| Author | Message | |||
     Michael Kinsey (mike_kinsey) New member Username: mike_kinsey Post Number: 4 Registered: 1-2009 |
I work with stearines on the NIR, and they have fundementally different NIR characteristics depending on their C18 ratios and C16 ratios, however I have split them into two one low Iv the other high, which can also be seen by the IV value range of the stearines, this can also be seen via GC analysis. The stearines I work with, some have very low IV below 0.5 mgKOH/g and some higher values of near 10 mgKOH/g. I don't use a pharmacopieal ID method as it is not required, I just use a quantatative method to distingush whether or not all of the stearines we do fit in with the general trend of what we produce. But I do believe this is possible, I have never fully tried doing this as my job doesn't warrant it, but I have had a little play with this in the past in my spare time and it was working. Occassionally we have samples that have been contaminated due to our ever changing raw materaisl and this can be visibly seen to be different in the NIR spectra, using SG15 2nder usually 2080 to 2200 range. So IF your raw materials stearines have a constant chemistry and not a random mixture fatty acids then this would be possible, however be warned that it will require constant updating with new spectra depending on the raw materials which were used in the manufacturer of the stearine in the first place. I.e. by the use of extenders due to raw materials costings. | |||
| Gary Rollason (nvisible) Junior Member Username: nvisible Post Number: 6 Registered: 2-2007 |
Stearates for pharma are predominantly palmitic and stearic acids with fairly tight tolerances on the ratios. Something else to look at is whether or not the pharmacopieal ID methods can distinguish between these related compounds. I have several 'groups' in my raw material ID libraries which cannot be distinguished (eg lactoses, microcrystalline cellulose)by the compendial methods, therefore it is pointless to try to get the NIRS to unambiguously identify them. Sometimes it may be possible to identify these by NIRS as belonging to that group, then performing a secondary chemical test to distinguish between the group members if necessary. e.g. I use the NIRS to identify the Hypromellose group, then perform viscosity test to differentiate between the several grades received. | |||
| David Adamson (davida) New member Username: davida Post Number: 2 Registered: 5-2009 |
Hi All. A few days ago I suggested to Fredrik Andersson that his various caprylate salt forms might not be distinguishable from each other using NIR. I expected that because the mid IR spectra of the three forms of stearates we use here, showed no discernable differences. I have since been able to make some NIR measurements and found to my surprise and delight that there are several visually discernable differences in the spectra. Doing a first or second derivative shows them up even more clearly. I have attached a pdf of a portion of the second derivative spectra to make my point. Now it might be that in this case it is the fatty acid composition of each material that is responsible for the differences and not the counter ion. I cannot comment further on that as its beyond my expertise to know. These "stearates" are actually a made from a varying mixture of fatty acids.
| |||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 228 Registered: 9-2001 |
Tony - maybe what you're trying to think of is Ron Rubinovitz's work on identification of inorganics. He presented a very nice summary of it a couple of years ago (I think at Pittcon), although I don't know if he ever published it. Jerry - Jerry Workman's handbook is titled "Handbook of Organic Compounds", published by Wiley in three volumes. \o/ /_\ | |||
| Jerry Jin (jcg2000) Member Username: jcg2000 Post Number: 12 Registered: 1-2009 |
Hello, Tony Actually, I found your column is quite understandable and insightful. However,You could use more reference. I think Fredrik's question was about solid salt, not its aqueous solution. I agree we can NIR to distinguish between (in)organic salts in their solid state, considering its application in pharmaceutical area for polymorphism study. I am more interested in how we can use NIR to differentiate aqueous NaCl from KCl, or sodium acetate trihydrate vs potassium acetate trihydrate. It would be rather difficult, if not impossible. For example, in Jerry Workerman's compilations of IR spectra (Handbook of Physcical Chemistry??), sodium palmitate and potassium palmitate in water solution have almost the same spectra. Although I can't find their NIR spectra, I assume they are also alike. It would take much more effort to tell them apart by using NIR. Best, Jerry Jin | |||
| Art Springsteen (artspring) Member Username: artspring Post Number: 13 Registered: 2-2003 |
