• Ei tuloksia

The spectra, after acquisition were processed and analyzed using MATLAB (Re-lease 2015b; The Mathworks Incorporation, USA). The file which was obtained as an SPC file format (.spc) file from the Wire 3.4 software consisted of spectra from

Figure 3.4: A simple setup of the Raman acquisition process (adapted from [58]).

five different points from the wafer. The built in average function was initially used to average all the five spectra. The spectra was subsequently denoized using the

’wden’ function, a one dimensional denoizing tool which is based on the wavelet decomposition and reconstruction method [92]. A threshold selection rule ’rigrsure’

which is based on the Stein’s principle of unbiased risk was selected. A level de-pendent estimation of noise was used to rescale the decomposed spectra using an orthogonal wavelet level 10. The soft threshold which is a preprocessing tool was used to reduce the background in the spectrum

The denoizing procedure can broadly be summarized as decomposition of the spectra, detailed coefficient thresholding and reconstruction of the spectra. The spectra was further smoothened using the ’sgolayfilt’ function which applied a Savitzky-Golay FIR smoothing filter to the data. Savitzky-Savitzky-Golay filters are preferred to stan-dard averaging FIR filters because they filter the signal high frequency components alongside the noise keeping the most important high frequencies. A polynomial of the order 4 was chosen with a frame size of 9. Since less fluorescence was observed in the original spectra, no fluorescence baseline was set.

Chapter IV

Results and Discussion

This chapter reports on Raman and SERS spectra of methemoglobin and hematin.

By varying the concentration of these analytes in NaOH solvent, a SERS platform is used to predict how sensitive our detection technique can be. Other factors like power of laser dependence on intensity of the Raman peaks is also looked at as well as the feasibility of our SERS system for malaria detection.

4.1 Raman spectra of methemoglobin and hematin

Methehemoglobin crystals were flattened to fine mirror crystals. The Raman spectra of the relatively higher concentrated fine crystals were acquired under an excitation wavelength of 514 nm. This excitation wavelength was chosen because the analytes were non-fluorescent. Figures 4.1 show the Raman spectra for a wavenumber range of 100−1700 cm1.The assignments proposed to the peaks in the resulting spectra are based on the labelling scheme developed by Abe et al. [93].

It is evident from Figure 4.1 that there are four distinct peaks at the low wavenumber region of 100 −1190 cm1. The peaks at 677 cm1 and 757 cm1 is attributed to porphyrin bands ν7 and ν15 respectively [94, 95]. They are modes of the pyrrole breathing and symmetric pyrrole deformation. Within the same low wavenumber region, there exist bands at 1168 cm1 and 1125 cm1 which are as-signed to the asymmetric pyrrole half-ring stretching vibrations ν30 and ν22 [50, 96].

In proteins the stretch is due to C-N and C-C bonds.

The work of Salmaso et al. [97] proved that there exist three main bands in the wavelength region 1200-1300 cm1. The appearance of some of these bands within this region is dependent on the excitation wavelength of the laser source used. This

(a) (b)

Figure 4.1: (a)Shows the characteristic fingerprints of Raman within the wavelength range of 100 −800 cm1 whereas (b) gives the spectrum between 800−1700 cm1 of methemoglobin with a laser power of 0.05 mW and spectra acquisition time of 60 s.The positions of the spectral fingerprints are indicated by arrows.

region is called the methine C-H deformation region and there appears a single distinct peak for methemoglobin at 1240 cm1 with an excitation wavelength of 514 nm. This peak is assigned toν13 or ν42 and is due to threonine rocking of CH3.

The spin state marker band region, which falls within the range of 1650−1500 cm1, has three main bands appearing in this region for our excitation wavelength.

These bands are 1640, 1586 and 1564 cm1 and are assigned to ν10, ν37 and ν2

respectively. The band assignment atν10is due to proteins such as amide I whereas the assignments ν37 and ν2 are due to heme. There exist a very feint amino acid contribution due to the phenylalanine mode at 1005 cm1. These Raman features are in accordance with what has been reported in literature, though there are slight shifts in the peaks which is dependent on factors like temperature, preparation of the crystals as well as the excitation wavelength of the laser.

Interestingly, the bands which appeared at 677 cm1 and 558 cm1 are normally

observed in oxygenated hemoglobin. This is as a result of iron-oxygen vibration resulting from Fe-O-O stretching. It can be inferred that these peaks were observed in methemoglobin spectra because of the oxidation of iron with air molecules during sample preparation. These bands are though very less appearing.

Figure 4.2 shows the Raman spectra of hematin which was obtained by finely granulating its crystals into fine mirrors as was done in the methemoglobin crystals.

The same measurement conditions employed in methemoglobin was used in the acquisition of the spectra of hematin.

(a) (b)

Figure 4.2: (a)Shows the characteristic fingerprints of Raman within the wavelength range of 100−800 cm1 whereas (b) gives the spectrum between 800−1700 cm1 of hematin with a laser power of 0.05 mW and spectra acquisition time of 60 s.The positions of the spectral fingerprints are indicated by arrows.

Relatively strong Raman peaks were observed at 757 cm1, 1372 cm1, 1569 cm1 and 1628 cm1. Similar observations also occurred for methemoglobin at 757 cm1, 1375 cm1, 1586 cm1 and 1640 cm1. The peaks 1375 cm1 and 1372 cm1 are similar peaks and the slight shift may have arisen because of factors like temperature and sample preparation. The comparatively intense peak at 1586 cm1,

that appeared in hemoglobin is entirely missing in hematin. Characteristic shoulders are also observed in hematin spectra at 1341 cm1 and 1399 cm1. In the low wavenumber region, Raman bands which are less intense and of the same height by visual inspection are observed at 675 cm1, 692 cm1 and 713 cm1.

The Raman peak at 1628 cm1 is assignedν10as a result of the combined stretch-ing vibration in the vinyl group and its pyrrole part. The vinyl group gives a good de-scription of the Raman spectra of porphyrins and are always present in hematin [98].

This assignment clearly validates experimental findings [94]. The Raman band at 1569 cm1 is assigned to two normal modes. One of these modes is located to one of the porphyrins. The second mode is a combined effect of porphyrin and the stretch-ing of the C=C vibration consiststretch-ing of the iron-carboxylate propionic side chain.

This peak has been very sensitive to interactions between hematin and chloroquine in aqueous environments causing a π−π stacking between them and subsequently changing its depolarization ratio [99, 100]. An effect that is strongly being studied to ascertain the binding relationship between anti malaria drugs and hematin. The Raman peak at 713 cm1 is also due to these skeletal modes of porphyrin and is assigned toν4. Two peaks 1185 cm1 and 1307 cm1, are due to stretching of C-O-C and C-C/C-N respectively.