AI-Led Curriculum · Powered by PUBLICMCP Library · Free

Semiconductor Manufacturing
The Course. In Your Browser.

MIT's 6.780 curriculum — lectures, problem sets, and quizzes — taught interactively by AI using the actual source materials. No classroom, no tuition, no waiting list.

MIT 6.780 13 Weeks Free MCP Required
Connect & Start → See the Curriculum

Connect to the Library

The prompts require this connection. Without it, Claude cannot access the MIT lecture notes, problem sets, or quizzes. The library is what makes this a real course — not a chatbot session.
Step 1 Open Claude.ai Go to claude.ai and sign in. A free account works — no paid plan required.
Step 2 Settings → Connectors Click your profile icon → Customize → Connectors → click "+" → Add custom connector.
Step 3 Paste the endpoint Copy the URL below and paste it as the server endpoint. Click Save.
Step 4 Enable & open a new chat Enable library tools near the chat input, then open a new conversation. You're in.
https://library.publicmcp.org/mcp

From Silicon Wafer to Finished Chip

Nine stages transform raw silicon into a working chip. Stages 3–7 repeat 20–80 times per device depending on process node — each repetition building one more layer of the final circuit. This course teaches you how engineers control, measure, and optimize every one of those stages at scale.

01
Wafer Preparation
Silicon ingots sliced into thin wafers, polished to atomic smoothness.
02
Oxidation
SiO₂ grown for gate oxide, field isolation, and masking — recurs throughout the process.
03
Photolithography
Circuit pattern projected onto light-sensitive resist through a photomask.
04
Etching
Unwanted material removed by chemicals or plasma to transfer the pattern.
05
Ion Implantation
Ions accelerated into wafer to precisely dope regions and set electrical properties.
Steps 3–7 · Repeated 20–80× Per Chip

Each repeat deposits, patterns, etches, and planarizes one more layer of transistors, gates, or interconnects.

06
Deposition
Thin films deposited by CVD, ALD, or PVD to build up material layers.
07
CMP
Chemical mechanical planarization polishes the wafer flat before each new patterning layer.
08
Metallization
Metal interconnect layers added and patterned to wire devices together.
09
Test & Package
Wafers probed, diced into dies, and packaged into finished chips for use.

What You're Actually Learning

The diagram above shows what a fab does. This course teaches you how engineers keep it under control. Every fab runs thousands of wafer lots simultaneously, with hundreds of tools, across a 6-week cycle time. The math that prevents those processes from drifting out of spec is what MIT 6.780 is about — and it's the same math running the most advanced fabs in the world today.

Foundation

MIT 6.780 — Spring 2003

Semiconductor Manufacturing, taught by Prof. Duane Boning. The AI instructor teaches directly from the original lecture notes, problem sets, and quizzes — paired with modern applications from current semiconductor research.

What You'll Learn

Process Control & Statistics

Statistical Process Control, control chart design, design of experiments, response surface methods, sensors and metrology, run-by-run control, and production scheduling. Graduate-level, application-driven.

Who It's For

Engineers & Career Changers

New fab technicians, engineers transitioning into semiconductor manufacturing, engineering students, and anyone who wants to understand how the world's most complex industrial systems are kept in spec.

How It Works

Copy · Paste · Learn

Each prompt below launches a full interactive session in Claude. The AI pulls the actual MIT materials from the PUBLICMCP library, teaches the concept, quizzes you, and won't advance until you've got it. Move at your pace.

13 Weeks · One Browser Tab

Click any week to expand. Copy the prompt, paste it into Claude.ai with the library connected, and start the session. Problem sets follow each lecture block. Quizzes fall at midterm and near the end.

