Dear Mina Solhtalab , water forms hydrogen bonds with amides, hence the changes you observe within and without water in your experiments.

Amide I vibrations overlaps with the H-O-H bending vibrational band at 1650 cm-1, this band is approximately one order of magnitude stronger than the one for amides I in proteins. One possible alternative would be to use D2O instead of H2O, deuterated water doesn´t show overlapping with amides I or II, what is good from a spectroscopic point of view, but use of D2O changes the interactions and therefore could have effects on the protein folding and also it will shift the amides bands. Working with water as solvent would require use of water background substraction and to improve the signal to noise use of thin films of solution (something you already have by using ATR).

May be the following papers would be useful to you:

https://watermark.silverchair.com/39-8-549.pdf?token=AQECAHi208BE49Ooan9kkhW_Ercy7Dm3ZL_9Cf3qfKAc485ysgAAAskwggLFBgkqhkiG9w0BBwagggK2MIICsgIBADCCAqsGCSqGSIb3DQEHATAeBglghkgBZQMEAS4wEQQM_OS9GyedZmmNg7mnAgEQgIICfNzxbYwGo7_bZlMGgbP8V2Z5852bPiyIcxYEpaC41xXTgItUqNc-s5-jwu-dCyotslOk7VG1M2Rln1J36cmce7dwyeyVqW5arnzEh-czRXJ5EBhpi3zJToeTqTFVnjDAbiMCEmROcaOzoC9Vx8gHbPhyuGNk7TUj4nKa5eX8GvC60IqbYvVd-f9H2can2-0ABbe9k1nEB_RAs-dTI0J5BmZCZoovW_Q93yZvFBi9MJBY2b57ZelVEfpaLc0_2ec3X8eGbHteDIIvavzHIUPivuX2-iKckPQPfNwlH2aLYPgqIP2bom4K2DsRFGZR8D8ZUxAgnAOiyMe7eOcNn1F8fc3Mbcngl1jhkmVshF5iRVtjxHiKkNXfn6o4KAui2cGTuPzu4Aj6Y23hgCKV7jGZJvY9zjvoz8dcfaHbO_pQGAbPvjXJ7LLyo5Qneo4HorcQ6LunHMoxyWW3_3plo2S3j1u3sgn6N2ihahRl0qWNriO343L2WAHcA_tpOkKRFb2LbPyYH3gSQS82e9fBjhR5gb1HeEPV-v-Fqv-YMHQYy2_Dag0NUHL1wUXqnZxqV2opZJ5j0MkrB6naS7Zz1t5NldjZdmo7VjxTfMj8kYrMygTPxRCUyjdAsXKUbuXVUJqHk9_DAebf3ERRrH2YCNIFhoqk0oP4-3W4NtUbf15fytJVH1MjN65eMPAiR63IVXDeaztkA8yGmBeqw6uTjTm6SefWyLaeKE0Tk3VRTeDDjExeFWdZrflFo-FI21Lh4En-wtVeyff2tMtKzJtW729x_9r1I6ozAxnbIVm-1QYdVefMdrb0f2Fmw9E48GxmV6oLX17TSsWCtv0QV9h0Ug

Article Amide-Amide and Amide-Water Hydrogen Bonds: Implications for...

Alternatively, you could also give a try to Raman spectroscopy, where the water bands are not a problem and amides I and II can be clearly identified.

May be the following application note from Horiba could be interesting:

https://www.horiba.com/fileadmin/uploads/Scientific/Documents/Raman/HORIBA_webinar_proteins.pdf

Hope this helps. Good luck with your research!

Unfortunately, all amide bands are rather heavily overlapped with water bands. However, there are some difference in frequency. So, here is what you can do.

1. Obtain a beautiful pure water spectrum using your ATR system.

2. Then, obtain your protein solution spectrum without changing any conditions, except for the sample.

3. Convert the spectra into the absorbance mode.

4. Subtract the pure water spectrum from the protein solution spectrum. The water in protein is slightly different structure than in pure water. However, they are already hydrogen bonded anyway. Therefore, the peak shift is not significant. You can see the difference in the maximum peak position of the protein solution spectrum as a result of the heavy overlap of the Amide I band (depending on the conformation of the protein, the frequency of the Amide I hand shift a bit, but it is near 1660 cm-1). The water band is at 1640 cm-1. Thus, you can subtract the water spectrum. The difficulty is that to know when to stop subtracting it. If you know exact concentration of water in your protein sample, since you are using the same (slightly different as the refractive index of the pure water and protein sample are different), you can reasonably well subtract by simply multiplying the water concentration to the water spectrum. For example, if the protein solution is 10wt%, then multiply 0.9 to the pure water spectrum and subtract. Make sure that the spectrum around 1640 cm-1 should never go below the baseline. In fact it should be slightly positive as it is near the Amide I band. As you subtract the water peak at 1640 cm-1 (increase the fraction of subtraction) you will be able to see the peak position would shift toward Amide I band position. If your protein solution is about 5-20wt%, you can use this approach relatively well.

In any measurement of an analyte in solution, the concentration of the analyte is important - you should try increasing the concentration, to see when the water spectrum starts to change. See for example the attached.

Extracting the relevant data is not trib=vial because of the overlap of the Water peak with the amide 1. This paper may give some hints

H. Yang, S. Yang, J. Kong, A. Dong, S. Yu, Obtaining information about protein

secondary structures in aqueous solution using Fourier transform IR spectroscopy,

Nat. Protoc. 10 (2015) 382–396, doi:http://dx.doi.org/10.1038/nprot.2015.024

Thank you all ( Manuel Gómez, Hugh J Byrne, and Hatsuo Ishida ) for your guidance and helpful answers to my question. I could finally get good amide peaks after increasing the concentration, increasing the signal-to-noise ratio, and step by step subtracting the exact background buffer from my sample. Thank you Hugh J Byrne for the very helpful "Nature Protocol paper" you shared, it helped me a lot.