XRF of Industrial alloy sheet metal 70 Cu 30 Ni
¿ªÔÆÌåÓýGeo, What beam current where you using? Based on the dead time (DT) of 25% the current could probably be backed off a bit. Note that most of this DT is dealing with the higher energy counts that you¡¯re not interested in due to the 50 kV tube voltage. The optimal approach would be to set the tube HV to just above the energy range of interest. In this case Cu, which has a binding energy of 8.8 keV. So a 10 or 15 kV HV would eliminate all the high energy counting and just optimize the energy range of interest. You would then set a current to keep the dead time in the 10 ¨C 15% range and you could play with the peaking time to optimize the resolution. For low count rates use a long peaking time which will get better resolution and use a shorter time for higher count rates but with lower resolution. When I run an unknown it¡¯ll be 50kV, then a run at 40, then a run at 20 which gets the HV near a general binding energy range which gives a much better resolution and count time and lowers the limit of detection. Dud ? From: [email protected]
[mailto:[email protected]] On Behalf Of GEOelectronics@...
Sent: Wednesday, October 14, 2020 12:21 PM To: [email protected] Subject: [XRF] XRF of Industrial alloy sheet metal 70 Cu 30 Ni ? This is a sample cut from a piece of sheet metal, the assay
printed on it was Cu-70% Ni-30%. |
|
All good advice Dud. And I agree, but in Feb 2014 I didn't know that yet. Thanks for helping us all get from there to here, and there's so far yet to go. Geo From: "DFEMER" <dfemer@...> To: [email protected] Sent: Wednesday, October 14, 2020 3:39:50 PM Subject: Re: [XRF] XRF of Industrial alloy sheet metal 70 Cu 30 Ni Geo, What beam current where you using? Based on the dead time (DT) of 25% the current could probably be backed off a bit. Note that most of this DT is dealing with the higher energy counts that you¡¯re not interested in due to the 50 kV tube voltage. The optimal approach would be to set the tube HV to just above the energy range of interest. In this case Cu, which has a binding energy of 8.8 keV. So a 10 or 15 kV HV would eliminate all the high energy counting and just optimize the energy range of interest. You would then set a current to keep the dead time in the 10 ¨C 15% range and you could play with the peaking time to optimize the resolution. For low count rates use a long peaking time which will get better resolution and use a shorter time for higher count rates but with lower resolution. When I run an unknown it¡¯ll be 50kV, then a run at 40, then a run at 20 which gets the HV near a general binding energy range which gives a much better resolution and count time and lowers the limit of detection. Dud ? From: [email protected]
[mailto:[email protected]] On Behalf Of GEOelectronics@... ? This is a sample cut from a piece of sheet metal, the assay
printed on it was Cu-70% Ni-30%. |
|
On Wed, Oct 14, 2020 at 03:39 PM, Dude wrote:
All correct. Here's a scan from today that was taken to illustrate another point concerning "Too-much-of-a good-thing-is-still-TOO-MUCH" Step one is to repeat a 50 kVp 10 uA run on pure Ag test stamp, which was originally done with an SDD sensor but now with a Si-PIN sensor. Notice the very high peaks in the Ag Ka1 and Ka2 area. The problem is with so many counts coming in, there are more chances of coincidence- two separate photons arriving at the exact same time and both adding to the photopeak energy- and the probe shows just that. Both Ka1+Ka1 and Ka1+Ka2 coincidence peaks are cluttering up the spectrum. And that's on LINEAR display, there must be much more garbage in the LOG display of same data. |
|
Now simply by reducing the beam KvP to the K-Edge plus ~3 keV, the scan is as good as the probe can make it. In this case we know it is silver, if you were searching for impurities or testing an unknown sample,? you would start with 50kVp and do a full sweep, then later narrow down to the area of interest. Data was obtained from http://physics.nist.gov/cgi-bin/XrayTrans/search.pl?download=column&element=All&trans=Kedge&lower=8000&upper=200000&units=eV See also: http://physics.nist.gov/PhysRefData/XrayTrans/index.html Remember that 12.3984191(11) keV is 1 Angstrom (http://physics.nist.gov/cuu/Constants/Table/allascii.txt)
? ? ? |
¿ªÔÆÌåÓýYou¡¯ll be ?getting sum peaks (Ka + Ka and Ka +Kb) when the sample is a pure metal or would have a high concentration (% level) and you have a very high count rate (dead time 28%) due to a high flux from the x-ray tube. Best way to avoid this is back off the tube current, use a low peaking time and set up the baseline restoration and filters to optimize for high count rates. ?You can also use a primary beam filter to reduce the tubes characteristic radiation which has the largest effect on the sample than the broad energy brem and will reduce the count rate. Beam filters must be tailored to the energy range of interest and the tube¡¯s anode composition. Dud ? On Wed, Oct 14, 2020 at 03:39 PM, Dude wrote:
Geo, What beam current where you using? Based on the dead time (DT) of 25% the current could probably be backed off a bit. Note that most of this DT is dealing with the higher energy counts that you¡¯re not interested in due to the 50 kV tube voltage. The optimal approach would be to set the tube HV to just above the energy range of interest. In this case Cu, which has a binding energy of 8.8 keV. So a 10 or 15 kV HV would eliminate all the high energy counting and just optimize the energy range of interest. You would then set a current to keep the dead time in the 10 ¨C 15% range and you could play with the peaking time to optimize the resolution. All correct. Here's a scan from today that was taken to
illustrate another point concerning "Too-much-of-a
good-thing-is-still-TOO-MUCH" |
