Run-to-Run Evolution of Fluorocarbon Radicals in C[sub 4]F[sub 8] Plasmas Interacting with Cold and Hot Inner Walls

Abstract

In the ultralarge scale integrated circuit (ULSI) multilevel metal-lization process, high-aspect-ratio contact/via hole etching are now studied intensively employing high-density fluorocarbon plasmas. However, the "etch stop" which occurs occasionally in long-term runs causes serious trouble in the production line. This phenomenon is presumed to result from changes of plasma parameters due to the fluorocarbon polymer film deposition on the inner wall of a reactor. Since the SiO 2 etching kinetics follows the ion-induced reaction, a slight increase of the polymer film thickness on the SiO 2 surface, even of several atomic layers, significantly reduces the etch rate, 1 causing etch stop. To date several studies have reported on plasma-wall interaction: (i) the temperature rise of the reactor inner wall increases CF x (x 1-3) radical densities, 2 and (ii) polymer deposi-tion on the wall does not occur when heated up to 200C. 3 (iii) Higher CF 2 radical density was observed by infrared laser absorption spectroscopy (IR-LAS) near the wall surface. 4 Laser induced fluo-rescence (LIF) also revealed that (iv) CF radical density was decreased on the heated and seasoned wall, 5 and (v) other polymer precursors in addition to CF x (x 1-3) radicals may exist in the gas phase. 6 (vi) Adequate control of the reactor wall temperature improves oxide-to-nitride/silicon etch selectivity. 7 In spite of many studies, time-dependent variations of plasma parameters in conjunction with wall temperature have not been studied sufficiently. In this paper we have systematically investigated the time evolution of wall temperatures, total pressures, and CF x (x 1-3) radical densities in C 4 F 8 plasmas interacting with both cold and hot inner walls, and discuss correlations between them. Experimental Experimental apparatus used in this study are depicted in Fig. 1. Inductively coupled plasmas (ICPs) were generated using either a bell-jar-type source or a planar type one. In the bell-jar-type source used in the first half of this paper, plasmas were generated by 13.56 MHz radio frequency (rf) power supplied to a single-turn antenna wound around a 135 mm quartz bell jar which was connected to a stainless steel (SUS) made reactor. A 150 mm inner wall SUS tube whose temperature was controlled by circulating water from room temperature to 80C was inserted into the reactor. The surface area ratio of the quartz bell jar to the inner wall was approximately 1:3. In the case of the planar-type source described in the latter half, an rf power was supplied by a single-turn antenna which was set on the quartz plate. The inner wall could be heated up to 200C by a resistive heater. Wall temperatures were measured by a thermocouple fixed on the wall. For radical measurements a quadrupole mass spectrometer (QMS; HIDEN EQP-300) was installed so that its sampling orifice was located 15 cm away from the antenna. QMS's head inside the chamber can be heated up to 200C as well as the reactor wall. After replacing the QMS with the load-lock type sample stage (see Fig. 1c), etching experiments were carried out at the same position as the radical measurement. The sample stage was water cooled and coupled to the 400 kHz HF generator through the blocking capacitor. Total pressure in the reactor was monitored using a diaphragm manometer. Preliminary experiments using infrared absorption spec-troscopy revealed that the molecular density in the reactor was inversely proportional to wall temperatures. Therefore molecular density was evaluated from both pressure and wall temperature values. Absolute radical density was estimated by appearance mass spec-trometry (AMS) 8 using reported ionization cross-sectional data. 9,10 As for the value of ionization cross section of CF 3 from C 4 F 8 , which proved to be overestimated, we adopted the value of 1.14 Å 2 which is calibrated via the evaluation of the apparatus function of the QMS. We assumed that the cross section is not affected by the temperature rise of the inner wall, while the temperature dependenc of both C 4 F 8 molecular density n C4F8 used for calibration and orifice conductance C were taken into consideration. These are expressed by n C4F8 1/T and C (T/Mw) 1/2 , respectively, where T and Mw are the wall temperature and molecular weight, respectively. Ionization voltage of the QMS was calibrated according to the ionization potential of Ar at 15.8 eV. 11 Radicals were ionized in the ionization cage just after passing through the sampling orifice, and voltages of several lens electrodes in the QMS were carefully tuned to realize collision-free transport of ionized radicals to the detector. Thus free radicals inside the plasma can be directly analyzed using this measurement system. In addition, for the compensation of the degraded sensitivity of the secondary electron multiplier of QMS, calibration was done using the fragment signal intensity from C 4 F 8 gas at the ionization electron energy of 70 eV before and after each measurement. Unless otherwise noted, oxygen plasma ashing was carried out before each experiment , and C 4 F 8 plasmas were generated under the condition of the rf In order to clarify the cause of poor reproducibility of the high-aspect-ratio SiO 2 contact/via hole etching using high-density plasmas , interactions between radical species in fluorocarbon plasmas and the inner wall of the reactor have been studied. In this paper time-dependent evolution of wall temperatures, total pressure, and CF x (x 1-3) radical densities in C 4 F 8 /ICP (inductively coupled plasma) generated in both cold and hot wall reactors have been investigated and their correlations are discussed. The temperature rise of the reactor wall during plasma operation causes the decomposition of fluorocarbon polymer films which are formed from higher order fluorocarbon radicals as well as lower order CF x (x 1-3) radicals into lower order radicals, and consequently plasma parameters such as the pressure and radical densities fluctuate. Though intensive water cooling of the whole reactor achieved good reproducibility, a thick polymer film deposition on the wall causes another problem of particle contamination. On the other hand, wall heating at high temperatures actually suppresses polymer deposition, but it has been found that the conversion from higher order radicals to lower order radicals still occurs on the wall. The CF 3 radical is predominantly produced in cold wall reactors, while the CF 2 radical appeared rapidly with wall temperatures of 220-240C, and the CF 2 density equaled the CF 3 radical density around 250C. This implies that an excess amount of both radicals possibly influences the gas-phase chemistry in the plasma and degrades the reproducibility even in a hot wall reactor.