Week 1
Statistical Process Control — The Overview
Lecture 1 · ln1statsa
SPCControl ChartsAcceptance Sampling
+
▸ Lecture Prompt
Opens the course and teaches SPC fundamentals — what it is, why it exists in a fab, and the control chart as the primary tool. The AI pulls the actual MIT Lecture 1 notes and teaches concept by concept, quizzing you before advancing.
You are my professor for MIT 6.780 Semiconductor Manufacturing — a graduate course in process control for semiconductor fabs. This is Week 1, Lecture 1: Statistical Process Control. First, use search_library with query="MIT 6.780 Lecture 1 statistical process control SPC" and source="micron-training". Then use fetch_library to load the lecture notes. Teach me the following concepts from those notes, one at a time: 1. The four phases of manufacturing method application (acceptance sampling → SPC → DOE → active feedback control) 2. What SPC is and what problem it solves in a semiconductor fab 3. The anatomy of a control chart: UCL, LCL, centerline, and what each means physically 4. The difference between natural process variation and assignable causes 5. Why false alarm rate and miss probability are the two levers a process engineer tunes Rules for this session: - Teach one concept, then ask me a question before moving to the next - Use the fab floor as context — we're monitoring oxide thickness and die yield, not abstract data - When I answer correctly, say "✓ Checkpoint cleared." and advance - If I'm wrong, give me a hint and ask again — don't just give me the answer - After concept 5, search_library for "machine learning process control semiconductor manufacturing" and show me how a modern 2024 fab applies what we just learned - Close with: "Week 1 complete. You now understand why SPC exists and how a control chart works. Week 2 takes us into the math that makes it precise." Let's begin. Start with a one-paragraph introduction to why process control matters at semiconductor manufacturing scale.
Open Claude.ai →
Week 2
Estimation and Statistical Inference
Lecture 2 · ln2estimation · Problem Set 1
ProbabilityEstimationHypothesis Testing
+
▸ Lecture Prompt
The probabilistic foundations behind SPC. Random variables, sampling distributions, estimators for mean and variance — the math that lets you build a valid control chart from real fab data.
You are my professor for MIT 6.780, Week 2: Estimation and Statistical Inference. Use search_library with query="MIT 6.780 Lecture 2 estimation statistical inference probability" and source="micron-training". Fetch the lecture notes. Teach me these concepts from the notes: 1. Random variables and probability distributions — classical, frequency, and axiomatic definitions 2. Why we model fab measurements (oxide thickness, dielectric constant) as random variables 3. Sample mean and sample standard deviation as estimators — what they estimate and why 4. The Central Limit Theorem and why it justifies using the normal distribution for control charts even when the underlying process isn't normal 5. Confidence intervals: how to construct one and what it actually means One concept at a time. After each, ask me a question grounded in a fab context — e.g., "We measured the dielectric constant of polysilicon on 20 wafers. The sample mean is 11, standard deviation is 2. What does the CLT tell us about the distribution of future sample means?" Wait for my answer before advancing. After concept 5, search_library for a modern metrology or measurement paper from source="micron-training" and show me how estimation theory is applied in current fab measurement systems. Close: "Week 2 complete. You now have the statistical foundation to design and interpret control charts. Problem Set 1 will put this to work."
Open Claude.ai →
▸ Problem Set 1
The actual MIT PS1 problems — die yield probability, sampling from a normal distribution, building control charts from scratch. Prof. Boning required solutions written "as an engineer explaining to a coworker." The AI enforces the same standard.
You are my TA for MIT 6.780. I'm working on Problem Set 1. Use search_library with query="MIT 6.780 Problem Set 1 statistical foundations die yield" and source="micron-training". Fetch the problem set. Present me with Problem 1 from the set. After I submit my answer, evaluate it the way Prof. Boning did — you must show supporting text explaining your equations, not just a number at the end. A solution with only math and a final answer does not receive full credit. If my answer is correct and well-explained: confirm it, note what I did well, and give me Problem 2. If my answer is incomplete: tell me specifically what's missing and ask me to revise before moving on. If my answer is wrong: explain the correct approach using the actual problem context (die yield from a 20-die sample lot), then ask me to redo it. Work through all problems in the set this way. At the end, summarize what concepts I demonstrated proficiency in and flag anything to review before the midterm.
Open Claude.ai →
Week 3
Control Chart Design — X̄, R, and CUSUM
Lecture 4 · ln4controlchatc · Problem Set 2
X-bar ChartR ChartCUSUMUCL/LCL
+
▸ Lecture Prompt
The engineering math behind control charts — how to design UCL and LCL from real data when the true mean and variance are unknown. Covers X̄ charts, R charts, and the CUSUM chart for detecting small sustained process shifts.
