Hi all. I received many responses to my questions about temperature
calibration .. thanks to everyone who replied. The concensus seems to be
that yes, the temperature read by the thermocouple deviates from the
actual temperature the further you get away from room temperature. Also
there were suggestions as to how to minimize the deviation.
One good suggestion was that since there is a temperature gradient from
the bottom of the sample to the top that one should use as small a sample
as possible. Another one is to use a high gas flow rate.
For those who are interested I've included a digest of responses to my
original query.
Cheers :)
Keith
----------------------------------------------------------------------------------------
I've done a lot of this over the years, and the difference between
display(thermocouple) temperature and the actual sample temperature is often
very dependent on the VT air-flow, specific probe characteristics, and will
even change with time if the position of the thermocouple in the air-stream
moves (or gets bumped or jarred). As you move away from ambient, the
difference between display and actual will always increase. Also, too much
airflow will increase this discrepancy (which may seem counterintuitive), so
I always try to find the "sweet spot" for the probes that are used a lot for
VT NMR.
I'm not sure about the non-linearity you are seeing at higher temperatures,
but you might consider experimenting with more/less air-flow. I'm not sure
what to make of the "discontinuity"... that seems very odd to me.
Just FYI, I have a little spreadsheet with the "formulas" for the various
calibration samples/standards that makes it easy to enter the peak
separation, and it displays the temperature (or vice-versa). It's slightly
more convenient that a calculator, and at least I know what coefficients are
being used (vs. Varians "tempcal" command, which is a little cryptic).
http://chemnmr.colorado.edu/manuals/temperature-calibration.html
Best regards,
-Rich Shoemaker
----------------------------------------------------------------------------------------
DMSO takes up water very quickly. Perhaps that would affect the data.
Is the tube sealed? For VT calibration I use the standard sealed tubes
Varian supplies which are pure methanol for low temperatures and ethylene
glycol for temperature above ambient. I never checked for
discontinuities.
Fred Morin, PhD NMR Facilities Manager Department of Chemistry McGill
University 801 Sherbrooke St. West Montreal, QC H3A 2K6
----------------------------------------------------------------------------------------
At a glance ... you seem to be saying that the methanol and ethandiol
calibration curves don't agree? Can you not overlap them by 20 degrees or
so?
I calibrated VT once using a digital thermometer. IT looked like a
voltmeter, with a thermocouple at the end of a 5-foot wire. We put the
thermocouple in an NMR tube with a mL of H2O and lowered it into the
probe. I was concerned the thermocouple might be affected by the field,
but confirmed that ice thawed at 0 C.
-+=+-+=+-+=+-+=
Roger Kautz
----------------------------------------------------------------------------------------
Accurate temperature is always a challenge (for me at least!).
You really have to be careful with shims, gas flow, volume of the samples,
stabilization time...
Shim should be very good if you want to measure an accurate temperature.
If the shim is bad, then the distance between the peaks will not give the
real temperature anymore (I do not really understand why, but it's true).
If the shim is very different between the 2 samples, it means that the
temperatures of the shim stack are different too, so temperature of the
probe and sample will change. It could explain the discontinuity. You
often have to use high gas flow for high and low temperatures to remove
temperature gradients. You have to be careful with stabilization time
after each temperature change.
I attached a useful excel file, which contains formulas for different
calibration samples. I found several equations for ethylene glycol in the
literature, which could give a difference of several degrees.
I hope it will help you and I hope you will post all comments.
Sincerely,
Dr. C?dric Malveau
----------------------------------------------------------------------------------------
I've always found the same sort of result as you. As over the years we
have done a lot of VT for which we wanted accurate readings, I have always
made my own calibration charts. At low temperature it is straightforward
(although very tedious); I have always double-checked the result by
watching tubes of deuterated solvent freeze. Like you, I have found the
two standards do not agree where the range overlaps. Because I have a
reasonable degree of confidence in my low temperature readings, I have
taken the low temp. value as the accurate one and applied a correction to
the high temperature chart, with the warning to users that it may not be
particularly accurate. Fortunately, nearly all of our need for accuracy
has been below room temperature.
If you hear of a method to double-check the values above ambient, I'd love
to hear about it.
Cheers,
Chris
----------------------------------------------------------------------------------------
No answer for you- but I see the same thing...
I've calibrated our probes with the 4% methanol sample- they all give a
very linear response from 5-25C- thermocouple reading higher at lower T,
just as you see. Most users are running around 20-35C, so I've just gone
with that slope as being fine, and noone has ever complained that their
spectra didn't match up with data from home specs at same temperature.
The other day I needed to run something at 60C, and accurate temp was
important- so I pulled out the ethylene glycol temperature standard from
our kit, just to check. Following my trendline from the methanol
calibration I should set the temperature at 57 to get 60C, in the end
according to the high temperature standard, a temperature setting of 66
gave a true reading of 60... I haven't had a chance to do the full range
from 30-70 with ethylene glycol yet, but I guess I will see the same
thing- and am left wondering how do I get a calibration for the whole
temperature range of our probes?
Looking forward to what the response is from the rest of the list,
Tara
----------------------------------------------------------------------------------------
Dear Dr.?Keith Brown, The first point that needs to be borne in mind is that
there is very little annular space between?
the tube and the probe insert. This restricts the flow of VT air and this may be
a cause of?nonlinearity /discontinuity in the graph. This can be overcome by using 3mm or
2.5 mm?microprobe NMR tubes on the 5 mm probe.
Second point is the equilibration time before performing the experiment it needs
to be?sufficient, especially when the experiments are being carried out at a
temperature far?away from ambient.
Third point is the flow rate of the VT air should be?sufficient?so that a
temperature gradient can?be avoided?in the sample.
Hope these tit bits will be useful. Should any more queries arise, feel free to
write to me.
----------------------------------------------------------------------------------------
Thanks for posting this.
I too am very curious what the consensus out there in the "real" world
is on temperature calibrations in NMR.
I know that the new "standard" for cryoprobes is to use just
deutero-methanol (Magn. Reson. Chem. 2007; 45: 175?178).
The 0.2% residual 1H-signal is used for temperature calibration because
of significantly less radiation damping.
I am totally swamped with other things right now, but I look forward to
reading your summary.
Cheers,
Bob.
----------------------------------------------------------------------------------------
> So, finally my questions. Why the non-linearity and why the large
>discontinuity? What are your experiences and techniques used for
>temperature calibration?
I think the problem with the liquid thermometers is that they aren't stable.
I've heard that methanol is better if its sealed without oxygen. We usually
just use a multimeter thermometer. A group recently bought an HH506RA
Multimeter with a TJ144-CPSS-116U-8-SB-SMPW-M thermocouple. There
was a slight problem with an iron strain relief coil near the thermocouple,
but they removed it. They have an NMR cap with a hole that the thermocouple
can fit through. The thermocouple cable is long enough that the multimeter
can sit on the console while the tube is in the magnet. Put about the same
amount of solvent in the thermocouple tube as in the sample tube. You do have
to remember to turn off the decoupler during the temperature measurement.
hth,
dave scott
----------------------------------------------------------------------------------------
Interesting set of issues. I can tell you best how we perform
calibrations, and less definite about your set of calibration data.
First, there shouldn't be a jump at ambient temps, and I'd suggest
looking into the sample preps to see if that problem isn't related to
that. Because Varian (at least used to) provides temp standards with
new systems, I've had enough of these carefully prepped samples so we
mainly use them here.
We also use thermocouples--fed into NMR tubes to the middle of the
sample region, and covered with a solvent to assist in thermal
conductivity--in conjunction with MeOH and EtGlyc standards. Very
useful when temps run outside the standards' ranges. Students are
instructed to use one of these two methods to check temp calibrations
regularly. Typically some calibration will be performed during every
session when temp accuracy is critical. The calibrations are very
dependent on gas flow rate (watch this closely!), and any work on the
probe can affect the calibrations via various routes; the most
problematic is slight changes in the thermocouple positioning in the
probe. Clearly, some restriction or change in the gas flow can also
changes the calibrations. In my experience, regular calibrations are
necessary for the more temp dependent work.
Hans Reich's group here has made up internal calibration compounds for
use in their studies so the temp can be measured for every 13C set of
data (see Sikorski etal Mag Res Chem 36 (1998) S118-S124; and Reich
etal, J Org Chem 74(2) (2009) 719-729).
I hope this helps. Be happy to continue a conversation; always new
things to learn.
Charlie
----------------------------------------------------------------------------------------
I have to confirm the observation (on Bruker probes) that the real
temperature is lower (below R.T.) and higher (above R.T:).
The deviatzions are up to 7..10 K at 370 K, depending on the probe
(placement of temperature sensor in the probe an the air flow).
You should keep in mind that there is a temperatur gradient within the
sample.
We put our own PT-100 sensor in a NMR-tube and measured a difference of
nearly 1 K between top and buttom of the sample (at 330 K).
So: there is not only *one* temperature in the sample! It is a
"temperature field" (in physical sense of "field").
At higher temperatures (low viscosity) you will get a convection with in
the tube (this can be observed on lock of the methanol sample at some
higher temperatures), which gives you some mean values.
The advise is: use high gas flow rates and short filling heights.
If you use the Bruker's au-macros "tempcalc": you should answer for both
samples, you used, the questions "pure yes/no ?"
with "no".
Yours
findy
---------------------------------------------------------------------
Dr. Keith Brown
Department of Chemistry/
Saskatchewan Structural Sciences Center
University of Saskatchewan
Saskatoon, Saskatchewan
306-966-1725
http://chem4823.usask.ca/kbrown.html
-----------------------------------------------------------------------
Quantum Mechanics: The dreams stuff is made of.
-----------------------------------------------------------------------
Received on Thu Jun 03 2010 - 07:55:30 MST