What

I have sat through four EPA audits in the last decade, two of them as the lab manager being questioned and two as a consultant called in after a notice of violation landed on someone else’s desk. The pattern is always the same. The auditor does not start with your instrument logbooks. They start with the method reference in your SOP, then walk backward through the sample chain to see whether what you actually did matches what the method says you must do. Most findings I have seen come from a gap of two or three steps in that walk, not from a bad calibration curve.

This article is about what EPA methods actually require of a lab, and what an auditor will ask you to produce to prove you met those requirements. It is not a summary of the methods themselves. It is a map of the checkpoints.

Under 40 CFR Part 136 for Clean Water Act work, the method is the method. You cannot substitute a different column, a different extraction solvent, or a different detector without going through the alternate test procedure process or documenting equivalence under the flexibility provisions. Under SW-846 for solid waste under RCRA, you get more room to maneuver, but you also carry more burden to prove that what you did is equivalent to what the method specifies. Either way, the auditor wants to see the method number and revision date on the bench sheet, not just on the SOP cover page.

Here is the failure I see most often. A lab writes its SOP in 2019 against Method 8270E, then in 2021 the method gets a revision, and the SOP never gets updated. The bench analysts are running the current method because the instrument software has it loaded, but the SOP still says 8270E. The auditor pulls a bench sheet from last quarter, sees the method reference, pulls the SOP, sees a different revision, and now you have a documentation finding that costs you three days of corrective action paperwork. The chemistry was fine. The paperwork was not.

I keep a one-page method cross-reference sheet taped inside the front cover of every SOP binder. It lists the method number, the current EPA revision, the date we last reviewed it, and the initials of whoever did the review. When the auditor asks, I hand them the sheet. It takes thirty seconds and it kills that whole line of inquiry.

The quality control checkpoints auditors actually pull

Every EPA method has a table of QC acceptance criteria. Method 8260D, for example, requires a continuing calibration verification every twelve hours, a method blank with each batch of twenty or fewer samples, a laboratory control sample with each batch, and a matrix spike and matrix spike duplicate at a defined frequency. Method 625 requires similar checks but with different acceptance windows. The auditor does not want to see that you ran these. They want to see that you ran them at the frequency the method specifies, that you evaluated them against the method’s criteria, and that you took action when they failed.

The action part is where labs get caught. I have reviewed dozens of QC packages where a continuing calibration verification came in at 18 percent deviation against a 15 percent criterion, and the analyst wrote “accepted” in the comment field with no explanation. That is an automatic finding. The method gives you a path forward (reanalyze, or document a specific justification), but “accepted” is not a justification. The auditor will ask you to explain it, and if you cannot, they will ask how many other times you did the same thing.

Calibration and continuing verification

For a typical GC-MS volatile method under SW-846, the initial calibration requires a minimum of five concentration levels, with the lowest standard at or below the reporting limit. The curve fit (linear, quadratic, or average response factor) has to meet the method’s criteria, usually a relative standard deviation of 20 percent or less for average response factor, or a coefficient of determination of 0.99 or better for a regression. Continuing calibration verification is typically a mid-level standard run at the start of each twelve-hour analytical shift, with an acceptance window of plus or minus 20 percent of the initial calibration response.

What the auditor checks: the timestamps on the CCV runs relative to the sample runs. If your shift started at 07:00 and your last sample was injected at 19:45, you have a problem, because the CCV was more than twelve hours old. I have seen labs fail this because the autosampler queue was set up without checking the clock. The fix is a hard stop in the sequence table, not a reminder on the whiteboard.

Blanks and contamination control

Method blanks are the quiet killer. A method blank is carried through the entire preparation and analysis sequence with the samples, and it must show no target analytes above the reporting limit. If your method blank shows methylene chloride at 8 micrograms per liter against a reporting limit of 5, you have a contamination problem somewhere in the chain: the reagents, the glassware, the vial septa, the purge-and-trap system, or the lab air itself. The auditor will ask you to identify the source and to demonstrate that you corrected it before the affected samples were reported.

I have personally traced a persistent methylene chloride blank to the septa on the sample vials. The liner facing was a silicone formulation that off-gassed under the purge conditions. Switching to a PTFE-faced silicone septum with a lower bleed profile eliminated the blank within one batch. If you are chasing the same problem, the septa on your HPLC vial closures are the first place I would look, before you tear apart the concentrator.

Holding times, preservation, and the chain of custody walk

Every EPA method specifies a maximum holding time from collection to extraction and from extraction to analysis. For volatile organics under Method 5030/8260, the holding time is 14 days from collection to analysis, with zero headspace in the vial and preservation to pH less than 2 with HCl. For semivolatile organics under Method 3510/8270, it is 7 days to extraction and 40 days from extraction to analysis. For metals under Method 3010/6010, it is 6 months for most analytes, with preservation to pH less than 2 with nitric acid.

