JP2002286704A - Data processing device for chromatograph - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data processing device capable of accurately determining the range in which impurities are included at a chromatogramic peak. SOLUTION: A peak in a chromatogram is judged, and the photo- absorptivesness at the reference point is calculated from the photo- absorptivesness spectrum obtained in the peak part at the i'th measuring time, and the photo-absorptivesness spectrum of the reference point (measuring time) where the mentioned photo-absorptivesness is obtained is determined as the reference spectrum (S4). The similarity r of the spectrum and the judging reference value t are calculated on the basis of the photo-absorptivesness spectrum data at the measuring time, reference spectrum data, and the data of photo- absorptivesness spectrum on the base line (S5 and S6). For each measuring time, the reference spectrum is determined anew, and the similarity r and judging reference value t are calculated and plotted in a graph (S8 and S9). Thereby the external variation of the spectrum at the peak top is unlikely to give influence, and it is possible to determine the similarity r and judging reference value t accurately.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing apparatus such as a high performance liquid chromatograph (hereinafter abbreviated as "HPLC"), and more particularly, to the purity of a peak in a chromatogram obtained by HPLC or the like. The present invention relates to a data processing device for determining whether a peak is caused by a certain single component or a peak obtained by overlapping two or more components.

[0002]

2. Description of the Related Art In a configuration in which a photodiode array detector is used as a detector in HPLC, information on the direction of spread of the wavelength can be obtained simultaneously at each measurement point, so that an absorbance spectrum is created at each measurement point. Further, in the time axis direction, it is possible to create a chromatogram in which, for example, attention is paid to a certain wavelength and absorbance data on an absorbance spectrum is plotted, or maximum intensity data on each absorbance spectrum is selected and plotted. Although peaks corresponding to sample components appear in such chromatograms, when a plurality of components have the same or close retention times, a phenomenon occurs in which the peaks of these components overlap. Therefore, in order to determine whether the peak appearing in the chromatogram is due to a single component or whether the peaks of a plurality of components are overlapped,
It is necessary to determine the peak purity.

Conventionally, as one of the methods for determining the peak purity, there is one described in Japanese Patent Application Laid-Open No. Hei 7-218491. This peak purity determination method will be described below with reference to the flowchart of FIG.

After the absorbance spectrum data at each measurement point within a certain period is collected and stored in a storage unit,
The analysis is performed according to the procedure shown in FIG. First, all the absorbance spectrum data is read (step S1).
1) The spectrum data at each measurement point is divided into baseline data and peak data (step S1).
2). Here, the peak portion data is, for example, data of an elution peak portion that appears when data of wavelengths at which peaks appear is arranged in chronological order. What is necessary is just to employ | adopt the method conventionally used in the analysis of a gram. As an example, the time when the slope of the tangent of the time-absorbance curve becomes a predetermined value or more is defined as the start time of the peak portion, and the time when the slope of the tangent decreases below the predetermined value is defined as the end time of the peak portion. It is.

[0005] The baseline portion is a portion other than the peak portion. After separating the data into baseline and peak,
Further, data at the time of measurement indicating the maximum absorbance in each peak portion is distinguished as peak top data (step S12). After dividing the data in step S12, the noise level N is calculated based on the data of the baseline portion (step S13). The data of the baseline portion is, for example, as shown in FIG. 7. The standard deviation σ is calculated from these data, and σ or 3 × σ is calculated as the noise level N.
And

Next, the spectrum data at each measurement point belonging to the peak portion of the target component is sequentially read from the storage unit, and the similarity r and the determination reference value t are calculated by a method described later. At this time, the absorbance spectrum at the time of measurement, which is classified as the peak top in step S12, is used as a reference spectrum. That is, first, the data of the absorbance spectrum at the start time T1 of the peak portion of the target component is read, and the similarity r between the two is calculated from this data and the data of the reference spectrum (step S14). next,
A determination reference value t is calculated from the spectrum data at the measurement time T1, the data of the reference spectrum, and the noise level N calculated in step S13 (step S1).
5).

