Calibration of spectrometer - Master sample as reference

[qsmso]

Hello friends,
I got confused with the calibration room in my company. They said to me during audit that they cannot calibrate The spectrometer use to analysis chemical composition of material. What they do is to buy master sample and use it as reference point when running the equipement once a month.

Is this correct concept for calibration? If not, what should they do to comply with the calibration concept?

Thank you for your feed back.

QSMSO

 

[SteelMaiden]

That's basically it, you don't really calibrate it, you standardize the readings. Now, that is a fine line to walk, but if you make sure that your records are good, you should be ok. We standardize once per shift.

hint: teach everyone to avoid the "C" word whenever possible.

 

[Craig H.]

qsmso

I am sure that there are more qualified people who will chime in here, but let me take a shot.

One suggestion would be to see if the "owner's manual" is available, either at your plant or from the manufacturer. It should have some advice on working calibration.

That said, we use a similar "master sample" system weekly on our most important QC tests, just to confirm that our testing is OK. It has worked well. The Western Electric handbook has advice on how to use this method with XMR charts to maintain testing systems, and it is a valid approach.

Hope this helps, and good luck.

 

[Bill Ryan - 2007]

I agree with Steelmaiden - we call it "profiling of spectrometer" and we "profile" it once per day (minimum) against certified standards (we alloy our own aluminum).

Just curious - what kinds of Gage R&R numbers has anyone seen/submitted for a spectrometer. Mine have been anywhere in the range from 7.5-145% of tolerance, depending on the element being checked. With respect to anything over 25%, we have been able to "correct" the spectrometer and/or operator methodology to have satisfactory numbers to submit.

Bill

 

[Cyndy B]

I have had 2 different experiences with Spectrophotometers. The first was in Blood Banking. The monthly maintenance required that a known set of standards be read by the spectrophotometer in 3 categories: Repeatability (20 times), Linearity and Drift (performance over a predetermined amount of time). These standards were supplied by the manufacturer and NEVER considered a calibration. They were an indicator of lamp stability and performance. The equipment went through a seperate validation prior to use.

The second is in the Chem lab I manage. We don't have a predetermined, colormetric standard to performance check the spec but we do run 5 standards (purchased material diluted to the appropriate concentration) with each assay performed. The readings are used to calculate a curve. The tested samples, of course, fall within this curve. This is done with each assay type being performed. Again, not a "calibration", just a performance check of how the lamp is reading and how well the material being assayed is extracted by the Chemist.

I don't think you can actually calibrate a spec unless there are some filters you can manipulate.

Hope I don't sound like a complete dolt. Have a good one.

 

[qsmso]

Thank you for all responses. However, the master sample, that used as reference standard, needs to be calibrated or not? I am considering how to track to national standard or equivalent as 4.11.2.b(QS-9000)
The reason I asked because when I audit the calibration room, they use gage blocks to do inside calibrate of vernier gages. Those gage blocks were sent out to calibrate periodically. They can show certificate result when we need to track. So, I think they should use the same concept with master sample of spectrometer. However, they said very difficult.

PLEASE advice !
QSMSO

 

[sgray]

I was just wondering what type of spectrometer you are talking about. For analysis of chemical composition of a material it could be several instruments. Is it an Emission spectrometer or XRF spectrometer? We happen to use both of these. We have standards for both instruments that we use to make calibration curves with. Due to the type of alloys we make we could only buy a couple certified standards. We cast the rest of the standards ourselves and then analyzed them on an AA. We ran at least one of the certified standards on the AA at the same time we analyzed our own standards and this gives us tracibility. The instruments are then drift corrected, (Standardized), on a regular basis and a known sample is then run as a QA/Check standard to verify the instrument was drift corrected properly.

Hope this helps.

 

[Joe Cruse]

I'll echo sgray's and Steel Maiden's comments here. Depending on what kind of spectrometer you're running, the term calibration may not be "useful". We use XRF, and calibration means setting up an analytical program for a material using certified reference materials to make a set of calibration curves. This process is "calibration", and is done one time, unless a major change takes place with the spectrometer. Day to day "drift" is corrected for by another procedure, which is NOT calibration. A check for instrument drift is also performed, but this is not a calibration check, per se, but a drift check. This is done per the manufacturer's recommendations, and we set up tolerances in the software to tell us when a drift correction must be performed. We do run certified reference samples for daily QC checks, along with the daily drift check.

We do something very similar with our ICP. The instrument is "profiled" daily with a standard solution, and each method used is checked with a QC standard and standardized, if outside the specified tolerances.

Don't let an auditor paint you in a corner with the term "calibration" when it comes to spectrometers. Our previous lab manager let the auditor make him set up a calibration check/correction that was totally inappropriate for XRF (dealing with pressed powders), and would actually have done the opposite of the intended end, most of the time. Always talk with your instrument maufacturer on this before setting up a "calibration" procedure for a spectrometer.

Bill, I've never submitted an R&R for the spectrometers. The manufacturer suggested calculating 3 SD after about 50 runs of the drift check sample, and setting that for the limits. Regular drift correction should ensure that the spectrometer stays in those limits, barring a catastrophic failure of a part (which can be replaced and then drift correction comes back in.) We've seen this work pretty well on our main spectrometer after implementing it.

Joe

 

[Ryan Wilde]

qsmso said:

Thank you for all responses. However, the master sample, that used as reference standard, needs to be calibrated or not? I am considering how to track to national standard or equivalent as 4.11.2.b(QS-9000)
The reason I asked because when I audit the calibration room, they use gage blocks to do inside calibrate of vernier gages. Those gage blocks were sent out to calibrate periodically. They can show certificate result when we need to track. So, I think they should use the same concept with master sample of spectrometer. However, they said very difficult.

PLEASE advice !
QSMSO

Yes, your master sample should be a known sample. They are known as "Certified Reference Materials", and they are generally available from your national lab, as well as several private companies.

Ryan

 

[Joe Cruse]

Ryan Wilde said:

Yes, your master sample should be a known sample. They are known as "Certified Reference Materials", and they are generally available from your national lab, as well as several private companies.

Ryan

Ryan, for many spectrometer types, that is true, but not for all. For XRF, to check the spectrometer for "drift" (or standardization check), all that is required is a sample with the desired elements in sufficient quanity, in a form that is "permanent". All you're doing is checking instrument count rates for drift from Day Zero, so if the check sample is unchanging, it will do fine without ANY certified #'s associated with it. That being said, it is still a really good idea to check the calibration curves with a CRM of the type analyzed by that set of curves.

Daily/shift/hourly correction for instrument drift is a must, but I think it's a misnomer to call it calibration with a spectrometer. It may or may not involve the use of CRM's; depends on the spectrometer type and manufacturer instructions. Calibration and recalibration (or calibration update or method standardization) DOES involve CRM's.

I just got through this subject with a supplier profile from one of our customers, who is a major Big 3 supplier. They wanted a copy of a certificate of calibration (by external calibration source) on a piece of equipment in our process that affects their product. They wanted a spectrometer calibration certificate. I couldn't give them one, because we don't get an official certificate of calibration from the manufacturer for any of our 2 different spectrometer types (XRF and ICP). When they come in to PM 2/year, they give a PM report, but they have no way to "calibrate" the spectrometer in the way a vendor like Midwest Balance would calibrate an analytical balance for us. The spectrometer vendors can't sit there and PM an instrument and then totally redo each set of calibration curves a customer has, and then run QC checks with a CRM.

Well that's what I told the customer, and our registrar, and they seem to be okay with that

qsmso, I'd be interested to hear what type of spectrometer(s) you are having to deal with. We investigated the proper route needed for drift correction/"calibration check", put it in our QMS, and then folowed it, with proper records, and the registrar is more than satisfied with this. That should be all that is necessary.

Joe