[Babcock_Hall]

I have been teaching a laboratory on spectrophotometry for a number of years.  It calls for a stock solution of 100 μM 4-nitrophenol (MW 139.1).  This year I made up two versions of this solution (one was 1 L using about 13.9 mg solid, and the other was 2 L using about 28.0 mg solid.  I mixed 0.5 mL of each stock with 0.5 mL of pH 9 buffer (the pKa is about 7).  In both cases, the absorbance is much lower than expected.  I used a molar absorptivity of 18,000, and I took into account the twofold dilution. The two absorbances yielded concentrations of stock solution of about 71 μM, not 100 μM.

For many years I used a bottle from Sigma, and the gravimetric concentration was always close to the one obtained by spectrophotometry.  Then last year I used a different supplier, and the concentration was low by almost twofold in concentration IIRC.  This year I used a third supplier, and the material looks crystalline.  I can use the material I prepared, but I would like to know what might be happening.  Any thoughts about what might be going on?

[discodermolide]

Did you get a certificate of analysis?
Perhaps the material is contaminated with another isomer?
It might be worth running a ¹³C-NMR and checking it out by TLC.

[Babcock_Hall]

Good ideas.  One isomer, 2-nitrophenol, has a much lower molar absorptivity, around 4800 versus 18,000 for the 4-nitro isomer.

[Babcock_Hall]

The H-1 NMR in CDCl3 looks like a para-substitution pattern, and the chemical shifts of the aromatic hydrogen atoms are about what they are in d₆-DMSO (spectral database).  The hydroxyl hydrogen is only 6 ppm vs. 11.something in  DMSO.  I don't see any evidence of a second isomer.  If there were three waters of crystallization, that would come close to explaining the absorbance results, but I have no other reason to believe that water is present.

[discodermolide]

Perhaps a TLC comparison of the three samples would help?

[Babcock_Hall]

Do you have a hypothesis of what might be going on?

[discodermolide]

Since using the Sigma material you have been having problems. So it must be something in the new batches from these suppliers.
Perhaps try and get an elemental analysis and a sulphate ash.

[408]

I second the idea of disco to get an EA.  I suspect either some moisture is present(not necessarily hydration waters), or that perhaps an amount of a metal salt of the 4-nitrophenoxide anion is present from an incorrect neutralization during synthesis.  Check ICP for metal presence.

Just recrystallize it once before use and it should get cleared up.

[Corribus]

Check the easy things first: when was your balance last calibrated? 13.9 mg isn't very much mass, a small error in your balance calibration can really throw your measurement off.

Also, where is your extinction coefficient value coming from?

[Babcock_Hall]

I checked the balance with a 500 mg standard.  The molar absorptivity is given as 18,300 in "Data for Biochemical Research," by Dawson et al.  They cite a paper from the Journal of Biological Chemistry 170 476 (1947).

[Corribus]

Ok. Having done many, many extinction coefficient calculations over the years, and especially watching many people do them incorrectly, I can tell you that the most frequent mistakes I've observed include: (1) weighing out too little material such that the weight error is large and (2) trying to calculate epsilons using solutions that are too concentrated. 

I am looking at your numbers and a few things stand out to me:

You made two stock solutions, one with 28 mg and one with ~14 mg, and reported that they both give you the same erroneous calculated concentrations using the extinction coefficient. I don't know what kind of balance you're using, but my rule of thumb for calculating extinction coefficients is never weigh out less than 100 mg. If your balance has an error of (for the sake of argument, let's go large) 5 mg, if you use 100 mg you'll have 5% error in your measurement.  If you're weighing out 14 mg, the error goes up to more than 33%! Point is that weighing out a lot of material reduces the error from the balance.  I'm not kidding when I say I've seen people try to weigh 5 mg out on a standard balance and then they wonder why their epsilons have errors of like 300%. Now, you did two measurements and got the same erroneous concentration value using a literature epsilon, which seems to suggest the balance isn't the issue, but still to be safe I'd recommend in the future you weigh out 100 mg minimum and then dilute the solution to get your desired concentration. Sometimes it's unavoidable to use less if that's all you have, but since you've bought this stuff, I presume you have it in abundance, so why risk it? (Plus, I'm not sure a 500 mg standard is that relevant when you're weighing out 13 mg).

Second, and definitely more important, I ran your numbers and using a 50 microMolar solution (after mixing with your buffer) with an epsilon of 18,000 M-1 cm-1 and a 1 cm cuvette (I'm assuming), I calculate an OD of 0.9.  This is usually WAY too high for a precise concentration measurement.  Here's why: in the Beer's Law formulation, the absorbance value in OD is equal to -log (X), where X is the ratio of the intensity after absorption to the intensity before. If your OD is 0.9, this means that almost 90% (87.4%) of your light is being lost by the time it gets to the detector.  Most UV/VIs detectors aren't very sensitive and have trouble measuring differences in light intensity when the intensity is pretty low. For a good concentration calculation I usually like to be at OD values of 0.1 or even less, depending on the sensitivity of your instrument. This ensures that your instrument will be able to perform optimally and give a precise and accurate result. Moreover, at higher concentrations, deviations from Beer's Law due to the analyte itself begin to become pronounced. This can vary depending on compound but I've never ran across one where Beer's Law breaks down in the 0.1 OD range: 1 OD, definitely.

Third, the buffer you are using could very well be affecting the extinction coefficient, which I'm assuming was not determined in a buffer solution by your literature source (maybe it was, in which case, disregard). Optically active chromophores are often pretty pH sensitive.

For these reasons, before I'd go running off doing elemental analysis under the assumption that a material purchased from a major chemical manufacturer is almost 30% impure, I'd do the following:

Weigh out 100 mg, add 100 mL of your solution (with 1:1 with buffer), or whatever it takes to dissolve everything, and then dilute to what you think should be 100 microMolar - about where you were before. Then do serial dilutions of about 20% each time and record points to ensure that you are in the linear range of OD. I'd back down until I was 0.1 or so OD anyway just to be safe.  Take about five points in the linear range and determine a new extinction coefficient using a linear regression. The best way to do this is prepare your different concentration using fresh stock each time, to eliminate errors from the balance, but this can be laborious.  Anyway, if your calculated extinction coefficient compares well to the literature value, then great.  You should be good to go.  If it doesn't compare well, and the literature value was taken in the same buffered solution you are using, then it may be time to start thinking about issues with the substance itself (purity or whatnot).

That's what I'd do, anyway. Otherwise you could be chasing an impurity that isn't there, when the problem all along is the procedure you're using to do your experimental determination of concentration. Even so, it may not be a bad idea to purify your material before you do this (recrystallize, chromatography), although electronic absorption experiments are typically forgiving of impurities unless they are in the multiple% range.  Incidentally, I'm not sure what lab you are setting up, but if you generate your own extinction coefficient from your material, an impurity in it may not even matter (depending on what you're using it for).

[Babcock_Hall]

Corribus, Each year the students construct a standard curve with 100-750 µL of the 4-NP solution, and I have never noticed a deviation from Beer's law.  Also two different spectrophotometers gave similar results.  On the other hand, the issue of weighing is a large potential source of error, as you point out.

[Corribus]

Well, I was speaking generally. Deviations from BL law vary from chromophore to chromophore.  If you have verified you're in a linear range, then you're fine. On the other hand, the % of transmitted light is only dependent on the OD value, not on the BL linearity, and your spectrometer's precision and accuracy may be poor when you are trying to measure values above 1 OD.  Each spectrometer is different, of course, and you should check with the manufacturer about what their recommended detection ranges are if you have reason to believe this kind of error may be to blame for poor results.

This is why I always recommend targeting the 0.1-0.4 OD range, because there are almost never deviations from BL and most spectrometers are reasonably accurate in this window.