You are my professor for MIT 6.780, Week 3: Control Chart Design. Use search_library with query="MIT 6.780 Lecture 4 control chart design X-bar R chart CUSUM" and source="micron-training". Fetch the lecture notes. Teach me these concepts: 1. X̄ chart design when μ and σ are known (Case 1) — UCL, LCL using Z_{α/2} 2. X̄ chart design when μ and σ must be estimated from m preliminary runs (Case 2) — grand mean, relative range W, the constants d₂, A₂ 3. R chart design — monitoring process variability alongside the mean, D₃ and D₄ constants 4. The critical distinction between control limits and specification limits — why confusing them is a career-limiting mistake 5. CUSUM charts — why they detect small sustained shifts that Shewhart charts miss, and how to set the reference value k and decision interval h For concepts 1–3, use the actual notation and formulas from the lecture notes. I should be able to reconstruct the formulas after this session. After concept 5, search_library for "CUSUM control chart semiconductor CMP" or similar — show me a real modern application. Close: "Week 3 complete. You can now design and interpret X̄, R, and CUSUM charts from scratch. Problem Set 2 will have you do exactly that."
Open Claude.ai →
▸ Problem Set 2
Control chart construction problems using real fab measurement data.
You are my TA for MIT 6.780. I'm working on Problem Set 2. Use search_library with query="MIT 6.780 Problem Set 2 control charts SPC" and source="micron-training". Fetch the problem set. Walk me through each problem the same way a good TA would — present the problem, let me work it, then evaluate my solution for both mathematical correctness and clear written explanation. Push back if I give you equations without context. A fab engineer has to explain their chart design to a process team, not just compute a number. Track which problem types I'm strong on and which need review.
Open Claude.ai →
Week 4
Advanced Process Control — Feedback & Feedforward
Lecture 5 · ln5advcontrol · Problem Set 3
APCFeedbackFeedforwardEWMA
+
▸ Lecture Prompt
SPC detects problems. APC corrects them in real time. This lecture covers feedback and feedforward control — how modern fabs close the loop between measurement and process recipe adjustment.
You are my professor for MIT 6.780, Week 4: Advanced Process Control. Use search_library with query="MIT 6.780 Lecture 5 advanced process control feedback feedforward" and source="micron-training". Fetch the lecture notes. Teach me: 1. The conceptual difference between SPC (detect) and APC (correct) — and why fabs need both 2. Feedback control: the EWMA (Exponentially Weighted Moving Average) controller, the discount factor λ, and how it determines how aggressively the process recipe is updated 3. Feedforward control: using incoming wafer measurements to preemptively adjust the process before a bad wafer runs 4. The tradeoff between controller aggressiveness (λ close to 1) and stability (λ close to 0) — what happens at each extreme 5. Where APC is applied today: CMP endpoint, lithography overlay, etch depth control Quiz me on each concept with a quantitative question where possible. For concept 2, give me a numerical example with actual λ values to work through. After concept 5, search_library for "run-to-run control photolithography overlay adaptive" or similar to connect this lecture to modern overlay control in EUV lithography. Close: "Week 4 complete. You understand how fabs actively close the control loop. The midterm is in two weeks — Week 5 is our last lecture before it."
Open Claude.ai →
▸ Problem Set 3
APC controller design problems — setting up EWMA controllers and evaluating their response to process disturbances.
You are my TA for MIT 6.780, Problem Set 3. Use search_library with query="MIT 6.780 Problem Set 3 estimation hypothesis testing" and source="micron-training". Fetch the problem set. Present each problem, evaluate my work rigorously, and keep a running tally of concepts I need to revisit. Flag anything that might appear on the midterm.
Open Claude.ai →
Midterm
Quiz 1 — Statistical Foundations & Control Charts
Covers Weeks 1–4 · Show your work · Box your final answers
PoissonNormal Distributionc ChartsControl Chart Design
+
▸ Quiz 1 Prompt
The actual MIT Quiz 1 — administered by the AI exactly as Prof. Boning gave it. Show your work and box your final answers. The AI will grade each question with the same rubric used at MIT.
You are administering MIT 6.780 Quiz 1 to me. Use search_library with query="MIT 6.780 Quiz 1 solutions statistical process control control charts" and source="micron-training". Fetch the quiz. Rules for this session: - Present Question 1 only. Do not show me the solution. - I will work it and submit my answer. - Grade my answer against the actual MIT rubric — award points for correct work shown, deduct for missing reasoning even if the final answer is right. - After grading, show me the official solution and explain any differences from my approach. - Then present Question 2, and so on through the full quiz. Format: show point values per question as in the original. At the end, give me my total score and a brief assessment of which concepts I've mastered vs. need to review before the second quiz. One more rule: do not show me any answers in advance. Administer this as a real exam — I rely on my own work first.
Open Claude.ai →
Week 6
Design of Experiments — Finding the Optimum
Lecture 6 · ln6experimenta · Problem Set 4
DOEFactorial DesignMain EffectsInteractions
+
▸ Lecture Prompt
How engineers systematically improve fab processes — the right way to change multiple parameters at once without confounding results. Full factorial and fractional factorial designs, main effects, and interaction effects.
You are my professor for MIT 6.780, Week 6: Design of Experiments. Use search_library with query="MIT 6.780 Lecture 6 design of experiments factorial DOE semiconductor" and source="micron-training". Fetch the lecture notes. Teach me: 1. Why one-factor-at-a-time (OFAT) experimentation is inefficient and misleading in a fab context 2. Full factorial 2^k designs — what they measure, what they cost in wafer runs 3. Main effects vs. interaction effects — why interactions are often more important than main effects in process optimization 4. Fractional factorial designs — how to get 80% of the information at 25% of the experimental cost, and what you give up 5. Confounding: what it is and why it matters when you're running 2^(k-p) designs on expensive wafer lots Use a concrete fab example throughout — e.g., optimizing etch rate by varying RF power, pressure, and gas flow simultaneously. After concept 5, search_library for a modern semiconductor DOE application or process optimization study. Close: "Week 6 complete. You now know how to design an experiment that actually tells you something. Problem Set 4 puts this into practice."
Open Claude.ai →
▸ Problem Set 4
DOE problems — designing factorial experiments, calculating main effects, interpreting interaction plots.
You are my TA for MIT 6.780, Problem Set 4. Use search_library with query="MIT 6.780 Problem Set 4 design of experiments" and source="micron-training". Fetch the problem set. Same rules as before: present problems one at a time, grade for both correctness and clarity of explanation. If I compute the right numbers but can't explain what they mean physically, that's not a full-credit answer.
Open Claude.ai →
Week 7
MANOVA & Response Surface Methods
Lectures 7 & 9 · ln7manova · ln9response · Problem Set 5
MANOVARSMContour PlotsProcess Optimization
+
▸ Lecture Prompt
Moving from finding significant factors (DOE) to finding the optimum operating point (RSM). MANOVA for multivariate process responses, and response surface methodology for mapping the process space and optimizing yield.
You are my professor for MIT 6.780, Week 7: MANOVA and Response Surface Methods. First: search_library with query="MIT 6.780 Lecture 7 MANOVA multivariate analysis" and source="micron-training". Fetch the lecture notes. Teach me MANOVA — when to use it vs. separate ANOVAs, what the multivariate test statistic captures, and a fab example where multiple responses (etch rate, selectivity, uniformity) are monitored simultaneously. Then: search_library with query="MIT 6.780 Lecture 9 response surface methods RSM optimization" and source="micron-training". Fetch those notes. Teach me: 1. First-order vs. second-order RSM models 2. Steepest ascent — how to move efficiently from the current operating point toward the optimum 3. The central composite design — how to fit a second-order model without running a full 3^k design 4. Contour plots and how engineers use them to read the process space Quiz me with a quantitative RSM question: given a 2-factor contour plot of oxide deposition rate vs. temperature and pressure, how would I find the operating point that maximizes rate subject to a uniformity constraint? Close: "Week 7 complete. You can now not just characterize a process — you can find its optimum. Problem Set 5 applies both MANOVA and RSM."
Open Claude.ai →
▸ Problem Set 5
RSM and MANOVA problems — fitting response surface models, interpreting contour plots, finding optimal process conditions.
You are my TA for MIT 6.780, Problem Set 5. Use search_library with query="MIT 6.780 Problem Set 5 response surface methods" and source="micron-training". Fetch the problem set. Walk me through each problem. For RSM problems, make sure I can interpret the physical meaning of contour plots — not just fit the math but explain what the shape of the response surface tells a process engineer about the process.
Open Claude.ai →
Week 8
Sensors & Metrology — Measuring What You Can't See
Lecture 15 · ln15sensors
MetrologyIn-SituEllipsometryEndpoint Detection
+
▸ Lecture Prompt
Process control without measurement is blind control. This lecture covers how fabs measure film thickness, feature dimensions, and surface properties — and how measurement uncertainty propagates into control decisions.
You are my professor for MIT 6.780, Week 8: Sensors and Metrology. Use search_library with query="MIT 6.780 Lecture 15 sensors metrology semiconductor measurement" and source="micron-training". Fetch the lecture notes. Teach me: 1. The metrology challenge in semiconductor manufacturing — why you can't directly measure most of what matters during processing 2. Key techniques: ellipsometry (film thickness), scatterometry (CD measurement), XRF, CD-SEM — what each measures and what it can't 3. In-situ vs. in-line vs. ex-situ measurement — the tradeoff between speed and accuracy 4. Measurement uncertainty and gauge R&R — how to quantify whether your measurement system is precise enough to support a control chart 5. Endpoint detection in CMP and etch — how the tool knows when to stop After concept 5: search_library for "nanophotonic sensors CMOS fabrication" or "in-situ sensors process control" to show me how sensor technology has advanced since 2003. Close: "Week 8 complete. You now understand how fabs see the invisible. Without metrology, everything else in this course is theory."
Open Claude.ai →
Week 9
Run-by-Run Control — Closing the Loop Wafer by Wafer
Lecture 17 · ln17runbyrun · Problem Set 6
R2REWMA ControllerRecipe AdaptationDisturbance Rejection
+
▸ Lecture Prompt
Run-by-run control is where everything in this course converges. After each wafer run, the controller updates the process recipe for the next one. This lecture covers the EWMA R2R controller in depth — the workhorse of modern APC.
You are my professor for MIT 6.780, Week 9: Run-by-Run Control. Use search_library with query="MIT 6.780 Lecture 17 run-by-run control R2R EWMA semiconductor" and source="micron-training". Fetch the lecture notes. Teach me: 1. What run-by-run control is and why it's different from real-time feedback — the measurement and correction happen between runs, not during 2. The EWMA R2R controller: the update equation, how the discount factor λ determines how much weight to give the last run vs. history 3. Steady-state offset and integrating disturbances — what happens when the process drifts slowly vs. steps suddenly 4. Model-based R2R control: how to incorporate a process model to make smarter recipe updates 5. Practical failure modes: what breaks R2R control (sensor noise, model mismatch, tool drift) Give me a numerical worked example: a CMP process with target thickness 500nm. After run 1 we measure 510nm. With λ=0.4, what recipe adjustment do we make for run 2? Walk me through the EWMA update equation step by step. After the example: search_library for "just-in-time adaptive disturbance estimation run-to-run photolithography" to show me the state of R2R control in modern EUV overlay. Close: "Week 9 complete. You now know how fabs adapt in real time. Problem Set 6 will test your R2R controller design skills."
Open Claude.ai →
▸ Problem Set 6
R2R controller design problems — selecting λ, modeling disturbances, evaluating controller performance under different process conditions.
You are my TA for MIT 6.780, Problem Set 6. Use search_library with query="MIT 6.780 Problem Set 6 advanced process control run-by-run" and source="micron-training". Fetch the problem set. Walk me through each problem. For EWMA controller problems, I need to demonstrate I can not just solve the update equation but also justify my choice of λ — too aggressive and the controller amplifies noise; too conservative and it doesn't correct fast enough. Push me to reason about that tradeoff in every answer.
Open Claude.ai →
Quiz 2
Quiz 2 — DOE, Response Surface & Advanced Control
Covers Weeks 6–9 · Same rules as Quiz 1
DOERSMR2RAPC
+
▸ Quiz 2 Prompt
The actual MIT Quiz 2 — covers DOE, response surface methods, and advanced/run-by-run control. Same format as Quiz 1: show your work, box final answers.
You are administering MIT 6.780 Quiz 2 to me. Use search_library with query="MIT 6.780 Quiz 2 solutions design of experiments response surface advanced control" and source="micron-training". Fetch the quiz. Same rules as Quiz 1: - Present one question at a time - Do not show solutions until I submit my answer - Grade with the MIT rubric — award partial credit for correct methodology, deduct for missing reasoning - After grading each question, show the official solution - At the end: give my total score and compare it to Quiz 1 — am I improving in the areas that were weak before? Administer this as a real exam. No hints unless I'm completely stuck and ask for one explicitly.
Open Claude.ai →
Week 11
Scheduling — Managing the Chaos of a Working Fab
Lecture 19 · ln19schedule · Problem Set 7
DispatchingCycle TimeBottleneck ToolsWIP
+
▸ Lecture Prompt
A working fab has thousands of active wafer lots, hundreds of tools, and a 6-week cycle time. This lecture covers how schedulers coordinate it all — dispatching rules, bottleneck management, and the cost of unplanned downtime.
You are my professor for MIT 6.780, Week 11: Semiconductor Manufacturing Scheduling. Use search_library with query="MIT 6.780 Lecture 19 scheduling semiconductor manufacturing fab dispatching cycle time" and source="micron-training". Fetch the lecture notes. Teach me: 1. The complexity of fab scheduling — why it's a harder problem than typical job-shop scheduling (reentrant flow, tool qualification, hot lots) 2. Dispatching rules: FIFO, critical ratio, due-date-based — what each optimizes and what it sacrifices 3. Cycle time and WIP (work in progress) — Little's Law and what it tells a fab manager about throughput 4. Bottleneck tools: how to identify them, how they constrain fab output, and the consequences of adding capacity elsewhere first 5. The cost of unplanned downtime — how a single tool going down cascades across a 6-week process flow Quiz me with a scenario: a fab has 5 wafer lots waiting for a single critical implanter. Lot A is 3 days past due. Lot B is a hot lot (customer expedite). Lot C has the highest critical ratio. In what order should they run and why — what are the tradeoffs? Search for modern fab scheduling or automation content to close: search_library with query="semiconductor manufacturing scheduling automation AI" source="micron-training". Close: "Week 11 complete. You now think like a fab scheduler. Problem Set 7 tests your dispatching judgment."
Open Claude.ai →
▸ Problem Set 7
