Vacuum ultraviolet (VUV) photons in plasma processing systems are known to alter surface chemistry and may damage gate dielectrics and photoresist. We characterize absolute VUV fluxes to surfaces exposed in an inductively coupled argon plasma, 1–50 mTorr, 25–400 W, using a calibrated VUV spectrometer. We also demonstrate an alternative method to estimate VUV fluence in an inductively coupled plasma (ICP) reactor using a chemical dosimeter-type monitor. We illustrate the technique with argon ICP and xenon lamp exposure experiments, comparing direct VUV measurements with measured chemical changes in 193 nm photoresist-covered Si wafers following VUV exposure.

We present a comparison of blanket 193 nm photoresist (PR) roughening and chemical modifications of samples processed in a well-characterized argon (Ar) inductively coupled plasma (ICP) system and an ultra-high vacuum beam system. In the ICP system, PR samples are irradiated with Ar vacuum ultraviolet (VUV) and Ar ions, while in the vacuum beam system, samples are irradiated with either a Xe-line VUV source or Ar-lamp VUV source with Ar ions. Sample temperature, photon flux, ion flux and ion energy are controlled and measured. The resulting chemical modifications to bulk 193 nm PR and surface roughness are analysed with Fourier transform infrared (FTIR) spectroscopy and atomic force microscopy. We demonstrate that under VUV-only conditions in the vacuum beam and ICP (with no substrate bias applied) systems 193 nm PR does not roughen. However, roughness increases with simultaneous high energy (>70 eV) ion bombardment and VUV irradiation and is a function of VUV fluence, substrate temperature and photon-to-ion flux ratio. PR processed in the ICP system experiences increased etching, probably due to release of H- and O-containing gaseous products and subsequent chemical etching, in contrast to samples in the vacuum beam system where etch-products are rapidly pumped away. The surface roughness structure and behaviour, however, remain similar and this is attributed to the synergy between VUV-photon and positive ions.

Vacuum ultraviolet (VUV) photons are known to modify the bulk chemical composition of 193 nm photoresist, typically penetrating ∼100 nm and depleting carbon–oxygen bonds. Fourier transform infrared transmission measurements as a function of VUV photon fluence demonstrate that VUV-induced bond breaking occurs over a period of time. We present a model based on the idea that VUV photons initially deplete near-surface O-containing bonds, leading to deeper, subsequent penetration and more bond losses, until the remaining near-surface C–C bonds are able to absorb the incident radiation. Fitted model photoabsorption cross-sections compare well with the literature values.

Articles you may be interested inWafer heating mechanisms in a molecular gas, inductively coupled plasma: in situ, real time wafer surface measurements and three-dimensional thermal modeling

We present a comparison of patterned 193 nm photoresist (PR) line width roughness (LWR) of samples processed in a well characterized argon (Ar) inductively coupled plasma (ICP) system to RMS surface roughness and bulk chemical modification of blanket 193 nm PR samples used as control samples. In the ICP system, patterned and blanket PR samples are irradiated with Ar vacuum ultraviolet photons (VUV) and Ar ions while sample temperature, photon flux, ion flux and ion energy are controlled and measured. The resulting chemical modifications to bulk 193 nm PR (blanket) and surface roughness are analysed with Fourier transform infrared spectroscopy and atomic force microscopy (AFM). LWR of patterned samples are measured with scanning electron microscopy and blanket portions of the patterned PRs are measured with AFM. We demonstrate that with no RF-bias applied to the substrate the LWR of 193 nm PR tends to smooth and correlates with the smoothing of the RMS surface roughness. However, both LWR and RMS surface roughness increases with simultaneous high-energy (⩾70 eV) ion bombardment and VUV-irradiation and is a function of exposure time. Both high- and low-frequency LWR correlate well with the RMS surface roughness of the patterned and blanket 193 nm PR samples. LWR, however, does not increase with temperatures ranging from 20 to 80 °C, in contrast to the RMS surface roughness which increases monotonically with temperature. It is unclear why LWR remains independent of temperature over this range. However, the fact that blanket roughness and LWR on patterned samples, both scale similarly with VUV fluence and ion energy suggests a similar mechanism is responsible for both types of surface morphology modifications.