Which Temperature Sensing Technique Offers the Highest Sensitivity?

[Excom]

Hi

which temperature sensing technique is the most sensitive one?

 

[f95toli]

Depends on what kind of temperature you want to measure and also what type of sample you have.
Also, you need to define what you mean by "sensitive" in this case.

 

[Bob S]

Hi Excom-
Analog Devices has a digital temperature sensor (ADT7410) that has resolution of +/- 0.0078 degress C and an absolute temperature accuracy of +/- 0.5 degrees C. Is this what you want? See
http://www.analog.com/static/imported-files/data_sheets/ADT7410.pdf
Bob S

 

[Excom]

If it is non-contact and around room temperature, which sensing principle is able to detect the smallest change in temperature? Thermocouples, thermistors, bimetallic cantilevers, resistance temperature detectors (RTDs), diodes, transistors, thermowells or etc?

 

[Bob S]

Do you have a need for an ultra-sensitive thermometer, or are you just doing a survey of the various types of non-contact temperature sensing? By non-contact, do you mean non-conduction, non-convection, or radiation sensing only?
Bob S

 

[Excom]

Hmm I am not sure if it is called contact or non-contact. The thing I have in mind is measuring temperature changes in a fluid.

The reason why I am asking is because I am going to make a thermometer for a micro-calorimeter.

 

[Bob S]

Hmm I am not sure if it is called contact or non-contact. The thing I have in mind is measuring temperature changes in a fluid.

The reason why I am asking is because I am going to make a thermometer for a micro-calorimeter.

You might look at the Analog Devices TMP05:
http://www.analog.com/static/imported-files/data_sheets/TMP05_06.pdf
which has a temperature resolution of +/- 0.025 deg. C. This is much easier to interface than the ADT7410 I referenced earlier. It will require 5 wires, probably 30 Ga. It probably cannot be immersed in the fluid. The output is a pulse width that changes 200 microseconds per deg. C. At the receiving end, you could use a 500 kHz (or higher) oscillator and count the width of the pulse in a decade counter.
Bob S

 

[f95toli]

Bob S: I don't think +-25 mK counts are "ultra sensitive".

There are a lot of different ways of doing this, but you have to be more specific with regard to HOW sensitive you want the sensor to be

How much space is there for the sensor? Presumably you need the sensor to be much smaller than the sample?

What is the temperature range?

Do you need absolute calibration or are you only interested in measuring the CHANGE in temperature?

What kind of read-out electronics do you have? Resistance bridge?

I don't really know that much about RT sensors (all my experiments are done at cryogenic temperatures), but I would guess that even something as simple as a bare silicon diode might work

Also, precision thermometry is always a bit tricky. Getting the right sensor is the easy bit, installing it correctly, setting up the read-out system etc is not easy.

 

[Bob S]

What about this?
http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=RSINAK000036000012001817000001&idtype=cvips&gifs=yes
"A major microcalorimetric problem, the development of a very sensitive thermometer, is considered. The theory of a pyroelectric thermometer is reviewed and verified calorimetrically. A thermostatic system utilizing both a layer of low thermal diffusivity material and a large heat sink was used to strongly attenuate external temperature fluctuations. A pyroelectric thermometer, used in the thermostat, measured temperature changes as large as 2.52×10−3 °C and as small as 6×10−6 °C. The sensitivity of the thermometer was limited by the temperature stability of the thermostat. Technically simple modifications of the calorimeter could make the detection of temperature changes as small as 5×10−7 °C possible. Detection of temperature changes smaller than 10−7 °C may be achieved by changes in size, shape, and material of the pyroelectric sensing element. ©1965 The American Institute of Physics"
Bob S

 

[Excom]

@ f95toil:
1. The plan is to measure around room temperature.
2. It is important that the sensor is small. The goal is to place it inside a container with a volume of 200 microliter.
3. It is okay if the temperature range is around 30 degrees.
4. Measuring changes in temperature of 10 micro Kelvin.

@ Bob S
Thanks for the article.

 

[0xDEADBEEF]

Well I guess you are more interested in the relative temperature change. I am sure that there are thermistors for that type of application, but the real problem is this: How "strong" is your source? And how much time do you have.

You can surely build a thermopile that accomplishes what you need, or a thermistor. But the problem is, that you will always have leads going there. If you are measuring let's say mercury than an ultra fine thermopile will work, but if you are measuring somethng with a bad thermal conductance your leads will transport too much of the outside heat into the system.
With a resistor the trouble is, that you need to send current through it. This can heat your sample, but this can be overcome by using very low current and some integration and averaging magic, or some lock in technique, this needs time. It also has some finite size and you have to make sure it is small enough that it follows your sample's changes.

So you have the triangle of speed, accuracy, and heat transport from the outside that you need to balance. If you need very high speed maybe you have to look into some other related property and measure that instead, preferably in optics.

 

[Excom]

Thanks for the help