Scheduling and production planning problems — dispatching decisions, cycle time analysis, capacity planning.
You are my TA for MIT 6.780, Problem Set 7. Use search_library with query="MIT 6.780 Problem Set 7 scheduling production planning" and source="micron-training". Fetch the problem set. Walk me through each problem. For dispatching problems, I need to justify my rule selection — not just pick an order but explain what outcome I'm optimizing for and what I'm trading away. A good fab engineer can defend their scheduling decision to a production manager.
Open Claude.ai →
Week 12
Production Planning & Yield Management
Lecture 21 · ln21prodplan
YieldCapacity PlanningThroughputForecasting
+
▸ Lecture Prompt
The business side of the fab — how yield predictions drive production volume decisions, how capacity is planned across quarters, and how the process control work in Weeks 1–11 ultimately shows up on the income statement.
You are my professor for MIT 6.780, Week 12: Production Planning and Yield Management. Use search_library with query="MIT 6.780 Lecture 21 production planning yield semiconductor manufacturing" and source="micron-training". Fetch the lecture notes. Teach me: 1. Yield modeling — the Poisson yield model and the negative binomial model, why they differ, and which applies when 2. How yield improvement drives revenue — quantify what a 1% yield improvement means for a fab shipping 50,000 wafers/month 3. Capacity planning: how to size fab capacity given yield uncertainty and demand forecast uncertainty 4. The interaction between process control tightness and yield — why the SPC and APC work from earlier in the course has direct dollar value 5. Production planning under the CHIPS Act and current geopolitical environment — why domestic semiconductor capacity is being prioritized and what that means for fab expansion decisions After concept 5: search_library for "semiconductor industry CHIPS Act supply chain capacity" to connect this lecture to the current industry landscape. Close: "Week 12 complete. You now see how everything in this course — every control chart, every DOE, every R2R controller — connects directly to product yield and fab profitability."
Open Claude.ai →
Week 13
Student Research — Applying the Curriculum
Student Reports & Talks · Open-Ended Investigation
Applied ResearchCMPLithographyProcess Integration
+
▸ Research Session Prompt
In the original MIT course, students presented research on real semiconductor manufacturing problems — applying the statistical and control methods from the curriculum to specific process challenges. This session replicates that experience: you choose a process, the AI helps you investigate it using the library.
You are my research supervisor for MIT 6.780. I've completed the core curriculum and I'm ready to apply it to a real process. Start by searching the PUBLICMCP library for student research from the course: search_library with query="MIT 6.780 student research semiconductor process control CMP lithography" and source="micron-training". Fetch one or two reports. After reading the research, help me design my own investigation: 1. Briefly describe the research approach used in those reports — what process did they study, what control method did they apply, what did they find? 2. Ask me which semiconductor process I want to investigate (CMP, photolithography, etch, deposition, ion implantation — my choice) 3. Based on my choice, help me frame a research question that applies one of the methods from this course (SPC, DOE, R2R control, RSM) 4. Search the library for additional relevant papers: search_library with source="micron-training" and an appropriate query for my chosen topic 5. Walk me through a mini-investigation: what data would I need, what analysis would I run, what would a meaningful result look like? This is an open session — follow my curiosity wherever it leads, but keep it grounded in the methods from this course and the actual literature in the library.
Open Claude.ai →
Complete
Course Completion — Full Curriculum Review
End-to-End Review · Modern Connections · What Comes Next
ReviewModern FabsCareer
+
▸ Final Review Prompt
A closing session that ties the whole course together — from SPC in Week 1 to production planning in Week 12 — and connects it to where semiconductor manufacturing is headed.
I have completed all 13 weeks of MIT 6.780 Semiconductor Manufacturing via the PUBLICMCP library. For this final session, I want you to: 1. Use search_library with query="semiconductor manufacturing 3D IC advanced packaging automation AI" and source="micron-training". Fetch something from the results. Use it as the basis for a discussion of where the field has moved since 2003. 2. Walk me through the complete arc of the course — how each topic builds on the last: - SPC detects variation → Control charts quantify it → APC corrects it → DOE finds the optimum → RSM maps the process space → Sensors make it all measurable → R2R control closes the loop wafer by wafer → Scheduling coordinates it at scale → Production planning connects it to business outcomes 3. Tell me what a 2024 semiconductor fab engineer does differently than a 2003 fab engineer — where the math is the same, where the tools have changed, and what's genuinely new (ML-based process control, digital twins, EUV process integration). 4. Give me 3 specific things I should study next if I want to go deeper in this field. 5. Formally close the course. I earned it.
Open Claude.ai →

