PHYSICAL REVIEW E VOLUME 51, NUMBER 6 JUNE 1995 Ion kinetics and sym~etric charge-transfer collisions in low-current, diffuse (Townsend) discharges in argon and nitrogen S. B. Radovanov, * R. J. Van Brunt, and J. K. Olthoff Electricity Divi8ion, Electronics and Electrical Engineering Labomtory, National Institute of Standards and Technology,Gaithersburg,Maryland 20899 B. M. Jelenkovict Joint Institute for Labomtory Astrophysics, University of Colomdo and National Institute of Standards and Technology, Boulder, Colomdo 80309-0,UO (Received 8 August 1994) Tcanslational kinetic-energy distributions of mass-selected ions have been measured in diffuse, low-current Townsend-type discharges at high electric field-to-gas density ratios (EjN) in the range of 1 x 10-18_2 X 10-11 Vm2 (1-20 kTd). The discharges were generaied in Ar and N2 under uniform-field conditions and ion energies were measured using a cylindrical-mirror energy analyzer coupled to a quadrupole mass spectrometer. The mean ion energies determined from measured energy distributions of Ar+ in Ar and N2+ in N2 are compared with the mean energies predicted from solutions of the Boltzmann transport equation based on the assumption that symmetric reso- nant charge transfer is the predominant ion-neutral interaction. The results for Ar+ and N2+ are consistent with predictions made using a constant (energy independent) cross section for which an effective ion temperature can be defined. However, for both ions, the measured mean energies tend to fall increasingly below the predicted values as EjN increases. The possible causes and significance of the differences between the measured and calculated mean ion energies are examined by consid- ering collisions other than charge-transfer that can affect ion energies as well as uncertainties in the charge-transfer cross sections used in the calculations. Measurements were also made of the relative contributions from N+ and Ar2+ to the ion ftux. Over the EjN range of interest, N+ accounts for less than 15% ofthe ion ftux in nitrogen and Ar2+ accounts for less than 5% of the ion ftux in argon. PACS number(s): 52.40.-w I.INTRODUCTION The importance of resonant charge-transfer processes in determining the kinetic-energy distributions of ions in radio-frequency (rf) discharges has become evident from numerous recent investigations [1-6,8). The results of these investigations for argon have stimulated discussion about the relative roles played by charge transfer ver- sus other elastic and inelastic ion-neutral collisions in af- fecting ion energies within the discharge sheath region [7,9-11). Charge transfer has also been invoked to ex- plain the observed energies of positive ions in the cathode fall region of dc glow discharges [12-17). Velocity distributions of Ar+ in Ar have been mea- sured by Ong and Hogan [18) in a uniform-field drift tube for electric field-to-gas density ratios (EIN) up to .Guest scientist from the Institute of Physics, P.O. Box 57, Belgrade, Yugoslavia. Present address: University of New Mexico, Department of Chemical and Nuclear Engineering, Albuquerque, NM 87131. tGuest scientist from the Institute of Physics, P.O. Box 57, Belgrade, Yugoslavia. Present address: NIST, Time and Fre- quency Division, Boulder, CO 80303. 1063-65lX/95/5 1(6)/6036(11)/$06.00 3.2 X 10-19 Vm2 (320 Td). They found that the mea- sured distributions showed increasing deviations from a Maxwellian distribution as E IN was increased. It has been shown [15,19-21) from solution of the Boltzmann equation in one dimension that if resonant charge trans- fer is the dominant interaction with a constant, energy independent cross section QCT then the ion-energy dis- tribution should be of a Maxwellian form from which an effective ion "temperature" T + can be defined and is given by (1) where e is the electron charge and k is the Boltzmann constant. The assumptions that lead to Eq. (1) will be examined in the next section. The corresponding ion drift velocity is given by - (2) where M is the ion mass. Makabe and Shinada [22) have pointed out difficulties with the ion sampling procedures used in drift-tube ex- periments such as those performed by Ong and Hogan [18), which can lead to distortions in measured ion- 51 6036 @1995 The American Physical Society