{
  "schemaVersion": 1,
  "id": "atlas-lesson-hbm-mr-muf-route",
  "title": "Mass reflow and molded underfill",
  "category": "learning",
  "format": "Atlas lesson",
  "description": "Separate stack placement, metallic reflow and molding in an MR-MUF assembly route.",
  "canonicalUrl": "https://semiaifoundry.com/learn/atlas/hbm-mr-muf-route/",
  "publishedDate": null,
  "revisionDate": "2026-10-07",
  "author": null,
  "topic": "Packaging",
  "series": null,
  "license": null,
  "rightsUrl": "https://semiaifoundry.com/ai-atlas/licenses.html",
  "version": "1.3.0",
  "status": "Published educational lesson",
  "evidenceScope": "SK hynix’s process explanation describes reflow and molded underfill as separate roles, with implementation-specific chip stabilization in its advanced route. No comparative vendor yield or cost is assumed.",
  "contentCoverage": "published-lesson",
  "artifactUrl": "https://semiaifoundry.com/ai-atlas/lesson-data/hbm-mr-muf-route-7a9c2967bc0f790f.json",
  "interactiveUrl": "https://semiaifoundry.com/ai-atlas/#hbm-mr-muf-route",
  "related": [],
  "supersedes": [],
  "supersededBy": [],
  "contentHash": "5a5376911eeed64d40a1fa9f962d2d7292bde492c3bb0fc16f5ac2bb8f59a37b",
  "contentLength": 4807,
  "jsonUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-hbm-mr-muf-route.json",
  "markdownUrl": "https://semiaifoundry.com/agents/publications/atlas-lesson-hbm-mr-muf-route.md",
  "sections": [
    {
      "id": "section-1",
      "title": "Mass reflow and molded underfill",
      "start": 26,
      "end": 382
    },
    {
      "id": "section-2",
      "title": "The mechanism",
      "start": 382,
      "end": 1557
    },
    {
      "id": "section-3",
      "title": "Science and assumptions",
      "start": 1557,
      "end": 1585
    },
    {
      "id": "section-4",
      "title": "Batch-limited station rate",
      "start": 1585,
      "end": 2141
    },
    {
      "id": "section-5",
      "title": "Engineering example",
      "start": 2141,
      "end": 2365
    },
    {
      "id": "section-6",
      "title": "Distinctions and tradeoffs",
      "start": 2365,
      "end": 2762
    },
    {
      "id": "section-7",
      "title": "Specifications and interfaces",
      "start": 2762,
      "end": 2983
    },
    {
      "id": "section-8",
      "title": "Interfaces",
      "start": 2983,
      "end": 3133
    },
    {
      "id": "section-9",
      "title": "Check your understanding",
      "start": 3133,
      "end": 3162
    },
    {
      "id": "section-10",
      "title": "A reflow station doubles its rate while molding remains slower. What happens to line capacity?",
      "start": 3162,
      "end": 3516
    },
    {
      "id": "section-11",
      "title": "Evidence and scope",
      "start": 3516,
      "end": 3889
    },
    {
      "id": "section-12",
      "title": "Sources",
      "start": 3889,
      "end": 4189
    },
    {
      "id": "section-13",
      "title": "Continue exploring",
      "start": 4189,
      "end": 4212
    },
    {
      "id": "section-14",
      "title": "Foundations",
      "start": 4212,
      "end": 4455
    },
    {
      "id": "section-15",
      "title": "Related lessons",
      "start": 4455,
      "end": 4807
    }
  ],
  "citations": [
    {
      "title": "Explore in Atlas",
      "url": "https://semiaifoundry.com/ai-atlas/#hbm-mr-muf-route"
    },
    {
      "title": "SK hynix — Small Size, Big Impact",
      "url": "https://news.skhynix.com/en/small-size-big-impact/"
    },
    {
      "title": "Published lesson data",
      "url": "https://semiaifoundry.com/ai-atlas/lesson-data/hbm-mr-muf-route-7a9c2967bc0f790f.json"
    },
    {
      "title": "Atlas releases and licensing",
      "url": "https://semiaifoundry.com/ai-atlas/licenses.html"
    },
    {
      "title": "HBM layer count and physical height",
      "url": "https://semiaifoundry.com/learn/atlas/hbm-stack-height-budget/"
    },
    {
      "title": "Microbumps, reflow and thermocompression",
      "url": "https://semiaifoundry.com/learn/atlas/bump-thermocompression-assembly/"
    },
    {
      "title": "Thermocompression with non-conductive film",
      "url": "https://semiaifoundry.com/learn/atlas/hbm-tc-ncf-route/"
    },
    {
      "title": "Selecting a bonding route under constraints",
      "url": "https://semiaifoundry.com/learn/atlas/hbm-bonding-process-selection/"
    },
    {
      "title": "Cycle time, ramp and effective capacity",
      "url": "https://semiaifoundry.com/learn/atlas/fab-ramp-cycle-time/"
    }
  ],
  "recordHash": "5f73f85452fdb9b841718ea4dc230884a6bc63a71b1230eee0c575de1bef3503",
  "contentFormat": "text/markdown",
