Anomalous Skin Effect in Thin Films

G. R. Henry
Citation: Journal of Applied Physics 43, 2996 (1972); doi: 10.1063/1.1661647
View online:
View Table of Contents:
Published by the AIP Publishing
Articles you may be interested in
Anomalous Hall effect in epitaxial permalloy thin films
J. Appl. Phys. 114, 163714 (2013); 10.1063/1.4827198
Surface Impedance of Thin Metal Films under Anomalous Skin Effect Conditions
Appl. Phys. Lett. 20, 417 (1972); 10.1063/1.1653998
Simple Approximate Analysis of the Anomalous Skin Effect in a Plasma
Phys. Fluids 14, 1582 (1971); 10.1063/1.1693646
Anomalous Skin Effect in a Gaseous Plasma
Phys. Fluids 12, 1290 (1969); 10.1063/1.1692665
Erratum: Anomalous Skin Effect in a Plasma
Phys. Fluids 10, 2288 (1967); 10.1063/1.1762031

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: Downloaded to ] IP: On: Mon, 28 Sep 2015 14:07:08

(2) size-limited dc. Z=l/at. M. I. we obtain 2 J. 28 Sep 2015 14:07:08 . For our range of densities. 773 (1969). In Fig. I. 9 (1966). Opt.+' neRIO=no wI " wQ o_l (w)f(w)41TW 2 dw . Thus. J. Phys. noting that if local thermodynamic equilibrium occurs. gl fo'" Ql_l(u+ut)e. Although three lengths are involved. Shelepin. we have a microreversibility between the collisional processes O=i and 1 =i. July 1972 [This article is copyrighted as indicated in the article. 1. film thickness. recombination is mainly collisional. the principal contribution to the two integrals of Eq. 7 1. Watson Research Center. RI1 RIO )f(w )41TW 2dw no fw: WQO_I (w)f(w)41TW 2dw = nl fwt WQl * Equipe de Recherche du Centre National de la Recherche Scientifique. a bridge is established between. Soviet Phys. IN.. with particular attention to the transition between the dc limit and the anomalous skin-effect limit. If we take the reduced variable U =mw 2/2kT e .R. Phys.133. a "map" is shown4 indicating regions where various asymptotic formulas apply. The most important restrictions in this treatment are that the conductor be metallic (i. 4The collision frequency 1J is. for example. Smirnov. (USSR) 21 (No. Abramov and B. Anomalous Skin Effect in Thin Films G. 43. A. respectively. and mean free path of the conduction electrons. Downloaded to ] IP: the ionization energy for levels 0 and 1. PEYRAUD AND N. and the remaining two (film thickness t and classical skin depth 15) then determine the important physical effects. Peyraud. (1) classical skin effect. and f is Maxwellian at Te. where the mean free path may be equal to or greater than the classical skin depth2 and/or the conductor thickness.)du For a cold plasma (Te:S 104 OK. The primary purpose of this paper is to examine in detail the behavior of the impedance of a film where the skin depth is about equal to the film thickness. We consider in detail the case in which the mean free path of the conduction electrons is comparable to or greater than the film thickness (thin films at low temperatures). a region where asymptotic formulas do not apply. Note that if one picks any point on the map initially and then the mean free path l is reduced. e. . All of these conditions are usually met by conductors in integrated circuits. I). that the Fermi surface be spherical or nearly spherical). one is carried into the upper right quadrant. corresponding to the classical limit (either dc or skin effect). Spectros. the anomalous skin-effect and the sizelimited dc regions. and we can evaluate RI1 and RIO. In calculations throughout this paper we choose the symmetric boundary condition that the electric field be the same on both sides of the film. and L. Vol.) . 533/68.30. Yorktown Heights. tmw~ and tm (w t)2 are. Doklady We obtain the ratio _I (w In the ratio R I1 /R IO. 3 Other boundary conditions are briefly discussed in Sec. The surface impedance per square 1 is considered as a function of electrical frequency. The impedance is approximately given by the following formulas 5 in the various regions: classical dc. and that the frequency be low enough to make the displacement-current and electron-relaxation effects negligible. the considerations of this paper are relevant to the possible use of integrated circuits at cryogenic temperatures.239 On: Mon. 10. 