- gzk@hnlcalloy.com
- +8615737205888
- Qugou Town, Anyang City, Henan Province
A certificate of analysis on which every value falls inside the standard range can still be the piece of paper that ruins your next heat.
The reason is simple enough: the certificate answers “does this conform to some general standard,” not “does this suit your furnace.” A standard sets upper and lower bounds, and those bounds are drawn at the lowest common denominator of an entire industry. Your steel grade, your addition practice, your slag system — none of them were considered when the limits were written.
This guide gives you a six-step method for reading a certificate, explains what each of eleven line items actually means, and provides a decision tree you can work through before releasing a lot. By the end you should be able to do three things: see where your money goes beyond the headline element, spot the risk that is absent from the document rather than present on it, and know what to compare when two laboratories disagree.
Start with something that is routinely conflated — which kind of document you are actually holding.
Three documents turn up on a purchasing desk, their names used interchangeably, though they carry very different weight once a dispute starts.
|
Document |
Common label |
Issued by |
Weight when a dispute arises |
|
Mill analysis sheet |
COA / Certificate of Analysis |
The producer’s own laboratory |
A statement by one party. Traceable to a heat number gives it traceability value, but it is issued by an interested party |
|
Mill test report |
MTR |
The producer (fuller format, often includes process data) |
Also a party statement, but with more content and therefore more to cross-check against |
|
Independent test report |
Third-party inspection report |
A laboratory independent of buyer and seller |
Independently sampled and tested — the strongest evidence of the three |
The difference is not how much detail each contains. It is whether sampling was carried out independently of interested parties.
One misunderstanding shows up repeatedly in receiving departments: that “a third-party report is attached” means “the shipment was independently inspected.” What matters is whether the independent laboratory tested a sample it drew itself or a sample the supplier sent to it. The latter certifies the piece of material that arrived at their door, not the material inside your containers. The gap between those two sentences is where most arguments on the claims table begin.
The sampling layer — how the top of a cast can read differently from the bottom — is covered in Ferroalloy Purchasing: Five Risks That Never Appear on the Quotation. This article does not repeat it and moves straight to the numbers.
Test ①: ask one question before you read anything else: “who took the sample this document describes?” The answer decides where it belongs — as a receiving record, or as evidence.
The largest number on the certificate (Si in ferrosilicon, Mn in ferromanganese, Cr in ferrochrome) sets your pricing basis. It also has a property that is easy to miss: it is a mass percentage, not the quantity of element you will recover.
Two practical problems keep it from being usable as money.
Problem one: a high main element does not mean cheap element.
There is only one conversion that matters:
Price per tonne of element = price per tonne of alloy ÷ main element content (as a decimal)
+ any composition adjustment called for by the contract
Its value is that it puts different offers on one scale. As an illustration (purely illustrative — replace the figures with your own quotes): suppose two ferrosilicon offers, one at A per tonne with 75% Si and one at B per tonne with 72% Si. Only once the relationship between A/0.75 and B/0.72 is established does “which is cheaper” mean anything. Usually the lower headline price also carries the lower main element, and under a complete contract part of that gap is recovered through the grade adjustment clause — so ranking offers by headline price is often simply wrong.
The same arithmetic governs grade selection and offer comparison generally; see FeSi75 vs FeSi45: How to Choose. Here it is applied to the act of reading a certificate.
Problem two: content means “contained in it,” not “delivered into the steel.”
A reading of 75% means three quarters of this alloy’s mass is that element. How much of it actually enters the steel, how much goes to slag, how much is oxidised away — none of that is the certificate’s business. It depends on sizing, addition point, temperature at addition, and slag condition.
One product category needs calling out separately: Mg in nodularizers. The Mg figure on a certificate is normally total magnesium, and part of it already exists as MgO. Magnesium oxide does no nodularizing; it is the residue of magnesium that has been oxidised. Two certificates both reading Mg 6% can therefore differ substantially in effective magnesium — the difference depending on how much of it sits as MgO, and whether MgO is even printed depends on whether your specification asked for it.
If you buy nodularizer, the specification should fix total Mg, an MgO ceiling, and the method for calculating metallic magnesium together. Without those, the Mg figure has no agreed authority.
Test ②: after reading this line you should be able to state two numbers — the actual price per tonne of element, and how much of that element is in an available form. Missing either, this step is unfinished.
The impurity column carries the highest information density on the page and gets read least carefully. The usual approach — “check nothing exceeds its ceiling” — misses two things.
First, impurities split by destination, and the two groups call for different responses.
|
Type |
Typical elements |
Where it goes |
What you actually care about |
|
Enters the steel |
C, P, S, some residual elements |
Stays in the product, affects composition |
Does it push against your grade ceiling, does it accumulate |
|
Goes to slag or sits inert |
Part of the oxide content |
Mostly leaves with the slag |
Mainly affects slag volume and chemistry, and recovery |
A single element can appear on both sides. So the useful question is not “how much of this impurity is there” but: will it consume my alloy, will it contaminate my steel, or both?
Second, impurity level is itself a costing question, and it is routinely charged to the wrong department.
Carbon illustrates it clearly. Whether carbon in a manganese alloy is feedstock or a liability depends entirely on what you are melting. In high-manganese steel it displaces part of your recarburiser and is a benefit you were handed for free; in ultra-low-carbon steel it is something you pay to remove, and you pay for that same carbon twice — once to buy it in, once to drive it out. The one number carries opposite signs on two different reports. This is laid out in full in HC / MC / LC Ferromanganese: How to Choose.
Reading certificates then yields a practical conclusion: impurity ceilings should be back-calculated from your steel grade, not copied from a standard.
Permitted contribution from the alloy
= (grade ceiling for the element − level already present before addition
− contributions from other charge materials)
÷ alloy addition rate (converted to consistent units)
The figure this produces is often stricter than the standard’s ceiling and occasionally looser. Following the looser result saves money you should not have spent.
Test ③: for every impurity that enters the steel, you can state “how much headroom my grade has left for it.” If you cannot, you are looking at the column without reading it.
This is the step most often skipped and the one that costs the most when it is.
What appears on a test report is decided by the contract, not by the product. Elements your specification did not ask for are normally not tested for, and there is no obligation to report them. The result is a certificate that looks reassuringly clean: every listed item within limits, everything else blank. The blank space does not mean “none present.” It means “not looked for.”
Steelmaking carries a group of elements usually called residuals or tramp elements — Ti, Pb, As, Sn, Sb, Bi among them, the exact list depending on your grade requirements. They have two properties that make them awkward to buy against:
That combination produces a failure mode that is very hard to attribute: certificates clean, composition within specification, yet certain grades keep failing final inspection — and tracing it back reveals a raw material that has been contributing a trace element for months. The money lost in that failure usually dwarfs whatever switching suppliers appeared to save.
The remedy is not at the reading stage. It is before you order:
Specify mandatory test elements by steel grade, not by alloy family. The alloy list tells you what this product is normally tested for; only the grade list tells you whether those elements do harm when they arrive. Put that list in your enquiry and state plainly that typical levels of unlisted elements are to be declared at quotation. Only then does blank space on the certificate carry meaning.
Test ④: you can name the residual elements your key product lines must control, and their corresponding charge limits. If you cannot, you cannot raise a testing requirement, and you will not see them printed.
What this step produces needs to reach paper next — how to write it into a binding clause is covered in Ferroalloy Purchase Contracts: Seven Points Your Specification Must Fix.
The sloppiest form a size specification takes is 10–50 mm. As written, it constrains nothing.
It fails to answer three questions: how much above the top limit is permitted, how much below the bottom limit, and how that percentage is counted — by mass, or by number of pieces or bags. One phrase, “90% min,” means fines held under 10% when counted by mass, and almost nothing when counted by pieces. Fines are light; piece count and mass share can differ by an order of magnitude.
A size clause that binds contains at least:
|
Item to agree |
Why it fails without it |
|
Target range (e.g. 10–50 mm) |
The foundation, but not sufficient to decide anything alone |
|
Oversize ceiling |
Oversized pieces affect feeding and melting rate, and can jam equipment |
|
Fines ceiling |
Fines are a principal source of oxidation loss, and affect recovery and dusting |
|
Basis for the percentage (normally by mass) |
Unstated, buyer and seller may be counting entirely different things |
|
Sampling and screening method |
Determines whether your re-check can reproduce their number |
The fines line deserves its own mention. Fines carry far more surface area than the same mass of lump, and oxidation loss rises accordingly; for some alloys — silicon-bearing ones especially — a high fines fraction means less element reaching the steel than the certificate’s content implies. Which means failing sizing can invalidate the price per tonne of element you calculated in step 1. Assaying at 75% is one thing; what is absorbed after degradation during storage and transit is another.
Test ⑤: you can point to precisely which item this certificate’s sizing column is missing. A missing oversize or fines ceiling is the most common gap and the one that most easily becomes an argument after arrival.
Moisture is often passed over as a transport detail. It is in fact the one physical figure on the page that changes how much you pay.
The mechanism is straightforward: ferroalloys are priced per tonne, the scale weighs what is there at the time, and that includes water. So one lot reads two weights at two points — at loading and at discharge. Moisture accounts for part of the difference.
Three things follow and all must be settled:
Test ⑥: you can state, without opening the contract, which time point governs settlement for this lot. If you cannot, the settlement clause has not actually landed.
The five steps above read line by line. This one sets the numbers against each other — and it catches a case that passes every earlier step while the certificate as a whole still does not hold together.
Three checks are available:
Check one: the list is implausibly short.
Add every element printed on the certificate, then look at what is left over. In silicon alloys, the largest component after silicon is iron, alongside aluminium, calcium, carbon, phosphorus and sulphur. If a certificate claims a very high main element while reporting every impurity near its floor and yet prints no iron figure, then those numbers cannot account for the mass of the lot. It does not necessarily mean falsification. It does mean the element list is incomplete and should be asked for in full.
Check two: elements from the same source move together.
Impurities do not arrive at random; they come from somewhere. What the ore carries is usually a different group from what the reductant carries, and elements sharing a source tend to run high together or low together. When a certificate shows one impurity reading anomalously low while its usual companion reads normally, ask for an explanation — either the process genuinely differs, or one item has not been listed as it stands.
Check three: when three pieces of good news arrive together, go back and look again.
Here is the case. A certificate reports the main element very high (near the grade ceiling), reports every impurity very low (near its floor), and belongs to a lot priced below comparable offers — three values that do not sit comfortably alongside one another. The cost structure of making a tonne decides it: high main element, low impurities, and low price rarely hold simultaneously. When that combination appears, either the basis differs between lines (main element dry basis, impurities as-received, say), or items are missing from the list.
Test ⑦: do the numbers form an account that makes sense as a whole? If even one of the three contradictions is present, the document’s credibility has to be reassessed rather than the shipment advanced.
The six steps compressed into something that can hang on a laboratory wall. Left column: how each item is commonly misread. Right column: how to read it properly.
|
# |
Item |
Common misreading |
Correct reading |
|
1 |
Main element content |
Higher is better |
Convert to price per tonne of element before comparing (§2) |
|
2 |
Mg (nodularizer) |
Mg content equals effective Mg |
Distinguish total Mg from MgO; fix the method for metallic Mg |
|
3 |
Carbon C |
An impurity, lower is better |
Grade-dependent; in high-Mn steel it is a resource (§3) |
|
4 |
Phosphorus P |
Only a ceiling matters |
In most grades it enters the steel and resists removal |
|
5 |
Sulphur S |
Only a ceiling matters |
Same, and it drives desulphurisation cost directly |
|
6 |
Aluminium Al |
Harmless |
Affects inclusion morphology and deoxidation products; grade-sensitive |
|
7 |
Calcium Ca |
Harmless |
An active component in some products, an impurity in others |
|
8 |
Residual / trace elements |
Not printed means not present |
Test scope is set by the contract; untested is not absent (§4) |
|
9 |
Size / sizing |
10–50 mm is enough |
Needs oversize and fines ceilings plus a counting basis (§5) |
|
10 |
Moisture |
A shipping detail |
Changes settlement weight; the governing time point must be named (§6) |
|
11 |
Heat number / issuer |
Any is fine |
Determines the document’s evidentiary weight (§1) |
Six steps done, now the decision. The table below goes from what you found to what you do.
|
What you found |
What it means |
What to do |
|
Heat number missing, or inconsistent with markings and packing list |
Cannot be traced to a melt |
Do not release. Establish the correspondence first, then look at numbers |
|
Main element below the agreed minimum |
Pricing basis has moved |
Apply the grade adjustment per contract; reject if no agreement is reached |
|
Main element within spec but well off the reference grade |
Pricing basis has moved |
Same: compute the adjustment and confirm in writing |
|
Main element off reference with no adjustment clause in the contract |
The pricing basis has a hole |
Close that gap in writing first, then settle this lot |
|
An impurity that enters the steel is pressing your grade headroom |
Composition may be affected |
Judge against your grade; change application or reduce the addition rate if needed |
|
Sizing lacks oversize and fines ceilings |
The clause has no authority |
Any acceptance under concession must record the deviation in writing, with screening results attached |
|
Moisture abnormal, and packing or container shows corresponding damage |
Possible wetting in transit |
Photograph as evidence, request re-weighing, and take it up with the carrier |
|
One of the three cross-checks produces a contradiction |
Data integrity in doubt |
Ask for the full element list, or request an independent re-test |
|
None of the above |
Complete and internally consistent |
Release, and archive the certificate number against the material ledger |
How to think about the worst case: release is not a binary pass or fail, it is “accepted under what condition.” Real risk control happens on the concession line — record the deviation, keep the re-weigh results, bind the certificate number to the heat number in your records. Those actions cost nothing, and three months later, when someone asks, they are the whole of your evidence.
Trap one: treating “meets the standard” as “suits my use.” This is the central argument of the article and the most common misjudgement. The standard’s ceiling is an industry compromise; your product requirement is usually stricter and occasionally looser. Copying the standard is wrong in both directions — too strict wastes money, too loose causes failures.
Trap two: when two laboratories disagree, arguing about who is more accurate. The right order is to compare sampling method first, then test method. Two labs returning different results on one lot are usually not disagreeing about competence; they are looking at different samples or working to different methods. That is exactly why a contract should fix an arbitration method and a permitted tolerance — with those two written down, a dispute has an end; without them, both sides simply repeat their own number.
Trap three: accepting a concession without recording the deviation in writing. Concession acceptance is the most common piece of practical flexibility on site and the one that most often leaves a loose end. “We’ll watch it next time,” agreed verbally, leaves nothing behind to show that this lot was accepted subject to a condition. Every concession needs a written record, even if it is only a confirming email.
Trap four: treating the test certificate as acceptance, rather than as input to incoming inspection. A certificate moves risk forward: it screens once, before the material reaches your door. The decision to use it should rest on your own re-verification after arrival (or the agreed independent test result) and on the charge calculation before it enters the furnace. Do that well and a claim has a starting point; skip it and trying to work backwards through paperwork later usually means working with gaps.