Hello Fredrik, While I'm almost as old as Tony, I haven't been doing NIR quite so long but... Being an old chemist, and a simple folk at heart, using NIR to determine if something is a K or Na salt seems to be to be killing a gnat with a thermonuclear device. How about a simple flame test (orange for Na, violet for K). Or don't they allow you to do flame tests any more? I'm surprised no one has mentioned this! | |||
| Tony Davies (td) Moderator Username: td Post Number: 187 Registered: 1-2001 |
Hello Fredrik, I think that this is the first time anyone has suggested that I was a statistician! I have been accused of being a chemometrician but I�m really an analytical chemist who has been a student of near infrared spectroscopy for the last thirty years! I also knew something about chemometrics before I became involved in NIR spectroscopy (and that�s how I came to be writing about it for nearly twenty years). I am disappointed that you found the article �very abstract�. I was trying to explain exactly what you find to be a problem. The reason why you find it difficult to decide how to challenge your calibration is that the method you are using is only appropriate when the ONLY samples that can be present have the same identity as ones in the calibration. The is no way to decide what are the correct samples to use as challenges when the calibration is being used where there is the possibility of testing a sample that was NOT PRESENT in the calibration. I know it has been used for years but I only recently realised the implications of its modern usage for testing incoming raw materials. I can only repeat my recommendation from the article. If you only have one container then you should (in my opinion) use compendium methods to establish the identity of its contents; not NIR spectroscopy. With regard to aqueous solutions of salts. The NIR spectrum of water is very complex and is an equilibrium that is very easily disturbed. I have investigated different anions of sodium salts which can be identified by the changes in the H2O spectrum. I believe that similar work has been published for different cations but cannot recall it at the moment. It is possible that NIR spectroscopy of a solution of your salts would reliably distinguish between K or Na cations. Best wishes, Tony | |||
| Fredrik Andersson (fredrik_andersson) New member Username: fredrik_andersson Post Number: 3 Registered: 5-2009 |
Howard: Ah, I think I might not have been perfectly clear with what I meant. What I meant was that it is not enough that the challenging is a chemical analogue, it also has to be of similar polymorphism and particle size as the substance it is challenging. Take for instance NaCl, which we tried to validate here, if you challenge NaCl with KCl that you just take from your ordinary lab chemicals, the challenge will probably not pass and you might conclude that the two substances have some apparent differences. But, if you order KCl from the same supplier (and same quality) that you get your NaCl from, the substances are suddenly so similar that the validation fails for NaCl. True story, very sad. Now, you might say that NaCl is not a good example, because it really isn�t. But the same case can be made for aminoacetic acid where polymorphisms invalidated some of my samples, or poloxamer, where particle size failed one of my samples, but passed a sample of different MW that I wanted to use as a challenge. Or D-sorbitol, where production quality and lab quality were so different I actually though it was a mix up for real, almost. Ian: Oh dear, I�m sorry, it was not my intention to start any battles, it was just that article seemed very abstract to me, also I think I might have conflated the authors credentials. In my defense though, they use quotation marks around real world, which is clearly indicative of a statistical mindset, also I�m Swedish (which is just always a great excuse for reckless conflation of identities). | |||
| Ian Michael (admin) Board Administrator Username: admin Post Number: 21 Registered: 1-2006 |
I'm not sure how Tony will react to being called a "statistician", but I'll let him fight his own battles!! | |||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 226 Registered: 9-2001 |
Frederik - I think you're missing a point when you say: "... any sample from a new supplier that we do not use as the raw materials supplier, might not actually be comparable to the wrong substance ..." I think that the point to note is that it also won't be comparable to the RIGHT substance, which is the indicator that you're looking for. If identifying WHICH wrong substance it is, is important, then you can do subsequent testing for that, if necessary. But the key to using the "quick" method is to determine if the material under test is the "right stuff" or not. So the only critical part is avoid a false positive for that; false positives for the "wrong stuff" can be handled other ways. Hopefully, getting "wrong stuff" will be a rare event, so that even if you have to use slower and more diffcult methods for that, the time saved by using a fast method for the majority of the incoming samples will justify having the capability. \o/ /_\ | |||