An example of the significance and calculation method of the similarity r and the determination reference value t is as follows (JP-A-1-161123).
No.). First, assume that there is no noise. Assuming that the absorbance data at each wavelength of λ1, λ2, ..., λn is A1, A2, ..., An in a certain absorbance spectrum, the absorbance spectrum data is an n-dimensional vector A (A1, A2, ..., A).
n). If the patterns a and b of the two absorption spectra are obtained from the same sample, the corresponding vectors Aa (Aa (λ1), Aa (λ2),..., Aa
(λn)) and Ab (Ab (λ1), Ab (λ2),..., Ab (λn)) should have the same ratio of the corresponding elements (absorbance data) (ie, Aa (λ1) / Ab (λ1). ) = Aa (λ2) / Ab (λ2) = ...
= Aa (λn) / Ab (λn)), so that both vectors Aa and Ab face the same direction in the n-dimensional space as shown in FIG. That is, both vectors Aa and Ab
Is θ, θ = 0 in this case. On the other hand, when the two spectral patterns a and b are obtained from different samples, the ratio of the corresponding elements of the corresponding vectors Aa and Ab is not always constant as described above, and FIG. As shown in FIG. 7, both vectors Aa and Ab point in different directions, and θ takes a non-zero value.

Therefore, considering the value (similarity) r = cos θ = (Aa · Ab) / ｛| Aa | × | Ab |｝ (1) where (Aa · Ab)
Represents the inner product of the vectors Aa and Ab), and the similarity r becomes 1 when the directions of both vectors are completely the same, and when the directions of both vectors are close (that is, when θ is a small value). Take a value close to 1. Therefore, when two spectral patterns are obtained, whether or not they are obtained from the same sample can be determined based on whether or not the similarity r is close to 1.

However, in practice, since each absorbance data of a spectrum pattern always includes a noise component, even if two absorbance spectra are collected from the same sample,
The directions of the two vectors do not completely match, and the similarity does not become 1. Thus, assuming a certain noise level, to what extent the similarity r is close to 1 will consider next whether it can be determined that both spectra are taken from the same sample.

As shown in FIG. 9C, a vector A of two spectral patterns a and b obtained from the same sample is obtained.
a and Ab have the same direction when there is no noise. However, noise components (N1, N2,..., Nn) are included in the elements of each vector.
Are superimposed, the vector of the measured spectrum pattern including the noise becomes (Aa (λ1) + N1, Aa (λ2) + N2,..., Aa
(λn) + Nn). Therefore, when the maximum value of the noise vector is estimated as | N |, the tips of the actually measured spectral pattern vectors are within the n-dimensional sphere of radius | N | from the tips of the vectors Aa and Ab, respectively. enter. That is, conversely, the angle θ between the two measured vectors Aa, Ab of the spectral pattern is the tangent Aa ′, A drawn from the origin to the outside of these two n-dimensional spheres.
If it is greater than the angle between b ′ (θ1 + θ2), then it is unlikely that both spectral patterns are of the same sample. As described above, since the similarity r between the two spectral patterns is represented by the cosine (cos) of the angle θ between the two, the criterion value t of the similarity r is also within this angular range (θ1).
+ Θ2). That is, the determination reference value t is calculated by the following equation, and when r <t, it can be said that there is no possibility that two spectral patterns are obtained from the same sample. t = cos (θ1 + θ2) = cosθ1 × cosθ2−sinθ1 × sinθ2 = (| Aa | ^2- | N | ² ) ^1/2 / | Aa | x (| Ab | ^2- | N | ² ) ^1/2 / | Ab | − (| N | / | Aa |) × (| N | / | Ab |) (2)

In steps S14 and S15, the similarity r and the determination reference value t with respect to the reference spectrum data are calculated from the spectrum data at the first measurement time T1 of the peak portion by the above equations (1) and (2). Then, it is checked whether the calculation of both values r and t has been completed for all the spectral data included in the peak portion (step S16).
4, the calculation of r and t is performed on the spectrum data at the next time T1 + ΔT. Thus, r and t for all the spectral data belonging to the peak portion
Is completed, the process proceeds to step S17, and the similarity r
Is determined in which the time range is equal to or less than the determination reference value t. The peak in this time range is not a single component peak,
It can be considered that the impurity peaks overlap.

[0012]

In the above-mentioned conventional method for determining peak purity, an absorbance spectrum at the time of measurement showing a peak top on a chromatogram is used as a reference spectrum. An absorbance spectrum is also used. However, there is no guarantee that such a reference spectrum will have the original spectral shape of its components without being affected chemically or physically. Actually, when the concentration of the components in the sample is high, the distortion of the absorbance spectrum due to the non-linearity of the detector appears greatly, and the above method causes a problem that the accuracy of the peak purity determination is significantly reduced.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a chromatographic data processing apparatus which can improve the accuracy of peak purity determination. .

