A study is undertaken of the loss mechanisms of H and Cl atoms in an inductively coupled plasma used for the etching of III-V materials for photonic device fabrication. A better understanding of these mechanisms may allow us to refine our previous kinetic models of Cl2/H2-based plasma used to anisotropically etch InP-based devices [1], and be useful for monitoring the reactor walls state. The study is also of interest for Cl2/HBr-based plasma chemistries.

The plasma phase is diagnosed using a time-resolved optical emission spectroscopy technique often referred to as pulsed induced fluorescence (PIF). In previous PIF studies, the plasma is pulsed with a standard TTL signal and a short probing pulse (0.05 to 1 ms) is added to scan the afterglow. In our case we extract the fluorescence signal directly from the rising edge of the plasma ignition and therefore only standard pulse operation of the RF generator is required.

In principle the evolution of various radical densities in the afterglow could be followed with the PIF technique by varying the duration of the interval between two successive pulses (the off time). In this study we monitor the evolution of hydrogen and chlorine radicals to deduce their surface recombination coefficients. The plasma is pulsed with an off-time ranging from 200 µs to 100 ms. The on-time is chosen for steady state conditions to be reached. The targeted pressure value lies in the range of 0.5 mTorr to 10 mTorr.

In the case of hydrogen, comparing the increase rate of H2 to the decay rate of H allows us to estimate if surface recombination mechanisms other than H(g) + H(s) -> H2 have to be considered. We therefore monitored both the H-alpha (656.3 nm) the Fulcher-alpha (602nm) line of H2. By pulsing the plasma with long off-times we can verify that emission from dissociative excitation of H2 is negligible under our experimental conditions.

The typical recombination coefficient of H in our reactor, with SiOAlCl passivated walls, has been found to be around 0.01. More interestingly the growth rate of H2 is at least two times higher than the decay rate of H. We will therefore discuss two possibilities that may explain this experimental observation:

1) the existence of a competing loss mechanism for H2;

2) the existence of another loss mechanism for H-atoms. This possibility was proposed in a study of the side-wall passivation of InP etched with Cl2/H2 chemistry [1], where H(g) + Cl(s) -> HCl was assumed to favor the removal of Cl from the passivation layer.

Finally, the PIF technique is evaluated for the first time in chlorine plasmas to deduce the surface loss coefficient of chlorine atoms.

 [1] L. Gatilova et al, JVST A 27 (2009) 262