The Gaussian output (in GView and as part of the output) report two bonds as:
0.963
0.958
Is the final significant as reported by Gaussian and uncertain term?
Should I report the bond length distance as:
1) 0.005
or
2) 0.01
I tried (1) but I was informed that this was incorrect.
Thanks
Dear Anthony,
You should report as experimental which is like 1.463 A (3 digits after the dot).
https://www.google.ps/search?q=bond+lengths+table&biw=1024&bih=588&tbm=isch&imgil=OKJlItcfxQBOnM%253A%253BU6pSLqB9DLusQM%253Bhttp%25253A%25252F%25252Fwww.chegg.com%25252Fhomework-help%25252Fusing-bond-lengths-table-92-p-340-assuming-ideal-geometry-ca-chapter-10-problem-95p-solution-9780073402659-exc&source=iu&pf=m&fir=OKJlItcfxQBOnM%253A%252CU6pSLqB9DLusQM%252C_&usg=__75mmCGOi4fv5d83YcxRjxR-KzJo%3D&ved=0ahUKEwimurWcusLLAhWKWhoKHZSEBlAQyjcIJA&ei=HuDnVqazLoq1aZSJmoAF#imgrc=5EU7MuJqyLD0rM%3A
Rafik
Hi Rafik,
I do agree with you, and this is how I reported, however a colleague suggested that this is well beyond the accuracy of the Gaussian calculation. Any thoughts?
Given that the coordinates in the Gaussian log file yields cartesian coordinates of values to six decimal places and a bond distance of up to 4 decimal places, I would say that 0.005 is accurate.
I'd say that the values are identical within the expected accuracy of the method, if it's any kind of DFT or MP2 in any basis set (the accuracy is around 0.1A, so both of your values are equal to 0.96A). Although Rafik has point that most papers nowadays give bond lengths values up to 3 decimal points, as in most xray studies.
Gaussian printing options (xyz with 6 digits) do not bear any information about the accuracy, it's just a convention.
Dear Anthony,
I totally agree with Bartosz. I recommend that you use such as in the experimental 1.254 A. There is no need for 4 decimal; three is enough as you wrote.
Rafik
I do have a bit of an issue then. I have been reliably informed, from literature, that deviation from bond length can be use to measure the presence of hydrogen bonding e.g., water to a hydroxyl group, you would expect to see a stretching of the O-H bond as the hyroxyl H is attracted to the O of water. This would be a comparison between two computational models, which from my experience consistently shows a bond stretch in the presence of water.
Dear Anthony,
Yes. Hydrogen bonding results in O-H stretching, though the difference is not so large.
If you have an effective hydrogen bonding in which the bond angle of H---O-H is about 180 degrees you will see a maximum stretching. This phenomenon can be seen also by FT-IR, stretching of the O-H group.
Rafik
Anthony,
I think there is a general confusion here about two related, but distinct properties - accuracy and precision.
http://cdn.antarcticglaciers.org/wp-content/uploads/2013/11/precision_accuracy.png
Your results are probably not too accurate, i.e. a good representation of the true bond lengths. But the calculations are precise enough that you will find similar trends across different systems, and (hopefully!) always the same values when you redo your calculations. From my perspective, it's okay to give three decimal places if you want to highlight the effect of a hydrogen bond within the calculations themselves (I've done it that way myself in the past).
In direct comparison to experiment, you may want to reconsider the significance with which you present your theoretical data, because then you're not looking at the relative changes, but the absolute values -- so you would care more about accuracy, not precision. If your experimental data has only two significant figures after the decimal point, then it's no use to give the calculated results with three. But either way, the precision of the calculations (not their accuracy!) should allow three decimal places.
Hi Matthias,
That was a fantastic and very helpful explanation. The second paragraph regarding comparing bond lengths between systems within the context of an identical theoretical approach, was exactly what I wanted to hear.
Thanks.
- Badges
- Science topic