Photoelectric effect and Planck's Constant

1. Oct 24, 2013

Gringema

1. The problem statement, all variables and given/known data
I'm trying to make sense of some data taken during a lab in which we were tasked with obtaining a value for Planck's constant by measuring different stopping voltages from varying wavelengths of light. The value I got was 3E-15 eV*s, but I know I can get a better value if I account for the error when calculating linear regression. What I need help figuring out is how to do that.

2. Relevant equations
V=hv-W

3. The attempt at a solution
http://tinypic.com/r/b7fm1u/5

2. Oct 24, 2013

Gringema

Didn't know you could attach files! Here's the actual spreadsheet.

Attached Files:

• Photoelectric Effect.xlsx
File size:
14.4 KB
Views:
81
3. Oct 24, 2013

UltrafastPED

There is a lot of variation in the stopping voltage for some of the optical wavelengths - I would attempt to reduce the variance here. This is the cause of the not-so-great r^2=.92.

Your value of 3e-15 eV*s is not too bad for an undergraduate lab; the standard value is 4.136 eV*s.

4. Oct 24, 2013

Gringema

Is there a way for me to give more weight to points with less error, or to account for error bars when calculating regression Excel?

5. Oct 24, 2013

UltrafastPED

For error bars with Excel:
http://office.microsoft.com/en-us/e...remove-error-bars-in-a-chart-HP010342159.aspx

You could exclude the data with the largest std. dev. - but you have to explain _why_ you did this in the lab report. I've never reported scientific or engineering data which has been "weighted" based on the error bars.

Usually if the error is inconsistent between the various cases it means that there are methodological, procedural, or instrumentation sensitivity issues. For example, trial 3 for 445 nm is way off the mean of the other four trials; could this be a transcription error? Was there something else happening with this trial?

Because that one point is inconsistent with the rest of the data (an outlier) document it, remove it, and run the regression without that one point.

6. Oct 27, 2013

Gringema

Thanks for your help. I have one more question for anyone who knows the answer. What's the method I used to find uncertainty in frequency? It is shown in the Excel file, but I can't remember what it's called.

7. Oct 27, 2013

UltrafastPED

From looking at your spreadsheet it looks like you used bandwidth from a spec sheet; or it may have been called "line width" in which case you had to convert it to a frequency.

8. Oct 27, 2013

Gringema

Yes, but do you have any idea what the process I used to change uncertainty in wavelength to uncertainty in frequency was? Something like:
deltax=((x+deltax)-(x-deltax))/2

9. Oct 27, 2013

UltrafastPED

Bandwidth is the range of frequencies in the source - your numbers look like lamps, not lasers.

If given as a linewidth (perhaps from a diffraction grating) then you have the relationship:

frequency = speed of light / wavelength

Then the difference in two frequencies is:

f2 - f1 = c*(1/w2 - 1/w1)