GPC/SEC is the established separation technique in all modern laboratories working with classical polymeric materials, construction products, electronic chemicals, pharmaceutical excipients, nutritional (hydro)colloids, bio-therapeutics, and many more.
The success story of GPC/SEC has multiple reasons:
- GPC/SEC delivers results which allow predicting product properties and application behavior directly from the chromatography lab.
- It is a very versatile technique with applications in R&D, product development, quality assurance and (online, atline) production control.
- It is a mature separation technique, well understood with companies that can make it work reliably and robustly.
- Instrumentation and software can be scaled from simple systems/applications to hyphenated systems with information-rich detectors in a Macromolecular Chromatography Data System (MCDS) environment.
- It is perfectly fit for regulated laboratories in pharma, food, cosmetic and related industries with established partners providing optimized hardware, software, quality support and expert service engineers to cover the complete system life cycle including comprehensive qualification services.
Figure 1 shows an easy to operate GPC/SEC system for routine work. Table 1 lists different types of systems and applications.
Challenges in Establishing Analytical Quality in the Laboratory
Laboratory processes have changed fundamentally with the requirement to operate more and more sophisticated information-rich techniques by the same or even reduced staff. In times of constantly increasing sample numbers this has led to an increasing demand in fast methods and lab automation for more output in shorter time. The lack of staff and, even more important, the lack of specific method experts in the labs, is a constant threat to delivering analytical quality.
This dilemma is shown in the quality triangle which relates the major contributions to the analytical process (Fig. 2).
Obviously, there has to be a balance between the required time, investment in infrastructure, staff and training, and the quality of the output which has to be matched to the criticality of the process or product.
Therefore, lab managers have to define which minimum analytical quality is required and adjust the necessary input in resources. This optimization can be found by iterating the effort/quality relationship as shown in Figure 3.
The Role of GPC/SEC Software for Analytical Quality
GPC/SEC results and their validity are crucial in many applications, e.g. as part of a QC/QA test for a product release. E.G. in pharmaceutical applications antithrombic heparin, dextran or hydroxy ethyl starch in emergency care have to meet very narrow specifications with help patients survive accidents and surgery. Equally important are the registration and accreditation of polymeric products or formulations with regulatory agencies like FDA, ECHA and many more, to ensure non-toxicity and save and environmentally friendly product use. Obviously, the accuracy and precision of the GPC/SEC results are an important issue in such a context.
Meaningful chromatographic results depend on the strict repeatability of separation conditions and instrument performance. The GPC/SEC separation process is governed by multiple parameters, just as for other liquid chromatography separation techniques. This is the reason for its versatility. On the downside the intrinsic multi-parameter processes require good understanding and adherence to strict laboratory processes by users. Additional parameters will influence the validity and accuracy of results as the instrumentation adds another level of complexity and its parameters need to be monitored.
Every chromatography user knows from own experience that many experimental details can influence the result and the final analytical quality of an experiment . There are always various systematic and random contributions to the accuracy and precision of the final result. High analytical quality is only achieved if systematic and random errors are avoided (Fig. 4).
Systematic errors can be avoided if users have sufficient training and time to perform their work. Random errors are much more difficult to identify and track.
Typical contributions to statistical (random) error are among others:
- Pump pressure fluctuation
- Old (noisy) UV lamp
- Unpurged RI detector
- Column (particle) shedding
- Insufficient degassing of eluent
- Poor calibration data
- Variations in multi angle laser light scattering (Malls) detector normalization
- Stability of viscosity data
- Precision of light scattering data
However, statistical models are very well suited to quantify random result deviations from the true (or generally accepted) value . Thus, if random errors are monitored, the result uncertainty can be determined.
Detector signal quality, calibration quality and system stability are main contributions to GPC/SEC result uncertainty. Advanced error propagation calculations have to be performed to get a reliable estimate of the final result uncertainty .
Modern chromatography software can be used for monitoring, evaluation and uncertainty calculations based on error propagation. This helps users to avoid potential pitfalls before a poor result enters a report.
Although result uncertainty tests have been required in many applications and by many users, even today there is only a single chromatography software  which allows the fully automated determination of result statistical errors for all types of results and all types of instrumentation being used in the lab. The measurement error determination in WinGPC is available for molar mass averages, molar mass fractions e.g. <500 Da, the crucial parameters for toxicity evaluation in FDA approvals and Reach registrations, peak areas, peak positions, viscosities, radii, Mark-Houwink constants. It can be used for all calibration methods from conventional, broad, universal, end group, viscosity or light scattering detection.
Figure 5 shows a screen shot of the molar mass distribution results with uncertainty calculation enabled. The result table shows the results for different properties in column 2 and the respective relative result uncertainty in column 3 in per cent. Additional detector signals are presented similarly.