| Fredrik Andersson (fredrik_andersson) New member Username: fredrik_andersson Post Number: 2 Registered: 5-2009 |
Wow, thanks for all the answers! I think you guys are right about the risk analysis for my challenges, it�s just that the mere possibility that a significant amount of potassium could get into the product would be so bad that any possible risk reduction is necessary. So if I could demonstrate that the substances do in fact look different I could get away with doing fewer additional id-tests. But, on the other hand, seeing as NIRS is supposed to provide information on: crystallinity, particle size, polymorphism, granulation process etc, any sample from a new supplier that we do not use as the raw materials supplier, might not actually be comparable to the wrong substance if it was accidentally produced by our supplier and then mislabeled, (which is my favorite �worst case�). In any case, I have to do a justification based on �The probability of their presence in the concerned pharmaceutical setting� anyway. Also what if my supplier suddenly decides to start producing any number of strange and dangerous substances that might throw my risk analysis, even with periodic updates? So yeah, I think I might be to paranoid for this job. Francesco: Yea, that�s true (especially about it not being my fault, I tell my boss that every single day), anyway we should just use high school kids to do the SCI for all our sodium-id needs, we get about 20 tons NaCl in 25 kg bags every other month or so  Jerry: Sorry, couldn�t see the attached spectra. But in water solution wouldn�t the ions not actually be associated with the organic moiety as they are in the crystal form, and thus even less likely to affect the spectrum? And NIR is supposed to give more information about crystallinity, polymorphism etc. so I don�t know if IR is directly comparable in this case. I also have sodium bicarbonate in my library and testing potassium bicarbonate against it hasn�t actually been a problem. I use wavelength correlation over the entire spectrum, and get a correlation of 90 for potassium bicarbonate and the acceptance limit is set at 99. I have two different suppliers of NaHCO3 so I�ve tried KHCO3 from both and it doesn�t seem to be a problem. I would assume that the different ion radii causes the difference in the crystal structure and thus in the NIR spectrum, is this a reasonable assumption? If so, it would make sense that the counter ion influences the spectrum for small molecules but less so for larger ones. Ian: Hmm, somehow I was more confused after reading that article than I was before. And then I realized that they are statisticians, so it makes sense that it makes no sense /Fredrik | |||
| Sheelagh Halsey (shalsey) Intermediate Member Username: shalsey Post Number: 16 Registered: 12-2006 |
The EMEA have published good guidance on how to set up and challenge identification methods, see website http://www.emea.europa.eu/pdfs/human/qwp/330901en.pdf All challenges need to be risked based, so you should only be challenging with samples that are possible. Based on your knowledge of the supplier, you can decide the worst case scenario of the wrong delivery. You should then test this. As to the different ions, I have seen cases where they impact the spectra. I haven't looked at this particular example, but it is easy enough to run the spectra and see if they are different. Sheelagh | |||
| Francesco Davini (franz) New member Username: franz Post Number: 1 Registered: 2-2009 |
I understand Fredriks problem and his approach to it. Nevertheless the whole discussion makes me think how absurd our business became. We are discussing an analysis (Na vs K) which is usually a basic high school exercise and can be done using the most simple devices and glassware. It's like using a large truck to go shopping at the local market, and wondering where could we find a parking... It's a crazy world, and it's not Fredrik's fault | |||
| Ian Michael (admin) Board Administrator Username: admin Post Number: 20 Registered: 1-2006 |
Tony Davies had some comments around this subject in his column in Spectroscopy Europe, see http://www.spectroscopyeurope.com/TD_21_2.pdf. | |||
| Gary Rollason (nvisible) New member Username: nvisible Post Number: 5 Registered: 2-2007 |
Fredrik wrote: 'the potassium version of the chemical is unavailable from my suppliers' This indicates that it is also unavailable from their suppliers. I understand where you are coming from, but following that logic you will need to challenge your model with every molecule known to man. Some degree of supplier certification should always be undertaken. Challenges to your model should be made on a risk basis. If you believe that there is no risk/ minimal risk, this should be documented in your justification for not performing that particular challenge. | |||
| Dennis Karl (dennisk) Member Username: dennisk Post Number: 11 Registered: 4-2004 |
It is risky to assume the supplier is supplying what they say it is. They might only be a middle man so assuming it will be a particular variant of a product is not a good assumption to make. They may have assumed they were supplied with the material they ordered. Often "Certificates of Analysis" are merely rubber stamped copies of previous analysis. All steps to make a positive identification is required, perticularly in the pharmaceutical, or related, industries. | |||
| Gary Rollason (nvisible) New member Username: nvisible Post Number: 4 Registered: 2-2007 |
If the potassium ion version is not available from your rawe material supplier why do you need to challenge your model with it? The chances of receiving this material is therefore non-existent. You should only need to challenge you model with similar substances if there is a chance that your supplier may mix them up. Quite often the spectra will visually look identical, but by the correct choice of math treatment you should/may be able to differentiate between organics with different counter ions as they will affect to some degree the positions and intensities of your absorbances. IR and NIR are two similar, but different, beasts. | |||
| Jerry Jin (jcg2000) Junior Member Username: jcg2000 Post Number: 9 Registered: 1-2009 |
Fredrik, You might find it difficult to use IR to distinguish between organic salts with different counter ions. I can't find IR spectra of the compounds you mentioned. But I found other organic salts. Attached please find IR spectra of sodium palmitate and potassium palmitate in water solution. You can see they are basically the same. Cheers! Jerry Jin | |||
| David Adamson (davida) New member Username: davida Post Number: 1 Registered: 5-2009 |
http://www.nirpublications.co.uk/cgi-bin/discus/show.cgi?tpc=5&post=3882#POST3882 Hi Fredrik. In my experience it will not be possible to make the distinctions you require. This is due to the fact that the counter ions will have little if any effect on the spectra of the organic moieties. You will have to use NIR to identify the organic part and some chemistry to identify the correct counter ion. Similarly you will probably have to test to ensure there is none of the wrong counter ion. I expect to be in a similar situation but with Magnesium Stearate, calcium stearate and aluminium stearate. I have not done any NIR analyses with these yet but the Mid IR spectra are identical so I expect the NIR spectra will do likewise. Good luck | |||
| Michael Kinsey (mike_kinsey) New member Username: mike_kinsey Post Number: 1 Registered: 1-2009 |
I work in a oleochemical company on Merseyside who use NIR for alot of there product analysis fatty acid side, gylcerine and also currently building ester product models, I have nearly five years worth of experience with NIRS, FOSS XDS instrument, from building calibrations, qualitative and quantatative, to long term validation of models, also general maintanence and instrument calibration, and i am the site expert. However the company whom I work for has recently announced that they are shutting the site and making everyone on site redundant. I am therefore looking for a job, prefereably NIR orientated as current roll involves general laboratory work also. Does anyone know of any NIR specific internet sites which advertise NIR analysts/technician jobs, I am willing to travel possibly move for a new job and a new challenge. Best regards Mike | |||
| Fredrik Andersson (fredrik_andersson) New member Username: fredrik_andersson Post Number: 1 Registered: 5-2009 |
Hello Everybody, I work for a pharmaceutical company in the QC raw materials department. I'm currently validating a method/model for single container identification of organic substances with NIRS. To what extent should I challenge my model with similar substances to prove its specificity? The problem in particular is the counter ion of organic salts, eg sodium caprylate vs potassium caprylate and sodium acetate trihydrate vs potassium acetate trihydrate, where the potassium version of the chemical is unavailable from my suppliers. Tricky, especially since my company produces parenterals, so potassium is a bad risk. Also, from your experience, to what extent is it possible to distinguish between organic salts with different counter ions? Any correlation to size of the compound, ie bigger organic part means lesser impact of the counter ion? Best Regards Fredrik Andersson |
Stearates 2 deriv