@Max Planck,
Whilst I do not disagree with your comment, it does need to be placed in the context of whether a thermal camera Microbolometer array will operate satisfactorily without the vacuum being present. I would argue that the performance in such a circumstance is so poor as to render the camera pretty much useless. If wishing to image only hot items is the only requirement, fine, but most thermal cameras are required to image both low and high temperatures.
[...]
Fraser
Fraser,
I just commented on the incorrect information given by Bill W.
I also think I commented earlier clearly on the vacuum importance.
If you are interested, the key element here is in fact the thermal conductance of the membrane (pixel) - which is the sum of the thermal conductance through the supporting legs Glegs and the thermal conductance through the air Ggas (if no vacuum) - and the thermal time constant of the pixel. Given typical geometry of actual microbolometers, if vacuum is lost, Ggas becomes roughly two orders of maignitude bigger than Glegs. You can find a good description in the articles below:
He, X., Karunasiri, G., Mei, T., Zeng, W. J., Neuzil, P., & Sridhar, U. (2000). Performance of microbolometer focal plane arrays under varying pressure. IEEE Electron Device Letters, 21(5), 233–235. doi:10.1109/55.841306
Eriksson, P., Andersson, J. Y., & Stemme, G. (1997). Thermal characterization of surface-micromachined silicon nitride membranes for thermal infrared detectors. Journal of Microelectromechanical Systems, 6(1), 55–61. doi:10.1109/84.557531
As a sidenote, the authors in the first article used a 20 um separation between the membrane and the substrate, thus lowering Ggas about 10 times comparing to typical designs which are using in most cases rougly a 2.5 um gap to create a quarter-wave Fabry-Perot resonator and boost the sensitivity around 10 um. Vacuum still remains the best option.