智能体
arxiv.org ·
用于 X 射线多层膜反射镜的智能体编写沉积配方：受模式约束的大语言模型控制磁控溅射系统并取得经反射率验证的结果
该 arXiv 预印本（cond-mat.mtrl-sci）研究让 LLM agent 在给定多层膜规格并读取实验室档案的条件下，用磁控溅射系统的原生配方语言编写沉积配方，并通过三层共 17 个工具的 Model Context Protocol 桥接提交，约束在控制应用中强制执行。目标为周期 6.85 nm 的 30 双层 Ru/C 多层膜。Agent 仅能写入自己的工作项目，无法改动所读校准数据或评分测量。以 X 射线反射率对照人工撰写参考评分，相对周期差 3% 的验收阈值在沉积前即固定。四次计分沉积中，相对周期差从按给定校准的 20% 降至一次实测反馈迭代后的 2.26%，再降至二次迭代后的 0.79%；同时识别出面向 agent 接口的 16 种失效模式。作者声明结论仅限所观测的周期、材料对与腔体。
AI推荐理由 arXiv 预印本，作者与摘要信息完整，实验设计有明确约束（3% 验收阈值预先设定、agent 只能写入自己的工作项目、无法修改校准与测量），并给出可验证的 X 射线反射率结果：相对周期差从 20% 经一次反馈迭代降至 2.26%、二次降至 0.79%。新信息在于用 schema 约束的 LLM agent 通过 17 个工具的 MCP 桥接直接编写磁控溅射沉积配方，并识别出 16 种接口失效模式。局限是作者自述结论仅限所观测的周期、材料对与腔体，且为预印本未经同行评审，不能外推为通用能力。
78
正文
Condensed Matter > Materials Science
arXiv:2609.12796 (cond-mat)
[Submitted on 11 Sep 2026]
Title:Agent-authored deposition recipes for X-ray multilayer mirrors: schema-bound LLM control of a magnetron sputtering system with reflectivity-verified outcomes
Authors:Oleksiy V. Penkov, Haoyu Fu, Yueying Zhan, Pengyuan Wu, Jiong Jin
View a PDF of the paper titled Agent-authored deposition recipes for X-ray multilayer mirrors: schema-bound LLM control of a magnetron sputtering system with reflectivity-verified outcomes, by Oleksiy V. Penkov and 4 other authors
View PDF
Abstract:Periodic multilayer mirrors for X-rays require a layer period held to about 0.02 nm, set by deposition rates held in the laboratory's own notebook. A large language model agent was given a target multilayer specification and read access to a laboratory archive. Deposition recipes were authored in the native recipe language of a magnetron sputtering system and submitted through a Model Context Protocol bridge of 17 tools in three tiers, with enforcement held in the control application. The target was a 30-bilayer Ru/C multilayer with a period of 6.85 nm, a pair chosen because each period is fitted as two layers. The agent could write only inside a working project of its own, so it could alter neither the calibration it read nor the measurements that scored it. Mirrors were scored against a human-authored reference by X-ray reflectivity, and the 3% acceptance threshold on relative period difference was fixed before deposition. Across four scored agent-authored depositions the relative period difference fell from 20% on the calibration as given to 2.26% after one measured-feedback iteration and 0.79% after a second. Sixteen failure modes of the agent-facing interface were identified. The result is bounded to the period, pair and chamber observed.
Subjects:
Materials Science (cond-mat.mtrl-sci); Instrumentation and Detectors (physics.ins-det)
Cite as:
arXiv:2609.12796 [cond-mat.mtrl-sci]
(or
arXiv:2609.12796v1 [cond-mat.mtrl-sci] for this version)
https://doi.org/10.48550/arXiv.2609.12796
Focus to learn more
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Oleksiy Penkov [view email]
[v1]
Fri, 11 Sep 2026 12:56:20 UTC (1,218 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Agent-authored deposition recipes for X-ray multilayer mirrors: schema-bound LLM control of a magnetron sputtering system with reflectivity-verified outcomes, by Oleksiy V. Penkov and 4 other authors
View PDF
view license
Current browse context:
cond-mat.mtrl-sci
new
|
recent
| 2026-09
Change to browse by:
cond-mat
physics
physics.ins-det
References & Citations
INSPIRE HEP
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
IArxiv recommender toggle
IArxiv Recommender
(What is IArxiv?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)