  "content": "Packaging · Atlas v1.3.0\n\n# Mass reflow and molded underfill\n\nSeparate stack placement, metallic reflow and molding in an MR-MUF assembly route.\n\n[Explore in Atlas](https://semiaifoundry.com/ai-atlas/#hbm-mr-muf-route)Stack alignment, reflow and polymer fill; schematic sequence\n\nLearning objectiveSeparate stack placement, metallic reflow and molding in an MR-MUF assembly route.\n\n## The mechanism\n\nMass reflow forms many solder interconnections within a thermal operation. Molded underfill subsequently fills and encapsulates the intended spaces with molding compound. The route therefore separates electrical joint formation from the bulk protective fill, although particular implementations can use additional stabilization during stacking. Polymer flow through narrow gaps, trapped gas, die bow and reflow behavior still matter. Increasing the number of parts processed together can move a bottleneck from the bond operation to alignment, molding, cure, inspection or test; a batch label does not establish the final factory output.\n\n1.  Align and place the compatible bumped memory dies while controlling thin-die shape.\n2.  Apply any stabilization required by the named implementation; do not assume all MR-MUF products share this substep.\n3.  Run the selected reflow operation to connect the metallic joints across the assembled stack.\n4.  Fill and encapsulate using the specified molded-underfill sequence; control venting, flow, pressure and cure.\n5.  Inspect the same traced stack for joint continuity, voiding, warpage and functional behavior.\n\n## Science and assumptions\n\n### Batch-limited station rate\n\n$$\nQ=\\frac{K a}{t_{\\mathrm{cycle}}}\n$$\n\nQ = K a / tcycle\n\nSymbols and units\n\nQ\n\nProcessed stacks per unit time\n\nstacks/s\n\nK\n\nStacks processed per modeled batch\n\nstacks\n\na\n\nProductive-time fraction\n\nfraction\n\ntcycle\n\nOccupied batch cycle time\n\ns\n\nK is processed stacks per batch and a is the productive-time fraction. Q is the station’s nominal processed-stack rate before yield.\n\nAssumptions Consistent time units, full modeled batch, no double counting overlap; the slowest compatible route station can dominate the line.\n\n## Engineering example\n\nSynthetic station: 20 stacks per 100 s batch at 80% productive time gives 0.16 stacks/s, or 576 stacks/hour before yield. A 400-stack/hour downstream screen caps this toy line below the reflow rate.\n\n## Distinctions and tradeoffs\n\nFactory output follows the limiting compatible stage, including its yield and availability.\n\nBatch processing can reduce per-stack occupancy at one station while narrow-gap filling, alignment and inspection determine end-to-end yield and capacity.\n\nMemory demand becomes qualified delivered stacks only after the whole assembly route can sustain its required rate.\n\n## Specifications and interfaces\n\nReflow\n\nForms metal joints\n\nDo not describe polymer encapsulation as the conductor.\n\nMolded underfill\n\nFills and protects\n\nFlow, voids and mechanical behavior require separate controls.\n\n### Interfaces\n\nPlaced bump stack Input\n\nAlignment and solderable interfaces\n\nEncapsulated stack Output\n\nConnected and mechanically supported memory\n\n## Check your understanding\n\n### A reflow station doubles its rate while molding remains slower. What happens to line capacity?\n\n1.  It can remain limited by molding\n2.  It necessarily doubles\n3.  It becomes independent of test time\n\nAnswer and reasoning\n\nIt can remain limited by molding\n\nFactory output follows the limiting compatible stage, including its yield and availability.\n\n## Evidence and scope\n\nSK hynix’s process explanation describes reflow and molded underfill as separate roles, with implementation-specific chip stabilization in its advanced route. No comparative vendor yield or cost is assumed.\n\nAdjust stack layers, thickness, joint route, interface bandwidth and conditional yield; presets are synthetic, not supplier process windows.\n\n## Sources\n\n-   [SK hynix — Small Size, Big Impact](https://news.skhynix.com/en/small-size-big-impact/)\n\n[Published lesson data](https://semiaifoundry.com/ai-atlas/lesson-data/hbm-mr-muf-route-7a9c2967bc0f790f.json) · [Atlas releases and licensing](https://semiaifoundry.com/ai-atlas/licenses.html)\n\n## Continue exploring\n\n### Foundations\n\n-   [HBM layer count and physical height](https://semiaifoundry.com/learn/atlas/hbm-stack-height-budget/)\n-   [Microbumps, reflow and thermocompression](https://semiaifoundry.com/learn/atlas/bump-thermocompression-assembly/)\n\n### Related lessons\n\n-   [Thermocompression with non-conductive film](https://semiaifoundry.com/learn/atlas/hbm-tc-ncf-route/)\n-   [Selecting a bonding route under constraints](https://semiaifoundry.com/learn/atlas/hbm-bonding-process-selection/)\n-   [Cycle time, ramp and effective capacity](https://semiaifoundry.com/learn/atlas/fab-ramp-cycle-time/)"
}