X-ray measurement apparatus and X-ray measurement method

Patent metadata

Field	Value
Publication number	US-11788976-B2
Application number	US-202117518612-A
Country	US
Kind code	B2
Filing date	Nov 4, 2021
Priority date	Nov 9, 2020
Publication date	Oct 17, 2023
Grant date	Oct 17, 2023

Abstract

In a preliminary measurement, spectrums obtained by detecting characteristic X-rays emitted from preliminary measurement points are transmitted to a spectrum processing unit via a noise filter unit. In a main measurement, a spectrum obtained by detecting characteristic X-rays emitted from a main measurement point is transmitted to the spectrum processing unit by bypassing the noise filter unit. The noise filter unit includes a machine learning type filter constituted of a CNN or the like. In a learning process, teacher data are generated using artificially-generated noise.

First claim

The invention claimed is:

1. An X-ray measurement apparatus, comprising:
a controller configured to set a group of preliminary measurement points on a sample in a preliminary measurement, and set a main measurement point on the sample in a main measurement after the preliminary measurement, wherein the group of preliminary measurement points is a two-dimensional array of preliminary measurement points;
a generator configured to generate, in the preliminary measurement, a group of X-ray spectrums based on a group of detected signals obtained by detecting a group of X-rays emitted from the group of preliminary measurement points, and generate, in the main measurement, an X-ray spectrum based on a detected signal obtained by detecting X-rays emitted from the main measurement point;
a noise filter unit having at least one noise filter configured to reduce noise included in each X-ray spectrum of the group of X-ray spectrums in the preliminary measurement and provided exclusively for screening the preliminary measurement points to determine the main measurement point;
a processor configured to process, in the preliminary measurement, respective X-ray spectrums that are output from the noise filter unit, and process, in the main measurement, the X-ray spectrum that has bypassed the noise filter unit; and
a map generator configured to generate a map showing a composition distribution of the sample based on a result of analysis of the group of X-ray spectrums, wherein the main measurement point is determined based on the map.

2. The X-ray measurement apparatus according to claim 1 , wherein
the at least one noise filter comprises a machine learning type filter that exhibits a noise reducing effect.

3. The X-ray measurement apparatus according to claim 1 , wherein
a measurement time for each of the preliminary measurement points in the preliminary measurement is shorter than a measurement time for the main measurement point in the main measurement.

4. The X-ray measurement apparatus according to claim 1 , comprising
an X-ray measurement unit, which comprises a plurality of wavelength dispersion devices that are selectively used, and which is configured to detect characteristic X-rays using a wavelength dispersion device selected from among the plurality of wavelength dispersion devices, wherein
the noise filter unit has a plurality of noise filters corresponding to the plurality of wavelength dispersion devices, and
from among the plurality of noise filters, a noise filter corresponding to the selected wavelength dispersion device is selected.

5. An X-ray measurement method, comprising:
a preliminary measurement process comprising generating a group of characteristic X-ray spectrums based on a group of detected signals obtained by detecting a group of characteristic X-rays emitted from a group of preliminary measurement points set on a sample, inputting the group of characteristic X-ray spectrums into a noise filter unit, and analyzing a group of characteristic X-ray spectrums that are output from the noise filter unit, wherein the group of preliminary measurement points is a two-dimensional array of preliminary measurement points;
a setting process of setting a main measurement point on the sample in a main measurement after the preliminary measurement based on a result of analysis of the group of characteristic X-ray spectrums;
a main measurement process comprising generating a characteristic X-ray spectrum based on a detected signal obtained by detecting characteristic X-rays emitted from the main measurement point, and analyzing or displaying the characteristic X-ray spectrum without transmitting the characteristic X-ray spectrum to the noise filter unit; and
a map generating process comprising generating a map showing a composition distribution of the sampled based on a result of analysis of the group of characteristic X-ray spectrums, wherein the main measurement point is determined based on the map.

6. The X-ray measurement method according to claim 5 , comprising:
a filter generation process of generating, before the preliminary measurement process, a machine learning type filter comprised in the noise filter unit, wherein
the filter generation process comprises:
generating a plurality of sets of teacher data; and
supplying the plurality of sets of teacher data to the machine learning type filter and causing the machine learning type filter to perform learning, and
each of the sets of teacher data is constituted of a characteristic X-ray spectrum serving as a correct answer data, and a noise-containing characteristic X-ray spectrum generated by adding artificially-generated noise to the characteristic X-ray spectrum.

7. A program stored in a non-transitory storage medium and configured to be executed by an information processing device, the program having:
a function of generating, in a preliminary measurement, a group of X-ray spectrums based on a group of detected signals obtained by detecting a group of X-rays emitted from a group of preliminary measurement points set on a sample, and generating, in a main measurement, an X-ray spectrum based on a detected signal obtained by detecting X-rays emitted from a main measurement point set on the sample in the main measurement after the preliminary measurement, wherein the group of preliminary measurement points is a two-dimensional array of preliminary measurement points;
a function of applying a noise reduction filter configured to reduce noise included in each X-ray spectrum of the group of X-ray spectrums in the preliminary measurement and provided exclusively for screening the preliminary measurement points to determine the main measurement point;
a function of processing, in the preliminary measurement, a group of X-ray spectrums to which the noise reduction processing has been applied, and processing, in the main measurement, the X-ray spectrum that has bypassed the noise reduction processing; and
a function of generating, by a map generator, a map showing a composition distribution of the sample based on a result of analysis of the group of X-ray spectrums, wherein the main measurement point is determined based on the map.

Assignees

• Jeol Ltd

Inventors

• Murano Takanori
• Uematsu Fuminori

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G01N23/2209 Primary

  using wavelength dispersive spectroscopy [WDS] · CPC title

• G01N23/223

  by irradiating the sample with X-rays or gamma-rays and by measuring X-ray fluorescence · CPC title

• G06N3/08

  Learning methods · CPC title