1) Simulated HPLC-DAD non trilinear data sets

Download this data set

Here you may find two examples of typical HPLC-DAD non-trilinear data sets. They consist of four matrices with four components [D1, D2, D3, D4]. Both data sets are constructed using the same concentration profiles and spectra, which reproduce non-trilinear structures which are due in practice to different run-to-run working conditions (mobile phase changes or pH variations, column ageing, ...). This non-trilinearity affects the concentration profiles in the different runs, causing non-regularly spaced shifts, peak shape changes and variations in the peak overlap. There is no selectivity in the spectral direction whereas the elution profiles are severely overlapped in runs D1, D2 and D3 and show local rank conditions for a good resolution in D4. The first data set is noise-free, whereas the second has a noise level equal to 1% of the maximum absorption with a heteroscedastic pattern proportional to the signal. Higher noise levels have not been applied because they would not be representative of those found in real HPLC-DAD data sets.

The two data sets are compressed in a *.zip file called ntdata.zip. The data sets are *.mat files compatible with MATLAB. The noise-free data set is in the file called nfhplc.mat and the noise-added data in nhhplc.mat. Below you have the explanation related to the variables you will find in each of these files.

Data file: nfhplc.mat. Noise-free data set.

Variables

D1 (51 x 96).     D2 (51 x 96).     D3 (51 x 96).     D4 (51 x 96).

Matrices of different chromatographic runs. 51 are retention times and 96 are wavelengths. The three-way data set can be formed appending the four data sets along any of the three modes of the data sets (row-wise, column-wise or tube-wise).

c1 (51 x 4).     c2 (51 x 4).     c3 (51 x 4).     c4 (51 x 4).

Matrices of the elution profiles used to simulate the data sets, D1, D2, D3 and D4, respectively.

s (4 x 96).

Matrix of spectra.

Each data set has been generated as the product: Di = ci * s

Data file: nhhplc.mat. Noise-added data set.

Variables

D1nh (51 x 96).     D2nh (51 x 96).     D3nh (51 x 96).     D4nh (51 x 96).

Matrices of different chromatographic runs. 51 are retention times and 96 are wavelengths.
The three-way data set can be formed appending the four data sets along any of the three modes of the data sets (row-wise, column-wise or tube-wise).

c1 (51 x 4).     c2 (51 x 4).     c3 (51 x 4).     c4 (51 x 4).

Matrices of the elution profiles used to simulate the data sets, D1, D2, D3 and D4, respectively.

s (4 x 96).

Matrix of spectra.

noise1 (51 x 96).     noise2 (51 x 96).     noise3 (51 x 96).     noise4 (51 x 96).

Matrices of noise added.

Each data set has been generated as: Di = ci · s + noisei

2) Real HPLC-DAD data set (A)

Download this data set

This is a real HPLC-DAD data set of organophosphorous pesticides in natural waters used in a laboratory intercomparison exercise1. Data set A is a three-compound system with two pesticides identified (azinphos-ethyl and fenitrothion) and one unknown interferent. The three-way data set is formed by one matrix with the three compounds and two matrices of standards with one known compound different in each.

Data file: adataset.zip (contains adataset.mat)

d1: mixture matrix with three compounds (two analytes and one unknown interferent).

d2: standard matrix of analyte 1. spd2: pure spectrum of analyte 1.

d3: standard matrix of analyte 2. spd3: pure spectrum of analyte 2.

1. R. Tauler, S. Lacorte and D. Barceló. "Application of multivariate curve self-modeling curve resolution for the quantitation of trace levels of organophosphorous pesticides in natural waters from interlaboratory studies". J. of Chromatogr. A, 730, 177-183 (1996).

3) Real HPLC-DAD data set (B)

Download this data set

Data set B

This is a real HPLC-DAD data set of organophosphorous pesticides in natural waters used in a laboratory intercomparison exercise1. Data set B is a system with three compounds with two different pesticides identified (diazinon and parathion-ethyl) and one unknown interferent; the related three-way array is composed by two matrices with all three compounds in each.

Data file: bdataset.zip (contains bdataset.mat)

d1: mixture matrix with three compounds (two analytes and one unknown interferent).

d2: mixture matrix with three compounds (two analytes and one unknown interferent).

sp1: pure spectrum of analyte 1.

sp2: pure spectrum of analyte 2.

1. R. Tauler, S. Lacorte and D. Barceló. "Application of multivariate curve self-modeling curve resolution for the quantitation of trace levels of organophosphorous pesticides in natural waters from interlaboratory studies". J. of Chromatogr. A, 730, 177-183 (1996).