2L. Appl.7. one length (chosen here as the mean free path l) merely establishes the over-all scale. Reuse of AIP content is subject to the terms at: http://scitation. in the first approximation. 6 z=t(1 +i)1/al5. New York 10598 (Received 22 February 1971) Numerical calculations are presented for the impedance per square of metallic films. No. Here we discuss theoretically some aspects of impedance relevant to thin metallic films at low temperatures. 3V. IV. 147 (1965). INTRODUCTION Electrical impedance has been studied in many materials under various conditions. nt/no is the Boltzmann ratio. A. (A1) is due to small energies tmw 2 in the region where the two functions (u+ut)Q1-l(u+ut) and (u+ul)Qo_l(u+ud verify l (U+Uf'jQl_I(U+Un» (u+U. Gudzenko. (Paris) 30. yielding RI1 »R lo . as- sumed to be constant.U(u+uf)du (A1) go fo'" QO_i(U+Ui)e-U(u+u. PEYRAUD supported in part by the Direction des Recherches et Moyens d'Essais under contract No.2996 J.aip. Henry* IBM Thomas J.)QO_I(U+U. we obtain APPPENDIX We show that recombination occurs mainly on the resonance level R Il » RIO.

30.619 =0. (6b) The latter expression indicates that the impedance minimum is lower by a factor of 1. CLASSICAL LIMIT The classical limit applies to situations where the electron mean free path may be neglected. It is interesting to note that as we go from the dc (large skin depth) region to the skineffect region.(1 + Y3i) .-1(12)1/6(1) . Figure 3 shows the phase of the impedance. therefore of 0) at a given film thickness t. 1. II we discuss the "classical" limit (where the mean free path of the conduction electrons is arbitrarily small) giving analytical results. 2. 8 (1)4/3 Z . a more convenient ordinate would be ao I ZI = (01 t)at I Z I . Reuse of AIP content is subject to the terms at: http://scitation.C.365. current flows on both sides of the film in the skineffect limit.sin(tl 0) 1 cosh(tjo) . aol zl =0..133.L-'--LJuJ 0. J. Appl.LLJ. considered as a function of frequency the impedance has no minimum. so that the impedance is half as large as that associated with the surface of a semi-infinite conductor.L. 43.LL_---"------"---.L. RESISTANCE FIG. LIMIT II. at I Z I ..1 10 8/t FIG.14 than the skin-effect In Sec. This expreSSion is exact. Neglecting the displacement current.4 7r2 al o' This result is of course well known. in dimensionless form applicable to any conductor.JLLLJ. In fact. however. Downloaded to ] IP: 139. III. No. as in the rest of this paper.6 and is presented only for comparison to a more involved case where the mean free path cannot be neglected.l_-L-L--'-. at which point the magnitude of the impedance is given by atl (4) zl =1.. at constant frequency. an inductive reactance is a positive imaginary impedance. we have assumed that any electron striking the face of the film is scattered in a completely diffuse (nonspecular) manner. as is suggested by direct measurement.C. I L---'--'-'-. 4(1 ) (ljt)2 Z=:3 al In(llt). D. 707/1. Vol.aip. Phys.14. explicit calculation shows that both the magnitude and phase oscillate about their asymptotes in a very heavily damped sine wave of wavelength 21T in the ratio tlo.46 (6a) or In the latter two expressions. both asymptotes are rotated counter clockwise by 45°.01 0.J. The magnitude of the impedance is plotted in Fig..239 On: Mon.cos(tjo) u (5) O. Numerical results in the more complicated case of long mean free paths (l comparable to t and/or 0) are presented in Sec. To consider the impedance for various thicknesses at constant frequency.. is particularly convenient when considering the impedance as a function of frequency (and. and applies for all values of tlo. an analytic expression for the impedance per square is easily derived in this limit: z= (2!a) [sinh(tj 0) + sin(tjo) 1+ i[sinh(t/o) . there is a film thickness which minimizes the impedance. and concluding remarks are made in Section IV. Remarkably enough. It is easy to see that in terms of this new ordinate. in terms of til (ratio of film thickness to electron mean free path) and oil (ratio of classical skin depth to electron mean free path).7. 2. both the magnitude and phase of the impedance "overshoot" the skin-effect asymptotes and then return to them. In other words. 28 Sep 2015 14:07:08 . The minimum occurs at tlo = 2. By convention. applies point by point throughout the conductor. The choice of ordinate in Fig. With the boundary conditions chosen. although there is no longer any size effect for dc. and the "overshoot" discussed above becomes an absolute minimum in impedance. Magnitude of impedance per square in the classical (small mean free path) limit. 2. Map indicating approximate regions of validity for various asymptotic formulas for the impedance of thin films.2997 ANOMALOUS SKIN EFFECT IN THIN FILMS where a denotes the conductivity of the metal. the anomalous skineffect impedance is hardly changed in making the opposite assumption8 (of complete specularity). 7 with the impedance given by (at)-1 irrespective of I. J = aE. In this case Ohm's law. July 1972 [This article is copyrighted as indicated in the article. ANOMALOUS SKIN EFFECT SIZE-EFFECT LIMITED D. (3) anomalous skin

28 Sep 2015 14:07:08 . J(z) = !~ i t K(Z ~ZI)E(Z')dZ" (8) with K(u) =f" (l/s _1/s3)e-slul ds.~ J:.239 On: Mon.. At some thickness. as implied by the 45° phase asymptote. The current density at a [E(r)])e-~/I dV r 4 . As the film becomes thicker. As stated above. Once the E field has been determined. Ii a dz (13) .r/~. owing to the heavy damping of the field within the film: The magnitudes of the field and current at the center of the film are substantially reduced from those at the surface. it is found that. Appl. the resistance falls in the usual dc manner. Pippard8 has quoted a very lucid. The size of the effect is fairly small. where ~ denotes the Fermi velocity of the conduction electrons. Thus instead of a sine-wave interference pattern typical of light in transparent films for example. o ~ given point now depends not only on the electric field at that point. (11) Equation (11) is a particularly simple type of integrodifferential equation. July 1972 [This article is copyrighted as indicated in the article. Downloaded to ] IP: 139. This is indeed found to be the case.. it is necessary to consider what happens at the faces of the sheet. we assume that the electrons are scattered diffusely from the faces of the film.0=0 where symmetric boundary conditions have been assumed. at a thickness t=rrli it is easy to show that the field at the center of the film is 90° out of phase with the surface field. the electric field is essentially constant in magnitude and phase through the film. for currents and fields tangential to the film. if t« Ii. HENRY 50r----------. 3. in this case electrons just scattered from a surface look like electrons arriving from a field-free region outside the film.. Reuse of AIP content is subject to the terms at: http://scitation. (7) where the origin is chosen to be the point at which the current is evaluated. Phys.R. (tK(Z' -Z) E( ')d ' dz2 -z2 li21Jo 1 Z z.. No. the field begins to vary appreciably inside the film.-----------. the result (which has also been obtained more formally) is found to be J(t 0) = ~f r(r· 20 (!) '4rrl Z <t W III ~ 10 a. In fact. (10) again. In the skin-effect limit. in applying the result to a sheet of finite thickness. both in phase and magnitude. and we may therefore drop the brackets. In). (7) extends over all space. First. may be understood as a rather muted thin-film interference phenomenon. It turns out to be somewhat easier to think about the resistance rather than the magnitude of the impedance. limit (for which alilZI =0. Phase of the impedance per square in the classical (small mean free path) limit. Should the thickness be increased a bit beyond this point. thus a small increase from t= rrli leads to an increase in the film resistance. we obtain from Maxwell's equations (10) so that 2 d E(z) _ . the resistive and reactive components of impedance are equal. the method used for numerical solution is indicated in the Appendix.30. 0. For a film of thickness t. J.1 10 SIt FIG. Vol. however. NONZERO MEAN FREE PATH The only essential complication introduced by consideration of nonzero mean free paths is that Ohm's law becomes effectively nonlocal. These qualitative features of the dc-to-skin-effect transition persist and become more pronounced as the mean free path of the conduction electrons become longer (see Sec. The integration in Eq. The brackets around E(r) indicate that it is to be evaluated at the time t . equivalently. restrict the integration to the volume occupied by the film.707). although it is more difficult to find the minimum with simple physical arguments. (9) Neglecting the displacement current. III.aip. and we would expect the resistance to be minimum at t= rrli. the impedance is easily computed. to give the impedance in terms of the E field and its first derivative at the surface of the film: Z= +[E (dE)-l] .. the resistive component of the current in the central region of the film will cancel some of the resistive current flowing near the surface. however.7. 43. This overshoot. Similar considerations apply to the magnitude of the impedance. Thus we may set E =0 outside the film or.----------~ III ~ 40 II: (!) W o ~ 30 N \J. or minimum. but also on the field that existed in the recent past within roughly a mean free path of the point in question. simple derivation (due to Chambers) of the generalized form of Ohm's law that applies when the mean free path is taken into account. perpendicular to the z axis (with the faces at 0 and t).org/termsconditions. we barely see the first half-wave of the interference phenomenon.133.2998 G. it is simply Z=E(O) [j/J(Z)dz}-l (12) and we can use Eq. the neglect of electron relaxation effects implies that the electric field is essentially constant over the time interval l/~.

--_ _ _ \ \ \ \ \ where ~ is the Fermi velocity of the metal. implicitly) for various choices of film thickness..:~-----II. Vol."\. July 1972 [This article is copyrighted as indicated in the article. (a) Phase of impedance per square as a function of classical skin depth (and therefore of frequency. It is difficult to give simple physical arguments for the more pronounced interference effects in the long mean free path case. 4 (a) Magnitude of impedance per square as a function of classical skin depth (and therefore of frequency.aip. "\ . it turns out that the E field is in some sense more lightly damped than in the claSSical case.6 3 1~0~-6~~~~10~-5~~~~IOL-4~~~~IOL--3~~~10-2 8/.133.=2-IO ' . In essence.1. For a film with t» l.=2. 5(a) to film thicknesses shown in Fig.----.1.~ >''''''·_·_·-1=. The phase of the impedance is also shown for a range of oil and til. No. The latter condition is equivalent to the dielectric relaxation time of the conductor being short compared to w. 4(b).------.-----.----¥-----'~-----\.. particularly at low temperatures.--------.1. It is clear that the behavior of the impedance in the long mean free path situa tion is qualitatively simila r to that in the classical region.------.7.1. It is reasonable to require (in order to make electron relaxation unimportant) 2rrj=w« ~/l.239 On: Mon.. 43. 4 and 5. and for most metals represents a less severe restriction than that involving electron relaxation. ~ ~ « if 20 ?0-6 10-4 8/..I. the magnitude of the impedance is given through the dimensionless quantity o-l!Z!. (15a) l>o. Reuse of AIP content is subject to the terms at: http://scitation. expression (14) turns out to be too restrictive at low temperatures where the mean free path becomes longer than the classical skin depth.01 10 8/.. (11) and (13) are presented in Figs. CONCLUSIONS .2999 ANOMALOUS SKIN EFFECT IN THIN FILMS o. although the "overshoot" or "minimum" tends to be somewhat more a more realistic condition is that8 I/. (15b) 11.30.__ o z -. (14) l<o 2rrj=w« ~/(lo)1/2. Phys. (a) ~80. 5.40 N u. The curves presented in Figs. under the restrictions that electron relaxation and displacement current can be neglected. 100 \ \~ '. The "overshoot" becomes more pronounced as conditions become more anomalous (as til becomes smaller).. ~. implicitly) for the film thickness shown in Fig... (b) FIG.=1/4 10 IV. The results of solving Eqs. which is of greatest importance for very thin films. between the 60° and zero limits. (b) Continuation of Fig.1 .----. 4(a) over a wider range. Actually. ~ ffiw 60r--. 4 and 5 correspond to transitions from the dc size-limited region directly to the anomalous skin-effect region. Again the "overshoot" is more pronounced than in the classical case. 0.