A compositional heterogeneity method to simplify LC data analysis
By Susan Goodchild and Jeff Hurlbut
A custom routine that automates calibration and curve fitting reduces
the time to analyze polymer compositional distribution liquid chromatography
data by 50%. Researchers at Solutia, Inc. (Indian Orchard,
MA) use the compositional heterogeneity method to determine the
weight percent composition of two copolymers, monomer A and monomer
B. The routine has largely automated what was previously a time-consuming
manual process that involved the selection of baselines from calibration
chromatograms, calculating statistical parameters, fitting a second-order
polynomial curve to the data, locating peaks on the sample chromatogram,
and reformatting the data for presentation purposes.
The company's Physical and Analytical Sciences Center (PASC) performs
basic and applied research to support synthetic fibers, plastic
interlayer for laminated glass, phosphorus-based food ingredients
and performance products, high-performance specialty chemicals,
and chemical intermediates businesses. The center uses a wide range
of polymer characterization techniques to meet the needs of product
development and laboratory-scale manufacturing operations. One of
its standby methods is the compositional distribution LC method,
which gives percent composition and a measure of the polydispersity
of a copolymer, based on the weight percentage of one of the monomer
units. This method differs from the traditional LC, in which the
components of the sample elute (move along the column) at varying
speeds based on their affinity to the stationary phase in the column.
Instead, the sample is deposited on the column, and elutes based
on its solubility with the mobile phase. Initially, the LC analysis
is performed on a series of standards with varying proportions of
two different monomers. The resulting chromatograms are used to
prepare a calibration curve that is used to assess the composition
of the sample.
Figure
1
Calibration data fitted to a second-order polynomial. The quadratic
is used to map retention time to weight fraction.
Figure
2 Sample
chromatogram shows calculated results for heterogeneity and
asymmetry indices. |
This method has provided more than satisfactory performance in
terms of accurately characterizing a number of copolymers. However,
the required statistical analysis was time consuming and so difficult
that it required the attention of a professional with an advanced
degree. In order to generate the calibration curves, the scientist
had to choose the baseline points at the beginning and end of the
peak for each calibration chromatogram. Then, with the statistical
analysis software used, data had to be entered on a number of different
screens to enter the calibration data. The process took about 10
min for each calibration chromatogram.
From this point, a statistical program took over and generated
the necessary output parameters. However, the output from the program
was available only in a rigid format that did not match the researchers'
needs. In particular, researchers usually need to create a single-page
report that shows the sample chromatogram along with the calibration
data and curves, statistical output, and other parameters. This
meant that several different plots had to be cut out and taped together
and finally photocopied to produce the final report.
Performing the statistical analysis on a typical run of eight samples,
which are run in duplicate, took a total of 16 hr. Since this analysis
was run on a regular basis, it occupied a considerable amount of
scarce researchers' time. In an effort to streamline the process,
researchers investigated commercial data analysis software to determine
whether one would have the necessary statistical routines, graphical
output capabilities, and a programming language that would allow
them to automate the data analysis process.
They discovered Origin™ version 5.0, a Windows-based technical
graphics and data analysis software package (Microcal Software,
Inc., Northampton, MA), which provides several crucial advantages
that make it well suited for this application. The software offers
a wide range of statistical capabilities, including a fitting function
category designed specifically for chromatography applications (Figure
1). It features a wide range of graphing capabilities, allowing
the user to adjust virtually any parameter of the graph simply by
clicking on it with a mouse. Finally, it utilizes an extremely powerful
programming language, LabTalk™, that provides access to virtually
every function in the program
How Origin Was Used
Working with Origin application programmers, the researchers developed
a LabTalk program that largely automates the statistical analysis
process. A custom Origin tool provides the user interface and manages
the process. The user starts the process by clicking the NEW CALIBRATION
DATA button to open a worksheet into which he or she can enter chromatogram
readings. The user then clicks the MEAN AND STANDARD DEVIATION button
to calculate the parameters for each standard and then simply clicks
the GRAPH button to plot the mean and standard deviation versus
retention time.
The user specifies a request number and date, which Origin uses
to data-stamp the plot. The software fits the data to a second-order
polynomial and later uses the coefficients to calculate the percent
composition based on the retention time of the sample. Next, the
user clicks the OPEN CHROMATOGRAM button and loads the sample chromatogram
in the form of an ASCII file. The routine parses the report file,
removes index numbers that are not needed, and saves the rate values
that are specified by the user. The rate values are used to calculate
the independent x-axis of the output curve as retention time.
After the software plots the curve, the user takes advantage of
specialized automatic and manual tools to zoom in and set a baseline,
which automatically snaps to the curve. He or she then clicks PROCESS
to generate the final graph, which contains the calculated values
for the analysis: the percentage composition of the sample, a heterogeneity
index (HI), and an asymmetry index (AI) (Figure 2). The percent
composition is a weight average of the total polymer. The HI and
AI refer to the shape of the composition distribution. The HI is
a measure of the breadth of the distribution of the polymer, and
the AI refers to the skewness of the distribution. An AI of 1 means
that the polymer is symmetrical with respect to monomer A distribution;
an AI of greater than 1 means that the polymer is skewed to high
A content, and an AI of less than 1 means that the polymer is skewed
to a low A content.
The routine significantly reduces the time required to produce
the needed output. It automatically places the calibration worksheet
and calibration graph on the same page, eliminating the need to
tape multiple sheets of paper together and photocopy. In addition,
researchers can interact dynamically with the output and customize
it in any way they wish using the program's point-and-click interface.
For example, they can easily select and customize color, size, fonts,
markers, ticks, text labels, line styles, and background colors.
Additional data sets can easily be added to the page when desired.
Conclusion
Origin's facility for custom programming has dramatically improved
the speed and ease of LC data analysis. Reducing the time required
to generate output for eight samples in duplicate from 16 to 8 hr
saves a considerable amount of researchers' time that can be devoted
to other tasks. The routine has simplified the data analysis task
such that it can be delegated to a technician in the future for
even greater time savings.
Ms. Goodchild is Senior Research Chemist and Mr. Hurlbut is Senior
Research Physicist, Solutia, Inc., 730 Worcester St., Sprinfield,
MA 01151, U.S.A.; tel.: 413-730-2402; fax: 413-730-2196.
AMERICAN LABORATORY : 44 : SEPTEMBER 1998