I wonder why you need these clean substrates. In the case of thin films deposited onto substrate the transmittance of the system film/substrate is not a product of transmittance of the substrate and transmittance of the film, so using blank glass in the reference beam does not lead to make a correction for the substrate.
For free standing films air should be reference but when films are deposited on substrates, measurements of absorbance with respect to air reference arise artifacts (as the substrates has some absorption in visible range, if though it is appeared transparent, or in some other range). This correction becomes vital when absorption of substrates and films are in same region. So in measurement of absorption of thin films deposited on transparent substrates correction for the absorption of substrates are necessary.
Use 2 slides manufactured in the same batch and cleaned together in the same way. This way, theoretically, you would have similar defects. Use one as blank and the other modified. The compensation should be pretty good and the absorbance should come only from your thin film.
Just make sure that if you are using transmission for the thin film, you are in the domain where glass is transparent. UV cuttof for glass is around 365 nm.
If you expect absorbance below 365 nm, use two quartz slides. I always use SUPRASIL quartz slides and they are virtually transparent all the way down to 200 nm.
My point was that if you want to know the transmittance of the free standing film it is not correct to measure the transmittance of the film/substrate system and divided it on transmittance of the bare substrate (this is what you are doing when you put the bare substrate into the reference beam). The reason is the muliple reflectances at each boundary and their interference. Hopefully it is more clear now
Just one simple example - SiO2/glass, the thickness of SiO2 is around 100 nm. Because of the particularly chosen optical thickness your sample (film/glass substrate) will have transmittance values higher (let's say around 96%) than the Transmittance of bare substrate (let's say 92%). If you put the sample in the objective beam and the bare substrate in the reference beam of the spectrofotometer the measured signal will be T=96/92 = 104%, that does not make sense. Regards