---~~----.I. the phase goes to zero as expected (no inductive reactance). and finally to the anomalous skin-effect region.1 0. Appl.1. Downloaded to ] IP: 139. and is moderately enhanced over the classical case. (a) ANOMALOUS SKIN EFFECT ASYMPTOTE The over-all features of the impedance of metallic films are established above.L. 2rrj=w« ~/l. (b) Continuation of Fig. (b) FIG. In this case the phase would show a plateau at 45°.ooo. while in the anomalous skin-effect region the phase is asymptotic to 60° in accord with the result that the inductive reactance is greater than the resistance by a factor of . 28 Sep 2015 14:07:08 . 1000 ANOMALOUS SKIN EFFECT ASYMPTOTE as the frequency increases from zero we pass first from the classical dc region to the classical skin-effect region. as a function of oil and parametrized by til. J. 4(a).f3. in a form independent of material. In the dc limit.

Since m takes on N + 1 values. This impedance is readily seen to be independent of the size of the square.30. f«5 x 109 Ts/3 Hz.2 variables. and two real functions by E=f+ig. 9 From the over-all similarity of the numerical calculations to the classical case. symmetric (or other) boundary conditions are easily realized. W.x')g(x') dx' . the symmetric case discussed above applies to a shielded strip line (planar center conductor between two ground planes). the complex equation (11) is first written as two real coupled equations.O>I/ 2(21Tw<T)-1/ 2 (MKS units). The appropriate boundary conditions depend of course on the specific problem addressed. with no approximation. The surface impedance of this square is defined as the ratio of the potential difference between the perpendicular edges (at the film surface) to the total current flowing through the perpendicular edges. c the velocity of light. Equations (A2) form a very convenient starting point for imposition of boundary conditions. and Konnerth. in a 32Kbyte "workspace". HENRY From Eqs. Phys.x')f(x') dx'.ym(2t). It would be efficient to use all of the equations to determine the best-fit solution on the basis of some sort of least-squares criterion. the equations in general have no solution at this point (they are homogeneous equations. Actually. ACKNOWLEDGMENTS The author wishes to thank R. 95114. Detailed consideration has been given to the operation of logic circuitry at low temperatures by Keyes. Finally. APPENDIX The techniques used for numerical solution of Eq.. on the way to the continuum limit N .. we have. and fn=f(x n). July 1972 [This article is copyrighted as indicated in the Downloaded to ] IP: 139. the results for planar films continue to hold approximately for cylindrical films (with the driving fields on the inner and/or outer faces) if the film thickness is small compared to the radius of the cylinder. (A2) where Kmn=K(x m -xn). In such intervals. Note that K(O) diverges. The numerical results presented above are therefore applicable to frequencies of interest in most circuit applications. with W the angular frequency. 43. (11) are essentially those used for integral equations and will only be briefly discussed here. The resulting set of linear algebraic equations is then solved. With this choice we now have 2N +2 inhomogenous equations for the remaining 2N . consider any square of the film with edges parallel and perpendicular to the current flow. For example.239 On: Mon. J. Reuse of AIP content is subject to the terms at: http://scitation. x=z/l. which turns out to be a fairly severe problem. We could not expect a unique solution anyway. In performing impedance measurements by introducing the film into a microwave cavity. one has in the classical case Zone olded(t) '" 2Z. (15) we find for copper at room temperature f« 3 x 1013 Hz. (A1) K(x .00. outgoing plane wave on the other side). We also thank A. we choose in this paper to make E real and unity on both faces of the film. In the work reported here the effective limit appeared to be reached at an average phase change (in the E field) per interval of about 15°. The "overdetermination" is an artifact of the numerical approximation technique: for N = 108 there is "relatively little" overdetermination. at each of the N + 1 points at the ends of the N intervals. San Jose. F. Keyes for pointing out the interest of this problem and for many helpful suggestions. Breaking the film thickness into N intervals.2 equations (by choosing different subsets one gets some feeling for the accuracy. gn=g(x n). Vmn and w" are appropriate weight factors for differentiation and integration. d 2f £lx2 (18) This result depends essentially on the great impedance mismatch between the conducting sheet and free space. respectively. Monterey and Cottle Roads. through the slight scatter in results). In circuitry.aip. and <T the bulk conductivity of the metal. Eq. (16) while at temperatures lower than about 30 oK. it may be possible to draw some rough conclusions about other geometries and boundary conditions from the classical case. Calif. Mayadas and M. Defining a dimensionless coordinate. Extended discussions with E. (17) where the temperature T is in degrees Kelvin.R. for example. or in other words. which ensures nearly perfect reflectivity at the faces. 6 If the boundary conditions are chosen as "one sided" (E specified on one side. the derivatives by polynomial differentiation formulas. increasing the number of intervals increases the number of equations to be solved. K-16. (18). Shatzkes for interesting discussions.2'3 d 2g 3 £lx2 = PIt 1)2 [2 0 (til 2' 1)2 10 II K(x . Harris. All computation was carried out in APL/ 360. and the integrals are approximated by sums. Appl. since no boundary conditions have been imposed. Vol. Dept. It may reasonably be expected that these classical results continue to apply very roughly in the anomalous region. and the relevant determinant need not be zero). approximately. = . In essence. 28 Sep 2015 14:07:08 . 2In nonmagnetic metals the classical skin depth is given by 0= C(21TW<T)-1/ 2 (cgs units). *Present address: IBM Systems Development Division. P. (11) becomes. gl =gN+I =0. Harris on both mathematical and physical aspects of this problem were invaluable. lIn a planar film with current flowing parallel to the faces of the film. As is characteristic in the numerical solution of integral equations. a strip line on a single ground plane involves the "one sided" boundary conditions mentioned above in Eq. we have N + 1 pairs of equations for a nominal 2N + 2 variables.7. No. but not explicitly solved in the anomalous skin-effect limit. in practice we found it convenient simply to solve a subset of 2N . K(O) was replaced by the average value of K over the interval.133. fl = fN +1 = 1. 0 = c(4m.3000 G.

pulsed electron accelerators capable of delivering up to hundreds of thousands of amperes of megavolt electrons for tens of nanoseconds have been developed. V experimental results are compared with 6.ANOMALOUS SKIN EFFECT IN THIN FILMS 3This boundary condition clearly applies in the DC limit. Phys. 1953). Konnerth. Equilibrium configurations for intense relativistic electron beams propagating through such media have been investigated by several authors. The beam drifts down the drift tube and is collected by a Faraday cup whose aperture is equal to the tube cross section. in final form 24 March 1972) Studies of the propagation of pulsed relativistic electron beams through initially un-ionized background gases are reported. a high voltage is applied between a field emission cathode and a thin foil anode. microwaves4 and neutrons. 25 kA) were transported through a variety of gases in order to check the validity of our calculations over a wide range of parameters. much of the beam is lost to the tube walls if the background pressure is too low (e. 13-18 In this paper we present the results of a study of one aspect of the beam transport problem. The theoretical calculations indicate that the low-pressure propagation characteristics of the lower JJ/Y beam are sensitive indicators of the beam's transverse energy. Pippard. 8 In a typical machine configuration. 