In this example the weight average molar mass Mw has a value of 4390 Da with an uncertainty of 3.84% which relates to 168 Da molar mass uncertainty. Consequently, the true Mw value for this sample will be between 4222 Da and 4558 Da with a confidence level of about 70%. In order to achieve 95% confidence for a result the error has to be multiplied by 2, which means the true weight average molar mass Mw will be within 4054 Da and 4726 Da. These result uncertainties also mean that the results of independent experiments cannot be considered different with a validity of 95% if the individual results are within the confidence limits of 4054 Da and 4726 Da.
If users want to decrease the error margin, the described software offers advice from its artificial intelligence (AI) engine. An assessment of result uncertainty contributions is displayed, which recommends which aspects of the analysis should be improved. In the case of the sample in Figure 5, the major contribution to overall statistical errors (despite already very small) is the detector signal quality; calibration quality and instrument stability play almost no role in this run as shown in Figure 6.
Obviously, there is a multitude of GPC/SEC methods which benefit from result uncertainty assessment.
Figure 7 shows results from Heparin product release tests which demonstrate that the numerical results (mw averages and high and low molar mass fractions), meet product specifications. However, the result uncertainty is so large that the product could also fail in a patient’s heparin injection not preventing thrombosis. MCDS software AI hint directs analysts to the root cause, which is this case has been poor calibration quality.
Accuracy and Precision with light scattering or viscometry detection
The more complex GPC/SEC setups become, the more important are metrics which will support users to understand the implications on results. Multi-detector setups with Malls and viscometry detection have become popular, but their implication on result accuracy and precision has not been attended accordingly. The importance of understanding the influence of systematic and random errors on results from light scattering and/or viscometry data is discussed in the following.
Figure 8 shows the GPC/SEC results obtained from a triple detector setup with Malls and viscosity detectors hyphenated to UV and RI concentration detectors.
The result uncertainty for all parameters is small and within expectations (refer to columns 6 and 7 in Table 4).
When we compare the measurement uncertainties of the same chromatographic run and the identical raw data set without using the LS and viscosity raw data, then we observe that results with conventional (narrow standard calibration) show significant smaller result uncertainties (see Table 4 columns 3 and 5).
The reason for larger statistical errors with viscosity (approx. factor of 2) and especially with light scattering (approx. by a factor 4 to 6) detections is caused by additional signals (e.g. 18 angles for Malls) and parameters which are required to determine the result. Consequently, this means highest probability for repeating and/or reproducing a result is with a GPC/SEC method which is based on the least complex setup.
Please note that the molar mass independent results are not changed, RI result uncertainty for peak position, Vp, and peak area, A, is the same as these results are not subject to additional calculations.
While there is a higher result uncertainty for results obtained with light scattering and viscometry, it is also important to remember that GPC/SEC is a relative method which separates based on the hydrodynamic volume of a molecule in solution and requires calibration. Accurate results can only be obtained, if the samples and calibrants are chemically and structurally comparable. Light scattering and viscosity detectors can overcome this shortcoming of GPC/SEC. However, the application will dictate which parameter is the most important to be optimized, accuracy or precision.
Any analytical method has its intrinsic inaccuracy. In order to interpret the results in the right way it is important to know the precision and accuracy of the analytical method. Determination of the result uncertainty will enhance analytical quality substantially because it can address the statistical errors. Results of sample comparisons can be interpreted more accurately as being identical or different within the uncertainty limits. Standard GPC/SEC software can calculate result precision without any additional burden on the user.
PSS Polymer Standards Service GmbH,
GPC/SEC/GFC web-semiar (on-demand)
1) a) D. Held, P. KilzS. Kromidas, H.-J. Kuss (eds.);, Qualification of GPC/GFC/SEC Data and Results, in: Quantification in LC and GC, Wiley-VCH, Weinheim, 2009
b) ISO Guide to the Expression of Uncertainty in Measurement, International Organisation for Standardisation, Geneva, 1995
c) ISO 5725: Accuracy of measurement methods and results, Geneva, 1997
d) S. Ellison, M. Rosslein, A. Williams, (eds.), EURACHEM/CITAC Guide: Quantifying Uncertainty in Analytical Measurement, London, 1995
e) A. Williams, S. Ellison (eds.), EURACHEM/CITAC Guide: Use of Uncertainty Information in Compliance Assessment, London, 2007
f) S. Ellison, M. Rosslein, A. Williams, (eds.), EURACHEM/CITAC Guide: Traceability in Chemical Measurement, London, 2003
2) P.R. Bevington, D.K. Robinson; Data Reduction and Error Analysis for the Physical Science, Mc.Graw-Hill, New York, 1992
3) B.P. Roe; Probability and Statistics in Experimental Physics, Springer, Berlin, 1992
4) PSS WinGPC UniChrom MCDS Specifications, PSS Polymer Standards Service, Mainz, 2009