The auditor will pull the chain of custody form, the cooler temperature record, the preservation check record, and the extraction and analysis dates, and they will do the arithmetic in front of you. If the sample was collected on March 3 at 14:00 and analyzed on March 18 at 09:00, that is 15 days and 19 hours. The method says 14 days. You have an exceedance, and the data are flagged as estimated or rejected depending on the program. The auditor will then ask how many other samples in the batch have the same problem, and whether your LIMS is set up to flag holding time exceedances automatically.

I have one strong opinion on holding times. Build the flag into the LIMS at login, not at report time. If the system warns the analyst when the sample is received that the clock is at day 11 of 14, you have a chance to prioritize the extraction. If the warning only appears when the report is generated, you have already lost.

Preservation and cooler temperature

Cooler temperature at receipt must be 4 degrees Celsius plus or minus 2 degrees, measured with a calibrated thermometer, with the measurement recorded on the COC or a receipt form. Preservation check for volatiles is a pH check on a separate aliquot, not on the sample vial itself, because opening the sample vial to check pH introduces headspace and loses the volatiles. The auditor will ask you to demonstrate that the pH check aliquot was separate. If your SOP says “check pH in the sample vial,” that is a finding.

Chemical preservation for metals is to pH less than 2 with nitric acid, verified with pH paper or a meter. For cyanide and sulfide, the preservation is different (NaOH to pH 12 for cyanide, zinc acetate for sulfide), and the auditor will check that the analyst used the right preservative for the right analyte group. I have seen a batch of cyanide samples preserved with nitric acid because the analyst grabbed the wrong bottle from the prep bench. The samples were destroyed. The audit finding was secondary to the data loss.

Instrument and consumable specifications that show up in audit findings

This is the part of the audit where consumables matter, and it is the part most labs under-document. The auditor is not going to ask you what column you used. They are going to ask you to prove that the column you used meets the method’s performance criteria, and that you have a record of when it was installed, what it was conditioned at, and when it was retired.

Columns and inlet consumables

For a Method 8260 analysis on a GC-MS, a typical column is a 60 m x 0.25 mm x 1.4 micrometer film thickness DB-624 or equivalent, and the method specifies the column or a performance-equivalent alternative. The auditor will ask for the column log: installation date, conditioning temperature and time, the initial performance check (usually a tune check and a calibration), and the retirement criteria (usually a failed calibration or a failed system suitability check). If you cannot produce that log, you have a finding. If you can, you have a clean line item.

The inlet consumables are where I see the most undocumented substitutions. The method specifies a liner type (for example, a 2 mm ID splitless liner with a deactivated glass wool plug for 8270, or a 1 mm ID liner for 8260 to minimize flashback). If the analyst swapped in a different liner because the correct one was out of stock, and the calibration still passed, the auditor may not catch it on the QC review, but they will catch it on the consumables inventory reconciliation. The liner part number on the bench sheet has to match the liner part number in the method. I keep a laminated card in the inlet maintenance kit with the approved part numbers for each method, and the analysts initial the card when they change a liner. It sounds fussy. It has saved me twice.

For the columns themselves, I buy from a small number of suppliers and I keep the certificates of analysis on file. Agilent’s GC column and accessory line is the reference standard for most of the methods I run, and Restek’s technical resources are where I send new analysts to understand retention index and bleed behavior before they touch a method. For sample handling and autosampler consumables, Thermo Fisher’s chromatography and sample handling catalog is a useful cross-reference when you are trying to match a vial and cap to a specific autosampler tray.

Vials, caps, and septa

The auditor will not usually ask about vials directly, but the data quality findings that come out of vial and septum problems will land in the audit report as blank contamination, carryover, or calibration drift. The specifications that matter:

  • Vial volume and geometry: a 2 mL, 12 x 32 mm vial is the standard for most autosamplers. The GPI 8-425 thread is the short screw thread used on many 8 mm caps; the GPI 13-425 is the 13 mm thread used on 4 mL vials. Match the thread to the autosampler tray, not the other way around.
  • Septa: silicone with a PTFE facing on the sample side is the default for volatile work. The PTFE facing is what keeps the sample from contacting the silicone, which is what causes the off-gassing I mentioned earlier. For semivolatile work, a pure PTFE or a PTFE-lined silicone is acceptable. For high-temperature work above 200 degrees Celsius, use a pure PTFE septum or a high-temperature silicone.
  • Puncture counts: a standard silicone septum is good for roughly 3 to 5 punctures from a 22-gauge needle before the reseal integrity degrades. Pre-slit septa are good for more, but they leak more on the front end. I replace septa on the autosampler every 100 injections regardless, because the cost of a septum is trivial compared to the cost of a reanalysis.
  • Cap torque: hand-tight plus a quarter turn is the working rule. Over-torque deforms the septum and creates a channel. Under-torque lets the vial leak during the purge. I keep a torque wrench set to 15 inch-pounds on the capping station for the volatile vials, and I check it monthly against a reference vial.

If you are sourcing vials and closures, the catalogs at Lab Vials and Chromatography Consumables are the two I check first for part-number cross-references against the major autosampler platforms. The analytical testing side of the house is where the method and compliance questions get answered, and I use it as a sanity check when I am writing a new SOP.

A comparison checklist for method compliance documentation

Here is the checklist I use when I am preparing a lab for an EPA audit. It is organized by what the auditor asks for, not by what the method says, because those are two different documents.

  • Method reference: SOP revision date matches the current EPA method revision. Bench sheets carry the method number and revision. Cross-reference sheet in the SOP binder. Finding if missing: documentation nonconformance.
  • Initial calibration: minimum five levels, lowest at or below the reporting limit, curve fit criteria met, raw data retained. Finding if missing: data quality, possible data invalidation.
  • Continuing calibration verification: run at the method-specified frequency (typically every 12 hours), acceptance window met, corrective action documented for failures. Finding if missing: data quality, batch rejection.
  • Method blank: one per batch of 20 or fewer samples, no target analytes above the reporting limit, source investigation documented for any detection. Finding if missing: contamination control failure.
  • Laboratory control sample: one per batch, recovery within the method window (typically 70 to 130 percent for most organics), corrective action documented. Finding if missing: accuracy verification failure.
  • Matrix spike and matrix spike duplicate: at the method-specified frequency (typically one per 20 samples or one per batch), recovery and relative percent difference within criteria. Finding if missing: precision verification failure.
  • Holding times: collection to extraction and extraction to analysis within the method limits, LIMS flag at login, documented corrective action for exceedances. Finding if missing: data invalidation.
  • Preservation and cooler temperature: pH check on a separate aliquot, cooler temperature at receipt within 4 plus or minus 2 degrees Celsius, calibrated thermometer. Finding if missing: sample integrity failure.
  • Consumables log: column installation and conditioning records, liner part numbers matching the method, septum replacement records, vial and cap lot numbers traceable to the bench sheet. Finding if missing: traceability failure.
  • Data review: second-analyst review documented, raw data retained, electronic audit trail enabled on the CDS. Finding if missing: data integrity failure.

The two items that trip up the most labs are the consumables log and the data review. The consumables log because nobody wants to write down a liner part number, and the data review because the second-analyst signature is often applied at report time rather than at raw-data review time. The auditor can tell the difference by looking at the timestamps.

Troubleshooting checklist for a failing audit prep

If you are two weeks out from an audit and you are not sure you will pass, here is the order I would work in.

  1. Pull the last six months of QC packages. Look for any CCV or LCS failure where the comment field says “accepted” without a justification. Those are your first findings. Write a corrective action memo for each one now, before the auditor asks.
  2. Check the SOP revision dates against the current method revisions. If any SOP is behind, update it, and document the review date. A one-page addendum is faster than a full rewrite and it satisfies the auditor.
  3. Pull the holding time records for the last 90 days. Identify any exceedances and confirm they were flagged and reported as estimated. If they were not flagged, you have a LIMS problem and a reporting problem.
  4. Walk the consumables inventory. Match the liner, septum, and column part numbers on the bench sheets to the part numbers in the method. Any substitution needs a documented equivalence justification.
  5. Check the cooler temperature records. Any reading outside 2 to 6 degrees Celsius needs a documented assessment of sample integrity. No assessment, no defense.
  6. Run a mock audit with someone who did not prepare the lab. Give them the checklist above and two hours. They will find the thing you have been walking past for six months.

I want to be direct about one thing. The audit is not a test of your chemistry. It is a test of whether your documentation can prove your chemistry. I have seen excellent analysts fail audits because their bench sheets were incomplete, and I have seen mediocre labs pass because their paperwork was airtight. The method is the contract. The documentation is the proof of performance. You need both.

Takeaway

EPA method compliance comes down to four things an auditor will check in order: the method reference on the bench sheet, the QC frequency and acceptance criteria, the holding time and preservation records, and the consumables traceability. Get those four right and the audit is a review, not an investigation. Get them wrong and no amount of instrument performance data will save the batch. The consumables side is the easiest to fix and the most often ignored, so start there: a liner part number on a bench sheet and a septum replacement log cost you ten minutes a week and they close the finding before it opens.

References & further reading

Related resources from our lab consumables network:

External references and standards: