Quantification using statistical parameters derived from signal intensity distributions in surface enhanced Raman...

Raman spectroscopy is a powerful method, which provides information on molecular structures, conformations, interactions etc. However, its applications are severely restricted because of low sensitivity. Although surface enhanced Raman scattering (SERS) significantly enhances sensitivity and enables single-molecular detection, quantification by this method is still challenging because of large signal fluctuations. In the present study, the signal intensity distributions (SIDs) in SERS of adenine and thymine on the silver nanoparticle (AgNP) platform are analyzed based on more than 10000 spectra to pursue the possibility of SERS quantification. The signals always involve large fluctuations but show statistically relevant patterns. SIDs are well represented by the exponentially modified Gaussian function, which is characterized by reproducible parameters. Thus, robust quantification is feasible using the parameters derived from the SIDs. At least 200 spectra for a given concentration are necessary to derive reproducible parameter values from the SID. The mean signal intensity determined from the SIDs is proportional to the adenine concentration in the range of 10-75 μM. However, this parameter becomes independent of the adenine concentration in the lower concentration range. In such concentrations, minor events, which give distinct SERS spectra, occasionally occur but have only marginal impacts on the mean signal intensity. The corrected standard deviation of the SID, which is estimated from the complementary error function, well represents the minor events and provides a clear correlation with the concentration in the range of 0.5-7.5 μM. Furthermore, the quantification in the nanomolar range is made possible by the incorporation of sample freezing, which enables to enrich target analytes and AgNPs in a liquid phase confined by ice.