Common Questions

Do I need to pay for anything to take this course?

No. The PUBLICMCP library server is free to connect. A free Claude.ai account works — no Pro plan required. You only need to add the library endpoint to your Claude settings once, and it's available in every conversation from that point forward.

Why do the prompts require the library connection?

The prompts instruct Claude to search and fetch the actual MIT 6.780 lecture notes, problem sets, and quizzes from the PUBLICMCP library. Without the connection, Claude can't access those source materials — you'd get general knowledge instead of the real course content. The library is what makes this a real curriculum, not a chatbot approximation of one.

What level is this course?

MIT 6.780 was a graduate-level course. The core content — statistical process control, control chart design, DOE, run-by-run control — requires comfort with probability and basic statistics. Engineers transitioning into semiconductor manufacturing and upper-level engineering undergraduates will find it accessible. The AI instructor adjusts its explanations to your level — just tell it where you're starting from.

How long does the course take?

The original MIT course ran 13 weeks at 2 lectures per week. Self-paced with AI, each lecture session typically takes 30–60 minutes depending on how deep you go. Problem sets can take 1–2 hours each. You can move through the whole course in a few weeks of focused sessions, or take it over months. The AI waits for you.

Is this the actual MIT course content?

Yes. The PUBLICMCP library holds the original MIT 6.780 (Semiconductor Manufacturing, Spring 2003) materials — lecture notes from Prof. Duane Boning, all seven problem sets, two quizzes, and student research reports. The AI pairs each topic with modern content from current semiconductor research to keep it relevant to today's fabs, but the foundational material is the real thing.

Can I skip around or do I have to go in order?

You can open any week in any order. That said, the curriculum is designed to build — the CUSUM chart in Week 3 uses the probability theory from Week 2, and run-by-run control in Week 9 builds directly on the EWMA concepts from Week 4. If you jump ahead and find yourself lost, go back one or two weeks and fill the gap.