Electronics and Semiconductor Manufacturing

01Purpose and scope

• This manual supports the course Electronics and Semiconductor Manufacturing. It describes how the work is actually done on production floors, in pilot lines, and in the engineering groups that support them. The scope covers wafer fabrication, back-end assembly and packaging, test, and the printed circuit board and surface mount lines that many electronics plants run alongside semiconductor work.
• The manual is written for process engineers, technicians, quality staff, equipment owners, and supervisors who move between these areas. It is also useful for learners who need a reference for how a plant is run, not only for the theory behind devices. It does not teach device physics in depth, and it does not replace the site procedures that govern a particular tool or product.
• Practice changes a great deal by context. A research pilot line may accept slower throughput and more manual handling in exchange for flexibility. A high volume consumer chip plant depends on tight control of recipes, automated material handling, and statistical monitoring. A contract assembly house may run many product families with short runs and frequent changeovers. The same principles apply in each case, but the tolerances, documentation load, and staffing model differ.

02Core principles

• Process control is the foundation. Every step in a flow, whether it is deposition, etch, lithography, cleaning, die attach, or final test, has a set of inputs, equipment states, and outputs that must stay inside known limits. A process that is in control produces results that are predictable and explainable. A process that is out of control may still produce good parts for a while, which is why trend data matters more than a single pass or fail result.
• Contamination and variation are the two main enemies. Particles, metallic traces, moisture, organic residue, and electrostatic discharge can each destroy or weaken a device. Variation in film thickness, dose, temperature, pressure, or timing shifts electrical parameters across a wafer or lot. Good practice reduces both sources at the point where they are created, rather than relying on inspection to catch them later.
• Traceability is not optional. Each wafer, lot, panel, or board should be traceable to the tools, recipes, materials, operators, and times that touched it. When a field return or test failure appears, traceability is what lets engineers narrow the cause within hours rather than weeks.
• Decisions should follow data and the approved change process. An engineer may see a promising recipe adjustment on a test wafer, but a change to a production recipe normally needs a documented experiment, a review, and a controlled release. Skipping that path is one of the most common routes to a yield loss that takes months to find.

03Planning and preparation

• Start by defining the product, the process flow, and the acceptance criteria. For a wafer process this means the device structure, the masks or layout data, the target electrical parameters, and the inline measurements that predict them. For an assembly line it means the bill of materials, the board or package drawings, the solder profile, and the test program. Missing or outdated definitions cause more rework than any single tool fault.
• Confirm that tools are qualified before running product. Qualification usually includes a preventive maintenance record, a particle or chamber condition check, a recipe verification on monitor wafers, and a review of alarms and recent excursions. A tool that has been opened for maintenance or has changed a consumable part should be requalified according to the site rule for that change.
• Plan materials and consumables with lead times in mind. Photoresists, process gases, specialty chemicals, solder paste, underfill, and test sockets all have shelf lives, storage conditions, and qualification records. Expired or mislabeled material is a frequent cause of unexplained shifts, so receiving inspection and clear expiry tracking are part of planning, not an afterthought.
• Set up the people side too. Operators need current work instructions, training records for the tools they run, and access to escalation contacts. Engineers need a way to request tool time for splits and experiments without disrupting production schedules. Plants that do this well treat engineering time as a scheduled resource rather than something squeezed in during a shift.

04Recommended workflow

1. 1
   : Review the product release package, including the process flow, critical parameters, inline monitor requirements, and any customer specifications. Confirm which steps are critical to electrical performance and which are controlled mainly for cleanliness or mechanical reasons.
2. 2
   : Verify tool readiness and material status before the lot is released to the line. Check the tool state, the most recent qualification wafer or board results, the consumable life counters, and the open maintenance items. Hold the lot if any of these are unclear.
3. 3
   : Execute each process step according to the controlled recipe. Operators should confirm the lot identifier, recipe version, and carrier or panel map at the tool before starting. Any manual intervention, such as a rework or a recipe override, must be logged with a reason.
4. 4
   : Perform inline measurements at the defined points. Typical measurements include film thickness, critical dimension, overlay, defect density, sheet resistance, and for assembly, solder joint inspection and dimensional checks. Compare results against control limits and against the expected trend for that tool and product.
5. 5
   : Respond to excursions using the site escalation path. A single out of control point may call for a hold and a quick check of the tool. A pattern across lots calls for engineering analysis and possibly a containment action on affected inventory. Record the decision and the evidence used to make it.
6. 6
   : Complete electrical or functional test and sort results according to the bin definitions. Ensure that the test program version matches the product release and that the test fixtures or probe cards are within their calibration or cleaning intervals.
7. 7
   : Close the lot with a data review. Confirm that all required measurements were taken, that holds are resolved, and that the yield and fallout are within expectations. Route the lot to the next operation or to shipping only after the review is signed off by the responsible role.