3 kA) and -1 (350 keV. I.L. 4The precise boundaries chosen are of course somewhat arbitrary and depend slightly on the parameter of interest (e. A useful measure of the effect of an electron beam's self-magnetic fields on the beam electrons' orbits is the ratio lilY. plasma heating. g. 2) are written so as to be valid in both cgs and MKS units. This paper gives an excellent discussion of the anomalous skin effect from first principles. New York. 1954).12 with or without guiding magnetic fields. the initial buildup of ionization and the development of radial force neutralization which occurs when a high-energy electron beam first enters an initially un-ionized background gas. the analytical approach gives essentially the same results and is advantageous because it consumes much less computer time (e.. IV the experimental apparatus is described and in Sec. Reuse of AIP content is subject to the terms at: http://scitation. andJ. Appl. Downloaded to ] IP: 139. it is equal to the ratio of the beam current to the Alfven19 current limit (17 00013y). Poukey Sandia Laboratories. Mathematical models are presented here which evaluate space-charge blowup losses as a function of drift-tube gas pressure. R.W. Numerical simulation and an approximate analytical model are used to describe this beam-loss process. Relativistic Electron Beam Propagation in Low-Pressure Gases* P. INTRODUCfION In recent years. Marton (Academic.B. 'VI. J.. only fractional space-charge neutralization is required for the beam to drift without expansion because the beam's self-magnetic fields tend to constrict the beam. 5 MeV.05 (1.). 1. 1-3 generation of x rays. No. Vol. Beams from these machines are of interest for material studies.H. Keyes. electrons are accelerated through the anode into a drift region in which they are transported to a target. The problem which we address may be described as follows: An electron beam is injected into a gas-filled cylindrical drift tube which is several times larger in diameter than the beam. in this manner. 1. 1 (1952).. 43.aip. Miller. g. Advan. the second consists of solving a differential equation for a single particle trajectory. pronounced beam front erosion occurs owing to the slow buildup of radial force neutralization of the beam electrons.133. Albuquerque. The transport medium which may provide electrostatic and magnetic neutralization of the beam fields may be simply an initially un-ionized gas or a preformed plasma 9. B. The calculations also indicate that this is not the case for the higher JJ/y beam.P. The first consists of a multiparticle numerical Simulation. lE. 9R. 6S. «0. magnitude of impedance. . Sondheimer. In Secs. Ramo and J. andK.239 On: Mon. The results of experiments with two different electron accelerators (p/y ~ O. 5All equations in this paper (except those of Ref. resistance. IEEE. July 1972 [This article is copyrighted as indicated in the article. Fields and Waves in Modern 3001 Radio (Wiley. 12 sec vs 20 min of CDC 6600 time for one waveform). At very low pressures. The front of the beam expands rapidly as the beam propagates due to spacecharge repulsion. New Mexico 87115 (Received 27 January 1972. E.g. In Sec. Beams from two different accelerators with lI/y -0.7 and ignition of fusion reactions. edited by L. The study shows that direct ionization alone is sufficient to explain the experimental results. 28 Sep 2015 14:07:08 . but a self-consistent model of its development which agrees with experimental results has not previously been presented. we choose to retain it at nonzero frequencies as well. 1914 (1970). 5 ion acceleration. Whinnery. p.30. namely. Phys. Actually.1 Torr) to provide rapid space-charge neutralization. Higher gas pressures allow more rapid radial force neutralization and thus more efficient propagation.A. and the calculated Faraday cup current waveforms are compared with measured currents. While it is felt that the Simulation is basically the more trustworthy of the two. 05 and 1) are shown to be in good quantitative agreement with the mathematical descriptions over the pressure range of ~ 1 1.7. Gerardo. Proc. The Larmor radius of an electron in a lilY == 1 beam (with no magnetic neutralization) is about equal to one-half of the beam radius. The necessity of space-charge neutralization for efficient transport has been well recognized for some time. etc. Advances in Electronics and Electron Physics. II and III of this paper we present two conceptually different theoretical analyses of the radial force neutralization problem.1 to ~ 1 Torr in a large variety of gases. 58. Propagation characteristics are found to scale in pressure from gas to gas inversely with the high-energy ionization cross sections of the gases. sA. Vol. Harris.W. New York.