mmmrad (MeV)“‘.

3.4. Thermodynamic negative surface ionization Atoms or molecules impinging on a hot metal surface may be emitted as positive or negative ions in subsequent evaporation processes after mean residence times 7, and ri, respectively. The process of direct surface ionization is statistical in nature, and therefore statistical and thermodynamic arguments can be applied to such systems to derive equations for the degrees of positive or negative ion generation. The subject has been reviewed in a rather comprehensive manner by Kaminsky, which contains many older review references on the subject [266]. The probability for arrival at a position far from the metal in a given state depends on the magnitude of the difference between the surface work function 4 and the electron affinity EA of the atom or molecule or 4 -EA. For thermodynamic equilibrium processes, the ratio of ions to neutrals which leave an ideal surface can be predicted from Langmuir-Saha surface ionization theory. The form of the Langmuir-Saha equation for the probability of negative ion formation for neutral particles of electron affinity EA interacting with a hot metal surface at temperature T and constant work function &Jis given by

X

EA-4

exp

(

kT

-I

)I

’

(75)

where r_, rO are the reflection coefficients of the particle at the surface and w_, o,, are statistical weighting factors for the negative ion and neutral, respectively. o_, wa are related to the total spin of the respective species given by w=2&+1,

(76)

where si is the spin of the electron. If the process takes place in the presence of an electrostatic field, E, then the possibility of negative ion formation is enhanced by means of the Schottky effect, provided the electric field is in a direction so as to remove the negative ions as

DISCHARGE CURRENT (A)

Fig. 61. Effect of adding cesium on the performance of the multi-cusp, magnetic-field H-/Dvolume-type source described in ref. [261].

they are formed. following form:

For this case, eq. (75) takes the

where E,, is the permittivity of free space. Eqs. (75) and (77) are more complex in reality because of the variation of 4 with crystalline orientation in cases where the metal is polycrystalline or the surface has uniformly or nonuniformly distributed surface contaminants. All of these effects can be taken into account by appropriately summing over the admixtures of existing work functions and statistical weighting factors in the respective expressions. From the relationships, it is evident that negative ion yields could be enhanced by lowering the work function 4, extracting in a high electric field E, or increasing the surface temperature T for elements where EA -=z4. Practical limitations imposed by the requirement of operational stability usually limit electric field strengths to values N lo4 V/cm or a lowering of the work function by - 0.038 eV. The greatest effect can be obtained in practice by effecting changes in the work function 4 or increasing the surface temperature T. The former can be effected by surface adsorption of minute amounts of low-workfunction materials such as the group IA and IIA elements. Calculated probabilities for direct surface ionization of selected high-electron-affinity atoms and molecules from a hot LaB, ionizer by use of eq. (75) are shown in fig. 62.

G.D. Alton / Ion sourcesfor materialsresearch

269

3.5. Nonthennodynamic ization

LaBe Ionizing Surface (+ q 2.36 eV) at 1373°K 11 1.6

I 2.0

2.6

3.0

3.5

4.0

4.6

5.0

5.5

6.0

E,, Electron Affinity (eV)

Fig. 62. Calculated probabilities for direct surface ionization of selected high-electron-affinity atoms and molecules from a hot LaB, surface by use of eq. (75).

3.4.1. Types of negative surface ionization sources Negative surface ionization has not been utilized frequently as a means for practical production of ion beams - principally due to the lack of chemically stable low-work-function materials. A few examples, however, can be cited. Sources based on the use of LaB, ioniz-

ers have been described in the literature [267-2691. The source described in ref. [267] utilizes a porous graphite ionizer which is impregnated with LaB,, while the source described in ref. [269] utilizes a solid spherical-geometry LaB, cathode which can be biased negatively with respect to the extraction electrode; this source is shown schematically in fig. 63.

HIGH VOLTAGE

equilibrium negative surface ion-

The processes for which the Saha-Langmuir relation apply are those that take place under thermodynamic or quasi-thermodynamic equilibrium. Such conditions are achievable for neutral particles incident on a surface at thermal energies where the sticking probabilities are approximately unity and consequently the residence time r is long enough to approach thermodynamic equilibrium before evaporation occurs. The probability that an atom will be ejected as an ion during sputtering from a low-work-function surface has been the subject of many investigations during the past several years. While considerable progress has been made toward a better understanding of negative ion formation by this process, the precise mechanism continues to be debated [270-2741. The key question is the role of the ejection velocity in the formation process. The probability for ion formation in models based upon local thermodynamic equilibrium (LTE) in the region of primary ion impact depends weakly on the ion ejection velocity [273,274]. On the other hand, theoretical treatments involving atomic excitation, leading to electron capture or loss, predict a strong velocity dependence for the secondary ion emission probability [270-2721. Theories have been developed which treat either negative ion formation in terms of an electron tunneling mechanism [271] or in terms of a velocity-dependent surface ionization mechanism [272]. Bond-breaking mechanisms have been proposed to explain secondary ion formation during sputtering of oxidized metal surfaces and atomic ion emission from

COOLANT

OUTLET GASEOUS MATERIAL

COOLANT

FEED INLET

INLET

ROUND ELECTRODE

Fig. 63. Schematic drawing of the negative surface ionization source described in ref. [2691.

270

G.D. Alton / Ion sourcesfor materialsresearch

molecular solids [275-2771. The simplest form of the local thermodynamic equilibrium (LTE) approach to secondary ion emission is conceptually different from either the bond-breaking, the electron tunneling, or the velocity-dependent surface-ionization mechanism. The simple LTE models predict weak or no velocity dependence for the probability of ion emission [274], while the bond-breaking mechanism [275-2771 electron tunneling [271], and velocity-dependent surface-ionization mechanisms [272] predict secondary ion emission that has an exponential velocity dependence. Although several independent and distinct negative ion formation processes may coexist during sputtering, particularly from compound and alloyed samples, there is convincing evidence that the mechanism of formation during sputtering of “clean” metal surfaces is a form of surface ionization. In sources based on the sputter principle, positive ion beams, usually formed by either direct surface ionization of a group IA element or in a heavy noble gas (Ar, Kr or Xe) plasma discharge seeded with alkali metal vapor, are accelerated to energies between a few hundred eV and several keV where they sputter a sample containing the element of interest covered with fractional layers of a highly electropositive adsorbate. Highly electropositive adsorbates such as the group IA elements dispersed on the surface of a metal greatly enhance the probability for negative ion formation by lowering the work function. This phenomenon makes possible the generation of a wide spectrum of negative ion beams in sources based on the sputter principle.

3.5.1. Electropositive adsorbate-induced changes

work function

It is well known that atomic adsorption of a dissimilar element on a clean surface affects the surface work function. The magnitude and sign of the change depends on the chemical properties of the adsorbed atom (adsorbate) and those of the host material (adsorbent). Electropositive atoms decrease the work function while electronegative atoms tend to increase the work function. The importance of surface adsorbates on negative ion formation processes has been realized since the discovery by Krohn that negative ion yields can be greatly enhanced by sputtering a material in the presence of an alkali metal [278]. The first direct correlation between negative ion yield and surface work function was made by Yu [271]. Semiempirical relations have been proposed which relate the work function change Ac$, in units of volts, V, to the surface coverage (T [279-2811. A simple analytical expression has been derived by Alton [281] which can be used to predict, with good accuracy, the value of the work function $J over the complete range of adsorbate coverage (W = 0 to u = 1). The equation

which expresses the functional be written as follows:

dependence

of 4 can

(78) where c$~ is the intrinsic work function of the sample and A&, is the maximum change in surface work function induced by the adsorbate at optimum coverage om. 3.5.2. The Nerskov and Lundqvist model Norskov and Lundqvist [272] determined expressions for positive- and negative-ion emission probabilities P+(v I) and P-(v I) as a function of the perpendicular component of the ejection velocity of the particle with respect to the surface, v I. For purposes of the present paper, we are only interested in P-(v I). In the prescription of ref. [272], the probability for negative ion formation can be cast into a simple energy-dependent form given by 2

P-( E,, e) = -exp

?r

-P@&W

-E, $12E

cos

+ 61

e (79)

where C#Jis the work function of the surface (4 depends on the relative adsorbate coverage (T), EA is the electron affinity of the ejected particle of mass M2 and energy E,, Vi is the image potential induced in the surface by the escaping ion, 0 is the polar angle of the sputtered ion with respect to the surface normal and /? is a constant. In eq. (79), (2E,/M,)‘/* cos 0 = v I , is the component of the velocity of the escaping particle perpendicular to the metal surface. The mechanism for ion formation based on this theory is a velocity-dependent form of surface ionization. Experimental evidence in support of the velocity dependence of the secondary ion formation process has been provided by Yu for negative ions [282] and by Vasile for positive ions [283]. 3.5.3. Types of cesium sputter negative ion sources 3.5.3.1. The Miiller-Hortig geometry source Several ver-

sions of negative ion sources have evolved since the discovery by Krohn [278] that the negative ion yield of sputtered particles is greatly enhanced by the presence of a thin layer of cesium on the surface of the material being sputtered. The first of the sources was developed by Miiller and Hortig 12841. The source consists of a continuously rotating wheel to which is attached an annular ring of the material to be sputtered. A thin layer of cesium metal is evaporated on the ring which is subjected to bombardment by approximately 20-keV

271

G.D. Alton / Ion sources for materials research

POSITIVE EXTRACTION

ION ELECTRODE NEGATIVE

ION

IONIZER

Fig. 64. Schematic diagram of the Middleton-Adams geometry negative ion source [285]. positive argon ions that impinge at an angle of 20” with respect to the surface. The negative ions produced in the sputtering process are then extracted by an electrode system. The source has produced a wide variety of negative ion beams, including 2.2 uA C-, 0.4 p,A MgO-, 1 p,A Cr-, 6 uA Cu-, 5 pA InO-, 14 FA Ag-, 4.6 ~.LATao;, 24 p,AO-, 12 p,A Au-. 3.5.3.2. Cesium sputter negative ion sources which utilize porous tungsten ionizers The Middleton-Adams geometry source: The negative sputter ion source of Middleton and Adams [285] also uses a cesium ion beam extracted from a cesium ion source of the porous-tungsten-ionizer type to sputter the material of interest. A schematic of the source illustrating the essential features is shown in fig. 64. The cesium surface ionization source is mounted at ground potential. A cesium beam of 0.1-l mA is accelerated by a potential of _ 20 kV and allowed to strike a conical surface of half-angle - 20”. The cesium serves as both a sputtering agent and an electron donor in the formation of the negative ion species. The negative ions are extracted through an aperture in the apex end of the cone. The source is equipped with an externally indexable wheel containing several samples, which permits rapid change of ion species. Gaseous material may also be introduced near the cone surface to enable formation of negative ions from these materials, therefore extending the range of capability of the source. The versatility of the ion source is exemplified by the number of negative ion species that the source has been used to produce. Negative ions have been produced from more than 20 elements, including: 20 p,AH-, 3 uALi_, 0.4 p.AB-, 40 uAC-, 100 uAO-, 30 p,A F-, 25 pA Si-, 40 p.A S-, 4 uA P-, 0.5 p.A CaH;, 0.8 JLA Fe H-, 10 uA Cu-, 0.3 p.A Bi-, and 20 l.tA Au-. The emittance of the Middleton-Adams source has typical values which range between 20~ and 309~ mmmrad (MeV)“’ depending on the size of the aperture in the conical base of the sample [286]. A version of the source that operates in the reflected beam mode has found practical use as a source of negative ions

from very small isotopically enriched samples and as a concept useful in archaeological dating experiments [287]. This mode of operation is illustrated in fig. 65. The modified Miiller-Hortig geometry source: A source based on the Miiller-Hortig geometry has been developed by, Alton [288]. A 20-keV cesium ion beam is extracted from a conventional porous tungsten cesium ion source and focused onto the sample surface at an oblique angle (10’) with respect to the sample surface by means of an asymmetric einzel lens. Negative ion beams are extracted through a spherical-geometry gridded electrode. The source is equipped with an indexable sample wheel which houses 18 or more circular samples. Among the ion beams which have been produced are: 0.2 p,A All, 3 p.A AlO-, 6 FA Au-, 25 uA C-, 20 pA C;, 100 FA Cl-, 0.5 +A Cu-, 0.8 pA CuO-, 40 p,A F-, 26 uA I-, 30 )LA 0-, 3 PA Pt-, 44 FA S, 2 uA Tao;. The multiple-sample AMS sputter source used at ETH:

A multiple-sample source specifically designed for use in accelerator mass spectrometry CAMS) has been developed which utilizes a porous tungsten cesium surface ionization source and is, therefore, similar to the

O’p

mm

Fig. 65. Illustration of the reflected Cs+ beam mode of operating the Middleton-Adams negative ion source [287].

G.D. Alton / Ion sources for materials research

272

CATHODE INSULATOR

COOLANT OUTLET EXTRACT INSULATOR

IR LOCK VALVE

SPUlTER CATHOD

-

FACE IONIZER

CESIUM OVEN

E’XTRACT APERTURE

/

\

EXTRACT ELECTRODE

Fig. 66. A schematic

drawing

COOLANT INLET

of the cesium-sputter

negative

ion source described

in ref. [294].

sten ionizer to form the positive cesium ion beam used for sputtering the sample material [290]. The cesium ion beam is accelerated toward samples mounted circumferentially on an indexable sample holder. The source can be used to produce a wide variety of atomic and molecular ion beams much like other sources based on the cesium sputter principle, and has a lower emittance than the Middleton-Adams source [286].

modified Miiller-Hortig source described above [289]; however, the target changing mechanism is designed for computer controlled transfer of magazines loaded with 25 samples through a vacuum airlock and into beam position. Chapman inuerted-geometry source: Chapman has developed an inverted form of the cesium sputter negative ion source which utilizes an annular porous tung-

HEATING ELEMENT,

__.._ -.-_.

__r

---‘_ITRY

EXTRACTION ELECTRODE \

CESIUM IONIZE R-

-3i -- +

‘NEGATIVE ION EXTRACTION APERTURE

’ Fig. 67. A schematic

drawing

of a cesium-sputter

negative

SPUTTER SAMPLE

ion source [295-2981.

equipped

with a spherical-geometry,

cesium -surface

ionizer

G.D. Alton / Ion sourcesfor materialsresearch

Another version of this source geometry has also been described by Yntema and Billquist [291]. 3.5.3.3. Cesium sputter negative ion sources which utilize the direct ionization of cesiam vapor

While sources which employ porous tungsten cesium surface ionizers have proved to be very versatile generators of many negative ion beams, this method of production rarely provides the optimum cesium layer coverage on the sample surface which is necessary for optimum negative ion production efficiency. With this approach, the cesium left on the surface is volume distributed in the region near to the surface and the steady-state surface concentration is determined by the saturation value for cesium for the particular material, which varies from material to material. (The saturation value is defined as the amount of cesium left on the surface after steady state ion bombardment conditions have been established.) Since the saturation value varies in inverse proportion to the sputtering ratio, high sputtering materials such as Ag, Cu, Au, etc. have low residual cesium surface content and thus do not produce optimum negative ion yields according to their capabilities. The sources described below overcome this handicap. In these sources, the sputter process occurs in a cesium vapor environment where adequate cesium coverage is present on the sample surface. Therefore, these sources are, in general, much more efficient than the sources which utilize porous tungsten ionizers because of the ability to control the amount of neutral cesium on the sample surface which is critically important for optimization of the efficiency for negative ion formation. HEATING

ELLIPSOIDAL

273

Several state-of-the-art negative ion sources based on direct surface ionization of cesium vapor have been described in the literature. Reviews of recent advances made in sources based on this technology are given in refs. [292,293]. These sources differ only in the geometry of the ionizer, its spacing in relation to the negatively biased sample, the spacing of the sample in relation to the ion exit aperture, and the aperture size. The source shown in fig. 66 [294] illustrates the principles of source types described below. Spherical geometry ionizer sources: A source equipped with a spherical geometry ionizer is shown schematically in fig. 67. The performance characteristics and mechanical design features of the source have been reported previously [295] and the emittance of the source has also been measured [296-2981. The positive cesium ion beam current density distribution at impact with the sample surface is - 0.75 mm full diameter when the sample is positioned at the focal point of the’system. This particular ionizer geometry does not exhibit a halo beam surrounding the highdensity distribution. Therefore, the high-density central region of the negatively biased sample serves as the sole region from which negative ion beams are generated within the source. The computed perveance for cesium in this electrode configuration is K = 2 x IO-’ [A/V3”]. Ellipsoidal-geometry cesium ionizer sources: To date, only one source equipped with an ellipsoidal-geometry ionizer has been developed. The source is illustrated in fig. 68. The source emittance and brightness have been measured, as well, and are reported in refs, [297,298]. The size and shape of the observed wear patterns for

ELEMENT

EXTRACTION ELECTRODE

GEOMI---”

\

’ SPUTTER SAMPLE

Fig. 68. A schematic drawing of a cesium-sputter negative ion source equipped with an e~psoidal-g#rnet~, [297,298].

cesium-surface ionizer

G.D. Alton / Ion sourcesfor materialsresearch

274

HEATING

ELEMENT \

CYLINDRICAL

EXTRACTION ELECTRODE \

GEOMF

\

NEGATIVE ION EXTRACTION APERTURE

’

Fig. 69. A schematic drawing of a cesium-sputter

SPUlTER SAMPLE

negative ion source equipped with a cylindrical-geometry [294,295].

this ionizer geometry, as well as those computationally predicted, are sensitive to sample position because of the strongly convergent nature of the cesium ion beam. When placed at the focal point of the electrode system, the sample wear pattern has a diameter 0 = 1.25 mm. As is the case with the spherical geometry source just described, halo beams surrounding the central highdensity region are not observed. This electrode configuration has a high perveance in relation to other focusing systems. The computationally determined perveance for the electrode system is found to be K = 17 x 1O-9 [A/V3’*]. Cylindrical-geometry ionizer sources: The cylindrical geometry ionizer source configuration shown in fig. 69 has been described in detail in refs. [294,295], and the emittance and brightness characteristics of the source have been reported in refs. [296-2981. The computed cesium ion current density distribution and observed sample wear patterns agree remarkably well. The observed wear pattern from this source is composed of two parts: a region of concentrated wear with full diameter of - 0.75 mm, and a low-density, uniformwear region with a diameter of - 4.5 mm. Thus, the negative ion beam extracted from the source is composed of beams of two distinctly different characters: (1) a very small source located on axis, and (2) a uniformly distributed halo beam surrounding the central high-density region. Both beams are expected to contribute to the emittance and perhaps increase its value over that of the previously described sources equipped with spherical and ellipsoidal geometry ioniz-

cesium -surface ionizer

ers. The computed perveance of this electrode geometry for cesium is K = 57 X 1O-9 [A/V3/*]. An ion source equipped with a spiral-wound, cylindrical ionizer has also been developed at the University of Pennsylvania; the performance of this source has been described in ref. [299], and details of the emittance and brightness of the source are reported in refs. [297,298]. The sample wear patterns from the source, shown in fig. 70, were found to be considerably more complex than those of the other sources. Typically, the central region of the sample was found to be strongly worn with a wear diameter of - 1 mm surrounded by a

SPIRAL WOUN CESIUM IONIZER

NEQATIVE ELECTRODE

/

\NEQATlVE EXTRACTION

ION ARERTURE

Fig. 70. A schematic drawing of a cesium-sputter negative ion source equipped with a spiral-wound cesium-surface ionizer [299].

G.D. Alton / Ion sourcesfor materialsresearch

large-diameter (d, = 8 mm) asymmetrically worn halo area. The perveance of the positive ion region of this source is more difficult to compute due to the spiral nature of the ionizing surface. When the source is equipped with a smooth cylindrical-geometry ionizer surface, a computed perveance of K = 38 X 10e9 [A/V3/*l is obtained. multiple-sample sources: For applications where both high efficiency and/or high frequency sample changes are desirable, as is the case for accelerator mass spectrometry CAMS) applications, the ability to process multiple samples is essential. A few sources have been designed which meet this requirement, including the sources described in refs. [289,300-3021. However, due to the lack of less than optimum cesium coverage on the sample surfaces for sources equipped with porous tungsten ionizers, such as the source described in ref. [289], the efficiency is rather low for most negative ion species. In order to improve the negative ionization efficiency and to reduce the time required for analyzing each sample, the source described in ref. [289] has been retrofitted to operate in a cesium vapor environment and has been equipped with a three-electrode, spherical-geometry ionizer patterned after the source of ref. [300]. The new source configuration is described in ref. 13021. Fig. 71 schematically illustrates the latter source (i.e., the source described in ref. [300]). The source is equipped with a sphe~cal-geomet~ ionizer and computer-controlled, sixty-sample, wheel-type sample indexing mechanism. The three-electrode structure is commensurate with good beam transmission for both the positive and negative ion beams. Negative ion beam intensity data: The negative ion beam intensities which can be extracted from the STEPPING MOTOR CONTROLLED MULTIPLE SAMPLE HOLDER (3O-.SAMPLES)>

SWIVEL LATCH

215

sources previously described depend on a number of factors. The rate of negative ion generation depends on the magnitude of the cesium ion current used to sputter the sample material, which, in turn, depends on the source operational parameters, e.g., the cesium oven temperature and sputter probe voltage. The space charge limited cesium current, I+, which can be accelerated at a given sputter probe voltage, I’, and subsequently used for sputtering the sample depends on the perveance, K, of the particular electrode configuration. The negative ion current which can be extracted from the total current generated in the sputter process depends on the size of the negative ion generation region on the sample surface, the angular distribution of the negative ion current at the ion extraction aperture, the spacing of the sample in relation to the aperture, the aperture size, the sputter probe voltage V, and the cesium oven temperature. Because of operational variables and the differences in the ionizer/ sample electrode configuration, the negative ion currents for a particular species will, in general, differ from source to source, and for a particular source vary from operational period to operational period. Negative ion yields for a particular species will depend on the chemical composition of the sample, as well. The versatility of the sources described above is reflected by the wide spectrum of momentum-analyzed negative ion beams that have been observed during their operation. A partial list of species and negative ion beam intensities realized from these sources under a variety of operating conditions are the following: 50 p,A H-, 8 uA Li-, 3 p,A Be-, 10 nA B-, 270 nA C, 280 nA O-, 30 PA F-, 3 p.A Na-, 1 uA MgH-, 2.8 ~.LAAl-, 100 uA Si-, 100 yA S-, 30 PA Cl-, 0.5 ~J.A

HOUSING

SLIDE UNIT Fig. 71. A schematic drawing of a multiple-sample cesium-sputter negative ion source equipped with a spherical-geometry cesium-surface ionizer [300].

G.D. Alton / Ion sources for materials research

276

[303]. The source

utilizes a single conical-geometry sample similar to the Middleton-Adams source 12851 and a microwave power source for ionizing neutral cesium vapor supplied from an oven. The microwavepowered cesium source generates l-10 mA of cesium ion beams at energies between 10 and 20 keV. Negative ions are extracted from a hole bored into the apex of the sample. Neutral vapor simultaneously impinges on the surface of the sample, which enables the source to produce intense beams of a variety of elements, including 740 FA for C-, 200 FA for C;, 320 FA for Cu-, 170 p,A for Si-, 59 p,A for Ni-, 15 FA for B-, and 20 PA for B;. No emittance information has been published for this source.

ELLIPSOIDAL GEOMETRY IONIZER A CYLINDRICAL GEOMETRY IONIZER

n

O[i+“““” 0

10

20

_ 30

40

50

60

70

60

90

100

PERCENT OF NEGATIVE ION BEAM INTENSITY ‘A PRODUCT OF GENERAL IONEX CORPORATION

3.5.4. Plasma sputter negative ion sources

Fig. 72. Normalized emittance E as a function of percent of negative ion beam intensity for spherical-geometry [296-2981, ellipsoidal-geometry [297,298], cylindrical-geometry [296-2981, and Model 860 [297-2981 cesium-sputter negative ion sources.

The advantage of the plasma-type source lies in the fact that, when operated in a high-density plasma mode, the negatively biased sputter probe containing the material of interest is uniformly sputtered. This characteristic makes it possible to take advantage of the large area spherical- or cylindrical-geometry optics of the spherical- or cylindrical-sector sputter probes and the plasma sheath surrounding the probe. Negative ions created in the process are accelerated and focused through the plasma to a common focal point which is usually chosen as the ion exit aperture. The sputter particle energy/ angular distributions and aberrations in the acceleration plasma lens system, of course, will increase the beam size at the focal point. Thus, high beam intensities can often be realized while preserving a reasonable emittance value. In sources which utilize hot cathodes (filaments) to generate the plasma for dc operation, the source lifetime can be limited by sputter erosion of the filament. This problem can, in part, be offset by making provisions for filament redundancy or use of rf or ECR plasma generation techniques.

K-, 0.4 FA CaH-, 2 p,A ScH-, 15 p,A Ti H;, 4 FA CrH,, 1 FA MnO-, 0.6 (LA FeO-, 1 p,A Co-, 6 PA CoC-, 150 p,A Ni-, 200 PA Cu-, 1 FA ZnO-, 4 p,A GaO-, 2 FA Ge-, 40 p.A As-, 75 FA Se-, 25 PA Br-, 0.5 PA Rb-, 0.6 FA YO-, 18 FA ZrH-, 0.4 PA Nb-, 1.0 PA NbO-, 1.0 PA MOO-, 0.5 FA Pd-, 18 FA Ag-, 0.5 PA CdO-, 0.5 p,A Sn-, 0.9 p_A Sb-, 30 p,A I-, 0.2 FA Cs-, 4 FA Tao-, 0.7 PA WO-, 1 PA ReO-, 0.3 PA OS-, 5 PA Ir-, 7.5 FA Pt-, 170 PA Au-, 0.5 FA TIO-, 0.4 FA PbO-, 1 FA PbO;, 0.6 FA Bi-, 0.6 FA UC;. Emittance data: Emittance and brightness measurements for the sources equipped with spherical, ellipsoidal, and cylindrical geometry ionizers, as well as for the Model 860 source, have been reported previously [296-2981. Average normalized emittance data for these sources are displayed in fig. 72 as a function of total percent of negative ion beam intensity; the emittance values are defined by eq. (1). The Kyoto University source: A high-intensity, heavy ion source has been developed at Kyoto University CATHODE TERMINAL

CESIUM INLET

3.5.4.1. Types of surface plasma-type H -/D - sources with conuerters Magnetron-&pe plasma sources: The generation

high-intensity,

ANODE CO/VER

of pulsed HP beams involving interactions CATHODE

CATHODE /

\

\

S’DES

7

E INSULATORS 0

1

2

3

4

I

I

I

5

lm

Fig. 73. A schematic drawing of the pulsed-mode magnetron H-/D-

negative ion source described in ref. [307].

277

G.D. Alton / Ion sources for materials research

between a high-density plasma seeded with cesium and a negatively biased surface (converter) was first reported by workers at Novosibirsk. Reviews of H-/Dsurface source development at this laboratory are given in refs. [244,245]. The magnetron-type plasma source was first developed by Belchenko et al. [304]. II- ions are extracted from the E x B produced plasma from a slit in the anode, elongated perpendicularly to the magnetic field. H- ion currents up to 22 mA, with the exceedingly high ion current density of 220 mA,/cm2, were extracted. Because of the small size and the very high arc power required, these results are available only for short pulse lengths of a few milliseconds. This type of source has been successfully used in accelerator-based applications and for use as a possible neutral particle injector for magnetic fusion energy devices. A number of sources based on this principle have been

developed, including those described in refs. [304-3071. A version of this source is shown in fig. 73 [307]. The plasma is confined with a l-3-kG magnet with direction parallel to the plane of the negatively biased converter. The mechanism of negative ion formation are essentially that as proposed in ref. 12721. Negative ions are principally formed by a nonthermodynamic surface ionization process as they leave the low-workfunction converter surface. The source is commensurate with slit-geomet~ extraction apertures. The cylindrical-geometry converter surface focuses the self-extracted beams through the slit aperture and thus greatly enhances the current densities which can be extracted from the source [305,306]. By utilizing cylindricalgeometry focusing properties, ion current densities up to 1 A/cm2 have been achieved [305]. Emittance measurements from this source type have been reported

Hp GAS

J

+ f -

MAGNET

SPUTTER PROBE P.S. -65OV, 1A (DC) -iOOOV, 1A (PULSED)

Fig. 74. A schematic drawing of the pulsed-mode, multi-cusp, magnetic-field H-/D-

source described in refs. [309,310].

278

G.D. Alton / Ion sources for materials research

[3061. Values of 1.9rr and l.%r mm mrad (MeV)l/* were recorded, respectively, for the x and y directions for a beam intensity of 40 mA.

ates pulsed H- beams with - 40 mA peak intensity 13091.An almost identical source (fig. 74) is used at the National Laboratory for High Energy Physics 13101for generation of pulsed H- beams for synchrotron injection applications.

High-intensity, multicusp, magnetic-field, H -/D -, negative ion sources: Multicusp magnetic field plasma

surface sources have been developed at a number of laboratories for the generation of high-intensi~ beams of H- and D-, including those described in refs. [308-3101. While most sources utilize spherical-geometry sputter cathodes, this source geometry is commensurate with cylindrical-geometry sputter cathodes which require slit apertures in the source. A high-intensi~ H-/Dversion of a source equipped with a cylindrical-geometry converter which produces 1.1 A in pulsed mode operation, has been described by Leung and Ehlers [308]. A source, based on this principle, which uses a spherical-sector-geomet~ converter, has been developed for use at the LAMPF. This source gener-

3.5.4.2. Types of heauy negative ion plasma sputter sources Radial-geometry plasma sputter sources: The radial-

geometry source was first developed by Tykesson and Andersen [311,3121 which was the precursor of a generation of sources which operate in a cesium vapor environment. These sources are, in general, much more efficient than sources which utilize porous tungsten ionizers because of the ability to control the amount of neutral cesium on the sample surface which is critically important for optimization of the negative ion formation efficiency. PERMANENT

-/ I

MAGNET

I~MENT

POSITIONABLE SPUTTER PROBE

!

,BORON

NITRIDE

INSULATOR

FILAMENT ALUMINA

FEE0

THROUGH

INSULATOR

COOLANT’ . ._-..._,.. ‘.

\

NEGATIVE

ION

__-

. ..__

_

FILAMENT

PROBE

SAMPLE

APERTURE

Fig. 7.5.A schematic drawing of the radial-geometry, plasma-sputter,

heavy negative i& source described in ref. [314].

G.D. Alton / Ion sources for materials research I.5 cm RADIUS SPHERICAL SPUTTER PROBE. -24 kV\

(0)

EXISTING

ION OPTICS

FOCUS ELECTRODE I

,= 0.2

mA/cn?;

2.3 cm RADIUS SPHERICAL SPUTTER PROBE -24 kV \

(bl

PROPOSED

DIVERGENCE

0 rrns=

f

ION OPTICS:

The radial-geometry sources described in refs. [313,314] embody principals employed in the Tykesson-, Urdersen source [311]. The source, shown in fig. 75, of ref. [314] is an improved version of the source described in ref. [313] and will be used as an illustration of this source type. A weak magnetic field (- 150 G) produced by a permanent magnet is used to collimate the primary electron beam which is thermonically emitted by a tantalum filament located at the end of the ionization chamber. The electron beam produces an approximately uniform plasma by collisional impact with neutral cesium vapor introduced into the chamber from the externally mounted oven. Auxiliary I I I I 1 ’ ’ ’ ’ SOURCE: RADIAL GEOMETRY PLASMA SPUTTER SPECIES: “‘Au-

14

0

10

20

30

40

50

60

70

80

90 100

PERCENT OF NEGATIVE ION BEAM INTENSITY

Fig. 77. Normalized emittance versus percent of negative ion beam intensity for the radial-geometry source described in ref. 13141.

EXTRACTION ELECTRODE

2.33’

ELECTRODE:

~=0.2

-20

mA/cm2;

Fig. 76. Negative ion extraction optics of the radial-geometry plasma-sputter fig. 75.

16

kV

/

/FOCUS

LOW ANGULAR

-20

279

kV

b’rms

= 0.79O

negative ion source of ref. [314] and displayed in

discharge support gas (usually Ar) is introduced into the chamber by means of a standard leak valve to supplement the cesium vapor; chemically active gases may also be introduced into the chamber for generation of atomic negative ions from the gas itself or for chemical combination with the sputter probe material in the formation of molecular negative ions. The sputter sample is cylindrical (typically 10 mm in diameter) with a concave spherical negative ion emission surface machined into the face of the material of radius p = 15 mm. Thus, the geometrical focusing properties of the spherical sputter sample/ plasma lens combination are utilized to increase the beam intensity of the source. The sputter probe is maintained at - - 1000 V relative to the discharge chamber. Fig. 76 displays the negative ion optics of the radial-geometry, cesiumplasma source described in ref. [314]. This particular source concept is, to date, among the most prolific and universal of sources based on the sputter principle, producing useful intensities of both atomic and molecular ion species. Among the ions and intensities that have been reported are the following: 175 p,A H-, 0.4 Li-, 4 uA BeH;, 25 FA BeO-, 0.6 uA B-, 20 PA C-, 20 j.t.A C,, 30 FA 0-, 20 uA F-, 12 pA S-, 12 uA MgH;, 20 uA Si-, 20 uA F-, 2.5 u.A Al-, 9 PA Al;, 2 PA CaH;, 2.5 FA TM;, 55 u.A Ni-, 50 PA Cu-, - 1 PA SrH;, 35 p,A Ag-, 3.1 uA TaN-, 1.4 PA W-, 80 FA Au-, and 12 IJ.A PbS. The normalized emittance [as defined in eqs. (l)] versus percent total negative ion beam current for a 48 PA Au- ion beam is shown in fig. 77. More detailed

280

G.D. Alton / Ion sources for

information concerning given in ref. [312]. The regal-geomet~,

the emittance

of this source is

p~sm~-s~~i~er, high-incenses, negative ion source: This type of source has also been

developed as a high-intensity, pulsed-mode, heavy negative ion source [315-3171. The radial-geometry, H-/Dion source, described in ref. [310] and displayed in fig. 74, has been modified for generation of heavy negative ion beams. The source is shown schematically in fig. 78. Readers are referred to refs. [316,317] for more comprehensive details on the design features, operational parameters, and performance characteristics of the source for heavy ion generation. For heavy negative ion generation, a high-density plasma discharge, seeded with cesium vapor, is produced by pulsing the discharge voltage of two seriesconnected LaB, cathodes maintained at _ 1450°C. For this application, the negatively biased spherical-geometry probe (converter) is made of the material of interest and as such is a consumable item, i.e., negative ions are

materials

research

formed by plasma discharge sputtering of the probe itself. In order to produce higher heavy negative ion beam intensities by sputter ejection at a given probe voltage, a chemically inert, heavy discharge support gas such as Ar, Kr, or Xe, is utilized. Cesium is introduced into the discharge from an external cesium oven. The sheath surrounding the negatively biased sputter probe (spherical radius, p = 140 mm and diameter, 0 = 50 mm) which is maintained at a negative voltage relative to housing, (typically 500 V) serves as the acceleration gap and lens for focusing the ion beam through the exit aperture (diameter, 0 = 18 mm). Under pulsed mode operation at the low duty factors utilized (typically, 2 x 10T3), the LaB, cathodes exhibit very little erosion after many hours of operation. With the combined long lifetimes of the sputter probe, LaB, cathodes, and low cesium consumption rate, the source can operate stably for a few thousand hours at constant peak beam intensity levels without maintenance or cleaning. Sources of this type have proved to be stably operat-

X, GAS MAGNET \

Fig. 78. Schematic drawing of the radial-geometry, high-intensity, heavy negative ion source described in refs. [315-3171.

281

G.D. Alton / Ion sources for materials research

Table 7 Total heavy negative ion peak beam intensities from the high-intensity plasma sputter negative ion source [315-3171

937 437 937 937 438 438 937 937 438 438 937 937 937 937

spherical spherical spherical spherical spherical flat flat flat spherical spherical spherical spherical spherical spherical

6.2 10.3 2.7 6 8.2 4.5 3.7 1.8 30 6.0 7.6 8.1 6 3.6

5om

Ni

g

40-

E

Ag- (91) Au- (73) Bi- (6); O- (42) C (36); C; (58) cu- (77) Cu- (40); 0- (60) As- (20); As; (52) P- (44) 0- (67) Ni- (87) Pd- (69) Pt- (71) Si- (75) Sn- (67)

ing with capabilities of providing a wide spectrum of negative ion beams suitable for a variety of uses, including tandem electrostatic accelerator, low-energy atomic physics, ion implantation, and isotope separator

3 i

30-

:

20-

5

0

20

PERCENTAGE

40

VALVE 1

WATER COOLED EXTRACTION ELECTRODE AlNlCo

\

m

f’

SmCo CUSP-FIELD MAGNETS / SmCo CUSP-FIELD MAGNETS

LAB, CATHODES THERMAL

-vdt

HIGH “OL+AGE INSULATORS

INSULATOR

SCALE (mm) SPUTTER

SAMPLE

120 (%)

applications. When operated in the pulsed mode, such sources hold considerable promise for use in conjunction with tandem electrostatic accelerator/ synchrotron injection applications for heavy ion research. Although principally tested in a low-duty-cycle (repetition rate: l-50 Hz) macropulsed mode (pulse width: SO-300 ps) suitable for heavy ion synchrotron applications, the

SPUTTER PRt INSULATOI

INSUiAiOR

IW

100

60

ION BEAM INTENSITY

Fig. 79. Normalized emittance versus percent total negative ion beam for the radial-geometry, plasma-sputter, heavy negative ion source of refs. [315-3171 and displayed in fig. 78.

\_

0

60

OF NEGATIVE

WATER COOLED COAXIAL CATHODE FEED-THROUGH

AIR-LOCK

2.5mA

l

5;

Sputter Sputter Geometry Total peak Species beam probe probe (%I intensity material voltage (mA) (VI Ag Au Bi C cu cue GaAs GaP MO Ni Pd Pt Si Sn

601

CESIUM

OVEN

Fig. 80. Schematic drawing of the axial-geometry, high-intensity, heavy negative ion source described in refs. [319,320].

G.D. Alton / fan sources for

282 ION ENERGY:

materials

51 keV; ION BEAM INTENSITY:

research

0’: 3.5mA; Au’: ImA 0

ACCELERATION ELECTRODE

PLASMA

25

So

MILLIMETERS

SPUTTER

O(kV)-51

-50

-50

-30

Fig. 81. The negative ion extraction optics of the axial-geometry, plasma-sputter, in fig. 80.

source also offers the interesting prospect of providing dc beams at mA intensity levels of the commonly used semiconducting material dopants (B-, P-, As-, Sb-, etc.), as well as O- for isolation barrier formation. The source can be used for the generation of high-intensity beams of O-; this is particularly attractive, since chemical reactions between the 0, feed gas and hot cathodes commonly used in volume discharge positive ion

-30

0

negative ion source of refs. [319,320] and displayed

sources, which lead to extremely short cathode lifetime, are avoided. Table 7 provides a partial list of total beam intensities, species, and probe materials utilized in the radial-geometry source of refs. [315-3171, 1,hen operated in pulsed mode. Ejramples of beam emittances versus percent total negative ion beam for this source are also shown in fig. 79 for Ni beams with peak pulse

Fig. 82. Schematic drawing of the axial-geometry, high-intensity, heavy negative ion source described in ref. [322].

G.D. Alton / Ion sourcesfor materialsresearch

intensities of 2.5 and 6 mA. The emittances are seen to increase in proportion to the ion beam intensity as expected from space charge considerations. The source has also been operated in dc mode [318]. In this mode, the source has demonstrated beam intensities of 1.5 mA of Cu-. This source has also been equipped with a cylindrical-geometry sputter probe which produces N 100 mA of Cu- at a duty factor of 10%. Axial-geometry, plasma-sputter, heavy negative ion sources: An axial-geometry, multicusp, ma~etic-field

form of the source, which is equipped with a spherical-geometry sputter cathode is under development at the Oak Ridge National Laboratory [319,320]. The source described is illustrated schematically in fig. 80. This source utilizes cylindrical coaxial LaB, cathodes and is equipped with provisions for fast interchange of the spherical-geometry sputter electrode. The computed negative ion extraction optics of the source are displayed in fig. 81. A similar source is also under development at the University of Kyoto which has been equipped with an rf plasma generation coil which is ~mmensurate with dc operation [321]. In this mode, dc beams of 2.5 mA of Cu- have been recorded. An axial-geometry source (fig. 82) is under development at the National Laboratory for High Energy Physics which utilizes ECR plasma-generation techniques to excite the plasma which again avoids the limited lifetime problems associated with hot cathode forms of the source [322].

Acknowkdgements The author expresses his gratitude to Ms. Jeanette McBride for typing and to Dr. C.M. Jones for editing the manuscript.

References [l] G.D. Alton, Nucl. Instr. and Meth. 189 (1981) 1.5. 121 G.D. Alton, Ionization Phenomena and Sources of Ions, Ch. 2, Vol. 4 in Applied Atomic Collisions Physics, ed. H.W. Massey (Academic Press, New York, 1983). 131T.S. Green, Rep. Prog. Phys. 37 (1974) 12.57. 141A. Th. Forrester, Large Beams: Fundamentals of Generation and Propagation (Wiley, New York, 1988). [5] LG. Brown ted.), The Physics and Technology of Ion Sources (Wiley, New York, 1989). [6] F.M. Penning, Physica 4 (1937) 71. [7] J. Backus, in: The Characteristics of Electrical Discharges in Magnetic Fields, eds. A. Guthrie and R.K. Wakerling, Chap. 11 (McGraw-Hill, New York, 1949). 181 F.B. Hooper, Jr., Advances in Electronics, Electron Phys. 27 (1969) 295. 191 J.R.J. Bennett, IEEE Trans. Nucl. Sci. NS-19 (2) (1971) 48.

283

WJlI. Langmuir, Phys. Rev. 2 (1913) 450. ml D. Bohm, in: Characteristics of Electrical Discharges in Gases, eds. A. Guthrie and R.K. Wakerling, Chaps. 3 and 4 (McGraw-Hill, New York, 1949). I121 J.D. Jackson, Classical Electrodynamics, Chap. 7 (Wiley, New York, 1%2). 1131 I. Langmuir, Phys. Rev. 33 (1929) 954. iI41D. Bohm, E.H.S. Burhop and H.S.W. Massey, Characteristics of Electrical Discharges in Gases, eds. A. Guthrie and R.K. Wakerling, Chap. 2 (McGraw-Hill, New York, 1949). 1151W.L. Rautenbach, Nucl. Instr. and Meth. 12 (1961) 196. 0. Almen and K.O. Nielsen, Nucl. Instr. and Meth. 1 1161 (1957) 302. 1171I. Chavet, Thesis, Univ. Paris (1965). 1181J.H. Whealton, R.W. Gaffey, and E.F. Jaeger, Appl. Phys. Lett. 36 (1980) 91. t191R. Becker and W.B. Herrmannsfeldt, Rev. Sci. Instr. 63 (1992) 2756. Nucl. Instr. and Meth. 189 (1981) 55. ml J.H. ~ealton, 1211W.B. Herrmannsfeldt, SLAC Rep. SLAC-166 (1973). [221 M. Von Ardenne, Tabellen der Elektronenphysik, Ionenphysik and ijbermikroskopie (VEB, Dtsch. Verlag Wiss., Berlin, 1956). [231 CD. Moak, H.E. Banta, J.N. Thurston, J.W. Johnson and R.F. King, Rev. Sci. Instr. 30 (1959) 694. [241H. Friihliche, Nucleonic (1959) 183. t251 H. Winter, Proc. Int. Conf. on Electromagn. Isot. Sep., Marburg, Germany, BMBW-FB70-28 (1970) 447. f261 R.A. Demirkhanov, H. Friihliche, V.V. Kursanov and T.T. Gutkin, Brookhaven Nat. Lab. Rep. BNL 767 (1962) 218. 1271J. Kistemaker, P.K. Rol and J. Politiek, Nucl. Instr. and Meth. 38 (196.5) 1. [281 C. Lejeune, Nucl. Instr. and Meth. 116 (1976) 417. LTJIR. Masic, R.J. Warnecke and J.M. Sautter, Nucl. Instr. and Meth. 71 (1969 339. 1301J. Illigen, R.K. Kirchner and J. Schulte, IEEE Trans. Nucl. Sci. NS-19 (1971) 35. [311O.B. Morgan, G.G. Kelley and R.C. Davis, Rev. Sci. Instr. 38 (1967) 467. [321N.B. Brooks. P.H. Rose, A.B. Wittkower and R.P. Bastide, Rev. Sci. Instr. 15 (1964) 894. [331J.H. Ormrod, Proc. Symp. on Ion Sources and Formation of Ion Beams, Brookhaven National Laboratory, BNL-50310 (1971) 151. 1341G. Gautherin, C. Lejeune, F. Prangere and A. Septier, Plasma Phys. 11 (1969) 397. G. Stover and F. Zajae, Proc. 1989 Particle Acceleration l3.51 Conference, IEEE Cat. No. 89, (X2669-0 (1989) 292. 1361F.M. Bacon, Rev. Sci. Instr. 49 (1978) 427. t371A. Septier, in: Focusing on Charged Particles, Vol. 2, ed. A. Septier (Academic Press, New York (1967) p. 123. [38] R.A. Demirkhanov Yu.V. Kursanov and V.M. Blagoveschenskii, Prib. Tekh. Eksp 1 (1964) 30. 1391J.E. Osher and G.W. H~ilton, Symp. on Ion Sources, Brookhaven BNL 50310 (1971) 157. 1401 O.B. Morgan, Jr., Symp. on Ion Sources, Brookhaven BNL 50310 (1971) 129. [41] W.L. Stirling, R.C. Davis, T.C. Jemigan, O.B. Morgan,

284

G.D. Alton / Ion sources for materials research

T.J. Orzechowski, G. &hilling and L.D. Stewart, 2nd Int. Conf. on Ion Sources, Vienna (1972) p. 278. [42] R.W. Bikes, Jr. and J.B. O’Hagan, Rev. Sci. Instr. 49 (1978) 435; R.W. Bikes, Jr. and J.B. O’Hagan, Rev. Sci. Instr. 53 (1982) 585. [43] H. Horiike, M. Akiba, Y. Arakawa, S. Matsuda and J. Sakuraba, Rev. Sci. Instr. 52 (1981) 567. [44] B.H. Wolf, Nucl. Instr. and Meth. B9 (1976) 13. [45] P.G. Weber, Rev. Sci Instr. 54 (1983) 1506. 1461 M.R. Shubaly, R.G. Maggs and A.E. Wooden, IEEE Trans. Nucl. Sci. NS-32 (5) (1985) 1751. [47] J.P. Ruffell, D.H. Douglas-Hamilton, R.E. Kaim and K. Izumi, Nucl. Instr. and Meth. B21 (1987) 229. [48] T. Taylor, J.S.C. Wills, E.C. Douglas and Tran Ngoc, Rev. Sci. Instr. 61 (1990) 454. [49] G. Sidenius, Nucl. Instr. and Meth. 151 (1978) 349. [50] Z. Song, J. Yu, Z. Yuan and R. Li, Rev. Sci. Instr. 61 (1990) 463. (511 A.T. Finkelstein, Rev. Sci. Instr. 11 (1940) 94. [52] H. Heil, Z. Phys. 120 (1944) 212. [53] K.O. Nielsen, Nucl. Instr. and Meth. 1 (19571 289. [54] M. Ma, J.E. Mynard and K.G. Stephens, Nucl. Instr. and Meth. A277 (1989) 279. [55] J.P. O’Connor, N. Tokoro, J. Smith and M. Sieradzi, Nucl. Instr. and Meth. B37/38 (1989) 478. [56] R. Hentschel and P. Henke, Rev. Sci. Instr. 63 (1992) 2590. [57] M.L. Mallory and E.D. Hudson, IEEE Trans. Nucl. Sci. NS-22 (3) (1975) 1669. [58] J.R.J. Bennett, Proc. 5th Int. Cyclotron Conf., Oxford, England (1969) p. 199. [59] PI. Vasiliev, N.I. Venikov, D.V. Zevjakin, A.A. Ogloblin, N.N. Khaldin, B.I. Khoroshavin, V.I. Chuev and N.I. Chumakov, Nucl. Instr. and Meth. 71 (1969) 201. [60] AS. Pasyuk, E.D. Vorob’er, R.I. Ivannikov, V.I. Zuzketsov, V.B. Kutner and Y. P. Tret’yakov, At. Energ. 28 (1970) 75. [61] T. Itahashi and S. Mine, Nucl. Instr. and Meth. A300 (1991) 1. [62] B.H. Wolf, J. Bossler, H. Emig, K.D. Leible, M. Miiller, P.M. R&h and P. Spadtke, Rev. Sci. Instr. 61 (1990) 406. [63] H. Baumann and K. Bethge, Nucl. Instr. and Meth. 122 (1974) 517. [64] J.H. Freeman, Nucl. Instr. and Meth. 22 (1963) 306. [65] D. Aitken, in: Ion Implantation Techniques, eds. H. Ryssel and H. Glawischnig, Springer Series in Electrophysics vol. 10 (1982) p. 23. [66] P.H. Rose, Rev. Sci. Instr. 61 (1990) 342. [67] S.E. Sampayan, L.E. Frisa, M.L. King and R.A. Moore, J. Vat. Sci. Technol. B6 (1988) 1066. [68] B. Cobic, D. Tosic and B. Perovic, Nucl. Instr. and Meth. 24 (1963) 358. [69] T. Jokic, G. Josimov, M.M. Tosic, B. Goncic and Z. Jurela, Nucl. Instr. and Meth. 178 (1980) 309. [70] Separation of Isotopes in Calutron Units, ed. H.W. Savage, Nucl. Energy Ser. 7 (1951). [71] E. Yabe, Rev. Sci. Instr. 58 (1987) 1.

[72] R. Bemas and 0. Nier, Rev. Sci. Instr. 19 (1948) 89. [73] I. Chavet and R. Bernas, Nucl. Instr. and Meth. 51 (1967) 77. [74] R. Pratap, K.B. La1 and V.P. Salvi, Rev. Sci. Instr. 61 (1990) 481. [75] J.L. Tuck. Picket Fence, USAEC Report WASH-184 (1954) 77. [76] R. Keller, P. Spadtke, and F. Niihmayer, Proc. Int. Ion Engineering Congress, Kyoto, Institute of Electrical Engineers of Japan, Tokyo (1983) p. 25. [77] B. Torp, B.R. Nielsen, D.M. Ruck, H. Emig, P. Spldtke and B.H. Wolf, Rev. Sci. Instr. 61 (1990) 595. [78] S.R. Walther, K.N. Leung and W.B. Kunkel, Rev. Sci. Instr. 61 (1990) 315. [79] R. Jancel and Th. Kahan, Electrodynamics of Plasmas (Wiley, New York, 1966). [80] Y.A. Getting, Phys. Rev. 59 (1941) 467. [81] P.C. Thoneman, J. Moffatt, D. Roaf and J.H. Sanders, Proc. Phys. Sot. London 61 (1948) 483. [82] CD. Moak, H. Reese, Jr. and W.M. Good, Nucleonics 9 (1951) 21. [83] P.M. Morozov, Dokl. Akad. Nauk SSSR 102 (1955) 61. [84] J. Moreau and R. Vienet, Rapp. CEA-AC-Fr. Comrnis. Energ. At. CEA-AC-4972 (1957). [85] L.K. Goodwin, Rev. Sci. Instr. 24 (1953) 635. [86] M.E. Abdelaziz, S.G. Zakhovy and A.M. Addelghaffer, Rev. Sci. Instr. 61 (1990) 460. [87] D. Blanc and A. Degeilh, J. Phys. Radium 22 (1961) 230. [88] S.N. Misra and S.K. Gupta, Nucl. Instr. and Meth. 122 (1974) 303. [89] I.H. Wilson, Radiat. Eff. 18 (1973) 95. [90] E. Regenstreif, Le Synchrotron a Protons du Cern, Chap. 5, Eur. Org. Nucl. Res., Geneva (1959). [91] F. Norbeck and R.C. York, Nucl. Instr. and Meth. 118 (1974) 327. [92] Y. Saito, Y. Mitsuoka and S. Suganomata, Rev. Sci. Instr. 55 (1984) 1760. [93] R. Geller, B. Jacquot and C. Jacquot, Proc. 4th Int. Conf. on Plasma Phys. Controlled Nucl. Fusion Res. (1972) Madison, Wis., CN-28, G6 (1971). [94] R. Geller, Trans. Nucl. Sci. NS-23 (1975) 904. [95] CM. Lyneis and T.A. Antaya, Rev. Sci. Instr. 61 (1990) 221. [96] T.A. Antaya, Rev. Sci. Instr. 63 (1992) 2794. [97] R. Geller, IEEE Trans. Nucl. Sci. CH2669-0 (1989) 1088. [98] P. Decrock, P. Van Duppen, F. Baeten, C. Dom and Y. Jongen, Rev. Sci. Instr. 61 (1990) 279. [99] P. McNeeley, G. Roy, .J. Soukup, J.M. D’Auria, L. Buchmann, M. McDonald, P.W. Schmor, H. Sprenger and J. Vincent, Rev. Sci. Instr. 61 (1990) 273. [loo] V. Kobecky, J. Musil and F. Zacek, Phys. Lett. 50A (1974) 309. [loll S. Pesic, Nucl. Instr. and Meth. 198 (1982) 593. [102] R. Nishimoto, S. Hagashii, M. Tanaka, A. Komori and Y. Kawai, Proc. 7th Conf. on Plasma Processing Studies, Tokyo (1990) p. 17. [103] N. Sakudo, K. Tokiguchi, H. Koike and I. Kanomata, Rev. Sci. Instr. 48 (1977) 762.

G.D. Alton / Ion sources for materials research

f104] N. Sakudo, K. Tokiguchi, H. Koike and I. Kanomata, Rev. Sci. Instr. 49 (1978) 940. [105] Y. Torii, M. Shimada and I. Watanabe, Nucl. In&r. and Meth. B21 (1987) 178. [106] Y. Torii, M. Shimada and I. Watanabe, Rev. Sci. Ins&. 61 (1990) 253. [107] J. Hipple, C. Hayden, G. Dionne, Y. Torii, M. Shimada and I. Watanabe, Rev. Sci. 61 (1990) 294. [108] Y. Matsubara, H. Tahara, M. Takahashi and S. Nogawa, Rev. Sci. Instr. 63 (1992) 2595. [109] T. Taylor and J.S.C. Wills, Nucl. Instr. and Meth. A309 (1992) 37. [llO] S.R. Walther, K.N. Leung and W.B. Kumkel, Rev. Sci. Instr. 57 (1986) 1531. [ill] J.M. Lafferty ted.), Vacuum Arcs - Theory and Applications (Wiley, New York, 1980). [112] A.J. Dempster, Rev. Sci. Instr. 7 (1936146. [113] V.S. Venkatasubramanian and H.E. Duckworth, Can. J. Phys. 41 (1963) 234. [114] R.G. Wilson and D.M. Jamba, J. Appl. Phys. 38 (1967) 1976. [115] K.V. Suladze and A.A. Plyutto, Sov. Phys.-Tech. Phys. (Engl. Trans.) 10 (1966) 1006. [116] H.J. Zwally, D.W. Koopman and T.D. Wilkerson, Rev. Sci. Instr. 40 (1969) 1492. [117] C.A. Anderson, J. Mass. Spectrom. Ion Phys. 3 (1970) 413. [118] D.A. Anderson and J.R. Hinthorne, Anal. Chem. 4.5 (1973) 1421. 11191 LG. Brown, J.E. Galvin and R.A. MacGill, Appl. Phys. Lett. 42 (1985) 338. [120] I.G. Brown, J.E. Galvin, B.F. Gavin and R.A. MacGill, Rev. Sci. Instr. 57 (1986) 1069. [121] H. Shiraishi and I.G. Brown, Rev. Sci. Instr. 61 (1990) 3775. [122] A.A. Vasilyev, Rev. Sci. Instr. 63 (1992) 2434. [123] R.H. Fowler and L.W. Nordhe~, Proc. R. Sot. (London) Ser. A 119 (1928) 173. [124] L.W. Nordheim, Proc. R. Sot. (London) Ser. A 121 (1928) 628. [125] R.H. Good and E.W. Miiller, Handb. der Physik 211 (1956) 176. [126] R. Gomer, Field Emission and Field Ionization (Harvard Univ. Press., Cambridge, MA, USA, 1961). [127] C.D. Hendricks, Proc. Int. Conf. on Electron Beam. Sci. Tech., R. Bakish, 1st ed. (Wiley, New York, 1964) p. 915. [128] R. Levi-Setti, Nucl. Instr. and Meth. 168 (1980) 1391. [129] S. Sato, T. Kato and N. Igata, Proc. 11th Symp. on Ion Sources and Ion Assisted Technology, ed. T. Takagi, Tokyo (1987) p. 91. [130] R.J. MacDonald (ed.), Proc. 4th Australian Conf. on Nuclear Techniques of Analysis, Lucas Heights, NSW, Australia (1985) p. 49. [131] R.A.D. MacKenzie and G.D.W. Smith, Nanotechnology 1 (1990) 163. [132] P.D. Prewett and G.L.R. Mair, Focused Ion Beams from Liquid Metal Ion Sources (Wiley, Chichester, 1991). [133] G.E. Taylor, Proc. R. Sot. (London) Seq. A 280 (1964) 383.

285

11341 P. Marriot and J.C. Rivibre, Emission Mechanisms in the Liquid Metal Ion Source: A Review, Rep. No. AERE-R 11294, Harwell Laboratory, Oxfordshire (1984). [135] R.G. Forbes, J. Phys. D. 15 (1982) 1301. [136] R. Gomer, Appl. Phys. 19 (1979) 365. [137] P.D. Prewett, G.L.R. Mair and S.P. Thompson, J. Phys. D15 (1982) 1339. 11381G.L. Allan, J. Zhu and G.J.F. Legge, Proc. 4th Australian Conf. on Nuclear Techniques of Analysis, ed. R.J. McDonald, Lucas Heights, NSW, Austraiia (1985) p. 49. [139] R.L. Seliger, J.W. Ward and R.L. Kubena, Appl. Phys. Lett. 34 (1979) 310. 11401T. Kashiwagi, Y. Gotoh, H. Tsuji, J. Ishikawa and T. Takagi, Proc. 11th Symp. on ton Sources and Ion Assisted Technology, ed. T. Takagi, Tokyo, Japan (1987) p. 87. [141] G.D. Alton and P.M. Read, Nucl. Instr. and Meth. B54 (1991) 7. [142] P. Marriott, Angular and Mass Resolved Energy Distribution Measurements with a Gallium Liquid-Metal Ion Source, Report No. AERE-R-12674, Harwell Laboratory, Oxfordshire (1987). [143] L.W. Swanson, G.A. Schwind, A.E. Bell and J.C. Brady, J. Vat. Sci. Technol. 16 (1979) 1864. [144] H. Boersch, Z. Phys. 139 (1954) 115. [145] K.H. Loeffler, Z. Angew. Phys. 27 (1969) 145. 11461W. Knauer, Optik 59 (1981) 335. 11471L.W. Swanson, G.A. Schwind and A.E. Bell, J. Appl. Phys. 51 (1980) 3453. [148] G.L.R. Mair, D.C. Grindod, M.S. Mousa and R.V. Latham, J. Phys. D16 (1983) L209. [149] G.D. Alton and P.M. Read, J. Appl. Phys. 66 (1989) 1018. 11501G.D. Alton and P.M. Read, J. Phys. D22 (1989) 1029. 11511A.E. Bell, G.A. Schwind, and L.W. Swanson, J. Appt. Phys. 53 (1982) 4602. [152] P. Wagner and T.M. Hail, J. Vat. Sci. Technol. 16 (1979) 1871. [153] P.D. Prewett and D.K. Jeffries, J. Phys. D13 (1980) 1747. 11541R. Clampitt, Nucl. Instr. and Meth. 189 (1981) 111. 11551 R, Clampitt, J. Vat. Sci. Technol. 12 (1975) 1208. 11561J. Ishikawa and T. Takagi, J. Appl. Phys. 56 (1984) 3050. [157] J. Ishikawa and T. Takagi, Vacuum 36 (1986) 825. [158] J. Ishikawa, H. Tsuji and T. Takagi, Rev. Sci. Instr. 61 (1990) 592. 11591Y. Gotoh and J. Ishikawa, Rev. Sci. Instr. 63 (1992) 2438. 11601 P.M. Read, J.T. Maskrey and G.D. Alton, Rev. Sci. Instr. 61 (1990) 502. [161] A. Wagner, Appl. Phys. Lett. 40 (1982) 440. [162] V. Wang, J.W. Ward and R.L. Selinger, J. Vat. Sci. Technol. 19 (1981) 1158. I1631 R.L. Kubena, C.L. Anderson, R.L. Selinger, R.A. Jullens and E.H. Stevens, J. Vat. Sci. Technol. 19 (1981) 916. [164] E. Miyauchi, H. Hashimoto and T. Utsumi, J. Appl. Phys. 22 (1983) L225.

286

G.D. Alton / Ion sources for materials research

[165] E. Miyauchi, H. Arimoto, H. Hashimoto, T. Furuya and T. Utsumi, Jpn. J. Appl. Phys. 22 (1983) L287. [166] K. Gamo, T. Ukegawa, Y. Inomoto, Y. Ochiai and S. Namba, J. Vat. Sci. Technol. 19 (1981) 1182. [167] A.E. Bell, G.A. Schwind and L.W. Swanson, J. Appl. Phys. 53 (1982) 4602. [168] R. Clampitt, Nucl. Instr. and Meth. 189 (1981) 111. [169] T. Okutani, M. Fukuda, T. Noda, H. Tamura, H. Watanabe and C. Sheperd, 17th Symp. on Electron, Ion and Photon Beam Technology, Los Angeles, California, May 1983. [170] R. Clampitt and D.K. Jeffries, Nucl. Instr. and Meth. 149 (1978) 739. [171] K. Gamo, T. Ukegawa, Y. Inomoto, K.K. Ka and S. Namba, Jpn. J. Appl. Phys. 19 (1980) L595. [172] K. Gamo, Y. Inomoto, Y. Ochiai and S. Namba, Proc. 10th Conf. on Electron and Ion Beam Science and Technology, 83-2 (The Electrochemical Society, Pennington, NJ, USA, 1982). [173] N.D. Bhaskar, CM. Klimcak and R.P. Fraukolz, Rev. Sci. Instr. 61 (1990) 366. [174] A. Wagner and T.M. Hall, J. Vat. Sci. Technol. 16 (19791 1871. I1751 L.W. Swanson, G.A. Schwind, A.E. Bell and J.E. Brody, J. Vat. Sci. Technol. 16 (1979) 1864. [176] J. Van de Walle and P. Sudrand, Proc. 29th Int. Field Emission Symp. (IFES), eds. H.O. Audros and Nordon (Almqvist and Widsell, Stockholm, 1982) p. 341. [177] M.F. Harrison, UK At. Energy Res. Establ., Rep. AERE-GP/R 2505. [178] H.L. Daley, J. Perel and R.H. Vernon, Rev. Sci. Instr. 37 (1966) 473. [179] V.T. Raiko, MS. Ioffe and V.S. Zolatarev, Prib. Tekh. Eksp. 1 (1961) 29. [180] G. Kuskevics and B. Thompson, Rev. Sci. Instr. 37 (1966) 710. [181] D.M. Jamba, Rev. Sci. Instr. 40 (1969) 1072. [182] G.S. Kino, NASA Contract NAS-34100 (1965). [183] R.G. Wilson, Proc. Int. Conf. on Appl. Ion Beams, Grenoble, (1967) p. 105. [184] R. Haug, Rev. Sci. Instr. 41 (1970) 670. (18.51CF. Cuderman, Rev. Sci. Instr. 49 (1969) 1475. [186] G.R. Brewer, Ion Propulsion (Gordon & Breach, New York, 1970). [187] H.L. Daley and J. Perel, Rev. Sci. Instr. 42 (1971) 1324. 11881 G.D. Alton, Rev. Sci. Instr. 59 (1988) 1039. [189] P. Dyer and R.G.H. Robertson, Nucl. Instr. and Meth. 189 (1981) 351. [190] H. Liebl and B. Senftinger, Rev. Sci. Instr. 59 (1988) 2174. [191] P.B. Johnson, A. Bolson and C.M. Henderson, Nucl. Instr. and Meth. 106 (1973) 83. [192] C.H. Kunsman, Science (Washington) 62 (1925) 269. [193] E.J. Jones, Phys. Rev. 44 (1933) 707. [194] J.P. Blewett and E.J. Jones, Phys. Rev. 50 (1936) 464. [195] R.H. Dawton, in: Electromagnetically Enriched Isotopes, ed. M.L. Smith (Butterworth, London, 1956) p. 37. [196] G. Couchet, Ann. Phys. (Leipzig) 9 (1954) 931.

[197] A. Septier and H. Leal, Nucl. Instr. and Meth. 29 (1964) 257. [198] D.W. Hughes, R.K. Feeney and D.N. Hill, Rev. Sci. Instr. 51 (1980) 1471. [199] D.R. Peele, D.L. Adler, B.R. Litt and B.H. Cooper, Rev. Sci. Instr. 60 (1989) 730. [200] G.D. Alton (unpublished). [201] A.E. Souziz, W.E. Carr, S.I. Kim and M. Seidl, Rev. Sci. Instr. 61 (1990) 788. 12021H.S.W. Massey, Negative Ions (Cambridge University Press, London, 1976). [203] B.M. Smirnov, Negative Ions (McGraw-Hill, New York, 1982). [204] G.D. Alton, Proc. Int. Workshop on Polarized Sources and Polarized Gas Jets, KEK Report 90-15, Nov. 1990, Tsukuba, Japan, p. 339. [205] H. Hotop and W.C. Lineberger, J. Phys. Chem. Ref. Data 4 (1975) 539. [206] T.J. Kvale, G.D. Alton, R.N. Compton, D.J. Pegg and J.S. Thompson, Phys. Rev. Lctt. 55 (1985) 484. [207] D.J. Pegg, J.S. Thompson, R.N. Compton and G.D. Alton, Phys. Rev. Lett. 59 (1987) 2267. [208] CF. Fischer, Phys. Rev. A39 (1989) 963. [209] H.S.W. Massey, Rep. Progr. Phys. 12 (1949) 248. [ZlO] J.B. Hasted, Proc. R. Sot. A205 (1951) 421; A212 (1952) 235. [211] P. Pradel, F. Roussel, A.S. Schlachter, G. Spiess and A. Valence, Phys. Rev. A10 (1974) 797. [212] G.D. Alton, T.J. Kvale, R.N. Compton, D.J. Pegg and J.S. Thompson, Nucl. Instr. and Meth. A244 (1986) 142. [213] R.H. McFarland, A.S. Schlachter, J.W. Stearns, B. Liu and R.E. Olson, Phys. Rev. A26 (1982) 775. (2141 AS. Schlachter, AIP Conf. Proc. No. 117 (1984) 180. [215] G.I. Dimov and G.V. Roslyakov, Instr. Expl. Tech. 17 (1974) 658. [216] R.J. Girnius, C.J. Anderson and L.W. Anderson, Phys. Rev. Al6 (1977) 2225. [217] B.A. D’Yachkov and V.I. Zinenko, Sov. Phys-Tech. Phys. 16 (19711305. [218] R.M. Ennis, Jr., D.E. Schechter, G. Thoeming, D.B. Schlafke and B.L. Donnally, IEEE Trans. Nucl. Sci. NS-14 (1967) 75. [219] R.J. Girnus and L.W. Anderson, Nucl. Instr. and Meth. 137 (1976) 373. [220] A.S. Schlachter, D.H. Loyd, P.J. Bjorkholm, L.W. Anderson and W. Haeberli, Phys. Rev. 174 (1968) 201. [221] J. Heinemeier and P. Hvelplund, Nucl. Instr. and Meth. 148 (1975) 65. [222] J. Heinemeier and P. Hvelplund, Nucl. Instr. and Meth. 148 (1975) 425. [223] T.J.L. Greenway, private communication (now deceased). [224] A.V. Bunge and C.F. Bunge, Phys. Rev. Al9 (1979) 452. [225] G.D. Alton, R.N. Compton and D.J. Pegg, Phys. Rev. A28 (1983) 1405. [226] L.M. Blau, R. Novick and D. Weinflash, Phys. Rev. Lett. 24 (1970) 1268. [227] R. Novick and D. Weinflash, Proc. Int. Conf. on Precision Measurements and Fundamental Constants,

