This guide is for the procurement engineer or quality manager who signs the purchase order for industrial adsorbent and who will, one day, have to answer a question from their own operations team: was this shipment actually what we ordered, and who checked? Pre-shipment inspection (PSI) is the most cost-effective quality control point in an imported adsorbent supply chain, because it is the only moment when an independent party can open the cargo, take a representative sample, run the tests and issue a finding while the goods are still at the loading port and can be refused, replaced or reworked at the seller's cost. Once a container has landed, sailed and been unloaded, the same quality question is worth roughly ten times more to answer, and often nobody answers it at all.
The problem with PSI on adsorbents is not that the tests are difficult. It is that most PSI scopes are written too loosely to be enforced. A scope that says "check quality and quantity" generates a rubber-stamped report saying exactly that, and the buyer ends up with a document proving a person was present, not that the material conforms.
What a pre-shipment inspection buys you, and what it does not
A PSI is a snapshot taken at one point in the production and packing chain. It buys you three specific things: confirmation that the shipment matches the purchase order at the level of grade, mesh size, packaging and quantity; an independent laboratory result on the characteristics you nominated, produced by a party with no commercial interest in the outcome; and most importantly leverage, because a documented discrepancy at the loading port is a lever you can use to demand replacement, credit or rework before the cargo is loaded onto the vessel.
What PSI does not buy you is any guarantee about performance in your plant. An inspector can confirm that a 13X lot has 95% relative crystallinity by X-ray diffraction and that the moisture is inside specification on the day of loading. That inspector cannot tell you whether the lot will still meet those numbers after six weeks at sea in a tropical climate, whether your loading procedure will break the pellets, or whether the bed you charge will give you the working capacity the datasheet promises.
Better to write a tight, executable scope and a separate retained-sample testing protocol, and to be explicit about which document answers which question. A useful companion reading is our molecular sieve quality verification checklist, which covers the test methods you would ask an incoming laboratory to run.
Why adsorbent shipments fail inspection more often than generic cargo
Adsorbents are hygroscopic, friable, graded by mesh rather than by weight, and sold in forms ranging from 1.6 mm extrudate to 200 µm powder. Every one of those properties creates a distinct inspection failure mode that does not exist for, say, a pallet of fasteners.
Moisture is the dominant risk. Molecular sieve and activated alumina are manufactured for the purpose of removing water from other people's streams, which means they hold water aggressively from the atmosphere. A 40 kg bag of 13X left open on a quay in a humid season can pick up enough moisture to push a 2% water specification to 4% or worse before it is even weighed. The classic discovery at inspection is not fraud, it is exposure: bags found with the outer layer of the stack damp, pallet wrap torn, or a container door opened for the tally and left open for two hours while the inspector went to lunch.
Mechanical degradation is the second risk. Formed beads and extrudate lose strength when dropped, when over-vibrated during stuffing, and when transported with a load that shifts. A container stuffed to the ceiling with bags that were not dunnaged or blocked adequately will arrive with a column of crushed product in the forward third. That damage is invisible from the door of a closed container Our guide to bulk crush strength against attrition index sets out which test predicts which failure.
Third, the count-versus-weight reconciliation is genuinely difficult. A 20-foot container of 3A 8x12 mesh nominally holds 450 to 500 bags of 25 kg, and the bulk density is itself one of the characteristics you are inspecting.
Six scope clauses your purchase order should contain
A workable PSI scope has six components.
1. Identification and traceability
Specify the exact product designation, grade and mesh size to be found on every bag: for example "Type 13X zeolite, 8x12 mesh (2.5 to 5.0 mm), supplied in 25 kg multilayer laminated aluminium-plastic bags". Require the inspector to record the production lot number from a sample of bags, to confirm the lot number is uniform across the shipment, and to photograph at least one legible lot label per layer of the stack.
2. Quantity and packaging
Require a full bag count by container, net weight verification on a stated minimum number of randomly selected bags, and a count of any damaged or leaking bags. Packaging specifications should be explicit: bag construction, whether a desiccant sachet is included and at what placement, pallet or not, and maximum stacking height.
3. The characteristic list with acceptance limits
This is the core of the scope and the part that is usually missing. Every characteristic must carry a method, a unit and a limit, and the method must be specific enough that two laboratories would run the same procedure. "Moisture: max 3%" is not sufficient. "Water content by Karl Fischer titration in accordance with ASTM E203, maximum 3.0% by mass" is sufficient.
4. The sampling plan
State the lot definition, the sample size, the selection method and the sample reduction method. If you leave this to the inspector's discretion, different inspectors will sample differently and the results will not be comparable between shipments.
5. Environmental and container conditions
Require the inspector to record the container's condition, whether it was dry and free of odour, whether the doors were open during stuffing, the humidity and temperature at the point of loading, and whether the bags were protected from rain.
6. Reporting and discrepancy handling
Fix the reporting deadline (typically before the bill of lading is issued), require photographic evidence, specify the report format, and state clearly what the inspector should do when a result falls outside the acceptance limits: withhold the clean report, issue a discrepancy notice, and notify the buyer before the vessel sails. A scope that does not say what happens on a discrepancy guarantees that you find out by email a week later.
Building the sampling plan on a statistical foundation
Two distinct statistical problems get confused in most PSI scopes, and separating them cleanly pays off. The first is selecting a sample from the lot, a probability sampling problem where the reference is ASTM E105, the Standard Guide for Probability Sampling of Materials. The second is deciding whether the sample passes: for pass/fail characteristics such as a broken seal or a foreign object, that is an acceptance sampling problem under ISO 2859-1, Sampling procedures for inspection by attributes; for characteristics measured continuously, such as moisture or bulk density, it is ISO 3951-1, sampling procedures for inspection by variables.
ISO 2859-1 gives you a plan where sample size and accept/reject numbers are chosen from a table indexed by AQL and lot size. At AQL 2.5 under normal inspection at General Inspection Level II, a lot between 501 and 1200 units requires a sample of 50 and allows up to 3 defectives; above 3000 units the sample rises to 125 and allows up to 5.
Understanding what an AQL actually promises is the difference between a useful inspection and a false sense of security. AQL is not a target quality level and it is not a guarantee. It is the quality level that, when the process average is at or below it, a plan will accept the large majority of the time. A practical and often misunderstood consequence: if your inspector takes 32 bags and finds zero defective bags, you have not proven the lot is defect-free. Under the usual 95% confidence interpretation, the upper bound on the lot defect rate is approximately three divided by the sample size, about 9.4% at n = 32 and about 2.4% at n = 125.
For bagged adsorbent there is a further wrinkle. The unit of the lot is the bag, but the characteristic you care about is the material inside the bag. You therefore need a two-stage plan: select bags from the shipment according to the attribute sampling plan, then reduce the contents of the selected bags to a laboratory sample. That reduction step is where most measurement error is introduced, and it must be specified. The riffle-splitting and rotary-division practice described in ASTM E105 is the appropriate reference, and we have covered the mechanical side in detail in our piece on pellet diameter against bed pressure drop. A scope that says "the inspector will take a sample" without naming the reduction method has left the most consequential step to whoever happens to be carrying a scoop.
Choosing an AQL for each characteristic
The table below is a defensible starting framework for bagged molecular sieve and activated alumina shipments, using the critical/major/minor classification convention widespread in industrial purchasing.
| Characteristic | Classification | Typical AQL (ISO 2859-1, normal, Level II) | Why |
|---|---|---|---|
| Wrong grade or type mixed into the lot | Critical | 0 (zero acceptance) | A single bag of the wrong zeolite type defeats the bed and cannot be screened out downstream. |
| Broken or leaking bag seal | Critical | 0.65 | An open bag is a direct moisture ingress path and invalidates the moisture result for the whole lot. |
| Foreign material, contamination, odour | Critical | 0.65 | Contamination of a molecular sieve charge is unrecoverable and can shut down a PSA unit. |
| Bagged, damaged, or short weight bag | Major | 2.5 | Causes a net-weight shortfall and indicates the packaging line was not under control. |
| Label missing or illegible lot marking | Major | 2.5 | Breaks traceability and makes the retained sample useless for arbitration. |
| Minor cosmetic marking or print defect | Minor | 6.5 | Does not affect performance or storage; only needs recording for trend purposes. |
Note the zero-acceptance rows. For critical characteristics you do not want the standard's tightened inspection levels; you want a plan with an accept number of zero, meaning any single non-conforming bag rejects the lot. The ISO 2859-1 machinery handles this naturally at AQL 0.65 with an accept number of zero for the smaller lot sizes, and for larger lots the plan gives a small accept number that you should override to zero in the purchase order if you mean it.
The characteristics to put on the certificate, and how to specify each test
Beyond the attribute characteristics inspected visually, a set of measured characteristics has to appear on the certificate of analysis and, where they are critical enough to justify the cost, on the PSI report itself. The point of specifying a method rather than a property is that adsorbent lots vary in the lab as much as in the field: a moisture figure is meaningless without knowing whether it was measured by Karl Fischer volumetric titration or by loss on ignition at 900 degrees, and those two numbers for a 13X lot can differ by a factor of two because the loss on ignition figure includes water released from the crystal structure as well as adsorbed moisture.
| Characteristic | Specified method | Indicative limit (typical industry range, set your own) |
|---|---|---|
| Water content | Karl Fischer, ASTM E203 (volumetric) or coulometric titration | 3A/4A/5A: 8 to 12% by mass as packed; 13X: 8 to 11%. Always state the basis. |
| Apparent / packed density | ASTM D4180 (vibratory packing density of formed catalyst particles) or ASTM D2854 (packed density of granular material) | 3A/4A 8x12 mesh: 0.60 to 0.68 kg/L; 5A/13X: 0.62 to 0.70 kg/L. Governs vessel sizing. |
| Particle size / mesh | ASTM E11 woven wire sieve cloth specification; analysis per ASTM D4513 for catalytic powders, or ISO 3310 for test sieves | 8x12 mesh nominally 2.5 to 5.0 mm; fines below the specified cut typically below 2% by mass. ASTM D4513 is written for 20 to 420 um catalytic powder, so for formed bead grades specify the sieve series and cut points directly. |
| Crush strength | ASTM D4179 (single pellet crush strength of formed catalysts) or ASTM D7084 (bulk crush strength) | Bead: 30 to 50 N average single pellet; extrudate: 90 to 150 N average radial crush. Grade-dependent. |
| Attrition / dust generation | ASTM D4058 (attrition and abrasion of catalysts and catalyst carriers) | Below 5% mass loss on the specified test cycle for commercial formed grades. |
| Adsorption capacity | Static water capacity per the grade-specific internal method, reported with the test conditions stated | 3A: 19 to 22 wt% at 25 degrees C; 5A: 21 to 24 wt%; 13X: 23 to 27 wt% at 25 degrees C under 10 mmHg. |
| Relative crystallinity | ASTM D5357 (zeolite sodium A relative crystallinity by X-ray diffraction) | Above 90% for 3A and 4A; above 85% for 13X, subject to the defect basis used. |
| Na2O / binder residue | XRF or ICP per the grade-specific method | Na2O below 2.0% by mass on the dry basis for most 3A/4A/5A grades. |
Two cautions on this table. First, the numeric limits are drawn from commonly encountered commercial grades and are offered as a starting point, not as our factory specification; your own purchase order is the contract and should state the limits explicitly rather than referring back to a supplier datasheet. Second, ASTM D2854 was written for granular activated carbon, not for molecular sieve, so if you invoke it for zeolite you are invoking it as an adapted practice and should say so, or specify ASTM D4180 instead, which was developed for formed catalyst particles and catalyst carriers and is closer to the material at hand.
Writing pass/fail rules that survive a borderline result
They are about a result that sits a little outside the limit, usually because the measurement carries real uncertainty and the limit was written with no margin. Three rules prevent that argument.
First, write the limit and the measurement uncertainty separately, and state how they combine. If your specification says water content below 3.0% by mass measured by ASTM E203, and the Karl Fischer result comes back at 3.08%, you do not have an objective argument until you have agreed the method's repeatability. ISO/IEC 17025:2017 clause 7.7 requires a testing laboratory to estimate and report measurement uncertainty for quantitative results, and clause 6.6 requires the equipment used to be calibrated and traceable. So the certificate of analysis from an accredited laboratory should carry an uncertainty statement, and your purchase order should say whether the limit applies to the raw result or to the result plus uncertainty. "3.0% max, excluding measurement uncertainty" and "3.0% max including uncertainty at k = 2" are the same specification on paper and materially different in practice.
Second, decide in advance what happens on a borderline result, because the inspector cannot decide it for you at the quay. The three outcomes are: reject and notify the buyer before loading, accept under a concession recorded in the report, or accept subject to a retained-sample test with a defined re-test protocol.
Third, define the lot and the sampling point in the same sentence. "Lot" should mean one production batch from one manufacturing campaign, not one container and not one purchase order. If your order covers 40 containers of 13X drawn from four production campaigns, that is four lots, and the inspection plan has to say whether each campaign is inspected separately. This is the single most common source of confusion in adsorbent PSI work, because the manufacturer is often shipping from several bags or several intermediate warehouses and the inspector sees one consolidated container load.
Who should perform the inspection, and how to qualify the inspector
There are three categories of PSI provider and the differences matter. A factory self-inspection is free and fast, and it is not independent; it works as a pre-shipment release check under the manufacturer's own ISO 9001:2015 clause 8.6 controls, but it carries no evidentiary weight when a dispute arises. A third-party inspection company (SGS, Bureau Veritas, Intertek, TUV and comparable firms) is the default for most cross-border adsorbent trades, because the buyer does not have to be in the country, the report is a commercial instrument usable in a dispute, and the firm already has a laboratory network in the origin region.
When you use a third party, qualify them against the work rather than against the brand. Is the laboratory that will run the chemical and physical tests accredited to ISO/IEC 17025:2017, and is the accreditation scope wide enough to cover the specific tests in your scope, since a general scope listing does not cover every method. Does the inspector have demonstrable experience with bulk solid adsorbents rather than only containers and cartons. Does the firm have a documented sampling procedure for bagged solids, and will they commit to the ASTM E105 and ISO 2859-1 approach in writing rather than taking a scoop. And can they open a container at the loading point and take a sample without disrupting the stuffing plan.
The retained sample: the cheapest insurance in the whole chain
The retained sample is a small, sealed quantity of the actual shipment material, taken at the same moment as the PSI sample, kept by a party both buyer and seller trust, and held for a defined period. It is the only physical evidence that survives if the argument about quality happens six months later, after the shipment has been consumed, and it is what allows a laboratory in the buyer's own country to arbitrate a disagreement with the origin country laboratory.
The practical protocol is straightforward. Take a quantity sufficient for two to three complete test sequences, typically 2 to 5 kg for bagged molecular sieve, from the same reduction as the PSI sample so it is representative. Seal it in a moisture-tight container, ideally with a desiccant sachet inside the closure but outside the product, and label it with the lot number, the date, the seal number and the sampler. Define a retention period matching the time your plant takes to consume the lot plus a margin for a dispute to escalate, and 12 to 18 months is the right order of magnitude for a process that consumes a lot over several months.
The value of a retained sample is disproportionate to its cost. A 5 kg sample that costs a few tens of dollars at the port can settle a dispute worth tens of thousands of dollars in replacement material, vessel downtime and air freight. It also serves a less obvious purpose: it lets you validate your own incoming test method against the seller's certificate over time. If your incoming screen and the origin certificate agree on every lot, your method is sound. If they diverge systematically on one characteristic, you have found either a method difference or a supply problem, and you would never have found it without a physical sample to test. Our piece on Karl Fischer against coulometric against gravimetric moisture measurement covers the method differences that most often explain a gap between two certificates.
Reading the report: what to check before you accept it
When the report arrives, five checks determine whether it is worth anything to you. First, does it name the actual container numbers inspected and the actual lot numbers seen, or does it describe the shipment generically? A report saying "one 20-foot container of molecular sieve was inspected" without a container number cannot be tied to your cargo. Third, are all the characteristics you nominated present with a value, a method reference and a unit, and is anything reported as "not tested"? Fourth, does the report state the ambient conditions during sampling and address the container's condition and seal status? Fifth, and most often overlooked, is there a discrepancy section, and if so, was the clean report issued or withheld as your scope required?
Common mistakes that turn a PSI into a wasted invoice
The most common error is writing the scope after the shipment is already loaded, when it is too late to be executed. A PSI scope has to be agreed at the proforma invoice stage, because the inspection must be booked and the inspector must be at the loading point on a specific day.
The fourth is failing to reconcile the inspection results against the certificate of analysis for the same lot. If the certificate reports a moisture content of 9.2% and the inspector's independent sample at the container reports 11.4%, that gap is real information about moisture pickup during handling, and it is available at no extra cost if you specified the comparison. Our guide to the QC checks to run before loading vessels sets out the same reconciliation from the buyer's side of the loading gate.
Frequently asked questions
What standard actually governs sampling for a molecular sieve pre-shipment inspection?
No single ASTM standard covers the whole activity, and it is worth knowing which reference does what. ASTM E105, the Standard Guide for Probability Sampling of Materials, governs how a sample is selected from a bulk material so that it represents the lot. ISO 2859-1 governs the acceptance decision for pass/fail characteristics, indexed by AQL and lot size, and ISO 3951-1 covers continuous characteristics measured by variables. A frequent misconception is that ASTM D5357 is a sampling standard; it is in fact the method for determining the relative crystallinity of zeolite sodium A by X-ray diffraction, which is a measurement you may include in the scope but which has nothing to do with how the sample is drawn.
How large should the sample be for a 20-foot container of bagged molecular sieve?
For a shipment of roughly 450 to 500 bags, an attribute inspection at AQL 2.5 under normal inspection at General Inspection Level II requires a sample of 50 bags, while AQL 0.65 requires 32 bags and AQL 6.5 requires 13. The sample must be taken from the bags selected under the plan, not from a convenient corner of the container, and it must be reduced to the laboratory sample using a riffle splitter or rotary divider rather than by hand, because that reduction step is where unrepresentative sampling is most often introduced.
What does an AQL of 0.65 actually guarantee?
Less than buyers expect. An AQL is a statistical risk threshold, not a quality guarantee and not a target. The useful interpretation is a confidence bound: if your inspector examines 32 bags and finds zero defective bags, the 95% upper confidence bound on the lot defect rate is roughly 9.4%, not zero. That is why sample size, and the classification of the characteristic, matter more than the AQL number itself, and why critical characteristics such as mixed zeolite type should carry a zero accept number rather than a small AQL.
Should the inspection happen before or after the container is stuffed and sealed?
After stuffing and sealing, without exception, and this is the point most often lost. The correct sequence is production, packing, stuffing, closure, inspection at the container, sampling, documented resealing with both the sampling seal number and the final shipping seal number recorded in the report, then departure.
Who should perform the inspection and pay for it?
The commercially important question is whether the inspection cost appears as a line on the commercial invoice. If the letter of credit is on FOB terms and covers the invoice value, adding the inspection fee as a separate invoice line creates an amount that does not match the credit, which is a discrepancy. Agree in the proforma invoice that the fee is either inside the FOB value or settled outside the documentary credit, and have that written into the instructions to the inspection company.
Can the buyer's own laboratory test the sample instead of using a third-party inspector?
You can test in your own laboratory, but you are then the only party checking the seller, which removes the independence that makes the result defensible in a dispute. A more common and effective arrangement is to have the independent party take the sample at the loading point and perform the physical and visual checks at the port, then send the sealed laboratory sample to your own accredited laboratory for the chemical and performance testing. This keeps the sampling and chain of custody independent while putting the measurements under your control, and it produces a second dataset that can be compared against the seller's certificate of analysis for the same lot.
What should happen if the inspection fails?
The scope must say what happens, because the inspector cannot decide it at the quay and will default to shipping the container while promising results later. The correct instruction is to withhold the clean report, issue a discrepancy notice to the buyer, and notify the buyer before the vessel sails. Three follow-on options are worth pre-agreeing in the purchase order: replacement of the non-conforming material, a price concession with the lot accepted, or acceptance subject to a defined re-test protocol on the retained sample with a stated number of retests and a stated expiry date for the concession.
How do I handle a shipment drawn from several production batches?
Define the lot explicitly in the purchase order and align the inspection plan to it. A lot should be one production batch from one manufacturing campaign, not one container and not one purchase order. If an order of 40 containers draws on four campaigns, there are four lots, and you should decide in advance whether each campaign is inspected separately or whether the campaign is the sampling unit and the container is only the shipping unit. This is a frequent source of confusion because the manufacturer frequently consolidates several bags or intermediate warehouses into one container load, so the inspector sees a mixed shipment that the certificate of analysis describes as a single lot.
Does the PSI replace our own incoming inspection at the plant?
No, and it should not be treated that way. The PSI answers the question of what condition the material was in when it was loaded, while your receiving inspection answers what condition it is in after transit, which is a different risk profile dominated by handling damage, bag breakage and moisture ingress through compromised seals. Specify both to the same standard and the comparison between them becomes informative: a systematic rise in moisture between the port certificate and your plant's incoming test points at handling, and a rise in fines and dust points at mechanical damage in transit rather than at the factory.
Why do the moisture figures on the certificate and the inspection report often differ?
Three causes account for most of the gap. First, the method: loss on ignition at 900 degrees captures not only adsorbed moisture but also water driven off the crystal structure and binder residue, so it routinely reads higher than a Karl Fischer result on the same sample. Second, the point of sampling: a sample drawn from the top of the stack has had more exposure to ambient humidity than one drawn from the bottom, and on a humid day at an open quay that difference can be significant. All three are reasons to specify the method and the sampling point precisely, and to reconcile the two numbers rather than to treat one of them as automatically correct.
Is a factory certificate of analysis enough on its own?
It is necessary but not sufficient. A factory certificate is a statement by the seller about its own product, produced under the manufacturer's ISO 9001:2015 clause 8.6 release controls, and it is entirely appropriate as the document that accompanies the shipment. What it cannot do is demonstrate the condition of the material in the specific container you are paying for, at the specific moment it was loaded. That is the gap the pre-shipment inspection closes, and the two documents together are what a serious importer uses: the certificate establishes what the factory believes the material is, and the independent inspection establishes what actually went into the container. Neither substitutes for the other, and the retained sample is what allows the question to be settled if the two ever disagree.
Molecular Sieve 3A / 4A / 5A / 13X
Type 3A, 4A, 5A and 13X zeolites supplied with a certificate of analysis and MSDS on every shipment, a documented retained sample from each lot, and lot-level traceability from the production campaign through to the container seal number.
View Product Details →Next steps
The most useful thing you can do this week is take the last adsorbent purchase order you placed and check whether it contains a PSI scope at all, and if it does, whether that scope names a test method and an acceptance limit for every characteristic you actually care about. If you want a second opinion, send us your current specification and the grade and mesh size you buy, and we will mark up which clauses are unexecutable as written, which are missing the method or the limit, and which characteristics are worth adding for your particular duty rather than copied from a generic template.
We supply 3A, 4A, 5A and 13X molecular sieve, activated alumina, alumina powder and pseudo-boehmite with a certificate of analysis and MSDS on every shipment, lot-level traceability, retained samples from each production campaign, MOQ 1 MT, lead time 15 to 25 days and free samples on request. For the trade-finance side of the same shipment, our guide to letter of credit documentation discrepancies covers the document set that has to accompany the inspection certificate.
Send us your PSI scope or your current purchase specification. We will review it against ISO 2859-1, ASTM E105 and your own process duty, and tell you which clauses a third-party inspector can actually execute at the loading point.