Braginsky, Gorodetsky, and Vyatchanin have shown that thermo-refractive fluctuations are an important source of noise in interferometric gravitational-wave detectors. In particular, the thermo-refractive noise in the GEO600 beamsplitter is expected to make a substantial contribution to the interferometer's total noise budget. Here we present a new computation of the GEO600 thermo-refractive noise which takes into account the beam's elliptical profile and, more importantly, the fact that the laser beam induces a standing electromagnetic wave in the beamsplitter. The use of updated parameters results in the overall reduction of the calculated noise amplitude by a factor of about 5 in the low-frequency part of the GEO600 band, compared to the previous estimates. We also find, by contrast with previous calculations, that thermo-refractive fluctuations result in white noise between 600 Hz and 39 MHz, at a level of $8.5\cdot 10^{-24}$Hz$^{-1/2}$. Finally, we describe a new type of thermal noise, which we call the thermo-chemical noise. This is caused by a random motion of optically-active chemical impurities or structural defects in the direction along a steep intensity gradient of the standing wave. We discuss the potential relevance of the thermo-chemical noise for GEO600.