JP2020071130A - Trace metal detection device - Google Patents

Abstract

To provide a guidance function based on a prediction of a change in concentration of trace metal in a trace metal detection device using laser-induced breakdown spectroscopy.SOLUTION: A data processing unit 190 receives spectrum data indicating light intensity at each wavelength of an emission spectrum 200 obtained by emitting laser light to a test material 51 including a sample 20. The data processing unit 190 detects the concentration of a predetermined metal element corresponding to a predetermined wavelength in the sample 20 based on the light intensity of the wavelength of the spectrum data and executes identification operation processing for calculating a coefficient of a predefined model function that indicates a transition of the concentration using a change history of the detected concentration. A display 195 displays prediction information obtained using the model function after the identification.SELECTED DRAWING: Figure 1

Description

  TECHNICAL FIELD The present invention relates to a trace metal detection device, and more particularly to a trace metal detection device using laser induced breakdown spectroscopy (LIBS).

  Japanese Patent Application Laid-Open No. 2016-95139 (Patent Document 1) describes a contained substance measuring device that detects a metal element (substance to be detected) in a liquid sample by applying laser-induced breakdown spectroscopy. In Patent Document 1, a test sample preparation in which a binder layer and a sample layer of a liquid sample are formed on a substrate made of silicon or metal (copper, nickel, etc.) is prepared, and laser light to the test sample preparation is prepared. From the emission spectrum obtained by analyzing the plasma emission due to irradiation, the substance contained in the liquid sample is specified and the concentration of the substance is measured.

JP, 2016-95139, A

  For example, the trace metal concentration in the cleaning liquid used in the semiconductor manufacturing process can be detected by a trace metal detector using LIBS. 2. Description of the Related Art In a semiconductor manufacturing line, a cleaning device is used for removing contaminants such as inorganic substances containing various metals adhering to a semiconductor wafer by a cleaning liquid. By cleaning the semiconductor wafer, Cu (copper), which is a wiring material, is mixed in the cleaning liquid. When the same cleaning solution is repeatedly used repeatedly, the Cu concentration in the cleaning solution gradually increases, and the cleaning solution becomes contaminated with Cu. As the Cu concentration (metal concentration) increases, the cleaning ability of the cleaning solution also decreases. Go It is feared that if the cleaning liquid with reduced cleaning ability is continuously used, the defective rate of the manufactured products increases (that is, the yield decreases).

  Therefore, from the user side (for example, a semiconductor maker), when the metal concentration rises above a certain limit, it is necessary to replace the cleaning liquid, but from the viewpoint of manufacturing cost and environmental load, There is a demand to use the cleaning solution to the limit within the range where quality can be maintained. Therefore, by detecting the metal concentration in the cleaning liquid each time with the trace metal detecting device using LIBS, it becomes possible to appropriately judge the necessity of replacement of the cleaning liquid.

  However, the replacement work of the cleaning liquid requires the preparation of a large amount of cleaning liquid and the replacement work for a long time. For this reason, it is difficult to suddenly incorporate the process of exchanging the cleaning liquid, especially during periods when the production line is continuously operating, such as during busy periods. Therefore, there is a need to not only detect the concentration of trace metals in real time, but also to predict future trends. It is understood that similar needs exist not only for cleaning liquids used in semiconductor manufacturing lines, but also for various liquids and solids that are targets of detection of trace metal concentrations as targets of measurement.

  The present invention has been made to solve such a problem, and an object of the present invention is based on prediction of a change in the concentration of a trace metal in a trace metal detection device using laser-induced breakdown spectroscopy. It is to have a guidance function.

  In one aspect of the present invention, a trace metal detection device includes a laser light source, a spectroscope, and a data processing unit. Light emitted by plasma obtained by irradiating a test material including a sample with laser light from a laser light source is input to the spectroscope. Spectral data indicating the light intensity at each wavelength obtained by the spectroscope is input to the data processing unit. The data processing unit includes a concentration detection unit, a storage unit, an identification calculation unit, and a display unit. The concentration detection unit detects the concentration of a predetermined metal element corresponding to the predetermined wavelength in the sample from the light intensity at the predetermined wavelength of the spectrum data. The storage unit holds the history of the density detected by the density detection unit. The identification calculation unit identifies the coefficient of a predetermined model function for indicating the transition of the concentration, using the concentration change history held in the storage unit. The display unit displays the prediction information acquired using the model function after identification having the coefficient identified by the identification calculation unit.

  According to the above-mentioned trace metal detection device, not only the concentration of the specific metal in the sample is detected in real time, but also the transition of the metal concentration can be predicted by identifying the model function from the history of the detected concentration. .. Then, by displaying the information based on the prediction, a guidance function based on the prediction of the change in the concentration of the trace metal can be provided.

