# Finding the structure based on spectra

I had to identify the structure of a drug based off the spectra I had. I suck at this. I gave it my best go and got some ugly structure, only to check it using a database and find out I was wrong. I gave up and searched for the chemical structure I was given, only to find out that the drug was diazepam and I was on the right track (my aromatic substitution was correct, and I got the correct degree of hydrogen defeciency, but I didn't know how to tie everything together and I was stuck with one more quaternary carbon that I didn't know where to place). I feel like the spectra I was given are hard to read (specifically the HSQC, HMBC, and COSY); do you all feel the same? I really want to know how to arrive at the answer.

The images are the spectra I was given. I was also given mass and IR strectra, but I cannot post too many pictures with my reputation. If you all need better resolution images, let me know. The molecular formula is C16H13ClN2O, and the groups I found with the IR were NH, C=O, and benzene. I am kind of afraid to ask the instructor at this point because I already found the answer. It wouldn't hurt to ask him anyway, would it? I just don't want to lose points.

EDIT: I took the original images down because I did not want the instructor to find it, if he hadn't already.

• Additionally, I would love any tips to help me get the structure correct the first time around. – Uchuuko Dec 14 '20 at 21:38

The most practical approach is to identify the structure of an organic compound by combination of spectroscopic techniques. This is why you were provided with UV-Vis, IR, NMR and MS data, because it equally reflects the eventual approach in the lab after product isolation and purification, too. Hopefully you have sufficient background what theses techniques assess (if not, consider consulting e.g., the spectroscopy section here on ChemSE, or on Chemistry LibreTexts).

Then, start training with simpler molecular structures in first place. You may use, for example, Field's training booklet from the shelves (either print or/and electronic resource) of your school's library. The obstacle here: for the majority of the sets, there is no answer in the book (the key is available only to instructors). You may however list structures likely to be in agreement with the data per set with public references such as the SDBS spectral database. You equally may train yourself picking an organic compound from Wikipedia and list the features a theoretical UV-Vis IR or MS or NMR spectrum would contain and again compare, e.g. with SDBS.

Yes, there are programs to predict the position and shape of NMR spectra, and you equally may find diazapam with a search via Hill formula on nmrshiftdb2.org, too. They may help you to internalize how structural features (e.g., CH vs CH2) show in the APT spectrum because you may select either of the type of nuclei to be displayed, or to highlight their attribution of a calculated position of a signal to an atom in a given structure, or their indication of a cross peak e.g. in the HSQC. Aside from an exam where these programs might not available to you, don't rely too much on them. E.g., the position of signals in the $$\ce{^1H}$$-NMR spectrum depends at least on the concentration of your analyte, the choice of solvent (e.g. chloroform or DMSO) may influence if exchangeable protons on alcohols and amines yield a broad signal, or split into a multiplet (if there are coupling nuclei).

The only way to get better in this is training. If the structure suggested is not in agreement with all of the spectral observations, the assignment is erroneous and one has to start again. It may be demanding, but don't give up.

As a starter:

• N-NMR: there are two types of N with different chemical environment
• H-NMR: there are two sets of alkyl-H; by integration possibly one $$\ce{CH3}$$ and one $$\ce{CH2}$$ or integer multiples of both.
• APT-NMR: again indication for a methyl, and a methylene group; count of CH: ... which are in the region expected for aromatic CH, count of C: ... etc.
• Thank you so much. – Uchuuko Dec 15 '20 at 12:11