Optics: measuring and testing – For size of particles – By particle light scattering
Reexamination Certificate
1999-09-28
2001-02-20
Font, Frank G. (Department: 2877)
Optics: measuring and testing
For size of particles
By particle light scattering
C356S337000, C356S338000, C250S575000
Reexamination Certificate
active
06191853
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for measuring particle size distribution and method to calculate particle size distribution of particles dispersed in a solvent.
2. Description of the Prior Art
Conventionally, there has been conducted a method for directing a laser light to particles dispersed in a solvent, detecting a scattered light from the particles by a detector, counting the detected scattered light to thereby measure the particle size distribution of the particles (which method is normally referred to as photon correlation spectroscopy (PCS)). With PCS, however, it is necessary to set some ranges of sampling time in advance and then measure the limits of the particle size ranges. In addition, to provide a wide measurement range for measuring large particles, it is necessary to make measurements while keeping the particles in a stable state for a long period. Besides, with this method, it is necessary to measure a particle size distribution in a diluted state so that scattered lights do not superpose one another.
To resolve these faults, there has been proposed measuring a particle size distribution by detecting an interference light due to scattered lights caused by the Doppler-shifted laser light directed to particles moving by Brownian motion and Fourier-transforming the detection signal of the interference light to thereby analyze the frequency. In general, in order to compute particle size distribution from power spectrum, an inverse problem for solving Category 1 Fredholm's integral equation which is a relative equation between the power spectrum, the response function, and the particle size distribution is performed.
For example, operation and display divisions by logarithmically dividing a measurement range into m sections are determined in advance. The particle size distribution F(D) (D=1 to m) indicating the number of particles of typical particle sizes in the operation/display section is obtained from light intensity frequency characteristics G. In other words, the relationship between frequency characteristics G(D) (D=1 to m) and the particle size distribution F(D) can be expressed as an equation of F=P
−1
G, where P is a response function. The inverse matrix P
−1
of the response function is obtained and F=P
−1
G is operated to thereby acquire the particle size distribution F(D).
In this method, by calculating detection signals by Fourier-transform to get a power spectrum, the particle size distribution can be measured even if the number of particles present in a solvent is large and scattered lights superpose one another. Also, since a wide range measurement can be made for relatively short measurement time, fast and stable measurement is made possible compared to PCS.
It is desirable that the detection signal of an interference light which has been sampled at high speed is used in the power spectrum. Since the resultant data is wide in range, mass storage memory is required. Further, when subjecting the sampled data to deconvolution or inverse problem and the data in a wide range is used as it is, an operation unit requires a mass storage memory and high processing speed.
In consideration of this point, for example, in Japanese Patent Application Laid-open Publication No. 170844/1991 (hereinafter to be referred to as known example), deconvolution of power spectrum was practiced in such procedures that the power spectrum is converted into the form of convolution integration which can be readily inverted by linear transformation by converting the power spectrum into a logarithmic equal distance form, followed by inversion of it. That is to say, the power spectrum is converted by treating the frequency by logarithmic frequency, and the particle size by logarithmic particle size, by which a shift variant response function is made on non-dimensional logarithmic frequency axis, by which there was realized deconvolution integration with which operation can be processed in more simple procedure.
However, in case of converting the power spectrum into a logarithmic equal distance form, it has been unavoidable to thin out the majority part of the sampled data in the stage of converting into logarithmic equal distance.
FIG. 12A-D
illustrate general processing methods for performing deconvolution operation or inverse problem from the detection signal of the measured interference light.
FIG. 12A
shows detection signals of scattered light,
FIG. 12B
shows a power spectrum (frequency distribution) which is calculated by the detection signals using Fourier transform,
FIG. 12C
shows a power spectrum calculated by a smoothing calculation from
FIG. 12B
after the scattered light has been smoothed. The power spectrum gives Lorentzian function indicating the intensity distribution of Brownian motion frequency in accordance with particle sizes. The power spectrum is subjected to deconvolution or inverse problem to thereby calculate a particle size or particle diameter distribution as shown in FIG.
12
D.
In the power spectrum shown in
FIG. 12B
, however, there are influences such as the vibration noise of the solvent, white noise such as noise included in the output of the detector, the amplifier, and the electric circuit, or noise from the power supply line. Normally, the level of vibration noise is almost proportional to the inverse number of the frequency, so that a high level of noise tends to appear in low frequency areas in the power spectrum.
Due to this, if a particle size distribution is to be obtained from the power spectrum including the noise as shown in
FIG. 12B
, a particle size distribution Fs having different particle sizes and shifted from a true particle size distribution Ft indicated by a virtual line in
FIG. 12D
may be outputted or an immaterial particle size distribution Fg which does not originally exist may be outputted.
Considering this drawback, there is proposed removing the influence of noise as much as possible by, for example, integrating the power spectrum obtained as shown in FIG.
12
C and the particle (diameter) size distribution and by performing smoothing. To remove noise with smoothing, however, it is necessary to carry out processing in a smoothing routine, which processing takes a lot of time. As a result, it has been difficult to make a real-time measurement with this method. Besides, with such smoothing performed, it is difficult to sufficiently remove the influence of noise in, particularly, a low frequency range and measurement accuracy has its limit.
In addition, if sampled data is selected at equal logarithmic intervals and the number of data points to be dealt with is reduced in order to increase processing speed, it is feared that errors may occur in the selection step as shown in FIG.
12
C. That is, the power spectrum, shown by &Circlesolid; marked points in
FIG. 12C
, has values of discontinuous intervals in accordance with sampling speed. Thus, while passing through true values indicated by a solid line, the &Circlesolid; marked points do not often superpose on the O marked logarithmic selection points. For that reason, it is necessary to obtain the data of selection points which do not superpose on the &Circlesolid; marked points by performing interpolation processing as indicated by a virtual line. The values obtained by this operation may be, however, shifted from actual values.
To avoid this drawback, the original data, which is converted into a power spectrum, may be sampled at a higher frequency. If the sampling frequency is high, it is required to employ components with higher-speed performance and the cost of the components increases accordingly. Thus, this method turns out to be impractical.
Moreover, as shown in an enlarged view B of
FIG. 12C
, where intervals of logarithmic selection points (O marked points) are wide, there often exist important portions for creating a power spectrum curve in the non-selected areas. Due to this, a polygonal line connecting selection points, as indicated by
Igushi Tatsuo
Yamaguchi Tetsuji
Font Frank G.
Horiba Ltd.
Nguyen Sang H.
Oppenheimer Wolff & Donnelly LLP
LandOfFree
Apparatus for measuring particle size distribution and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus for measuring particle size distribution and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus for measuring particle size distribution and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2613320