  The prediction information is related to the predicted value of the concentration after the present time according to the model function after identification, or the time when the concentration exceeds a predetermined threshold value when the concentration changes after the present time according to the model function after identification. Information to be included.

  As described above, it is possible to provide the user guidance by directly displaying the predicted concentration value after the present time using the identified model function or by displaying the information for predicting the time when the metal concentration exceeds the threshold value. is there.

  Alternatively, the data processing unit can be configured to further include a selection input unit. The selection input unit is provided to input a selection instruction of at least one metal element to be monitored from among the plurality of metal elements. The concentration detection unit detects the concentration of the metal element from the light intensity of the wavelength corresponding to each metal element selected as the monitoring target in the spectrum data, and the identification calculation unit calculates the coefficient of the model function for each metal element. Can be calculated individually.

  With this configuration, the calculation load of the data processing unit can be reduced, and the guidance function for the user can be enhanced by monitoring the metal concentration by narrowing down the monitoring target.

  According to the present invention, a trace metal detection device using laser-induced breakdown spectroscopy can be provided with a guidance function based on prediction of changes in trace metal concentration.

It is a block diagram explaining the composition of the trace amount metal detection device concerning this embodiment. FIG. 2 is a block diagram illustrating an example of a hardware configuration of the data processing device shown in FIG. 1. It is a functional block diagram of a data processor in a trace amount metal detecting device concerning this embodiment. 4 is a chart showing an example of stored contents of a density history table in the storage unit shown in FIG. 3. FIG. 4 is a conceptual diagram for explaining the functions of an identification calculation unit, a prediction unit, and a determination unit shown in FIG. 3. It is a conceptual diagram explaining an example of the display content on the display part shown in FIG. It is a functional block diagram of the data processor in the trace amount metal detecting device concerning the modification of this embodiment.

  Embodiments of the present invention will be described below in detail with reference to the drawings. In the following, the same or corresponding parts in the drawings will be denoted by the same reference numerals, and the description thereof will not be repeated in principle.

FIG. 1 is a block diagram illustrating the configuration of a trace metal detection device according to this embodiment.
With reference to FIG. 1, the trace metal detection apparatus 100 according to the present embodiment detects a trace metal in a liquid sample 20 to be inspected. For example, the sample 20 is a cleaning liquid for semiconductor wafers used in the cleaning device 10 in the semiconductor manufacturing process.

  The trace metal detection device 100 includes a heater 60, a laser light source 110, a control circuit 120, a spectroscope 130, a lens 140, a condenser 150, a fiber 160, and a data processing device 190. The trace metal detection device 100 detects a trace metal in the sample 20 by laser-induced breakdown spectroscopy (LIBS). Alternatively, the arrangement of the condenser 150 may be omitted.

  A predetermined amount of cleaning liquid is collected as a sample 20 from the cleaning device 10 and is dropped on the base material 50. In this embodiment, a chip made of silicon (Si) is used as the base material 50. The heater 60 heats the base material 50 onto which the sample 20 has been dropped to execute the evaporation and dry treatment of the sample 20. The evaporation to dryness treatment can be performed, for example, in the same manner as in Patent Document 1. Thereby, the test material 51 in which the sample 20 is fixed on the base material 50 is obtained.

  It should be noted that the sample 20 can be collected from the cleaning liquid by manual work as appropriate. Alternatively, as shown in FIG. 1, while the cleaning device 10 is in operation, by providing a cleaning agent path 25 in which a part of the cleaning liquid is introduced into the trace metal detection device 100 and then returned to the cleaning device 10, the cleaning liquid is provided. It is also possible to automatically take the sample 20 from.

  The laser output from the laser light source 110 is condensed by the lens 140 and is applied to the sample 20 of the test material 51. As a result, initial plasma is generated on the surface of the test material 51. The intensity and direction of the laser output from the laser light source 110 can be adjusted by the control parameter input to the control circuit 120.

  At the location irradiated with the laser, the atoms contained in the sample 20 are excited, and the laser light is absorbed by the initial plasma to grow the plasma. The light emission of the grown plasma is input to the spectroscope 130 through the concentrator 150 and the fiber 160. Alternatively, in the configuration in which the concentrator 150 is omitted, the plasma emission is input to the spectroscope 130 through only the fiber 160. A spectrum is output from the spectroscope 130. From the spectrum obtained by the spectroscope 130, data indicating the light intensity at each wavelength (hereinafter, also referred to as “spectral data”) can be obtained. The peak wavelength in the spectrum varies depending on the metal component. Therefore, the metal element contained in the sample 20 can be identified from the peak wavelength appearing in the spectrum.

