F. Bayrak c¸ eken et al. / Spectrochimica Acta Part A 67 (2007) 1276–1280
1277
include some measurement noises because of insufficient num-
ber of samples. This type of noise can be removed by applying
digital filtering techniques like smoothing, averaging [15,26,28].
uum systems in the set-up the absorption spectrum was recorded
between 190 and 750 nm band.
The image of the original absorption spectrum of -methallyl
radical was scanned by HP-Scanjet 7400c scanner at 600 dpi res-
olutiontoobtainahighresolutiondigitalimage. Theimageofthe
2
. Experimental

-methallyl spectra having 11 time intervals and 6 main absorp-
tion bands recorded from the measurements was also examined
by using image analysis techniques in order to obtain the exact
absorption band places from the spectra. For that reason the
original spectrum image was transformed to gray level image
as:
The 2-methylbut-1-ene used was supplied by Koch Light,
pure grade. The flash photolysis apparatus consisted of two
parallel photo-tubes arranged in series, contained in a reflec-
tor which was flushed with nitrogen. The photoflash energies
used were 780–1125 J, and the flash duration was 2 s. All
quartz components were spectrosil grade. Free radical absorp-
tion spectra were recorded on Hilger small and medium quartz
instruments using Ilford XK fast blue sensitive plates sensi-
tized with sodium salicylate. The plates were photometered on
a Joyce–Loebel recording microdensitometer with step wedge
calibration. Product analysis was carried out on a Perkin-Elmer
F11 chromatograph fitted with a gas sampling valve. Peak areas
were electronically integrated and retention times were checked
by direct calibration. The 2-methylbut-1-ene was further puri-
y(i, j) = Gray-level{I(i, j)}
(1)
where I and y are the original spectrum image and resulting
256 gray-level image, respectively. The intensity levels of the
spectrum image was normalized by using MATLAB functions
in order to reduce the computational costs as follows:
y(i, j)
yn(i, j) =
(2)
{
y(i, j)}max
−
7
fied by repeated vacuum (background pressure was 10 Torr),
distillation before use. The range of pressures studied was:
where yn is the normalized image. Each horizontal time interval
from “before” to “after” in Fig. 1 were marked and separated
from the original spectrum image to define the region of interest
(ROI) being analyzed as follows:
2
-methylbut-1-ene (0.5–20 Torr) + argon (100–700 Torr), thus
maintaining isothermal conditions. All kinetic spectroscopy was
carried out for single flashes giving <1% decomposition, as
multiple flashes resulted in the appearance of a continuous
absorption spectrum in the wavelength region of the free radical.
The electronic spectra of the -methallyl radical has been
recorded in the ultraviolet region, and its assignments discussed.
The intense regions of absorption of the ␣-methallyl and -
methallyl radicals (238 nm) lie 13 nm to long wavelength of that
of the allyl radical, and are easily distinguished from it. Although
the two methallyl radicals have their intensity maxima in the
same wavelength region, the individual vibrational structures
are quite characteristic, and furthermore the -methallyl radical
absorbs in two sharp bands at 225.1 and 223.1 nm, which are eas-
ilyidentified. Inallruns, theknowntransitionsofthe -methallyl
radical was observed at the shortest delay times, and a typical
result is shown in Fig. 1. The flash source emits continuous radi-
ation at wavelengths down to 170 nm, but paramagnetic oxygen
cuts the emissions below 190 nm, therefore without using vac-
yb(i, k) = y(i, j), k ∈ ROI , b = 1, 2, . . . , 11
(3)
b
where y s show the blocks to be tried. Each time intervals
b
cropped as in Eq. (3) was defined as a 1-dimensional density
row vector x(n) by computing the averages through the columns
of the matrix y (for each vertical axes) as:
b
M
ꢀ
1
x(n) =
yb(i, n)
(4)
M
i=1
Each element of the vector x(n) defines a distinct wavelength
and the amplitude value of that element gives its optical den-
sity in the spectral image. The elements of the vector x(n) were
smoothed by curve fitting in order to avoid noise effects and
obtain a better visual graphical representation for all absorption
bands than the original. Each band was identified from den-
sity information and plotted again as shown in Fig. 2, which is
Fig. 1. Flash photolysis of iso-butene (2 Torr) and N2 (700 Torr) showing the spectrum of the -methallyl radical and its decay. Flash energy = 1125 J.