What “regulated method” actually means in chromatography--Lab Chromatography Supplier

I still remember the first time a client sent back a method validation package because our calibration curve failed the EPA 8000D acceptance criteria. The RSD on the mid-point standard was 21%, the correlation coefficient was 0.994, and the method blank had a phthalate peak that we spent three days chasing through the solvent lot, the septa, and finally a bottle of PTFE-lined caps that had been stored next to a drum of plasticizer. That experience taught me something I now repeat to every lab manager I work with: in regulated chromatography, the method is not the document. The method is the entire chain from the sample container to the integration parameters, and every link in that chain has a specification that someone will audit.

This article covers what actually matters when you run chromatography under EPA, ISO, or AOAC methods. I will focus on the parts that trip up experienced analysts: acceptance criteria that differ between methods, consumable specifications that are buried in footnotes, and the failure modes that show up in audits rather than in day-to-day runs. If you are building or transferring a regulated method, the details below are the ones I have seen cause the most rework.

A regulated method is one where the procedure, the acceptance criteria, and the reporting requirements are defined by a body outside your lab, and where a third party can inspect your compliance. That is a different animal from an internal SOP. Under an internal SOP, you can change the column, adjust the gradient, and update the document. Under an EPA method, a column change may require a demonstration of equivalency. Under an ISO method, you may need to re-validate the affected performance characteristics. Under an AOAC method, you may need to run the method as written unless the method explicitly permits modification.

The three families most chromatography labs encounter are:

U.S. EPA SW-846 methods for waste and environmental samples, including the 8000 series for chromatography, the 3500 series for sample preparation, and the 6000/7000 series for specific analyte classes. These are published as individual methods with their own QC criteria, and they are updated on a rolling basis. The current compilations are available from the U.S. EPA SW-846 test methods for waste and materials page.
ISO methods for water quality, air quality, and food, such as ISO 6468 for organochlorine pesticides in water or ISO 11890-2 for VOC content in paints. ISO methods tend to specify performance requirements rather than fixed consumables, which gives more flexibility but shifts the validation burden onto the lab.
AOAC methods for food, feed, and agricultural products, such as AOAC 2007.01 for pesticide residues using QuEChERS. AOAC methods are often performance-based but include specific procedural details that are treated as normative.

The practical difference is where the flexibility lives. EPA methods often tell you exactly which column, which detector, and which calibration model to use, and then tell you how to prove an alternative is equivalent. ISO methods tend to tell you what performance you must achieve and let you choose the tools. AOAC methods sit in between. Knowing which regime you are under determines how much freedom you have with consumables, and that determines how much validation work you will do when a column is discontinued.

Consumables specifications that auditors actually check

Most audit findings I have seen in chromatography labs are not about the instrument. They are about consumables that were not controlled, not documented, or not qualified. The auditor does not usually care that you used a different brand of vial. The auditor cares that you cannot show the vial was suitable for the method and that it did not introduce contamination or analyte loss.

Columns and the equivalency question

Under EPA methods, the column is often specified by phase, dimensions, and sometimes by part number. Method 8270E, for example, specifies a 30 m x 0.25 mm x 0.25 µm 5% phenyl methylpolysiloxane column for semivolatile organics, with a defined temperature program. If you substitute a column with the same nominal phase but a different manufacturer, you are making a change that requires demonstration of equivalency under the method’s provisions. In practice, that means running the method’s performance evaluation mixture, meeting the tailing factor and resolution criteria, and documenting the comparison.

I have seen labs get this wrong in both directions. One lab replaced a specified column with a “better” one and failed an audit because the equivalency documentation was a single injection of a check standard. Another lab kept using a discontinued column from a dwindling stock rather than doing the equivalency work, and ended up with a method that could not be run once the stock ran out. The right approach is to build the equivalency data before you need it. Run the performance mixture on both columns, document retention time windows, resolution, and tailing factors, and keep the data in the method file.

For GC columns specifically, the phase is not the only variable. Film thickness affects retention and bleed. Internal diameter affects capacity and linear velocity. A 0.25 mm column and a 0.32 mm column with the same phase and film thickness are not interchangeable for a regulated method without demonstration. I keep a comparison table for every regulated method in the lab, with the specified column, the qualified alternatives, and the performance data that supports each alternative. Vendors like Agilent and Restek publish useful technical data on phase equivalence, but the demonstration is still yours to run and document.

Sample containers, septa, and the contamination pathway

This is where I have seen more regulated-method failures than anywhere else. A method blank with a phthalate peak, a semivolatile result with a silicone artifact, or a volatile analysis with a systematically low recovery often traces back to the container, the septum, or the cap liner.

For volatile organic analysis under EPA 5030/8260, the sample container is specified as a 40 mL amber glass vial with a PTFE-lined septum and a screw cap. The septum is not a trivial component. A silicone-faced septum with a thin PTFE film will bleed silicone into the headspace, and the film can be punctured through after a small number of autosampler injections. I have measured septum coring at as few as 30 to 50 punctures on a low-quality septum with a 23-gauge needle, which shows up as a ghost peak or a loss of pressure integrity. A PTFE-faced silicone septum from a reputable supplier will typically hold up for 100 or more punctures, but I still replace them on a schedule rather than waiting for a failure.

The cap torque matters too. Under-torqued caps leak during storage and transport, and over-torqued caps can crack the vial or deform the septum. For a 24 mm screw cap on a 40 mL vial, I target a torque of about 1.5 to 2.0 in-lb (roughly 0.17 to 0.23 N·m), applied with a calibrated torque wrench or an automated capper. This is not a number most labs check, and it is a common finding when volatile recoveries are inconsistent across analysts.

For semivolatile and metals work, the container material and the cleaning procedure matter more than the closure. I have seen method blanks fail because amber glass bottles were cleaned with a detergent that left a residue, or because the bottles were stored in a cabinet with solvent vapors. The fix is a documented cleaning procedure, a dedicated storage area, and a lot-level blank check. If you are running COD or other colorimetric methods alongside chromatography, the same logic applies to your digestion vessels, and a dedicated COD vial supply with lot traceability saves a lot of investigation time.

Filtration and the artifact problem

Filtration is one of the most under-controlled steps in regulated chromatography. A filter that sheds particles, adsorbs the analyte, or leaches extractables will fail a method blank or a recovery check, and the failure may not show up until the method is in routine use. For HPLC methods, I specify a 0.45 µm or 0.22 µm membrane with a low-extractable material, and I qualify the filter against the method’s blank and recovery criteria before adopting it. PTFE is a common choice for organic solvents because it is chemically inert, but not all PTFE is equal. A sintered PTFE with a loose structure can shed particles, and a laminated PTFE can delaminate under pressure. I look for a PTFE membrane filter with a defined pore size distribution, a documented extractables profile, and a housing that will not contribute plasticizers. For aqueous mobile phases and biological samples, a different membrane chemistry may be appropriate, but the qualification logic is the same.

For mobile phase filtration, the filter is protecting the column and the instrument, not the sample. A 0.45 µm filter on the mobile phase reservoir removes particulates that would otherwise accumulate on the frit and cause a pressure rise. I have seen a 200 psi pressure increase over a week traced to a mobile phase that was not filtered and a reservoir that was not kept under helium. A HPLC filter with a compatible membrane and a clean housing is a low-cost way to avoid that failure mode. The key is to qualify the filter for the mobile phase, not just for water.

Calibration, QC, and the acceptance criteria that differ

The acceptance criteria in regulated methods are not harmonized. EPA 8000D specifies a initial calibration with a minimum of five concentration levels, a correlation coefficient of at least 0.99 for most analytes, and a percent relative standard deviation (RSD) of 20% or less for the calibration factors, with some analytes allowed 30%. ISO 17025 does not specify calibration criteria directly; it requires that the lab demonstrate the calibration is fit for purpose, which usually means adopting the method’s criteria or a documented alternative. AOAC methods often specify a minimum number of calibration levels and a maximum residual.

The practical consequence is that a calibration curve that passes one method may fail another. I have seen a curve with an RSD of 22% pass an EPA method that allows 30% for a specific analyte and fail an internal SOP that adopted a blanket 20% limit. The fix is to build a method-specific QC table rather than a generic one, and to make sure the analysts know which criteria apply to which method.

Calibration model selection

Most regulated chromatography methods default to a linear calibration through the origin or a linear calibration with a non-zero intercept. The choice matters. A forced-through-origin calibration can bias low-level results if the detector response has a non-zero intercept, and a non-zero intercept can produce negative results near the detection limit. EPA 8000D allows both, but requires that the calibration model be justified by the data. I use a linear model with a non-zero intercept for most GC-MS methods and check the residuals at the low end. If the residuals show curvature, I move to a quadratic model only if the method permits it and I can demonstrate the fit with a lack-of-fit test.

The number of calibration levels is another point of divergence. Five levels is common, but some methods require six or more for a wide concentration range. I have seen a method fail a performance audit because the calibration range did not bracket the sample results, even though the curve itself was acceptable. The rule I use is that every sample result must fall within the calibration range, and any result above the highest standard must be diluted and re-run.

Continuing calibration and drift

Continuing calibration verification (CCV) is where routine methods live or die. EPA methods typically require a CCV at the beginning of each analytical batch and every 10 to 20 samples, with an acceptance criterion of ±15% of the true value for most analytes and ±20% for some. A CCV failure triggers a batch rejection and re-analysis, which is expensive. The most common causes I have seen are column degradation, detector drift, and a standard that has degraded in storage.

Column degradation shows up as a change in retention time, a loss of resolution, or an increase in tailing. I track retention time windows and tailing factors on every CCV, not just the concentration. A shift of more than 0.05 minutes in retention time or a tailing factor above 2.0 is a warning sign, even if the concentration is within criteria. Detector drift is often a sign of a dirty source or a failing electron multiplier, and it usually shows up as a gradual decline in response rather than a sudden failure. A standard that has degraded in storage is the easiest to fix and the easiest to miss; I check standard logs against preparation dates and storage conditions at every audit.

Method detection limits and the consumable contribution

Method detection limits (MDLs) are a regulated number in EPA methods. The MDL is defined as the minimum concentration of an analyte that can be measured and reported with 99% confidence that the analyte concentration is greater than zero, and it is determined by analyzing seven or more replicate spiked samples at a concentration near the expected MDL. The standard deviation of the replicates is multiplied by the Student’s t-value for n-1 degrees of freedom (3.14 for seven replicates at 99% confidence).

What most labs miss is that the MDL is a property of the entire method, including the consumables. A filter that adsorbs 10% of the analyte will raise the MDL. A vial that contributes a background peak will raise the MDL. A septum that bleeds will raise the MDL for volatile analytes. I have seen an MDL study fail because the lab used a different lot of filters than the one used during validation, and the new lot had a higher extractables profile. The fix is to qualify consumables at the MDL level, not just at the mid-point calibration level, and to control lot changes.

The same logic applies to the calibration standard and the internal standard. A standard that is prepared from a different source or a different lot can shift the response factor and change the MDL. I keep a standard qualification record for every regulated method, with the source, lot, purity, and preparation date, and I re-run the MDL study whenever a critical consumable or standard changes.

Comparison: EPA, ISO, and AOAC method requirements

The table below summarizes the differences that matter most in day-to-day operation. The numbers are typical rather than universal, and the specific method should always be checked.

Calibration levels: EPA methods typically require five or more; ISO methods specify performance rather than a fixed number; AOAC methods typically require five or more with a defined residual limit.
Calibration acceptance: EPA 8000D requires RSD of 20% or less (30% for some analytes) and r ≥ 0.99; ISO methods require a documented fit-for-purpose criterion; AOAC methods often require a correlation coefficient of 0.995 or better.
Continuing calibration: EPA methods require a CCV every 10 to 20 samples with ±15% recovery; ISO methods require a documented drift criterion; AOAC methods typically require a CCV every batch with ±10% to ±15% recovery.
Method blanks: EPA methods require one per batch of 20 or fewer samples; ISO methods require a documented blank frequency; AOAC methods require one per batch.
Matrix spikes: EPA methods require one per batch with 70% to 130% recovery for most analytes; ISO methods require a documented recovery criterion; AOAC methods typically require 70% to 120% recovery.
Column specification: EPA methods often specify the column; ISO methods specify performance; AOAC methods often specify the column or the phase.
Documentation: EPA methods require a defined data package; ISO 17025 requires traceability and uncertainty; AOAC methods require method-specific documentation.

If you are running a method under more than one regime, the safest approach is to adopt the most stringent criterion for each parameter and document the choice. That avoids the situation where a method passes one audit and fails another for the same data.

Operational precautions that prevent rework

The following practices have saved me more time than any single instrument upgrade. They are not glamorous, but they are the difference between a method that runs and a method that gets rejected.

Control consumable lots. Record the lot number for every column, filter, vial, and septum used in a regulated method, and qualify a new lot before it enters routine use. A lot change is a method change until you prove otherwise.
Qualify filters at the method blank and MDL level. A filter that passes a mid-point recovery check can still fail a method blank. Run the blank and the low-level spike before adopting a new filter lot.
Torque caps to a defined value. Use a calibrated capper or torque wrench and record the value. For 24 mm caps on 40 mL vials, 1.5 to 2.0 in-lb is a working range. Under-torqued caps leak; over-torqued caps crack.
Replace septa on a schedule. Do not wait for a failure. A 100-puncture replacement interval is conservative for most autosamplers and avoids the coring and bleed problems that show up as ghost peaks.
Track retention time and tailing on every CCV. Concentration alone is not enough. A shift in retention time or a tailing factor above 2.0 is an early warning of column degradation.
Store standards and solvents away from plasticizers. I have seen a method blank fail because the solvent cabinet was next to a drum of plasticizer. The contamination pathway is real and it is not always obvious.
Document equivalency before you need it. If a column is discontinued, you will need the equivalency data. Build it while both columns are available.

Troubleshooting checklist for regulated method failures

When a regulated method fails, the investigation should follow the sample pathway rather than the instrument. The checklist below is the one I use, in order.

Method blank failure: check the solvent lot, the filter lot, the vial and septum lot, and the storage area. Run a blank with a new lot of each consumable to isolate the source.
Low recovery: check the filter adsorption, the vial adsorption (especially for hydrophobic analytes), and the standard preparation. Run a recovery check with and without filtration.
High recovery: check for contamination in the consumables, the solvent, and the standard. A high recovery in a matrix spike can also indicate a matrix effect that the method does not correct.
CCV failure: check the column (retention time, tailing, resolution), the detector (response, drift), and the standard (age, storage). Replace the column only after the standard and detector are ruled out.
Retention time shift: check the mobile phase preparation (pH, buffer concentration), the column temperature, and the column age. A shift of more than 0.05 minutes is a warning sign.
Ghost peaks: check the septum, the vial, the solvent, and the needle wash. A ghost peak that appears at a fixed retention time is usually a consumable, not the instrument.
Pressure rise: check the mobile phase filter, the guard column, and the column frit. A 200 psi rise over a week is usually a filtration or precipitation problem.
MDL failure: check the consumable lot, the standard, and the replicate preparation. An MDL study that fails on one analyte is often a consumable problem, not a method problem.

Recommendation

For a lab running regulated chromatography under EPA, ISO, or AOAC methods, the highest-return investment is a consumable qualification program, not a new instrument. The program should include a written specification for each critical consumable (column, filter, vial, septum, standard), a lot-level qualification procedure that includes a method blank and a low-level spike, and a record that ties each lot to the method and the analytical batch. The specific parameters I recommend are: a 0.45 µm or 0.22 µm PTFE membrane filter qualified at the method blank and MDL level for HPLC methods; a 40 mL amber vial with a PTFE-lined septum torqued to 1.5 to 2.0 in-lb for volatile methods; a septum replacement interval of 100 punctures or fewer; and a column equivalency data package that includes retention time, resolution, and tailing factor for the method’s performance mixture. Vendors like Thermo Fisher publish useful specifications for sample handling, but the qualification data has to come from your lab.

If you are building a method from scratch or transferring one between labs, start with the consumable specification and the qualification data, then build the instrument method around them. The instrument is usually the most reliable part of the system. The consumables are where the variability lives, and in a regulated method, variability is the thing you have to control. A well-documented consumable program will pass more audits and cause fewer investigations than any single piece of equipment, and it costs a fraction of what a new detector costs. The full range of chromatography consumables and technical resources is available through Lab Chromatography, and the method-specific details are always in the method document, not in the catalog.

References & further reading

Related resources from our lab consumables network:

External references and standards: