Experimental measurements taken in a large magnetoplasma show that a simple double half-turn antenna will excite mϭ1 helicon waves with wavelengths from 10-60 cm. Increased ionization in the center of the downstream plasma is measured when the axial wavelength of the helicon wave becomes less than the characteristic length of the system, typically 50-100 cm. A sharp maximum in the plasma density downstream from the source is measured for a magnetic field of 50 G, where the helicon wave phase velocity is about 3ϫ10 8 cm s Ϫ1 . Transport of energy away from the source to the downstream region must occur to create the hot electrons needed for the increased ionization. A simple model shows that electrons in a Maxwellian distribution most likely to ionize for these experimental conditions also have a velocity of around 3ϫ10 8 cm s Ϫ1 . This strong correlation suggests that the helicon wave is trapping electrons in the Maxwellian distribution with velocities somewhat slower than the wave and accelerating them into a quasibeam with velocity somewhat faster than the wave. The nonlinear increase in central density downstream as the power is increased for helicon waves with phase velocities close to the optimum electron velocity for ionization lends support to this idea.

No abstract

Pt patterns of the 0.25 µ m design rule were etched at 20° C using a magnetically enhanced reactive ion etcher. The main problem of this device integration process is the redeposition of the etch products onto the pattern sidewall, making it difficult to reduce the pattern size. In both cases using a photoresist mask and an oxide mask, the redeposits of the etch products onto the sidewall were reduced by the addition of Cl2 to Ar, although the etch slope was lowered to 45°. Using the oxide mask, by adding O2 to the Cl-containing gas, the etch slope was increased up to 70°, and the redeposits were removed by an HCl cleaning process.

We developed a new method to enhace the photoresist selectivity in SiO2 etching by modulating both the source and bias powers and by controlling the phase difference between the modulation functions. Enhancement of mask selectivity was observed in the pulse plasma, especially in the out-phase condition. To understand the heavy polymerization in the out-phase pulse plasma, we analyzed the ion energy distributions of CF x +(x=1, 2, 3) ions using the energy-spectroscopic quadrupole mass spectrometer (QMS) and measured the waveforms of the bias power with a high-voltage probe which was connected directly to the wafer. Two distinct plasma potential distributions were obtained in the pulse plasma and the dc bias voltage (V DC) was maximum in the out-phase condition. The heavy polymerization in the out-phase condition was explained as a result of high V DC. We also investigated the emission intensity of the C2 (516.5 nm) line, and found that C2 species were precursors of the polymerization and contributed to the heavy polymerization in the out-phase condition.

The instabilities caused by the reflected rf power in a pulsed-plasma operation employing modulated rf power was studied. By suppressing the side-band modes in the frequency domain, the pulsed plasma became more stable and produced less reflected power. The mode-suppressed pulsed plasma showed almost the same plasma characteristics as the conventional step-function-modulated pulsed plasma. The mode-suppressed plasma was applied to etch a polysilicon pattern. The etched polysilicon profile showed no charge-up defects, suggesting that the mode-suppressed plasma can be utilized for controlling the electron temperature in a more stable operation.