[0014]

Means for Solving the Problems and Effects The present invention has been made to solve the above problems, and a chromatographic apparatus provided with a detector for sequentially detecting the absorbance spectrum of a sample eluted from a column with the passage of time. A data processing device for judging the purity of a peak appearing in a chromatogram obtained by plotting absorbance at each measurement time point in a time axis direction, wherein: a) a starting point to an ending point of a target peak; A reference spectrum calculating means for determining an absorbance spectrum at a time point separated by an appropriate time as a reference spectrum corresponding to the measurement time point for each measurement time point up to b), at each wavelength of the absorbance spectrum detected at each measurement time point Absorbance data and a reference spectrum corresponding to each measurement time point determined by the reference spectrum calculation means Means for calculating a similarity r between a detected absorbance spectrum at each measurement time point and a reference spectrum corresponding to the measurement time point based on the absorbance data at each measurement time point; c) each of the absorbance spectrums detected at each measurement time point Based on the absorbance data at the wavelength, the absorbance data at the reference absorbance spectrum corresponding to each measurement time point determined by the reference spectrum calculation means, and the absorbance data at the baseline absorbance data, the determination reference value t at each measurement time point And d) determining means for determining the magnitude relationship between the similarity r at each measurement time point and the determination reference value t.

That is, in this chromatographic data processing apparatus, when the similarity calculating means and the reference value calculating means calculate the similarity r and the criterion value t at each measurement time, respectively, the similarity calculation means and the reference value calculation means are different at each measurement time Use the reference spectrum. The reference spectrum calculating means is configured to calculate absorbance spectrum data constituting an absorbance spectrum at each measurement point collected for the target sample, or absorbance at each measurement point collected for a standard sample measured in advance. Based on the absorbance spectrum data constituting the spectrum, the absorbance spectrum at a point at a certain distance from the measurement point (this is referred to as a "reference point" corresponding to each measurement point) at each measurement point is determined as a reference spectrum. I do. Here, if the measurement time point and the corresponding reference point are too close, there is a high possibility that the shapes of the absorbance spectra of both are too similar, and if the measurement time point and the corresponding reference point are too far, there is a complete similarity. It will not be. Therefore, the distance, the time interval, or a parameter corresponding to the distance, the time interval, or the like may be appropriately determined by an experiment or the like. However, the distance is not always constant,
It can be determined adaptively according to the algorithm.

As described above, according to the chromatographic data processing apparatus of the present invention, the absorbance spectrum at the peak top is not uniformly selected as the reference spectrum for calculating the similarity and the determination reference value. At each measurement time point, an appropriate reference spectrum is selected depending on the absorbance spectrum at the measurement time point. Therefore, even if the absorbance spectrum for the peak top or the absorbance spectrum for the standard sample does not have the original spectral shape of the target component due to chemical or physical factors, without being affected by such factors, It is possible to determine the peak purity with high accuracy.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an example in which a data processing apparatus of the present invention is applied to a high performance liquid chromatograph (HPLC) will be described with reference to the drawings.

FIG. 1 is an overall configuration diagram of the HPLC of this embodiment. The sample liquid to be analyzed is injected by the injector 3 into the eluent (mobile phase) sucked by the pump 2 from the eluent tank 1 and sent to the separation column 4. Since the time during which each component in the sample passes through the separation column 4 (retention time) varies depending on the component, each component of the sample is temporally separated and eluted in the eluate that has passed through the separation column 4. The detector 5 is a photodiode array detector. In the detector 5, the eluate from the separation column 4 is passed through the flow cell 51, and white light from the light source 52 is transmitted through the flow cell 51. The white light is absorbed in the flow cell 51 at a wavelength specific to the components contained in the eluate. The transmitted light is dispersed in wavelength by the spectroscope 53, and the light intensity for each wavelength is detected by the photodiode array 54.

The detection signal from the detector 5 is amplified by an amplifier 6, converted into a digital signal by an A / D converter 7, and then input to a personal computer (hereinafter referred to as "PC") 8. The personal computer 8 stores programs for various data analyses, and performs qualitative analysis and quantitative analysis of components contained in the sample by executing analysis processing on the acquired data in accordance with the programs. Execute. The result is output to a display or a printer attached to the personal computer 8. The personal computer 8 has not only data processing but also a function of controlling the operation of each part of the HPLC.

Next, the characteristic processing of the data processing executed by the personal computer 8 will be described with reference to the flowchart of FIG. First, after a sample is injected into the injector 3 of the HPLC, data of the absorbance spectrum at each measurement time point is collected every minute within a predetermined time including the retention time of the component to be analyzed contained in the sample. The collected data is as shown in FIG. The data of the absorbance spectrum collected at each measurement time point is sequentially stored in the memory (RAM) of the personal computer 8 or an external storage device (hard disk or the like).

When the collection of the absorbance spectrum data within the above-mentioned predetermined time is completed, the data analysis processing relating to the peak purity determination is performed according to the procedure shown in FIG.

First, all the absorbance spectrum data is read from the storage unit (step S1), and the absorbance spectrum data at each measurement point is divided into the baseline data and the peak data (step S2). Step S
After the data is divided in step 2, the noise level N is calculated from the data of the baseline (step S3). That is, as shown in FIG. 7, the standard deviation σ is calculated from the data of the baseline portion, and the standard deviation σ or 3 × σ is set as the noise level N. This procedure corresponds to step S1 of the above-mentioned conventional method.
Same as 1 to S13.

Next, the starting point of the peak of the target component (time T
The following processing is performed on the absorbance spectrum data at each measurement point during a predetermined period from 1) to the end point (time T2). Here, as shown in FIG.
The peak start point T1 is the first measurement point and the peak end point T2 is the n-th measurement time point.

First, the absorbance spectrum data at the i-th (i = 1 to n) measurement time is read from the storage unit, and a reference spectrum corresponding to the measurement time is obtained (step S).
4). The specific procedure is as follows: (1) Absorbance spectrum Si at the i-th measurement time point
Is assumed to be as shown in FIG. 5, the variable X is determined based on the maximum intensity M in the absorbance spectrum Si.
To determine. Here, X is appropriately changed according to the value of M,
The relationship is determined by preliminary experiments.

(2) Now, let HTOP be the peak intensity at the peak top on the chromatogram as shown in FIG. 4, and determine this value and the peak intensity Hi at the i-th measurement time point.
Assuming that the value of the ratio Hi / HTOP with respect to I is I, the peak intensity Hr at the reference point such that the ratio with the peak top intensity HTOP at the peak portion is I + X is obtained by the following equation. Hr = HTOP × (I + X) This Hr is present at two points, a position preceding and succeeding the peak top in time, and a point closer to the measurement time point on the time axis at that time. Select as Hr.
(3) If the reference point giving Hr does not match the actual measurement point (for example, if there is a desired reference point between the measurement points j and k in FIG. 4), the measurement points j and j It is possible to use those obtained by synthesizing the absorbance spectra Sj and Sk for k by internal division. In this manner, the absorbance spectrum Si at the measurement time corresponding to the i-th time
Is obtained.

Next, the similarity r and the determination reference value t at the i-th measurement time point are calculated using the reference spectrum thus obtained. That is, the data of the i-th absorbance spectrum Si is read, and the similarity r between the two is calculated from the spectrum data and the data of the reference spectrum Ri obtained in step S4 (step S4).
5). Next, a determination reference value t is calculated from the data of the absorbance spectrum Si and the data of the reference spectrum Ri at the same measurement time point (i-th) and the noise level N calculated in step S1 (step S6). As the method of calculating the similarity r and the determination reference value t, the method described above can be used except that the reference spectrum data is different.

Then, it is checked whether or not the calculation of the similarity r and the determination reference value t has been completed for all the spectrum data included in the peak portion (step S7).
That is, the processes of steps S4 to S6 are executed from i = 0, and the process ends if the processes of steps S4 to S6 are executed for i = n. If not, the process returns to step S4 to calculate r and t for the absorbance spectrum data at the next measurement time T1 + ΔT. Different reference spectra Ri are determined corresponding to the respective measurement time points, and the spectrum data is used for calculating r and t.

After the calculation of r and t is completed for all the absorbance spectrum data belonging to the peak portion, a time range T3 to T4 in which the similarity r becomes equal to or less than the determination reference value t is obtained (step S8). ). Further, a graph as shown in FIG. 6 is created from the similarity r with respect to the time T and the change of the determination reference value t in the peak portion (T1 to T2) including the range T3 to T4, and the display of the personal computer 8 is displayed. (Step S9). When the similarity r exceeds the determination reference value t in the entire period in the predetermined peak portion, it is determined that there is no impurity.
Is smaller than the criterion value t, it indicates that there is an impurity. Thus, the measurer can immediately recognize the presence or absence of the impurity.

The graph obtained as described above can be used, for example, as follows. That is, r
Within the range T3 to T4 where <t is satisfied, there is no possibility that the collected absorbance spectrum pattern is the same as the spectrum pattern of the target component (spectrum pattern at the peak top) even if the maximum of actual noise is considered. (Actually, when the magnitude of the noise level is set to σ and when the noise level is set to 3 × σ, the likelihood of this determination is different.) ~
By setting the section of T4 not to be fractionated, only the target component can be reliably fractionated.