  Furthermore, by previously acquiring the light intensity of the peak wavelength in the spectrum for the metal in the form of a calibration curve or the like, the type of the metal contained in the sample 20 can be determined from the light intensity for each wavelength in the obtained spectrum data. And its concentration can be detected. The data processing device 190 can obtain the emission spectrum 200 of the sample 20 by performing arithmetic processing on the spectrum data output from the spectroscope 130.

  FIG. 2 is a block diagram illustrating a hardware configuration example of the data processing device 190. The data processing device 190 can be typically configured by a personal computer.

  Referring to FIG. 2, data processing device 190 includes a CPU (Central Processing Unit) 191, a memory 192, an I / O unit 193, a user input unit 194, a display unit 195, and a display driver 196. ..

  The I / O unit 193 has an A / D conversion function and a D / A conversion function, and functions as an input / output interface for analog signals to the CPU 191. For example, the spectrum data SD indicating the emission spectrum 200 output from the spectroscope 130 is input to the CPU 191 as digital data via the I / O unit 193.

  The CPU 191 executes arithmetic processing by software processing using programs and data stored in the memory 192.

  The display unit 195 is composed of, for example, a liquid crystal display, and can visually display information and data. The display content of the display unit 195 can be controlled by the CPU 191 via the I / O unit 193 and the display driver 196.

  The user input unit 194 is composed of, for example, a keyboard or a mouse, and receives an instruction input by the user. When the display unit 195 is configured by a touch panel, the user input unit 194 can be configured by using a soft switch displayed on the touch panel. The instruction input to the user input unit 194 is input to the CPU 191 as digital data via the I / O unit 193.

  Next, the arithmetic processing using the spectrum data by the data processing device 190 will be described in detail. FIG. 3 shows a functional block diagram of the data processing device in the trace metal detection device according to the present embodiment. For example, the function of each functional block in FIG. 3 can be realized by the software processing by the CPU 191 described above. Alternatively, it is possible to realize some or all of the functional blocks or some of the functional blocks by using dedicated hardware (electronic circuit).

  With reference to FIG. 3, the data processing device 190 includes a density detection unit 210, a storage unit 192a, an identification calculation unit 220, a prediction unit 230, a determination unit 240, and a display control unit 250.

  Spectral data SD indicating the emission spectrum 200 is input to the concentration detection unit 210 each time the test material 51 including the sample 20 is measured by the trace metal detection device 100.

  In the present embodiment, the trace metal detection apparatus 100 is configured so that a predetermined measurement condition for the cleaning processing time by the cleaning apparatus 10 and / or the number of cleaning processings (chips) by the cleaning apparatus 10 is established. First, it is assumed that the metal concentration of the cleaning liquid (sample 20) is measured by the trace metal detection device 100.

  The spectrum data SD is composed of data indicating the light intensity at each wavelength of the emission spectrum. The presence of the metal element corresponding to the peak wavelength in the sample 20 is detected by showing the peak where the light intensity at the specific wavelength exceeds the predetermined threshold. For example, when Cu (copper) is present, a peak appears at a wavelength of 324.7 (nm), and when Al (aluminum) is present, a peak appears at a wavelength of 396.1 (nm).

  Further, with respect to the light intensity at the peak wavelength corresponding to the specific metal, a calibration curve corresponding to the concentration is created in advance. Data indicating the calibration curve and the threshold value are stored in the storage unit 192a formed by using a partial area of the memory 192.

  The concentration detection unit 210 uses the data read from the storage unit 192a and the spectrum data SD to calculate from the spectrum data SD the presence or absence of a specific metal in the sample 20 and the concentration of the detected metal. You can The concentration detection unit 210 sequentially stores the metal concentration calculated for each measurement for a specific metal element in the concentration history table in the storage unit 192a.

FIG. 4 is a chart showing an example of stored contents of the density history table in the storage unit 192a.
Referring to FIG. 4, in the layer degree history table, for example, the detected metal concentrations of Ti (titanium), Cu (copper), and Al (aluminum) used as metal wiring materials at the time of manufacturing a semiconductor device. Is memorized.

  The metal concentration is calculated according to the calibration curve data from the light intensity at the specific wavelength in the spectrum data SD at each measurement time of the sample 20 by the laser irradiation measurement of the test material 51 by the trace metal detection device 100. The metal concentration calculated for each measurement is accumulated for each metal. Thereby, the history of the metal concentration with the passage of time can be retained.

  Referring again to FIG. 3, the identification calculation unit 220 calculates the coefficient of the model function of a predetermined format using the data held in the concentration history table shown in FIG. The model function predicts the transition (change) of the concentration of each metal element over time.

  The model function can be defined in advance as an arbitrary function, but in the present embodiment, the model function is represented by a quadratic function. Therefore, the concentration change model function can be expressed by the following equation (1).

Y = a · x ² + b · x + c (1)
For example, according to a known regression analysis method (typically, the least squares method), by substituting the data indicating the history of the past metal concentration held in the concentration history table into x in the equation (1), The coefficients a, b, c of the quadratic function model can be identified. In general, the coefficients a, b, and c are calculated so that the error between the metal concentration (actual value) and the value (Y) obtained by the function model formula in each measurement so far is the minimum.

  The prediction unit 230 calculates the predicted value of the metal concentration after the present time point according to the model function after identification, which has the coefficients a to c obtained by the identification calculation unit 220. As a result, the metal concentration Mc (i) at the i-th (i: natural number) measurement time for the specific metal is obtained. As a result, the metal concentration Mc (i) can be obtained for any i including i> m from the model function defined using the metal concentrations detected in the past 1 to m-th measurement times. it can.

  The determination unit 240 executes a determination process by comparing a predetermined threshold value Rth with the metal concentration Mc (i). For example, the threshold value Rth is set so as to determine whether or not the cleaning liquid used in the cleaning device 10 in the semiconductor manufacturing process needs to be replaced. As described above, the threshold value can be set in correspondence with the upper limit value within the range in which the quality of the final product device can be maintained. As a result, when the metal concentration Mc (i) increases according to the model function identified by the identification calculation unit 220, the limit number Nch can be obtained as the minimum value of i that satisfies Mc (i)> Rch.

  FIG. 5 is a conceptual diagram for explaining the functions of the identification calculation unit 220, the prediction unit 230, and the determination unit 240.

  With reference to FIG. 5, the identification calculation unit 220 uses the metal concentration data C1 to C3 for three times of i = 1 to 3 stored in the storage unit 192a (concentration history table) to calculate the equation (1). The model function 225 represented by a quadratic function is determined by identifying the coefficients a to c.

  Thereby, the prediction unit 230 can predict the metal concentration Mc (i) according to the dotted line portion of the model function 225 based on the identified coefficients a to c even in the region of i> 4. Further, the determination unit 240 can obtain the limit number of times Nch by comparing the metal concentration Mc (i) in the region of i> 4 with the threshold value Rth.

  Referring again to FIG. 3, the display control unit 250 controls the display content on the display unit 195 using the metal concentration Mc (i) from the prediction unit 230 and the limit number of times Nch from the determination unit 240. .. The display content designated by the display control unit 250 is displayed on the display unit 195 as visual information by the display driver 196 shown in FIG.

FIG. 6 is a conceptual diagram illustrating an example of display contents on the display unit 195.
With reference to FIG. 6, for example, together with the measured concentration L = Mc (N) at this time (i = N), the predicted concentration M = Mc (N + 1) at the (N + 1) th measurement, which is predicted according to the model function, is calculated. Can be displayed. Even after the (N + 2) th time, it is possible to display the predicted value of the metal concentration according to Mc (i).

  Furthermore, the display unit 195 can directly display the limit number of times Nch obtained by the determination unit 240, and can also display the remaining usable number T (times) based on the limit number of times Nch. For example, after the Nth measurement, it is possible to display T = Nch-N-1 (times) using the limit number of times Nch. The predicted density M and the remaining usable number T (or the limit number Nch) described in FIG. 6 correspond to an example of the “prediction information”.

  As described above, according to the trace metal measuring apparatus according to the present embodiment, not only the concentration of the specific metal in the sample is detected in real time by the emission spectrum acquired by LIBS, but also the history of the detected concentration is used. By identifying the model function, the transition of the metal concentration can be predicted. Then, by presenting the prediction information to the user, it is possible to predict the replacement timing of the substance (for example, the cleaning liquid of the semiconductor chip) from which the sample is collected, and the guidance based on the prediction of the change in the concentration of the trace metal. It can have a function.

  Note that the number of metal elements to be stored in the history of the metal concentration and the target of monitoring the transition of the concentration by the model function is an arbitrary number with respect to the metal elements included in the wavelength region detectable by the spectroscope 130. Is possible. When there are a plurality of objects to be monitored, different model functions are identified for each metal element. That is, for the coefficients a to c in the equation (1), different values are obtained for each metal element. Alternatively, when the behavior of the concentration transition is different between the metal elements, it is possible to define the model function by using a different function between the metal elements, such as using a quadratic function and a cubic function, for example.

