12 - - 21 . 3 - 12 d :: * . . . . 1.: : E 0. L . : 2011, ::. . .. . .. . ' .. .. OP . . . 14 . 1 23 * I OF I ORNL P 1539 . + . 1 ri . 171 . homenagement a 1,0. 1 l • } E . * . . T ...! . w i A stres s on the that t h ese lines in the intervenir en cuenta 10 4 7 • . - 1 · 1 • . i O - . - - - - . 1 14 . . 1 . I' . . - 1 A. : . 36 . . N S ? * 7. .... MICROCOPY RESOLUTION TEST CHART NATIONAL BUREAU OF STANDARDS -1963 '. .. . BRN VA-1539 SEP 20 1965 MOVING STRIATIONS PRODUCED BY A SIMPLE THERMAL INSTABILITY* Igor Alexeff Oak Ridge National Laboratory ABSTRACT A simple thermal instability is investigated as a possible cause of moving striations in discharge tubes. The mathematical treatment predicts which gases should and which should not exhibit the insta- bility. The frequency of oscillation of a cylindrical discharge, due to the instability, also is predicted. Not discussed is the velocity with which the instability appears to propagate longitudinally down the discharge. The theoretical predictions are compared with the experimental striation data of several authors, and reasonable agreement is found. -LEGAL NOTICE - I This report sa prepared as an accoud of hororubent sponsored work. Neither the Vatted de Manus wa nurudy or prestation, and or lapu do wa rupect to demor may, con Mon, a wobes at the marimidos com bond in de report, or the doma ad my wa, mount, where a proces declarda due mu y box bedring petredeky omal rights; or 8. And y Wa Madina se rupect to be made of her deg med tre 200 od ng balormatics, apperstes, method, or grocese diacioned in this report Mwadudu won, partan xtera mea leden helt m maryna contro de Conducta, at maphored me conductor, bu o de soch saployee or contractor of the Commission, or employee of such contractor properes, domain, or provides second to, uy tornada perb We progna contract vol du Coulalahon, a HI toplog via ma contractue, RELEASED FOR ANNOUNCEMENT IN NUCLEAR SCIENCE ABSTRACTS *Research sponsored by the U.S. Atomic Energy Commission under contract with the Union Carbide Corporation. ta This paper was rubmitted for publication in the open Uiterature at least months print to the issuanco date of this Micro- an. Since the U.S.A.B.C. has no evi. der.ce that it baw boon published, the pa- per is being distributed in Microcard form as a proprint. 1 I. INTRODUCTION Moving striations are bright concentrations of light that move at high speed through discharge tubes. Typical moving striations * are shown in the image-converter photograph of Fig. 1. Although moving striations bare been observed for many years and have been theoretically discussed by numerous authors, a simple, complete theory apparently Fring here does not exist. The most accepted theory is that of Pekerek,“ who describes a striation as a progressive ionization ox breakdown front. On this basis he obtains a dispersion relation which qualitatively predicts the striation phase and group velocities. In this paper we do not discuss the velocity of moving striations. Instead, mathematical expressions are obtained that predict which VA geses should exhibit a thermal instability and what the frequency of the instability should be. The theoretical predictions concerning the instability are compared with the experimental observations of striations of several authors, and fair agreement is found. II. THEORETICAL DISCUSSION The basic, new assumption in this investigation is that moving striations are primarily & radial, not a longitudinal phenomenon. This assumption is based on the striation structure, as shown in Fig. 1. The spacing of the striations is generally greater than several radii of the discharge tube. Therefore, the main plasma loss must be radially to the wall, since the plasma is not confined in any manner. Consequently, a radial oscillation process is assumed, in spite of the apparent longitudinal motion of the striations. If striations are caused by radiai cscillations, then an investigation of a small section of the 43charge tube should be sufficient to describe the phenomenon. Consider a section of the discharge tube so short in length that within it the plasma conditions can be considered uniform. For the sake of simplicity, we replace functions that depend on radius by suitable average values. The rate . . of heat input per unit volume to the electron gas is given by jºo , where oe is the electrical conductivity of the discharge. The rate of heat loss