G.D. Alton / Ion sources for materials research

Gaithersburg, MD 1970, eds. D.N. Langenberg and B.N. Taylor, Natural Bureau of Standards Spec. Publication No. 343 (1971) 403. [228] T.J. Kvale, R.N. Compton, G.D. Alton, J.S. Thompson and D.J. Pegg, Phys. Rev. Lett. 56 (1986) 592. [229] W. Gentner and G. Hortig, Z. Phys. 172 (1963) 357. [230] E.B. Hooper, Jr., P.A. Pincosy, P. Pot&en, C.F. Burrell, L.R. Grisham and D.E. Post, Rev. Sci. Instr. 51 (1980) 1066. [231] G.D. Alton, unpublished. [232] T.J.L. Greenway, Oxford University Report 85/77 (1977). [233] J. John, C.P. Robinson, J.P. Aldridge, W.J. Wallace, K.R. Chapman and R.H. Davis, Nucl. Instr. and Meth. 57 (1967) 105. [234] A. Vermeer, N.A. Van Zwol, and B.A. Strasters, Proc. 2nd Int. Conf. on Ion Sources, Vienna, Austria (1972) p. 901. [235] G. Phillipp, V. Scheib and A. Hoffmann, Nucl. Instr. and Meth. 115 (1974) 507. [236] J. Franck, Trans. Faraday Sot. 21 (1925) 536. [237] E.V. Condon, Phys. Rev. 32 (1928) 858. [238] CD. Moak, H.E. Banta, J.N. Thurston, J.W. Johnson and R.F. King, Rev. Sci. Instr. 30 (1959) 694. [239] M. Kobayashi, K. Prelec, and Th. Sluyters, Rev. Sci. Instr. 47 (1956) 1425. [240] A.P. Lawrence, R.K. Beauchamp and J.L. McKibben, Nucl. Instr. and Meth. 32 (1965) 357. [241] H.T. Richards and G.M. Klody, Proc. 1st Int. Conf. on Ion Sources, Vienna (1972) p. 804. [242] J. Aubert, G. Gautherin and C. Lejeune, Proc. 2nd Int. Conf. on Ion Sources, Vienna (1973) p. 797. [243] K.W. Ehlers, Nucl. Instr. and Meth. 32 (1965) 309. [244] V.G. Dudnikov, Nauka 1 (1975) 323. [245] G.I. Dimov, G.Y. Dereviahkin and V.G. Dudnikov, IEEE Trans. Nucl. Sci. NS-24 (3) (1977) 1545. [246] K.W. Leung, P.A. Pinrosy and K.W. Ehlers, Rev. Sci. Instr. 55 (1984) 1064. [247] R.P. Bastide, N.B. Brooks, A.B. Wittkower, P.H. Rose and K.H. Purser, IEEE Trans. Nucl. Sci. NS-12 (1965) 775. [248] E. Heinicke, K. Bethge and H. Baumann, Nucl. Instr. and Meth. 58 (1968) 125. [249] V.H. Smith and H.T. Richards, Bull. Am. Phys. Sot. 18 (1973) 618. [250] Y.I. Belchenko, G.I. Dimov, V.G. Dudnikov and AS. Kupriyanov, Revue Phys. Appl. 23 (1988) 1889. [251] V.G. Dudnikov, Rev. Sci. Instr. 63 (1992) 2660. [252] H.V. Smith, Jr., N.M. Schnurr, D.H. Whitaker, and E. Kalash, IEEE Conf. Proc. CH2387-9 (1987) p. 301. [253] P.W. Allison, IEEE Trans. Nucl. Sci. NS-24 (3) (1973) 1594. [254] J.G. Alessi, J.M. Brennan, A. Kponou and K. Prelec, IEEE Conf. Proc. CH2387-9 (1987) 304. [255] M. Bacal, Physica Scripta T2/2 (1982) 467. [256] M. Bacal, Nucl. Instr. and Meth. B37/38 (1989) 28. [257] K.N. Leung, K.W. Ehlers and M. Bacal, Rev. Sci. Instr. 54 (1983) 56. [258] M. Bacal, A.M. Brunefeau and M. Nachman, J. Appl. Phys. 55 (1984) 15.

287

[259] Y. Mori, T. Okugama, A. Takagi and D.H. Yuan, Nucl.

Instr. and Meth. A30 (1991) 1. [260] K.N. Leung, K.W. Ehlers, C.A. Hauck, W.B. Kunkel and A.F. Liefzke, Rev. Sci. Instr. 59 (1989) 453. [261] K.N. Leung, CA. Hawk, W.B. Kunkel and S.R. Walther, Rev. Sci. Instr. 60 (1989) 531. [262] K.N. Leung, G.J. DeVries, W.F. DiVergilio, R.W. Hamm, CA. Hauck, W.B. Kunkel, D.S. McDonald and M.D. Williams, Rev. Sci. Instr. 62 (1990) 100. [263] R. McAdams, A.J.T. Holmes and M.P.S. Nightingale, Rev. Sci. Instr. 59 (1988) 895. [264] T. Okuyama and Y. Mori, Rev. Sci. Instr. 63 (1992) 2711. [265] D.H. Yuan, R. Baartman, K.R. Kendall, M. McDonald and D.R. Mosscrop, 4th Symp. on Prod. and Neutralization of Negative Ions and Beams, AIP Conf. Proc. No. 158 (1987) p. 346. [266] M. Kaminsky, Atomic and Ionic Impact Phenomena on Metal Surfaces, P.W. Schmor (Springer, Berlin and New York, 1965). [267] N. Kashihira, E. Vietzke and G. Zellermann, Rev. Sci. Instr. 48 ‘(1977) 171. [268] B. Vosicki, T. Bjornstad, L.C. Carraz, J. Heinemeyer and H.L. Ravn, Nucl. Instr. and Meth. 186 (1981) 307. [269] G.D. Alton, Rev. Sci. Instr. 63 (1992) 2453. [270] A. Blandin, A. Nourtier and D.W. Hone, J. Phys. (Paris) 37 (1976) 369. [271] M.L. Yu, Phys. Lett. 40 (1978) 574. 12721J.K. Norskov and B.I. Lundqvist, Phys. Rev. B19 (1979) 5661. [273] G. Blaise and A. Nourtier, Surf. Sci. 90 (1979) 495. [274] Z. Sroubek, Phys. Rev. B25 (1982) 6046. [275] M.L. Yu, Nucl. Instr. and Meth. B18 (1987) 542. [276] P. Williams, Surf. Sci. 90 (1979) 588. [277] G. Slodzian, Surf. Sci. 48 (1975) 161. [278] V.E. Krohn, Jr., Appl. Phys. 33 (1962) 3523. [279] J. Topping, Proc. R. Sot. (London) All4 (1962) 67. [280] E.P. Gyftopoulos and J.D. Levine, J. Appl. Phys. 33 (1962) 67. [281] G.D. Alton, Surf. Sci. 175 (1986) 226. [282] M.L. Yu, Phys. Rev. Lett. 47 (1981) 1325. [283] M.J. Vasile, Phys. Rev. B29 (1984) 3785. [284] M. Miiller and G. Hortig, IEEE Trans. Nucl. Sci. NS-16 (1969) 38. [285] R. Middleton and CT. Adams, Nucl. Instr. and Meth. 118 (1974) 329. [286] G. Doucas, H.R.M. Hyder, and A.B. Knox, Nucl. Instr. and Meth. 124 (1974) 11. [287] K. Brand, Nucl. Instr. and Meth. 154 (1978) 595. [288] G.D. Alton, IEEE Trans. Nucl. Sci. NS-23 (1976) 1113. [289] R. Balzer, G. Bonani, M. Nessi, Ch. Stoller, M. Sufer and W. Wiilfli, Nucl. Instr. and Meth. B5 (1984) 204. [290] K.R. Chapman, Nucl. Instr. and Meth. 124 (1975) 299; K.R. Chapman, IEEE Trans. Nucl. Sci. NS-23 (2) (1976) 1109. 12911J.L. Yntema and P.J. Billquist, Nucl. Instr. and Meth. 199 (1982) 637. [292] G.D. Alton, Proc. Particle Accelerator Conf., IEEE Cat. No. 89CH2669-0, Vol. 2 (1989) 1112. [293] G.D. Alton, Nucl. Instr. and Meth. A287 (1990) 139.

288

G.D. Alton / Ion sources for materials research

[294] G.D. Alton, Nucl. Instr. and Meth. A244 (1986) 133. [295] G.D. Alton and G.D. Mills, IEEE Trans. Nucl. Sci. NS-32 (5) (1985) 1822. [296] G.D. Alton, J.W. McConnell, S. Tajima and G.J. Nelson, Nucl. Instr. and Meth. B24/25 (1987) 826. [297] G.D. Alton, Proc. 11th Symp. on Ion Sources and IonAssisted Technology, Tokyo, Japan (1987) p. 157. [298] G.D. Alton and J.W. McConnell, Nucl. Instr. and Meth. A268 (1988) 445. [299] R. Middleton, Nucl. Instr. and Meth. 214 (1983) 139. [300] G.D. Alton, Rev. Sci. Instr. 63 (1992) 2450. [301] I.D. Proctor, Nucl. Instr. and Meth. B40/41 (1989) 727. [302] T.R. Niklaus, W. Baur, G. Bonani, M. Suter, H.A. Synal and W. Wiilfli, Rev. Sci. Instr. 63 (1992) 2485. [303] H. Tsuji, J. Ishikawa, T. Maekawa and T. Takagi, Rev. Sci. Instr. 61 (1990) 427. [304] Y.I. Belchenko, G.I. Dimov and V.G. Dudnikov, Zh. Eksp. Teor. Fiz. 43 (1973) 1720. [305] J.G. Alessi and Th. Sluyters, Rev. Sci. Instr. 51 (1980) 1630. [306] H.V. Smith, Jr. and P.W. Allison, Rev. Sci. Instr. 53 (1982) 405. [307] C.W. Schmidt and CD. Curtis, IEEE Trans. Nucl. Sci. NS-26 (1979) 4120. I3081 K.N. Leung and K.W. Ehlers, Rev. Sci. Instr. 53 (1982) 803. [309] R.L. York and R.R. Stevens, Proc. 3rd Int. Conf. on Production of Negative Ions and Beams, ed. K. Prelec (American Institute of Physics Conf. Proc. No. 111, New York, 1984) p. 410.

[310] Y. Mori, A. Takagi, K. Ikegami, and S. Fukumoto, Proc. 4th Int. Conf. on Production and Neutralization of Negative Ions and Beams, ed. J.G. Alessi (American Institute of Physics Conf. Proc. No. 158, New York, 1987) p. 378. [311] H.H. Andersen and P. Tykesson, IEEE Trans. Nucl. Sci. NS-22 (3) (1975) 1632. [312] P. Tykesson, H.H. Andersen and J. Heinemeier, IEEE Trans. Nucl. Sci. NS-23 (2) (1976) 1104. [313] G.D. Alton and G.C. Blazey, Nucl. Instr. and Meth. 166 (1979) 105. [314] G.D. Alton, R.M. Beckers and J.W. Johnson, Nucl. Instr. and Meth. A244 (1986) 148. [315] G.D. Alton, Y. Mori, A. Takagi, A. Ueno and S. Fukumoto, Nucl. Instr. and Meth. A270 (1988) 194. [316] Y. Mori, G.D. Alton, A. Takagi, A. Ueno and S. Fukumoto, Nucl. Instr. and Meth. A273 (1989) 5. [317] G.D. Alton, Y. Mori, A. Takagi, A. Ueno and S. Fukumoto, Rev. Sci. Instr. 61 (1990) 372. [318] Y. Mori, A. Takagi, A. Ueno, K. Ikegami and S. Fukumoto, Nucl. Instr. and Meth. A278 (1989) 605. [319] G.D. Alton, Production and Neutralization of Negative Ions and Beams, AIP Conf. Proc. No. 210, ed. A. Herschcovitch (American Institute of Physics, New York, New York, 1990) pp. 411-419. [320] G.D. Alton, Rev. Sci. Instr. 63 (1992) 2455. [321] H. Tsuji and J. Ishikawa, Rev. Sci. Instr. 63 (1992) 2488. [322] Y. Mori and A. Takagi, Rev. Sci. Instr. 63 (1992) 2669.