  Alternatively, in the modification of the present embodiment shown in FIG. 8, in addition to the configuration of FIG. 3, a selection input unit 194a configured by using the user input unit 194 shown in FIG. 2 is further arranged. be able to.

  The selection input unit 194a is configured to receive a user input for designating a metallic element to be monitored. The user input may be, for example, a mode in which at least one is selected from a plurality of metal elements included in the detectable wavelength range and having a calibration curve prepared in advance.

  The selection input unit 194a outputs a selection signal Smt for specifying the metal to be monitored according to the user input. The selection signal Smt is input to the concentration detection unit 210, the identification calculation unit 220, and the determination unit 240. For example, regarding the detection of the metal concentration by the concentration detection unit 210, the storage in the storage unit 192a (concentration history table), the identification calculation of the model function in the identification calculation unit 220, and the determination process in the determination unit 240, the metal according to the selection signal Smt is used. It can be executed by limiting to only elements.

  As a result, it is possible to reduce the calculation load on the data processing device 190 and to enhance the guidance function for the user by monitoring the concentration of the monitoring target. In the configuration of FIG. 9, the detection of the metal concentration by the concentration detection unit 210 and the storage in the storage unit 192a (concentration history table) are currently selected so as to cope with the case where the monitoring target is increased by the user input. It is also possible to perform each measurement time for each of a plurality of user-selectable metal elements in addition to the metal elements inside.

  In the present embodiment, the transition (increase) of the metal concentration over time is predicted in units of the number of measurements, but other units (for example, in the present embodiment, the cleaning processing time by the cleaning device 10 or the cleaning process). It is also possible to predict the transition (increase) of the metal concentration over time according to the number of cleaning processes performed by the apparatus 10).

  Further, in the present embodiment, the example in which the cleaning liquid of the cleaning device used in the semiconductor manufacturing process is used as the sample has been described, but the sample for detecting the trace amount of metal is arbitrary. That is, for any liquid or solid, the trace metal detection apparatus according to the present embodiment or the modification is used to determine the future transition of the metal concentration from the history of the metal concentration obtained from the spectrum data of the sample by LIBS. It is possible to realize the prediction function.

  The embodiments disclosed this time are to be considered as illustrative in all points and not restrictive. The scope of the present invention is shown not by the above description but by the claims, and is intended to include meanings equivalent to the claims and all modifications within the scope.

  10 cleaning device, 20 sample, 25 cleaning agent path, 50 base material, 51 test material, 60 heater, 100 trace metal detection device, 110 laser light source, 120 control circuit, 130 spectroscope, 140 lens, 150 concentrator, 160 Fiber, 190 data processing device, 192 memory, 192a storage unit, 193 I / O unit, 194 user input unit, 194a selection input unit, 195 display unit, 196 display driver, 200 emission spectrum, 210 concentration detection unit, 220 identification calculation Part, 225 model function, 230 prediction part, 240 determination part, 250 display control part, Mc (i) metal concentration, Nch limit number of times, Rth threshold value, SD spectrum data, Smt selection signal.

Claims (4)

Laser light source,
A spectrometer into which light emitted by plasma obtained by irradiating a test material including a sample with laser light from the laser light source is input,
A data processing unit, into which spectral data indicating light intensity at each wavelength obtained by the spectroscope is input,
The data processing unit,
From a light intensity at a predetermined wavelength of the spectral data, a concentration detection unit that detects the concentration of a predetermined metal element corresponding to the predetermined wavelength in the sample,
A storage unit that holds a history of the concentration detected by the concentration detection unit;
An identification calculation unit that calculates a coefficient of a predetermined model function for indicating the transition of the concentration using the change history of the concentration held in the storage unit,
A trace metal detection device, comprising: a display unit that displays prediction information acquired by using the model function after identification having the coefficient calculated by the identification calculation unit.   The trace metal detection device according to claim 1, wherein the prediction information includes a predicted value of the concentration after the present time according to the model function after the identification.   The trace metal detection according to claim 2, wherein the prediction information includes information related to a time when the concentration exceeds a predetermined threshold value when the concentration changes after the present time according to the model function after the identification. apparatus. The data processing unit,
A selection input unit for inputting a selection instruction of at least one metal element to be monitored from among the plurality of metal elements,
The concentration detection unit, from the light intensity of the wavelength corresponding to each of the metal elements selected as the monitoring target in the spectral data, detects the concentration of the metal element,
The trace metal detection device according to claim 1, wherein the identification calculation unit individually calculates the coefficient of the model function for each metal element.

Images