from the electron gas 18 given by a Of T, where a 18 a. constant, Op is the thermal conductivity of the electron gas, and T is the electron temperature. Let B be the ratio of heat input to heat loss, For o and Ome one substitutes the theoretical values found in the literature, 3 Here e is the electron charge, n is the electron density, L is the electron's mean free path in the gas, m is the electron mass, v is the electron average velocity, and k is Boltzmann's constant. For L we substitute (NO), where N is the gas density, and O is the electron- gas molecule scattering cross section. The quantity B then is expressed 88 (2) Within the discharge, two quantities are conserved: first the current density ), because the discharge tube is fed by a high-impedance source; and second, the neutral gas density N, because the degree of ionization 18 low, and because striation velocities can exceed sonic speeds giving the gas no time to pove. In the above equation for B the quantities in the parenthesis are constants. The scattering cross section O is a direct function of electron velocity, and hence electron temperature. However, for temp- . erature increases of short duration compared to ion loss processes, the electron density n in the discharge plasma in proportional to the lategral with respect to time to the ion generation rate (T), tist 26/0 (T) dt. (3) . Thus if the electron temperature T increases suddenly, o responds immediately while a responds more slowly. Conversely, if I decreases O again responds immediately, while n responds only at a rate : corresponding to ion loss processes. We are now ready to discuss the conditions for thermal instability in the discharge tube. Assume that o increases more rapidly than T/. Equation (2) then reveals that the discharge is unstable. A slight perturbation that increases the local temperature T results in more heat input relative to heat 1088, and the temperature risee rapidly 10 a "runaway" fashion. Conversely, a slight decrease in the local temperature results in a catastrophic drop in temperature. However, 18 o increases less rapidly than T/C, any temperature perturbation corrects itsell and the discharge 18 stable. . Buable. Knowing the basic cause of the thermal instability, we can infer the rest of the oscillation cycle. The electron temperature T first rises in a runaway fashion until it reaches some limiting value, perhaps caused by n eventualiy becoming large enough to bring B equal to unity. However, even when T ceases to rise the integral form of n given by Eq. (3) means that a still rises. Therefore Eq. (2) shows that ß must finally become less than unity. The thermal instability now reverses. The lower T becomes the less the heat input relative to the heat loss, and a catastrophic cooling occurs. The result is a cold, dense plasma. This cold, dense plasma persists until the elec- tron density a drops sufficiently for B again to become greater than unity. The cycle then repeats. Thus, the cycle of the thermal instability is a sudden rise and fall of the electron temperature, coupled with a sudden rise and slow Fig. 2 hori decay of the electron density ag is illustrated in Fig. 2. This theoretical cycle corresponds closely to the striation cycle . experimentally observed. 4,5 The frequency of oscillation of the thermal instability is determined by the slowest process present in the cycle. This appears to be the rate of loss of the ions. For pressures less than a few torr, the main ion loss is by radial ambi-polar diffusion. The basic differential equation is : . : Here k is Boltzmann's constant, T is the electror temperature, m, is the lon mass, a is the lon density, v. is the lon-molecule collision rate, and t refers to time. For a cylindrical tube of radius ro, the time t required for the lon density to drop x factors of e 18 given by t = X 5.78 (5) kr. . The frequency of oscillation is the reciprocal of the above quantity. In comparing this equation with experiments for which x was not measured, a reasonable approximation is to set x equal to unity. In summary, the discharge tube exhibits a strong thermal instability if operating in a regime where the scattering cross section for electrons on gas atoms increases more rapidly with electron temperature then T/". The frequency of oscillation v is given by 1* here III. COMPARISON WITH EXPERIMENT A. Stability of Various Gases The discussion presented above gives clear-cut predictions as to which gases should and which should not show the thermal instability. However, one must not be surprised if these predictions only separate those gases which easily produce striations from those which produce striations with difficulty. Our simplified treatment does not contain - WD such complicating effects as the production of metastable atoms and . contamination of the gas. Another possible complication is that . . . . earlier workers oxy cave coufused moving striations with oscillatory phenomena originating at the anode of the discharge. Finally, the - - - thermal instability may be only one of several striation mechanisms. Predictions concerning the stability of various gasses are presented in Table 1, along with comments concerning the observations of moving striations. Diatomic gnses ere not discussed to avoid difficulties caused by molecular dissociation. For determining o. a standard reference of cross section data' suffices, except where noted. & Table I here Discussich of Comments from Table 1: 1. In helium we bave never observed clear-cut moving striations, although weals, strongly damped fluctuations were seen moving away from the anode. On the other hand, Astou and Kikuchiº have noted striations in helium. In neon the curve of o vs T rises in a stepwise fashion in the energy range of a few electron volts. Thus, one has regions of stability and instability, depending on the average electron temperature at which the discharge operates. In mercury vapor Fouldsu bas observed moving striations, although they were irregular. On the other hand, Crawford has not:: observed moving striations in mercury vapor 'under: the same conditions. 4. In cesiun vapor Robertson has never observed moving striations. This observation agrees with the theoretical prediction of stability as obtained from the latest croso section data. 13 B. Predicted and Observed Frequencies Predicted frequencies of oscillation vn are compared with observed frequencies v. in Table 2. The experimental data in Table 2 is intended to be a sample of the more recent and more complete published material. The ion-atom collision rate v. used in Eq. (6) 18 computed from the ion-atom collision cross section, assuming that both the ions and the gas atoms are at room temperature. * Table & bine Discussion of Comments from Table 2: 5. The paper of Stewarts contains the most complete set of data : for the requirements of Eq. (6). 6. The results of Sicha'' show a fluctuation of electron density during the striation cycle of only 20%. This result is smaller by a factor of about 50 than the results of Stewart," Paik, et al., ') and Sodomsky. Only if we assume that & . density fluctuation of e occurs will the theoretical frequency be in fair agreement with the experimental frequency. 7. 'At the highest pressure used by Donahue and It?ke, the main loss of plasma is possibly by recombination in the gas rather than by wall 1088e8, Recombination in the inert gases has been observed at pressures above a few torr. In the : presence of increased plasma losses, the observed oscillation frequency should be higher than that predicted in Eq. (6) as 18 indeed observed. A quantitative measure of the agreement between theory and experiment is obtained by computing an averape factor by which yo differs from w. Since one does not know if the predicted or the observed frequency is correct, the frequency difference is compared to the average frequency to obtain the frequency ratio f, f = (x - ) / 3/2 (7) + %)] . Omitting the anomalous-appearing results of Sichat and the highest pressure result of Donahue and Dieke, 15 we find the rms value for f to be + 0.383 and the most probable value to be $ 0.258. Using the most probable value for f we work backwards through Eq. (7) to find 1.7 > V/V> 1/1.7. The above statistical discussion shows that the predicted and observed frequencies shown in Table 2 agree fairly well. Remember that, in computing the predicted frequencies from Eq. (6), we must generally estimate some of the variables. In general, we do not know the ion decay constant x, the electron temperature T, and the 100-gas atom collision rate v. (which depends on the rarely measured gias terperature). . -10 -10 IV. SUMMARY This paper presents a simple model of a thermal instability which may drive moving striations. The model predicts both which gases should be unstable and what the frequency of the instability should be. Fair agreement is found between the predictions of the model and the exper- imental observations of moving striations by various authors. No attempt is made in this paper to derive the apparent longitudinal velocity of propagation of the instability. However, earlier experi- mental work by usº now suggests that the instability generally progresses down the discharge tube at a rate determined by ambi-polari diffusion (see Eq. 4). . . . ACKNOWLEDGMENT The author acknowledges several very useful discussions with Dr. W. D. Jones. «ll. Table 1 THE EXISTENCE OF THE THERMAL INSTABILITY IN VARIOUS GASES Average Electron Temperature Stability Prediction Striation Production Gas Comments stable difficult 3 eV I ev i ev indefinite easy unstable easy J. ev unstable easy 1 eV . unstable.. easy stable difficult 1 ev 0.2 ev ma stable not observed -12- Table ? THE PREDICTED FREQUENCY OF THE THERMAL INSTABILITY COMPARED WITH THE OBSERVED FREQUENCY OF MOVING STRIATIONS D torr Author Ges со ev CP8 cp8 Comments Author_ Stewart5 Gas A o p om torr .856 e eV 1.1 x 2.2 vp cps 920 20 c ps Comments 1,000. 5 Paik, A 0.8 1.3 1.0* 1.0+ 2.3 2.3 17,500 10,800 31,000.- 3,800 et al.15 Sodomsky 16 Siche i7 NE Ne Ne 1.05 zl.O 21.0 0.5 1.04 1.9 2.9 1.04 0.2 1.1 1.0% 0.2 22,000 8,100 21,300 12,000 1,600. 16,500 . 6 6 Donahue, A ; 6,800 et al. 18 c.75 0.75 0.75 2.1 4 12.0 1.04 ?.0* 1.0* 1.04 1.0* 1.0* 3,250 A 1,190 5,000 1,333 1,500- 3,500 2,500 0.75 30 1.04 1.0* 477 - 7 Bletzinger, Ne 1.25 1.0 1.0* 1.0* 15,000 3,200 ' et al. 19 - *Value not published, assumed by us. -13- FIGURE CAPTIONS · Fig. 1 Image-Converter Photograph of Moving Striations. The striations move from the anode on the right to the cathode on the left. The gas 18 krypton at about 30w pressure. Fig. 2 Deduced Variation of Electron Temperature and Ion Density During the Thermal Instability Cycle. in a Fixed Element of Volume. -14. . REFERENCES 1. Druyvesteyn, M. J., Physica 1, 273 and 1003 (1934). Armstrong, E. B., K. G. Emeleus and T. R. Neill, Proc. Roy. Irish Acad. 54A, 291 (1951). Watanabe, S., and N. L. Oleson, Phys. Rev. 99, 1701 (1955). Robertson, H. 8., Phys. Rev. 105, 368 (1957). Nedospasov, A. V., J. Tech. Phys. (USSR) 3, 153. (1958).;::. Chapnik, I. M., Soviet Phys. (JETP) 34, 1033 (1958). Wojaczek, K., Beiträge aus der Plasmaphysik (E. Germany), 1., 1 (1962); 2, 50 (1962); 2, 56 (1962); 2, 122 (1962); 2, 134 (1962). General Review: Iruyvesteyn, M. J., and F. M. Penning, Revs. Modern Phys. 12, 87 (1940). Francis, G., Handbuch der Physik XXII, 137 (1956). - * 2. Pekarek, L., and V. Krejci, Czech. J. Phys. B 12, 450 (1962); B 13, 881 (1963); Proc. 5th Int. Conf., Ionization phenomena in Cases, Vol. 1, H. Maecker, Ed. (North-Holland, Amsterdam, 1962), p. 573. 3. Joos, G., Theoretical Physics, 2nd Ec. (Hafner, New York, 1950), pp. 445-447. 4. Pupp, W., Physik 2. 36, 61 (1935). 5. Stewart, A. B., J. Appl. Phys. 27, 911 (1956). 6. Cooper, A. W., J. Appl. Phys. 35, 2877 (1964). 7. McDaniel, E. W., Collision Phenomena in Ionized Gases (Wiley, New York., 1964), pp. 116 and 117. 8. Aston, F. W., and T. Kikuchi, Proc. Roy. Soc. (London) A 98, 50 (1920-21). 9. Normand, C. E., Phys. Rev. 33, 1217 (1930). 20. Foulds, K.W.H., J. Electron. Control (GB) 2, 270 (1956-57). 21. Crawford, F. W., and H. R. Pagels, 'M. L. Report No...962, Microwave Laboratory, Stanford University, October, 1962 (unpublished). A - 2 19 17 12. Pobertson, H. 8., and M. A. Hakeen, proc. 5th Int. Conf.: Ionization Phenomena in Cases, Vol. 1, H. Maecker, Ed. (North- Holland, Amsterdam, 1962), p. 550. 13. Garrett, W. R., and R. A. Mann, Phys. Rev. 130, 658 ! 1.963). 14. McDaniel, E. W., Collision Phenomena 1.n Ionized Gases (Wiley, New York, 1964), p. 439. 15. Palk, 8. F., J. N. Shapiro, and K. D. Gilbert, J. Appl. Phys. 35 2573 (1964), 16. Sodomsky, K. F., J. Appl. Phys. 34, 1860 (1963). Sicha, M., Czech. J. Phys. B 13, 499 (1963). 18. Donahue, F., and G. . Dleke, Phys. Rev. 81, 248 (1951). 19. Bletzinger, P., and A. Garscadden, J. Electron. Control (GB) 16, 269 (1964). 20. Mulcahy, M. J., and J. J. Lennon, Proc. Phys. Soc. (London) 80, 626 (1962). Jones, W. D., and... I.Alexeft, Bull. Am. Phys. Soc. 9, 469 (1964). 21. 4 . . . 1 FIG. 1. . , ... : PION DENSITY C 1/ - ELECTRON TEMPERATURE t . . Y 3 TIME DENSITY AND ELECTRON TEMPERATURE . AS A FUNCTION OF TIME ::: i. .Fia. 2;. : .... : .:: . . € . C . END - + = DATE FILMED 9 / 13 / 66 : L ')