We demonstrate a simple method to identify noise sources in electron-beam systems and accurately quantify the resulting errors in feature placement. Line gratings with a 46 nm average pitch were patterned with electron-beam lithography (EBL) and measured with transmission x-ray diff raction (XRD) and scanning electron microscopy (SEM). All SEM micrographs were analyzed in Fourier space to facilitate comparison with XRD data. Di ffraction profi les and Fourier transforms of SEM micrographs contained numerous "satellite" peaks, meaning weak peaks adjacent to the strong primary nodes, that are characteristic of periodic extensions and compressions in the grating pitch. The wavelength and amplitude of these pitch variations were calculated with a simple scaling law by comparing the positions and intensities of satellite peaks relative to their neighboring primary nodes. This approach is remarkably easy to implement because it does not require any modeling of electron density pro les. Data were used to calculate the frequency of each noise source and the resulting variations in grating pitch. Two persistent noise frequencies were detected in the tool studied, (62 2) Hz and (86 +/- 3) Hz, and the tool manufacturer identifi ed likely noise sources as electromagnetic and mechanical in nature, respectively. The 60 Hz noise produced errors in a 46 nm grating pitch of 3s = 1.5 nm, where s is the standard deviation in the grating pitch. Errors due to the 86 Hz noise ranged from 3s = 1.5 nm to 2.5 nm. Variations of these magnitudes can be expected to have adverse e ffects on coupling efficiencies, cavity quality factors, and center wavelength values in photonic devices.
Citation: Journal of Vacuum Science and Technology B
Pub Type: Journals
electron-beam lithography, noise, photonics