05Quality checks, risk controls and compliance

• Quality checks should be placed where they catch problems early and cheaply. Inline monitors on a wafer line, automated optical inspection on a surface mount line, and in-circuit or functional test at the end of assembly each serve a different purpose. Overreliance on final test is expensive, because scrap and rework costs grow with every added step a bad unit passes through.
• Risk controls cover electrostatic discharge, chemical handling, high voltage and vacuum equipment, and ergonomic hazards during manual assembly. ESD programs typically include grounded workstations, wrist straps or heel grounders, humidity control, and ionization where needed. Chemical handling is governed by site hazard communication programs and by regulations that vary by country and state. Check current requirements with the occupational safety authority in your jurisdiction and your site safety office rather than relying on memory.
• Compliance and customer requirements shape many controls. Automotive, medical, aerospace, and defense customers often impose their own quality system requirements, change notification rules, and traceability retention periods. Some product lines carry export control restrictions on technology or on certain devices. Export and classification questions should go to the trade compliance function, which can interpret current rules.
• This guidance does not replace qualified professional judgment, approved site procedures, or required authorization. Where a decision affects product safety, worker health, or a regulated customer requirement, the responsible qualified person must make it under the approved process.

06Collaboration and documentation

• Process, equipment, quality, and test groups must share a common view of each lot. Daily or shift-based review meetings work well when they use the same data sources and the same defect or excursion codes. Without common codes, one group may call a problem contamination while another calls it a lithography issue, and the investigation stalls.
• Documentation should be short, current, and easy to find at the tool. Work instructions, recipe change records, deviation requests, and qualification reports should each have a clear owner and a revision history. Paper travelers have largely given way to electronic records, but the principle is the same: the record must show what was done, by whom, on which equipment, and with what result.
• Engineering change requests need a defined path. A typical path includes a description of the proposed change, the reason, a risk assessment, the validation plan, the approval signatures, and the effective date for production. Changes that affect electrical performance or reliability often need customer notification or approval under the contract terms.
• Supplier collaboration matters as well. Chemical, gas, substrate, and packaging suppliers can change their processes or formulations. Incoming material specifications, certificate of analysis review, and a clear point of contact for supplier notifications help a plant detect such changes before they show up as yield loss.

07Measures and continuous improvement

• Useful measures include first pass yield, final test yield, defect density per layer or per step, tool availability, mean time between failures for critical tools, cycle time per lot, scrap and rework rates, and the number of holds per thousand lots. Each measure should have an owner, a target, and a defined review frequency. Measures that nobody reviews tend to drift out of accuracy because the underlying data is not cleaned or checked.
• Yield analysis should separate systematic causes from random ones. Systematic losses repeat on the same tool, layer, pattern, or location and usually respond well to engineering correction. Random losses are spread out and often trace to contamination or handling. Pareto charts, wafer maps, and bin distribution analysis are standard tools for telling the two apart, and they should be used together with the process history.
• Continuous improvement works best when it is tied to a small number of projects with clear targets. A team might focus on reducing a specific particle adder on one etch chamber, or on shortening solder reflow profile variation across a board family. Projects should end with a controlled change and a monitor period that confirms the improvement holds across tools and products.
• The limits of measurement should be respected. Small sample sizes can make a yield change look large or small, and metrology tools have their own repeatability and reproducibility. When a result is near a decision threshold, repeat the measurement or collect more data before changing the process.

08Common mistakes and how to prevent them

• One common mistake is releasing a lot after a tool has been changed without requalification. The fix is a hard interlock in the tool status system that blocks product until the requalification record is complete, along with a clear rule for which maintenance actions trigger it.
• Another mistake is treating a recipe override as a quick fix. Operators or engineers sometimes adjust a time or temperature to recover a lot, and the change is not recorded or is not reversed. Preventing this requires locked recipes on production tools, a logged override process, and regular audits comparing running recipes with the approved versions.
• Teams also make the error of changing several variables at once during troubleshooting. The result is a fix that may work but cannot be explained, and it often fails when the tool is later changed. A designed experiment that changes one factor at a time, or a structured design of experiments where appropriate, produces knowledge that carries forward.
• Finally, plants often underestimate the cost of contamination from people and materials. Gowning lapses, uncleaned carriers, and unlabeled chemical containers cause defects that look like equipment problems. Regular gowning audits, carrier cleaning schedules, and material labeling checks are inexpensive ways to prevent these losses.

09Practical checklist

• Confirm the product release package and critical parameters are current before starting any lot.
• Verify tool qualification status, maintenance closure, and consumable life before loading product.
• Check that recipe versions on the tool match the approved release and that any override is logged with a reason.
• Confirm material lot, expiry, storage condition, and labeling at point of use.
• Run the required inline measurements and compare each result against control limits and recent trend.
• Review all excursions and holds with the responsible engineer and record the containment decision.
• Confirm ESD controls, gowning, and chemical handling requirements are met at each station.
• Verify test program version, fixture calibration, and bin definitions match the product release.
• Close the lot with a data review, confirm traceability records are complete, and obtain the required sign-off.
• Review yield, fallout, and hold metrics weekly, and assign owners for any systematic loss patterns found.

★Learning sources

• Micro Degree: Electronics and Semiconductor Manufacturing
• Micro Degree: Solar Power Systems
• Micro Degree: Energy Management and Audit
• Micro Degree: Industry 4.0 Technologies