TEST 1
For Test 1 we use a holmium filter. This is not described in the EP. Therefore I think it would be better to use a holmium oxide solution. I was told the holmium filter has a disadvantage compared to a solution. Can you explain that to me?

The perceived disadvantage of a holmium oxide glass is that the peak positions will vary from melt to melt, and within the melt. However, in our experience, we do not see this variation. The data shown above is the summary finding from some 100+ measurements on different glasses (from various sources). However, the difference in absolute peak values to your reference values is interesting?

When using a holmium oxide solution I've got 2 options: I can use your sealed cells or I can use for instance ampoules from other suppliers.
You don't supply a 10%v/v perchloric acid blank. as do some suppliers. Can you explain to me why I should use a perchloric acid blank for this test or why an air blank is also OK? The EP does not prescribe a blank.

The primary reference for this particular material is NIST SRM 2034. Fortunately, for most of these UV-Visible SRM's NIST have Special Publications in their 260-xx series, which detail the background to the production and certification of the appropriate SRM. These are invaluable sources of reference information, and can be freely downloaded as .pdf files from the NIST SRM web site.
In this particular case, SP 260-102 quotes on page 5 -section 3.1 'Influence of the Blank Cuvette, Solvent, and Water' '…no measurable influence was detected in the location of the wavelength ….'.

Therefore, no blank required.

TEST 2

Although the EP doesn't prescribe the use of a blank, we use 0.005 sulphuric acid as a blank for the photometric accuracy test. The EP prescribes a potassium dichromate solution in 0.005 M sulphuric acid. some suppliers offer such a reference solution and also ampoules containing 0.005 M sulphuric acid for blank measurement. However, Starna does not supply a potassium dichromate solution in sulphuric acid. Instead Starna can supply dichromate standards in 0.001M perchloric acid. Can you explain to me why this was done?

The latest revision(s) of the European Pharmacopoeia (Supplement 2000 and now European Pharmacopoeia 4) has revised the dichromate check, in relation to earlier versions. This allows a tolerance band of 57 - 63 mg, with an associated A 1%1cm calculated band - also revised in EP 4.

The text detailed below is taken directly from a recently revised publication, originally published in 1981 [1].

" P55. The NBS points out that 0.005 M H2SO4 has two disadvantages compared with perchloric acid at pH = 3: sulphuric acid has a greater ionic strength and hence greater salt effects; and there is the possibility of forming mixed chromium (VI)-sulphate complexes.".

As you will see from the above text, sulphuric acid has two disadvantages when compared to perchloric acid. These disadvantages have been confirmed not only by ourselves in the 30 years we've been working with dichromate solutions, but also by other manufacturers. From personal experience, countless laboratories also experience the problems associated with the use of these sulphuric acid based dichromate solutions, not least in the actual preparation of the solution as detailed in the E.P.
The main two advantages of sulphuric acid, availability and safer handling are cancelled out by using sealed reference materials. In addition, certified values on 'sealed cell' reference can be used to qualify the performance of the spectrophotometer in isolation of the laboratory; something which isn't possible when preparing the solutions in-situ?
Tabulated below is the direct comparison using the published calculated molar extinction coefficients in the E.P. and NIST Special Publication 260-54. 
As you can see the Absorbance differences are relatively small, and of the same order of magnitude one would expect when measuring the cells on a day to day basis, and still fall within the allowed tolerances if constant weight is assumed.

Calculated difference between 60.06 mg Potassium Dichromate in 0.005 Sulphuric Acid vs 60.06 mg Potassium Dichromate in 0.001M Perchloric Acid 

The key fact is that you are providing 'evidence of control' with a dichromate solution, at European Pharmacopoeia strength i.e. 60 mg/l, where the traceability path to the NIST primary SRM (SRM 935a) is well established by:
· Using SRM 935a to prepare the solutions as directed in NIST Special Publication 260-54.
· Certified values are obtained on a reference spectrophotometer qualified using a range of primary references, and where the calibration process has been accredited by the United Kingdom Accreditation Service (UKAS) to ISO/IEC 17025.

[1] Standards and Best Practice in Absorption Spectrometry - Edited by Chris Burgess and Tom Frost UVSG. ISBN 0-632-05313-5 Blackwell Science

In your brochure, you state that the use of potassium dichromate solvated in dilute perchloric acid is an established and well recognised method.

See NIST SP 260-54, which details the use of NIST SRM 935a (solid potassium dichromate), and the preparation of solutions in 0.001M perchloric acid, in the range 20-100 mg/l.

	Nominal	Min	Max
E, 235 nm	0.747	0.737	0.757
E, 257 nm	0.867	0.857	0.877
E, 313 nm	0.292	0.282	0.302
E, 350 nm	0.644	0.634	0.654

	Nominal
E, 235 nm	0.730
E, 257 nm	0.849
E, 313 nm	0.284
E, 350 nm	0.631

Does this mean the standards don't comply with the EP: The nominal value for the absorbance at 235 nm is already outside the acceptance criteria of the EP.

In previous editions of the E.P. the dichromate check specified manufacturing a 60.06 mg/l solution, and quoted the expected values (as shown above), together with a +/- 0.01 A tolerance. In more recent additions, the E.P. allows a tolerance of 57.0 to 63.0 mg/l, and calculates the specific absorbance A1% 1cm value, and quotes the range you have already listed.

In practice:

1. We allow a manufacturing range of 59.0 - 61.0 mg. This typically results in a 'sealed cell' concentration of between 59.4 - 60.0 mg. Hence your observation above.
2. We calculate the actual concentration in the sealed reference, and detail this on the certificate. As a Q.A. check we also ensure that the calculated A1% 1cm values meet the E.P. criteria.

Can you tell to me if I'm correct? Does the absorbance spectrum of potassium dichromate in perchloric acid differ from the spectrum of potassium dichromate in sulphuric acid? By a very small amount, reflected in the slight difference in the molar extinction coefficients, already documented. This is not sufficient to cause a problem with using materials of different spectral characteristics. In practice, if an instrument is found to be 'out of control' using potassium dichromate solution in perchloric acid, then the same will be true if potassium dichromate in sulphuric acid is used, and v.v. Maybe the typical values given in your brochure are just examples and are not those obtained using a standard containing exact 60.0 mg/l potassium dichromate but a value between 57.0 and 63.0 just as the EP prescribes? Correct. Is it OK to use you standard or not?

For the reasons already described, in practice, on a day-to-day basis, our CRM's will assist you in proving control of your instrumentation, using a certified traceable reference. However, you may wish to establish 'linkage' to the E.P. potassium dichromate check in your laboratory by running both solutions in parallel for a short space of time to provide the necessary data. Bearing in mind of course that your uncertainty budget calculation for the potassium dichromate in sulphuric acid must detail all the laboratory preparation contributions from balances, glassware, 'operative variability', etc.

Starna can supply standards with different concentrations of potassium dichromate for linearity testing. Why should I test for linearity using these standards? Isn't linearity dependent on the properties of a particular compound or sample? Please explain.

Yes, you are correct. Theoretical most liquid systems will obey the Beer-Lambert law (over at least part of their concentration range). However, in practice instrumental stray light will cause a negative deviation from linearity, to a point where increasing concentration will cause no further increase in measured Absorbance - the so called 'stray light limit'. In some instruments, the point at which this deviation begins to occur is at a surprisingly low Absorbance value! Obviously, this phenomenon can be estimated using a stray light check, as in test 3, but many laboratories still like to have a independent check on linearity, especially if they are involved in quantitative measurement work.

TEST 3

For the testing of stray light Starna can supply us with a separate water blank. For testing of their instrument some suppliers use HPLC-grade water (e.g. Milli-Q water obtained at the test site) as a blank. We use demineralised water to dissolve the potassium chloride and as a blank The EP prescribes "water" as a blank.

Can you explain why a blank should be used for this test?

Principally because at these wavelengths, there could be a significant contribution to the overall absorbance from the water - depending on its quality, oxygen content, etc.
Therefore, if you measured the potassium chloride solution reference against air, you will get a 'false' higher absorbance reading.

Can you also explain
why it would be better to use HPLC grade water instead of demineralised water and why it would be best to use the water blank supplied by Starna together with the potassium chloride Standard.

Dissolved oxygen has a negative effect on the value, and in ASTM E387, there is a specific note that states; "Apparent absorbance is strongly affected by dissolved oxygen. Bubble pure nitrogen through liquid for several minutes immediately before use. Use only recently distilled (not demineralised) water. We use commercial HPLC grade water, AR grade KCl and all our references are purged and sealed under high purity argon.

When looking at the some test procedure, it is apparently not necessary to use the same water to dissolve the potassium chloride and as a blank; is it correct that it really doesn't matter what type of water is used because water doesn't absorb the light in the 190 to 210 nm range, or could contamination or minerals in the water cause problems?

Absolutely not true, as per above statement.

As an interesting exercise can I suggest that you scan your suppliers references against a freshly prepared 1.2% w/v potassium chloride solution, that has been degassed using any available inert gas, e.g. helium, nitrogen, argon, etc., using HPLC water (treated in a similar manner), as the blank.

It has been reported to us that whilst (not surprisingly) the spectrum obtained from our sealed KCl is similar to a freshly prepared solution, the spectrum from your suppliers solution is somewhat different? I would be interested in your findings, should you decide to undertake this experiment, with other supplier solutions and I'll leave you to draw your own conclusions?

If it is just as good to use our own demineralised water instead of the Starna water as a blank when using the Starna potassium chloride standard to measure stray light, why does Starna then supply the water blank in a sealed cell? Is it just for convenience of the user? Is the water also a certified reference material or is a certificate not supplied with it? Is the Starna water blank the same batch of water that was used to prepare the Starna potassium chloride standard? What type of water is the Starna water blank and the water used to prepare the Starna potassium chloride standard?

See above.

Resolution TEST

I would like to ask you if I should perform the EP Resolution Test (using a toluene in hexane solution as a standard) when I use my spectrophotometers to calculate match factors? I'm comparing the spectra of my samples to the spectrum of a reference standard for sample identification.
Should I also perform the resolution test when I only need to confirm the presence of absorbance maxima and/or minima? Please explain.

Spectral maxima and minima should always be reported with the spectral bandwidth (SBW) used to generate the spectrum, because depending on the associated environment to the peak it may shift depending on the SBW used. When developing a scanning method, you should always select a range of SBW's, and observe the change in peak characteristic. The bandwidth to choose is the largest bandwidth, after which no further change in spectral information occurs.

Therefore, a resolution check is always a good idea, if only to confirm the SBW of the instrument. From personal experience, instrument slit mechanisms can go wrong!

System suitability TEST

Maybe you can advise me on the system suitability testing of spectrophotometers. As I mentioned we validate our spectrophotometers every 3 months. When after 3 months the spectrophotometers does not pass the EP tests, we need to explain that all results obtained within the time frame between the last validation and the current validation are reliable. Sometimes this is not so easy. Very true - I know of some auditor that would say that your instrument could have gone out of control the very second after your last check, and as you don't know when this actually happened, all data for this period in null and void. In case of using our HPLC we always perform a system suitability test every day to see if the spectrophotometers is suitable for its intended use. Maybe there are ways to perform such a system suitability test for a spectrophotometer to ensure my results are reliable. I could perform the EP tests every day using the Starna standards.

Exactly. However, because (by their very nature) these CRM's are not cheap, many people adopt a philosophy somewhat analogous to the metrological use of calibration weights for balances; i.e. an organisation will retain 'under lock and key' a set of reference weights for monthly checks, but will have a 'working set'. in the laboratory. In the case of spectrophotometric standards, once you have verified that the instrument is under control, then you could produce a 'working reference(s)' from any of your production materials, that you know to be stable. This way you can maintain your three monthly checks using the Starna references, but have some evidence of control from your working reference(s). In our catalogue we can supply PTFE screw-capped cells (1/ST/Q/10), which can be used for exactly this purpose.

Maybe it's an idea to dilute a standard solution and calculate the response factors for both the diluted and undiluted standard. In this way I'll get an idea of the linearity and the absorbance level for a certain application. Maybe you have some advise on how to perform system suitability and what test I need to perform to achieve the following:

1. Verify that the system is suitable for its intended use (a particular application).

2. Verify that the system is functioning correctly.Use the above two-level philosophy.
For some spectrophotometer systems self-tests can be performed such as the determination of the wavelength accuracy using the lines of the deuterium lamp. Maybe these kinds of tests are useful?

Theoretically, there is nothing better to check wavelength accuracy than the emission lines from an elemental lamp such as deuterium. The wavelengths of these physical references have been determined experimental to an accuracy of several orders of magnitude greater than is required in a conventional spectrophotometer, and are therefore deemed to be primary physical constants, with zero uncertainty associated with their value. However, deuterium has the drawback that it only has two useable lines in the visible region at 486.00, and 656.10 nm. Much more useful is an argon filled mercury 'pen' lamp, which has a range of useable lines from 200 - 810 nm.
This is the reason why the more enlightened manufacturers (no pun intended), are including the ability to run a mercury lamp as an alternative source in their new instruments.