Use thermistors close to the sample. They come in various shapes and sizes. Some have holes in the middle (e.g. ), so your sampling could be done above and in the middle of one, which will act as a black body and irradiate your sample with an essentially constant temperature, if controlled via a Wheatstone Bridge and a stepped-decade resistor box. Do your own calibrations.

I hope this helps, Singgih. -Steve-

It is not easy to measure temperature if environment is at consistent different temperature than the test item and the probe is exposed to radiation.

First you must ensure "perfect" thermal contact.

Second you must prevent the probe to exchange radiatively with the environment.

A solution to mitigate is to sputter low emissivity material on the probe (silver or alluminum) and/or provide low emissivity probe radiative shield.

If improperly installed, the temperature measured by the sensor

may have little relation to the actual temperature of the object being measured.

http://www.lakeshore.com/Pages/PackagesSI.aspx# have a look at DT-BR diode type

It is not easy to calculate surface temperature of a sample directly from radiation balance. To do that one have to know: 1) spectral irradiation provided by radiation source in every point of the sample surface; 2) spectral absorptivity of the sample surface, generally depending on temperature and generally varied from point to point of a sample surface; 3) from 1) and 2) one can calculate surface distribution of the energy flux from radiation source absorbed in the surface; 4) thermal conductivity of a sample together with sample holder and ambient gas; 5) spectral emissivity of the sample, generally depending on temperature and so may be different from point to point of the sample surface; 6) resolve thermal conductivity problem together with radiation balance problem to calculate temperature distribution along the surface of the sample. In other words to calculate temperature distribution one have to take carefully into account all sources of incoming and all leaks of outcoming heat in every point of a sample.

So, I agree that more reliable way is direct measuring of a temperature.

For non-contact reliable temperature measurements I can recommend 2-color IR temperature transducer available under Mikron trademark (see link attached). They are able to measure temperature of the surface smaller than 1 mm in diameter. 2-color technology removes to a great extent the problem of temperature changes of spectral emissivity of the measuring surface during non-contact temperature measurements.

More budget, but not more reliable, nor precise, nor simple solutions are contact temperature measurements with thermopile or thermistor type contact sensor, as stated in the above comments.

https://www.lumasenseinc.com/EN/products/temperature-measurement/infrared-thermometers/mikron-pyrometers/series-m67/pyrometer-m67s.html

If you can collect Raman spectra for both Stocks and anti-Stocks peaks and take the ration of the intensities for the two peaks, then it will allow you to determine the surface temperature with and without the UV light source.  Also, peak position for Raman spectra shifts with temperature (moving to lower wavenumbers for higher temperature due to mode softening).  If you have the data on Raman peak position as a function temperature, then you can determine the surface  temperature based on the Raman peak position.   The following article is an example for fuel cell temperature measurement using Raman spectroscopy:   http://pubs.acs.org/doi/pdf/10.1021/jp076361j