Spectroscopy: Determining Phenol Concentration using Calibration curve

[chemnerd666]

Hello,

My problem is as follows:

The lab I am working on requires the construction of a calibration curve from the measured absorbance of samples of known phenol concentration to intrapolate the phenol concentration of two unknown samples. I have constructed the calibration curve and determined the equation of the line. The problem is that when I attempt to solve for the concentration of the unknown samples (in ppm), I get a negative value. I am not too sure how to correct this, I attempted changing the y-intercept to 0 but I do not think this is correct as it manipulates the "line of best fit". Is there anyone who knows what I am doing wrong?

 

[Borek]

Hard to tell anything without seeing curve and the numbers.

 

[chemisttree]

If your best fit equation is in the form of y = mx + b, examine the constant 'b'. Is it a large negative value? If so, you have a large negative offset which usually indicates a problem in your math or your experimental setup. Ideally this number is close to zero.

 

[chemnerd666]

The b value isn't large, 0.4229. The math is correct because I did it all using spreadsheet. The problem lies in solving for the phenol concentration of the unknown samples, the equation of the line is y=0.7916x +0.4229, the y intercept is at 0.5 for absorbance. Both of the absorbances of the unknown samples are 0.2254 and 0.2833 thus I cannot solve for their phenol concentration using that equation of the line as I get a negative value. What I tried was setting the y-intercept manually to zero, at least then I can solve for both of them but then I am unable to propagate for uncertainty since the equation isn't y=mx +b but rather y=mx.

I am attaching a word file with the graphs as well as the table of values.

 

[chemisttree]

Your absorbances are WAAAAAAY too high. Anything greater than about .8 is in the non-linear portion of the Beer's Law zone. You have absorbances greater than 3.5! Points off for technique. You should dilute all your samples so that you are measuring absorbances <= 0.8.

Your unknown must have an absorbance less than 0.4229? All of your calibration samples should have been with concentrations low enough to bracket this absorbance. Based on your graph, you should have made up calibration standards in the range of 0.05 to 0.8 ppm.

 

[epenguin]

I agree with the last comment.

Fitting to ax + b falsifies if the thing you want to measure is then right near the bottom of the curve - it is in fact not what you said: "intrapolating".

You might then think to fit to y = ax, forcing the curve through the origin. But then if the deviation from linearity is systematic as it is, that is rubbish too, all the higher points become disinformative and you might as well use only the first point.

If when you re-do the calibration with lower abosrbances the curve is still systematically non linear you might think of not making a fetish of linearity - just read your unknowns off the curve drawn freehand. But I shall probably have the statisticians down on me for saying that.

Or worse, I shall have the statisticians enthusiastic about it and saying how to do it best statistically!

 

[chemnerd666]

I cannot re-do the absorbances of each sample as this lab is for school. I followed a procedure made up by the lab instructor and she actually verified my calculations when making the standard solutions for the calibration curve, assuring me they were correct. So I guess I am S.O.L. for determining the concentration of the unknown samples as I am unable to do so with the calibration curve. I guess I have a lot of explaining to do. Hopefully I don't get an F on this...

 

[chemnerd666]

Thank you all for your assistance, I probably would have gotten zero on this lab if I just did my own thing haha.

 

[chemisttree]

Best thing you can do now is just use the first datapoint (1ppm) and the origin (assume b=0). Find the concentration of the unknown and note it. Then recalculate the equation of the line using only the first two datapoints and determine the equation of the line for them (1 and 2 ppm data). Recalculate the concentration of your unknown and compare it with the the single point calibration. If they are off by more than 5% use the single point calibration, otherwise use the data from the first two calibration standards for your report.