micro balance

[JM]

thanks Hans /Dr.

As suggested by Peter , The theory put to practice and data obtained after subtraction, the 2 mg weight meets the 3X SD criteria but still fail in measurement uncertainty as per USP.

By the way i forgot to mention the balance, it is MX5-Mettler.

The argument put to rest !!!! time to buy more impurity standard ( more money to USP ) now i know why USP have these strigent requirment

Thanks everyone for this Interesting discussion.

Regards,

JM

[DR]

You could also consider a secondary (cheaper) source for the material, then assay it against USP material so that you can have a defensible purity factor for the secondayr material.

In a regulated lab, there need to be procedures and acceptance criteria in place in addition to the ability to answer any stability questions that may arise. (In other words, it isn't worth the effort unless the material is really super expensive).

[unmgvar]

JM, your suggested approach is actually correct.

this is exactly the approach used to perform loss on dry tests.

you get one of those glass thingy (can't remember the english name for it), you weight a certain amount of substance, after you dry the substance you reweight and then you substract to know the amount of water that was in the undried sample.
it is possible because you do all of this inside the working range of the balance and you make sure that you weight enough substance in such a way that the balance is sensitive enough to tell you the amount of water lost.
example: LoD spec is NMT 1% let's say.
your glass thingy weighs around 50000.00 mg
you weight 1000.00 mg of sample

total weight 51000.00 mg
you dry it and weight again. new weight 50999.90 mg

water content is 0.01%

DR you analogy of the Queen Mary is faulty. if the system used to perform the weighing is sensitive enough and in the weigh range accuracy then yes you could do it. in your Queen Mary's case it is not because the amount of deck chairs represent a negligible change in the total weight and the system to weight the Queen mary not sensitive enough.
but what if we decided to transport an amount of deck chairs in the Queen Mary that was about 20% of the mass of the Queen Mary (like in JM' case)? then we could tell the weight of those deck chairs. The main question is to what degree of precision we would be able to tell the precision of the weight.

HW Mueller, as you can see from the LoD example the balance can safely handle weight differences that are of very high magnitudes.

the main think to remeber is that any balance does not know that it is weighing 2 things A+B; sample+paper; it tells you the weight. we are influenced here by the fact that we know that we are weighing to things together at the same time.
As long as you are working within the balance specs you are fine.

you would need to have a good SOP for the procedure since you must declare the precision of weigh you will use. personnally i would use 0.00 mg and not 0.000 mg. you must as well specify the minimum amount sample that you allow your self to weigh in such a way, do yourself a favor and don't go below 1 mg.
because you are puting the container with the sample afterwards, i would also specify specific types of containers that must be used since they were checked for adding neglible volume changes.

[HW Mueller]

Again, the argument was that increasing the dead weight can not INCREASE the efficiency of weighing, which was the original suggestion. In your case, unmgvar, you could easiliy find a scale which could give you higher precision if you used a "container" that weighed a few mg instead of 50g, or: you could use the scale which is good at 50g and put 50 g sample to dry without container.......

[Peter Apps]

Can we draw an analogy with analytical limit of detection (LOD) and limit of quantitation (LOQ) ?

The limit of detection is the smallest quantity of analyte that gives a signal which can be disriminated from the signal that occurs in the absence of the analyte (aka the blank) with a specified probability. For a balance reading to 1 mg the limit of detection is 1 mg with a 50% probability of detection (you either see 0.001 or 0.000 in repeated weighings of exactly 1 mg).

The LOQ is the smallest quantity of analyte that can be measured with a specified uncertainty. In a typical analysis LOQ depends on repeatability, the slope and uncertainty of the calibration, and bias (which is sometimes expressed as "recovery"). An LOQ is always higher than an LOD.

The 3 mg lower limit on JM's balance is an LOQ - according to the USP (and probably tested with the actual balance on site) it is the smallest mass that can be measured with a relative uncertainty of 0.1 %.

If this analogy is sound, weighing 2 mg as a difference between 5mg and 7 mg is the same as determining 2 mg of analyte in a sample as the difference between blank matrix spiked with 5mg of analyte and sample spiked with 5 mg of analyte.

Enjoy the weekend Peter

[HW Mueller]

Peter,
do you mean that one might run into trouble (in UV detection for an example) if one spikes a sample, which contains lots of analyte, with the same amount of spike used in a blank which has nothing (no peak) at the analyte rt? One could indeed.

[Bruce Hamilton]

For the poster asking about the first article Peter pointed to, it was July 2004, and is available as a pdf at the site. I think his links both pointed to the second part.

http://www.lcgceurope.com/lcgceurope/da ... rticle.pdf

For those curious about what agony the USP is about to inflict on balance calibration, there's a neat little note at R@D magazine by Mark Ruefenacht and Tim Studt. Not sure if the long URL will translate OK here, but if not use Google and search on " USP <41> balance " and look for R&D magazine. I don't think that all proposals were accepted by the USP.

http://www.rdmag.com/ShowPR.aspx?PUBCOD ... PRODLETT=R

The critical aspect is that ongoing determination of balance performance will be required, and balance uncertainty and minimum weight determinations will be more based on users and daily use.

Bruce Hamilton

[Peter Apps]

Hi HW

No that is not what I had in mind. The discussion has been about measuring small quantities, not large ones. I was trying to draw an analogy between weighing a small mass by adding extra mass (JM's procedure) and analysing a low concentration by spiking with extra analyte.

Peter

[HW Mueller]

In what LC detection could one improve quantitative determination of an unknown by swamping it with a known? I thought one spikes to verify the rt of a peak in a complicated matrix.

[Peter Apps]

Hi HW

Just to be absolutely clear (I hope) - my intention was to show that loading a balance with extra weight is not a valid procedure. I was trying to draw an analogy between weighing small masses by adding extra mass (which some posters seem to think is acceptable) and analysing low concentrations by spiking the sample with extra analyte, which I thought that most analysts would recognise as being self evidently absurd.

I am certainly not advocating either loading a balance with extra weight, or spiking samples with analyte.

Peter

[HW Mueller]

Great! I am very much relieved, that makes it three in agreement.

[unmgvar]

My apologies for the delay,

as i was saying in my opening remark, i am basing myself upon already existing procedures which are daily used and approved both by the FDA and the EP Phr. the Loss on dry and residu on ignition procedures.
the USP procedures are 731, 732.

when we come to deal with a loss on dry we are faced with one major aspect: we cannot directly know the weight of the water in our sample. it is found indirectly by weighing the sample itself before andafter drying it.

i am also adding a link to a loss and dry procedure to better explain myself. this procedure is very good in my view because it takes into accounts various important subjects relating to our discussion.
http://www.ecn.nl/docs/society/horizont ... annex2.pdf

sections 8-10 at pages 5-6 will be the most relevant.

section 9: "Depending on the expected water content, weigh into the evaporating dish or crucible (8.3) a suitable amount
of sludge, mb, so that the dry matter obtained has a mass of not less than 0.5 g."
this restriction is forced on us so that we can get with a better degree of certainty for two important parameters:
1. the weight of the sample
2. a sustannable amount of weight change between the undried and dried sample weights.

both parameters must be within the working range of our balance of course.

section 9: "The mass (mc – ma) shall be regarded as constant if the mass obtained after another hour of drying does not
differ by more than 0.5% of the previous value or 2 mg, whichever is the greater (3.3)."

this part also set us what is the minimum uncertainty parameter for our test or what we call our 3X SD criteria.

most importantly we can see that we follow a correctly set procedure we can get a weight by using indirect measurements of the changes occuring to the balance. remember we never weight the water itself, we weight the "wet" and the "dry" sample and from the changes that of the sample that we record with the help of our balance we can know the weight of the water.
this is permissible because we are working within the allowed range of of weigh of our balance and we are making sure that the change occuring is within the accuracy range of the balance as well.


JM starting weight point is 5mg within the weight rangeof his balance. he is inducing a weigh change wich is around 2 mg uppward which is also within the working parameters of his micro balance, therefore he can use the indirect result he has to know how much sample he has in order to do his work.

we must not forget that the balance has absolutly no idea what we put on it. it sees only mass, and mass changes. as long as we are within the specs we can safely do it.

[unmgvar]

Peter Apps let's take a hypothetic situation in which i am doing some research on some defiecency disease in rats.

my sick rats produce not enough of my product but they still produce some of it. unfotunately the results are always below LOQ but for the sake of simplicity above LOD.

what if i take a sample from my rats and spike it in a controlled manner once with a certain concentrsation of my std and what if i do it a second time with a second known concentration in such a way that my response will be above LOQ but of course not too much above?
what i get is a linear curve F(x)= a(x) + b that of course does not pass through the 0,0 point.

if i go and look at 0= a(x) + b i will get a result of (-X) which is the amount of my compound naturally occuring within my rats.

if i can show good RSD, good linearity and that all other parameters that could make up B in my linear curve are negligible, can i used my result to show the amount of naturally occuring compound in my rats?

[unmgvar]

HW Mueller wouldn’t agree with me that one of the coolest things about democracy is that it gives the minority the chance to let its side be heard and maybe hope to change the decision of the majority which would otherwise be wrong?

also between USP 731, 732, their 2 analogs in the EP Phr, me, myself and I we get 7, which is a nice and very cool prime number don't you think?

[HW Mueller]

There is nothing more democratic than trying to stay with the facts or elucidating facts.
Any further statements could only be repetitions.