Scope the audit before you book the flight--Reagent Bottle for Sale

I have run supplier audits on three continents, and the pattern is always the same: the vendor who sends you a glossy PDF with a certificate of analysis attached is rarely the one who survives a two-hour walk through the plant. The audit is not a document review. It is a physical inspection of how glass is annealed, how closures are torqued, and how a lot number travels from a mold cavity to your receiving dock. What follows is the checklist I actually carry, with the numbers and failure modes I have caught in the field.

An audit of a reagent bottle vendor is not one audit. It is at least three, and they have different pass criteria. I split them this way because a supplier that molds borosilicate glass well may buy in its closures from a third party with no incoming inspection, and a distributor that warehouses beautifully may have no idea which cavity made a given bottle.

Tier 1: the molder. Owns the glass, the annealing lehr, and the mold cavities. Audit this if the bottle is the critical item.
Tier 2: the closure and liner supplier. Owns the PTFE-faced liner, the silicone backing, the screw cap geometry. Often a separate plant, often not audited by the bottle vendor.
Tier 3: the distributor or repackager. Owns traceability, incoming inspection, and storage. This is the audit most procurement teams actually perform, and the one that catches the fewest problems.

If you are buying through a distributor, ask which tier they are and get the name of the molder. A distributor who will not name the molder is telling you something. I have walked away from two of those and never regretted it.

Bring the specification you intend to enforce. Not a wish list, a document with tolerances on it. If your spec says “borosilicate 3.3” and nothing else, the audit has nothing to measure against. The Reagent Bottle range on this site gives you a starting baseline for dimensions and materials, but the audit needs your numbers, not the catalog’s.

Glass: composition, annealing, and the checks that expose a bad lehr

Confirm the glass type by test, not by label

Borosilicate 3.3 means a coefficient of linear expansion of 3.3 × 10⁻⁶ K⁻¹. That number is what lets the bottle take an autoclave cycle at 121 °C without a stress crack. Ask the vendor for the incoming glass certificate from the tubing or vial supplier, then ask to see the refractive index or expansion test they run on incoming lots. A molder that only keeps the supplier’s certificate is trusting, not verifying.

I ask for the hydrolytic resistance result under ISO 719 or ISO 720, depending on the glass class. Type I borosilicate should land in the hydrolytic class 1 range. If the vendor cannot produce a test report and only cites the raw material grade, treat the claim as unverified. The failure mode here is slow: sodium leaches from the glass into your reagent over months, pH drifts, and nobody connects it to the bottle until an assay fails.

Annealing is where cheap bottles hide their defects

Every molded bottle carries residual stress from cooling. The lehr exists to relieve it. A lehr running too fast, or with a blocked burner, produces bottles that pass visual inspection and fail in service. Ask to see the lehr temperature profile and the last month of recorded data. Look for a stable curve, not a sawtooth.

The field check is a polariscope. Two crossed polarizing films and a light source will show stress bands as colored fringes. I carry a small hand-held unit. A properly annealed bottle shows uniform, faint color. A badly annealed one shows bright bands concentrated at the shoulder and base, exactly where cracks start. This costs about sixty dollars and has caught more problems than any document review I have done.

Dimensional checks that matter

Run these on a sample of at least 30 bottles per lot from at least three mold cavities:

Neck finish diameter. For a GL45 bottle the screw thread must mate with the cap to a defined torque. Measure the thread major diameter and compare to the drawing. A drift of 0.3 mm is enough to cause cap wobble and a leak at the liner.
Body diameter and height. Relevant for autosampler carousels and for rack fit. A 2 mm height variation across a lot will jam a gravity-fed rack.
Wall thickness. Use an ultrasonic gauge at four points around the circumference at mid-height. Variation above 15% between the thin and thick side predicts a stress crack at the base.
Base flatness. A bottle that rocks on a bench will rock on a balance pan. Check with a feeler gauge; anything over 0.2 mm of rock is a reject.

Ask to see the vendor’s own incoming and in-process inspection records for these. A molder that measures only the neck finish is not measuring the problem areas.

Closures and liners: where the leak actually starts

Most reagent bottle failures in service are closure failures. The glass is fine; the cap leaks, the liner creeps, or the PTFE face separates from the silicone backing after a few autoclave cycles. Audit the closure supply chain as carefully as the glass.

Liner construction

A standard reagent bottle liner is a PTFE face bonded to a silicone or EPDM backing. The PTFE face gives chemical inertness; the backing gives the compressibility that seals. Ask for the liner thickness and the bond method. A liner under 1.5 mm total thickness has little compression reserve; one over 3 mm can bind in the cap and prevent full seating.

The failure I see most often is delamination after autoclaving. The bond between PTFE and silicone fails, the PTFE face curls, and the seal is gone. Ask the vendor for autoclave cycle data: how many cycles at 121 °C for 20 minutes before the liner shows separation. Fewer than ten is a problem for a bottle that gets reused.

Torque and the cap seal

Cap torque is a specification, not a feel. For a GL45 cap with a standard liner, the applied torque that gives a reliable seal without over-compressing the liner is typically in the 1.0 to 1.5 N·m range. Ask the vendor whether they specify a torque and whether they check it with a torque tester on the filling line. If the answer is “the operator tightens by hand,” you have no control over seal integrity across lots.

Over-torque is its own failure mode. It compresses the liner past its elastic limit, the liner takes a set, and on the second opening the seal is compromised. I have seen a filling line running at 2.5 N·m because the caps “felt loose,” and the leak rate on the leak tester went up, not down.

GPI and thread standards

The Glass Packaging Institute (GPI) publishes the thread finish specifications that most reagent bottle caps are designed to. A GL45 finish is a defined thread profile; a cap made to a different profile will thread on but will not seal at the specified torque. Ask the vendor to confirm the finish against the GPI drawing number and to provide the cap drawing for comparison. Mismatched finish and cap is a common source of “it threads on but weeps” complaints, and it is invisible until you put liquid in it.

Traceability and the lot number chain

An audit question that separates serious vendors from the rest: “Show me the lot number on a bottle in your warehouse and walk me back to the mold cavity and the glass lot.” A vendor who can do this in under ten minutes has a system. One who cannot has a hope.

Mold cavity identification. Every cavity should be identifiable, either by a cavity number molded into the base or by a documented mapping from lot to cavity set. Without it, a defect that comes from one cavity cannot be contained.
Glass lot linkage. The incoming glass lot must be recorded against the production lot. If a raw material problem surfaces later, you need to know which finished bottles are affected.
Retention samples. Ask how long the vendor keeps a retention sample per lot. I expect at least 12 months for reagent bottles, longer for anything used in regulated work.
Records retention. The audit should confirm records are kept for the period your quality system requires, typically 5 years for pharmaceutical customers.

The Thermo Fisher chromatography and sample handling documentation set is a useful benchmark for what a complete traceability package looks like. If your vendor’s package is thinner than that, ask why.

Cleanliness, contamination, and the wash line

Reagent bottles are often sold “ready to use,” which means somebody washed them. Audit the wash line even if the vendor says the bottles are for general use. The wash line is where trace contamination gets introduced or removed.

What to look for on the wash line

Final rinse water quality. Ask for the conductivity and TOC of the final rinse. Type II water at <1 µS/cm is a reasonable floor for general reagent bottles; anything destined for trace analysis needs better.
Drying. Residual water in a capped bottle is a contamination source. Ask how bottles are dried and how dryness is verified. A 105 °C oven dry followed by a cool-down is standard; a forced-air dry without a temperature record is not.
Handling after wash. Are bottles capped immediately, or do they sit open? An open bottle in a wash room picks up whatever is in the air.
Extractable testing. For bottles used with HPLC-grade solvents, ask for the extractables data. Total extractables by gravimetric method after a defined rinse is the minimum. The Restek technical resources library has method guidance for extractables testing if you need to specify it.

I once found a vendor washing bottles in the same room where they washed greasy machinery parts. The audit caught it because the wash line was visible from the loading dock. Documents would not have caught it.

Packaging, storage, and the receiving inspection you should demand

The audit does not end at the plant gate. How the vendor packs and stores determines whether the bottles arrive in the condition they left in.

Packaging checks

Carton compression strength. Ask for the box crush test result. A carton that crushes under stacking load transfers that load to the bottles.
Dividers and cushioning. Bottles should not touch each other. Ask to see a packed carton opened on the dock.
Pallet pattern. Overhang above 25 mm on a pallet invites corner damage in transit.
Closure protection. Caps should be tight enough not to loosen in transit but not so tight the liner takes a set. If the vendor ships caps loose and expects you to torque them, that is a specification gap.

Storage conditions at the vendor

Ask where the finished goods sit and for how long. A bottle stored in a hot warehouse for eight months before shipping has a different service life than one shipped from a climate-controlled stock. Ask for the storage temperature range and the actual recorded maximum over the past year. If the vendor cannot produce it, the storage claim is unverified.

Receiving inspection plan

Write a receiving inspection that matches the risk. For a general reagent bottle, an AQL of 1.0 on major defects and 2.5 on minor defects is a reasonable starting point. For bottles going into regulated analytical work, tighten to 0.65 on major. Define major and minor explicitly: a cracked bottle is major, a scuffed label area is minor, a dimensional drift beyond the drawing is major.

The inspection should include a leak test on a sample. Fill with water, cap at the specified torque, invert for 30 minutes, and check for weeping at the liner. This is a five-minute test that catches closure problems before they reach the lab.

Comparison checklist: what to score and how to weight it

I score vendors on a 100-point scale across five categories. The weights below are for a reagent bottle used in general lab work; shift weight toward traceability and cleanliness for regulated use.

Glass quality and annealing (25 points). Glass certificate present, lehr data stable, polariscope check clean, dimensional capability study within tolerance. A vendor scoring below 15 here is a risk regardless of price.
Closure and liner control (20 points). Liner autoclave data, torque specification and tester on line, GPI finish confirmation, incoming liner inspection. Below 12 is a leak waiting to happen.
Traceability (20 points). Cavity identification, lot linkage, retention samples, records retention. Below 12 means you cannot contain a defect.
Cleanliness and contamination control (15 points). Final rinse water data, drying records, handling after wash, extractables data where required. Below 9 is a contamination source.
Packaging and storage (10 points). Carton strength, pallet pattern, storage records, receiving inspection support. Below 6 is transit damage you will pay for.
Commercial and responsiveness (10 points). Lead time reliability, change notification, corrective action closure time. Below 6 is a supplier who will surprise you.

A vendor scoring 75 or above goes on the approved list. 60 to 74 is conditional, with a corrective action plan and a re-audit in six months. Below 60, I do not place the order, whatever the unit price says. The U.S. EPA SW-846 test methods documentation is a useful reference for the level of method control that a bottle used in regulated work has to support.

Troubleshooting: what the audit findings predict in service

The point of the audit is to predict failures before they happen. Here is how audit findings map to the symptoms you will see in the lab, with the fix.

Symptom: caps weep at the liner after autoclaving. Audit finding: liner delamination data missing or fewer than ten cycles. Fix: specify a liner with documented autoclave cycles and a bonded construction that survives the count you need.
Symptom: bottles crack at the base in service. Audit finding: polariscope shows stress bands, lehr data sawtooth. Fix: reject the lot, require lehr records with each shipment, re-audit the annealing line.
Symptom: reagent pH drifts over months. Audit finding: no hydrolytic resistance test, glass grade unverified. Fix: require ISO 719 or 720 test data per glass lot.
Symptom: bottles jam in a rack or carousel. Audit finding: height variation above 2 mm across the lot, no capability study. Fix: add height to the drawing tolerance and measure 30 pieces per lot at receiving.
Symptom: trace contamination in HPLC blanks. Audit finding: final rinse water unverified, bottles capped late. Fix: specify rinse water quality and require immediate capping; run an extractables check. If the contamination is at the injection end, check the HPLC autosampler vial supply as well, because the bottle is not always the source.
Symptom: lot-to-lot variation in leak rate. Audit finding: no torque specification, hand tightening. Fix: specify torque in the purchase order and require a torque tester on the filling line.
Symptom: breakage in transit. Audit finding: carton crush strength below stacking load, pallet overhang. Fix: specify carton test data and a pallet pattern in the contract.

What to put in the purchase order after the audit

The audit findings are worthless unless they end up in the contract. I attach a specification sheet to every reagent bottle PO with these fields filled in:

Glass type and expansion coefficient, with the test method for verification.
Dimensional tolerances: neck finish diameter, body diameter, height, wall thickness, base flatness.
Closure type, liner construction, liner thickness, and autoclave cycle count.
Cap torque at filling, in N·m, with the tolerance band.
GPI finish reference number.
Cleanliness: final rinse water conductivity and TOC, drying temperature and time.
Traceability: cavity identification method, retention sample period, records retention period.
Packaging: carton crush test minimum, pallet pattern, overhang limit.
Receiving inspection AQL levels for major and minor defects.
Change notification: the vendor must notify you before changing glass supplier, liner supplier, or mold.

The change notification clause is the one people forget. A vendor who switches liner suppliers without telling you has invalidated your audit and your autoclave validation in one move. Put it in writing.

The Agilent chromatography consumables and columns documentation is a good model for how a consumable specification sheet should read: every claim tied to a test method and a number. If your reagent bottle spec does not read like that, it is not a specification, it is a description.

Where the distributor audit differs

If you buy through a distributor, the audit shifts. You are not inspecting a mold; you are inspecting a warehouse and a receiving process. The questions change:

Does the distributor run incoming inspection on the bottles it receives, and to what AQL?
Can it produce the molder’s lot number for any carton in stock, on demand?
What are the storage conditions, and are they recorded?
Does it hold retention samples, or does it rely entirely on the molder?
What is the process for a field complaint, and how fast does it close?

A distributor that passes these is worth a premium. One that cannot name the molder and has no incoming inspection is a pass-through, and you are paying for a phone number. I have used both. The pass-through is fine for a non-critical bottle; it is not fine for anything that touches a validated method.

If your work spans both reagent bottles and the vials that feed the instrument, the same audit discipline applies at the Sample Vial end of the supply chain, and the Chromatography Products range gives you a sense of how the categories overlap in a real lab order. Audit the whole chain, not one item.

The audit is a filter, not a formality

Run the audit against a written specification, score it on the weights above, and put the passing findings into the purchase order. A vendor who scores 75 or above, can walk you from a warehouse carton back to a mold cavity in ten minutes, and has a polariscope-clean bottle and a documented liner autoclave count is a vendor you can build a method on. One who cannot is a vendor you will be re-auditing after the first leak complaint. The checklist is the cheap part. The failure in service is the expensive part, and it always arrives at the worst possible time.

References & further reading

Related resources from our lab consumables network:

External references and standards: