Building a Semiconductor Materials Procurement Dossier for a Second Source Review

What the dossier is for, and what it is not

A second-source review answers one question: if the current supply route for a material, a reclaimed wafer stream or a power module stops, can another route carry the requirement without introducing unacceptable risk? The dossier is the written case for that answer. It is not a supplier brochure collection and it is not a purchase order. It is the file a reviewer who was not in the meetings can read and either approve, reject or send back with specific gaps.

Treat the dossier as a decision record with three jobs. First, it states the requirement in your own words, so every supplier is measured against the same yardstick. Second, it separates what the supplier has stated from what you have verified. Third, it names the open items and who owns them. A dossier that hides its gaps is worse than one that lists them, because the reviewer will find them later at a less convenient time.

For Singapore-based teams, the practical framing is usually a mix of continuity of supply, documentation discipline and the ability to explain the decision to an internal audit or a customer. Keep the language plain and keep the evidence attached. Where a supplier makes a claim you cannot check, record it as a supplier statement rather than as a finding.

One useful habit is to give the dossier a version number and a date on every page. Second-source reviews often run across several weeks and several departments, and a versioned file prevents two people from reviewing different documents.

Section 1: Write the requirement before you contact anyone

Start with a requirement sheet that you own. It should list the material or component, the specification or acceptance criteria, the annual and monthly volumes you are willing to share, the packaging and handling expectations, the documentation you need with each lot, and the commercial terms that matter to you. Keep it to one or two pages.

A worked hypothetical example helps here. Imagine a Singapore operations team that buys a silicon-based material for a packaging line and wants a second route. Their requirement sheet might say: material type and grade, a certificate of analysis per lot, a defined packaging format, a lead time ceiling, and a named technical contact for deviation queries. None of these are supplier claims; they are the team's own conditions, and they belong in the dossier before any supplier response is filed.

The requirement sheet also protects you from scope drift. When a supplier offers something adjacent, such as a reclaim service instead of virgin material, you can note it as an alternative route and decide deliberately whether it belongs in this review or a separate one. Related material streams, including silicon recovery and wafer reclaim, are often reviewed alongside primary supply because they change the risk picture, but they should be assessed on their own criteria rather than folded into the same line item.

Finally, decide who signs the requirement sheet. A named owner makes the later approval step much faster.

Related resource: products.

Section 2: Build the supplier profile from sourced facts only

The supplier profile is where dossiers most often go wrong, because it drifts from evidence into impression. Restrict each entry to something the supplier has published or confirmed in writing, and label it as a supplier statement.

As an example of how a sourced profile reads, San Techno is a Singapore-based company working across semiconductor and solar materials, silicon recovery, specialised power module development and botanical extracts under the ST GREEN brand. The company was incorporated in 2006 and began as a wholesaler and distributor of semiconductor- and solar-grade silicon materials. Its silicon recovery work covers non-usable silicon from integrated-circuit fabs, silicon wafer manufacturers, solar cell companies and chip packaging facilities, and those materials can be processed into reusable raw materials for solar wafers and cells. If the scope also covers power modules, keep the design questions in a separate evidence pack. Detailed architecture, materials, dimensions and assembly processes for power modules remain confidential, which is a normal boundary to record rather than a gap to argue about.

Notice what this profile does not do. It does not score the supplier, rank it against others or imply that any of these capabilities match your requirement. That comparison belongs in a later section, after your own criteria are on the page. If a supplier's public material is thin on a point you need, write "not stated" and move on.

Section 3: Assemble the document pack and mark its status

List every document you hold and give each one a status: received, reviewed, verified or outstanding. A simple table with columns for document, date, source and status is enough. Typical entries include a company profile, a quality or environmental certificate where one exists, a material data sheet, a sample evaluation report, a commercial quotation and a signed confidentiality agreement.

Be strict about the difference between received and verified. A certificate that has been emailed is received. It becomes verified only when someone on your side has checked the scope, the issuing body and the expiry date against your requirement. The same applies to test data: a supplier's internal result is a supplier statement until your own incoming inspection or a mutually agreed third-party test confirms it.

Where a supplier's published material is deliberately limited, record the limit and the reason given. For power modules, for instance, architecture, materials, dimensions and assembly processes may be held confidential, so the dossier should note that the technical comparison will rely on the supplier's stated validation approach and on any data shared under agreement. San Techno describes thermal modelling, finite-element analysis, power cycling and infrared measurement as methods supporting ongoing design validation, and a reviewer can reasonably ask which of these will be evidenced for your application.

Keep the pack in one folder with a contents page. Reviewers who have to hunt for attachments tend to approve slowly or not at all.

Related resource: resources.

Section 4: Score against your criteria, not the supplier's strengths

Build a comparison table with your requirement sheet as the rows and each candidate as a column. Use a small scale, for example zero to three, and define what each score means before you start. As an author heuristic rather than an established method, a three might mean evidence held and reviewed, a two means supplier statement only, a one means partial or unclear, and a zero means no evidence. The scale is a communication device, not a measurement instrument.

This structure keeps the review honest when a supplier is strong in an area you did not ask about. A company may have deep experience in solar-grade silicon and silicon recovery while your requirement is a specific packaging material; the table will show that mismatch plainly without anyone having to argue about it.

Add a short note column for anything that needs a follow-up question. Examples include asking how a supplier handles deviations, what notice it gives for specification changes, and which contact owns technical queries. For power module candidates, a fair follow-up is to ask how the supplier's development direction accounts for the complete IGBT thermal path, material-layer expansion, packaging stress and assembly integrity, and what evidence it can share under confidentiality.

Finally, record who scored each row and when. A second reviewer should be able to see whether a score came from a document, a meeting or an assumption.

Related resource: about.

Section 5: Plan the sample and qualification step

A second source is not approved by paperwork alone. The dossier should contain a qualification plan that states what will be tested, on which line or in which evaluation, how many lots or units are involved, what pass and fail look like, and who signs the result. Keep the plan proportionate to the risk; a consumable material and a power module will not need the same depth.

Write the acceptance criteria before the sample arrives. This is the single most useful discipline in the whole dossier, because criteria written afterwards tend to bend towards whatever the sample did. If a criterion cannot be met yet, mark it as pending rather than lowering it.

For material streams that involve recovery or reclaim, the plan should also state what happens to material that fails evaluation, who holds custody at each stage, and what records travel with each lot. Those questions are easier to settle in the dossier than at the receiving dock.

Close the section with a decision box: approved, approved with conditions, or not approved, plus the date and the approver. A dossier without a decision box tends to circulate indefinitely.

Section 6: Keep the dossier alive after the decision

A second-source dossier has a useful life beyond the first approval. Set a review date, typically tied to your own supplier review cycle, and note what would trigger an earlier review: a specification change, a change of manufacturing site, a quality incident, or a shift in volume. Record the trigger list in the dossier itself so the next owner inherits the reasoning.

Maintain a short change log at the front. Each entry should say what changed, who changed it and why. This is especially valuable when a supplier's public information is updated, because a reviewer can see whether the earlier assessment still holds.

For teams that want to understand a candidate's wider material and engineering context before the next review cycle, the supplier's own product and resource pages are a reasonable starting point, provided anything taken from them is filed as a supplier statement and checked against your requirement sheet. Where a conversation is needed, San Techno lists direct contact details for enquiries, and it is worth agreeing in advance which questions you will put in writing so the answers can be filed.

A dossier that is reviewed on schedule and updated with evidence stays useful. One that is filed and forgotten becomes a document nobody trusts.

Related resource: San Techno.