Suggested new forum: Karl Fischer

Amidst all the myriad topics covered on this wonderful site, I cannot find any discussion of Karl Fischer moisture analysis. In fact, I could not find any forums anywhere devoted to this topic. I thus propose that a new forum be created under the Analytical Chemistry grouping.

Perhaps you could start a topic/post on this topic in the assay development forum in Analytical Chemistry?

We have created this new forum in response to your request. Spread the word around, and let's start an active discussion.

For the determination of water content, do you follow the USP criteria for sample quntity during measurment

If you are using a USP method, then you should always follow the suggestions for sample size.  These recommendations are usually based on the sensitivity of the method and the range of water content that you are trying to measure.

If you have a very dry sample, then using a larger quantity would be possible.  If your sample has more water, then using less would probably be acceptable, but there are always concerns when you reduce the size of the sample.

We have an FDA auditor on site and she is stating that we must analyze five or six replicates of sodium tartrate as a standard every time we do an automated Karl Fischer titration and define acceptance criteria for RSD. Is this standard practice?

Our current practice is to establish a titer with duplicate injections of water and then verify with a single determination of sodium tartrate with acceptance criteria of 15.5 to 15.8.

--Lorraine

Featherland:
This can be a tricky issue.  Typically, an analytical method is validated for a certain instrument/sample/analyte combination.  This validation should include the usual experiments for accuracy, precision, etc.  Once validated, the method should state what system suitability is required to verify performance.

It is typical to have a precision requirement of three to six replicates and a %RSD precision specification; many pharma LC and GC methods use this approach.  However, I do not think that this is written in stone for every piece of equipment.  There are, in fact, USP methods that do not even mention precision (See USP <467>). 

The issue probably revolves around the importance of the measurement.  If this is a very important number (e.g., assay %), then more verification is necessary.  If this is a limit test, then less is required.  I would think that a moisture content would be a less important value, but I don't know your situation.  You have some QC that you are already doing, which is a good start.  If you also have enough performance data for your instrument, then you can maybe make an argument that what you are doing is sufficient.  But also get a reference from the auditor of where in GMP it says that this is required, and then report that back here.

Good luck.

I checked with a QA Director at a pharma company, and confirmed that there is no specific GMP requirement that you must do six (or any number of) injections as a part of system suitability.  Now, you MUST justify whatever you choose to do, but as long as you have a valid scientific reason, and some data to back it up, you can write your own rules.

Of course, you may want to take a somewhat softer tone with the auditor. 

Hi all,  have been working on Karl Fisher methods for some "new" materials in my lab.  What would be a reasonable %RSD specification for Karl Fisher ?  I have a material that is on the order of .5 % water.  Whether I use 100 mg samples or 1 gram samples,  my RSD is around 7%,  seems high to me.  Any thoughts or advise from any experts out there.  I need to have some reasonable acceptance criteria and I am not sure if that is acceptable.  It is not for an assay,  however the value may be used to adjust for water in a material and then consequently would be contributing to the assay.

For those interested, RT Corp offers both Proficiency Testing (an ISO 17025 requirement) and Certified Reference Materials for Karl-Fisher <921>.  The product can be found at:  http://rt-corp.com/products/n16378_c356_p2_s1.aspx

For those who do not know what Proficiency Testing (PT) is:  PTs are blind samples that are provided to the user, and the user must report the data to the provider.  The provier will then indicate whether the user passed or failed the test.  PTs are a required part of ISO 17025.  The PTs can also be used to demonstrate training proficiency or to test contract laboratories proficiency.  FDA Laboratories are now ISO 17025 accredited.

What kind of KF analysis are you doing (I am assuming coulometric...but are you making a solution, introducing the sample directly into the vessel or using an oven)?

At 0.5% water, it becomes increasingly important to control water in ambient air.  If you are making a solution, I suggest you dissolve your sample in the vessel solution that is titrated to dryness (after pretitration).  In fact, this would probably be the best way to do it.  Anytime you open the vessel to add sample, you are letting in ambient moisture, and not very well controlled.  In addition, it would be a good idea to prepare your samples under a dry nitrogen atmosphere.  A blank and drift determination are important. 

If you are using a drying oven, it is important to have a tightly regulated flow to the oven (like a 0-3 psi regulator).  The nitrogen should also be very dry.  Run it through several drying tubes and jars (mine goes through six of them).  If your nitrogen is not dry and/or you are not using a tight regulator, your drift can vary GREATLY.  The drift is subtracted out and you can get relatively large variation in your results.

One final thing...make sure you are using the approriate solvent system for your compound.  If there are aldehydes or ketones in the compound, you need to use a vessle solution that will not have side reactions with the methanol in traditional vessel solutions.

I hope this helps.

Aaron

Another question.   Has anyone verified coulemetric KF as listed in the USP.   Running a compound that list LOD but we need to run KF instead for moisture due to limited material.   Customer has requested we verify KF for this product.    Any ideas?

It was recommended to spike water into a sample at two concentrations and run KF multiple times each and evaluate the recovery and RSD.     Not sure I can get a homogenous sample on small scale spiking.

Anyone have ideas or done something similar?

Scott

I'm interested in other labs' procedure for automated Karl Fisher analysis (we are using Mettler Toledo DL31). We are a contract lab and our clients aren't willing to pay for a full method validation of USP <921> in each product they send us, as "compendial methods don't need to be validated". Our current procedure:

1) three determinations of ~50 ug samples of water (amount of water determined gravimetrically) to standardize the KF reagent. Acceptance criteria RSD < 0.5%.
2) standardization verified with 150-200 mg of sodium tartrate (CRM). Acceptance criteria <1% error.

The FDA'er that was here had no problem with that approach, but we have client auditors who want to see a calibration curve established with three different amounts of sodium tartrate (if water <1% in their sample) or three different amounts of water (if sample has a water content of >1%) with each analysis.

Trying to establish industry standard here...

--Lorraine

Working in a contract lab can be very rewarding, but this is one of those other cases where you are caught in the middle. 

If you have "verified" that you can perform the compendial method properly (with supporting data) then I think most people will accept the result.  That's the whole purpose of publishing these methods; follow the procedure and the results are reportable.  (I won't get into a discussion on whether this is a good idea or not; it's just the way it is!)  As long as your normal operating procedures include calibration and QC that are consistent with the standard, you should not have to do more.

Now, if a customer wants something extra done (more calibration or QC) beyond the standard, then that would normally involve an extra charge.  Whether or not you charge for routine QC is an individual choice.  Some labs automatically bill for the extra sample(s), while others build it into the cost of the samples.

I'm not sure if this answers your question, but from my experience there isn't a single best or right way to do this.  Each lab has their own policy.

Good afternoon,

I am a PhD student, and I work hard with mineral oil. In this case  I have to determine the water content of the oil with Karl Fischer method (coulometric determination).
Would you be so kind to tell me what kind of water can this method determine: free water from oil or bounded water (bounded with oil molecules).

Thank you,

Best regards