Further, this result is used to determine the detection limit of a chromatographic apparatus, which is how much impurity component can be identified when an impurity component having a retention time close to the target component is mixed into the sample. Can also be used. That is, when a graph as shown in FIG. 6 is created for a sample containing only a pure target component, the similarity r exceeds the determination reference value t in the entire range of the peak portion (T1 to T2), but is mixed in the sample of the target component. As the amount of the impurity component gradually increases, a range (T3 to T4) where r <t appears when the amount of the impurity reaches a certain value c1. This impurity amount c1 can be used as a detection limit of an unseparated peak of the chromatographic apparatus.

As described above, in the data processing apparatus of the present embodiment, when calculating the similarity r of the spectrum and the determination reference value t, a reference spectrum corresponding to the intensity at each measurement time and the like is created and used. I use. For this reason, unlike the related art, only the absorbance spectrum corresponding to the peak top is not used as a reference, and the influence of external fluctuation of the absorbance spectrum is reduced, and the calculation accuracy of the similarity r and the determination reference value t is improved. As a result, the purity of the peak on the chromatogram can be more accurately determined.

The above embodiment is merely an example, and can be modified or modified as appropriate within the scope of the present invention. Specifically, for example, when the reference point does not coincide with the measurement time point in step S4, instead of synthesizing the reference spectrum from the absorbance spectra at both adjacent measurement time points, of the two adjacent absorbance spectra, A spectrum closer to the measurement time may be selected.

Further, in the above-described analysis processing for determining the peak purity, the method of obtaining the reference spectrum Ri in step S4 may be as follows. First, measurement points 1 to n
And select the respective maximum intensity data on the absorbance spectrum and plot the data on the time axis to create a chromatogram showing the maximum intensity (that is, in this case, the wavelength giving each data is the same) Not exclusively). Now, suppose that FIG. 4 is the chromatogram created in this way. Next, Hr is obtained by adding the variable X to the peak intensity Hi at the i-th measurement time point on this chromatogram. That is, the peak intensity Hr of the reference point
Is determined by Hr = Hi + X. Here, X depends on the value of Hi, and the relationship is determined by preliminary experiments. This Hr exists at two points, a position preceding and succeeding in time with respect to the peak top, and a point closer to the measurement time point on the time axis may be selected.

Furthermore, in addition to those described above, the algorithm for calculating the reference spectrum Ri or the reference point can be appropriately changed.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing an embodiment of a high-performance liquid chromatograph to which a data processing device of the present invention is applied.

FIG. 2 is a diagram showing data acquired by a detector of the device according to the embodiment.

FIG. 3 is a flowchart showing a procedure of data analysis processing relating to peak purity determination in the apparatus of the embodiment.

FIG. 4 is an explanatory diagram of a method for obtaining a reference spectrum in data analysis processing.

FIG. 5 is a diagram showing an example of an absorbance spectrum.

FIG. 6 is a diagram showing an example of a graph obtained as a result of the data analysis processing by the present apparatus.

FIG. 7 is a diagram showing a noise portion in data acquired by a detector of the present apparatus.

FIG. 8 is a flowchart showing a procedure of a conventional data analysis process related to peak purity determination.

FIG. 9 is an explanatory diagram for explaining the significance of a similarity r and a determination reference value t in peak purity determination.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Eluent tank 2 ... Pump 3 ... Injector 4 ... Separation column 5 ... Detector 51 ... Flow cell 52 ... Light source 53 ... Spectroscope 54 ... Photodiode array 6 ... Amplifier 7 ... A / D converter 8 ... Personal computer

Claims (1)

[Claims] 1. A chromatographic apparatus equipped with a detector for sequentially detecting the absorbance spectrum of a sample eluted from a column as time elapses, wherein a peak appearing in a chromatogram in which absorbance at each measurement point is plotted in a time axis direction. A data processing device for judging purity: a) For each measurement time point from the start point to the end point of a target peak, an absorbance spectrum at a time point separated by an appropriate time corresponds to the measurement time point A reference spectrum calculating means for determining as a reference spectrum; b) absorbance data at each wavelength of the absorbance spectrum detected at each measurement time point; and a reference spectrum corresponding to each measurement time point determined by the reference spectrum calculation means. Based on the absorbance data, the detected absorbance spectrum at each measurement point and its C) similarity calculating means for calculating the similarity r with the reference spectrum corresponding to the fixed time point; c) data on the absorbance at each wavelength of the absorbance spectrum detected at each measurement time point; Reference value calculating means for calculating a determination reference value t at each measurement time point based on the absorbance data of the reference spectrum corresponding to the measurement time point and the absorbance data of the absorbance spectrum at the baseline, d) similarity at each of the measurement time points A determination means for determining a magnitude relationship between the degree r and a determination reference value t, a data processing device for a chromatograph.