w wyl123ly 的博客
MD

01-06-AI如何设计稳定新材料

28 KB 2026.10.13 原文件

AI helps design new materials that work in the real world

本篇导读

生成式 AI 能在几分钟内"画"出上百万种新材料的结构,但其中绝大多数在化学上根本不稳定,做不出来也用不上。MIT 团队的做法很聪明:不是事后筛选,而是在生成之前先用语言模型把价电子规则"钉"进去,于是稳定材料的比例大幅提高,算力成本也降下来了。

这篇文章把机器学习、材料化学和产业需求(芯片、数据中心散热)串在一起,含引语、数字和类比。语言难度不高,但物理化学名词密集,是练习"读科研报道时区分结论与背景"的好材料。

一、英文正文

Today, anyone with a large enough artificial intelligence model can generate millions of new material designs in minutes. Unfortunately, that hasn't led to a huge leap in the number of new materials being used to improve the performance of products like computer chips and rockets.

One reason for the translation gap is that current models don't reliably factor in the chemical stability of the materials they generate, and unstable materials aren't very useful in the real world. That forces industries to allocate huge computational budgets to screening out all the unstable materials they generate, in some cases leaving behind a tiny fraction of usable options.

Now, MIT researchers have developed a framework that can be applied at the beginning of the materials generation process to vastly improve the stability rate while achieving targeted material properties. It works by ensuring every design satisfies certain key rules of chemistry relating to the electrons around the materials' atoms before the expensive generation step begins. The researchers call their approach "crystal generator with valence-constrained design, or CrysVCD."

In a paper published today in Nature Computational Science, the researchers show how CrysVCD allowed several commonly used material models to meet those valence shell rules more often, and used it to achieve high lattice-dynamics stability — a stringent stability test — in nearly 70 percent of computational material generations. They also showed the approach could support the creation of materials with specific desired properties, like high thermal conductivity or high dielectric constant, which is important for computer chips and data centers.

A hint of how the researchers envision people using their system is in the name.

"If material-generating models are like DVDs, we are like the DVD player," says associate professor of nuclear science and engineering Mingda Li. "You can plug this into any kind of model, not only existing diffusion models but also future models, where people can't generate enough stable materials, and it can improve stability."

Joining Li on the paper are Mouyang Cheng SM '26 and Weiliang Luo, MIT doctoral students in materials science and engineering and chemistry, respectively; Hao Tang PhD '26, a recent graduate in materials science and engineering; Bowen Yu, a senior undergraduate in physics; Yongqiang Cheng, a staff scientist at the Oak Ridge National Laboratory; Weiwei Xie, an associate professor at Michigan State University; Ju Li, MIT's Carl Richard Soderberg Professor in Power Engineering; and Heather Kulik, MIT's Lammot du Pont Professor of Chemical Engineering.

More efficient materials

Computational approaches to materials design have been around for decades, but recent advances in artificial intelligence have increased excitement about their potential. Of particular interest are models that can start with a desired material property and work backward to deliver a material that achieves that goal.

Some of those models use an AI technique known as diffusion, which is commonly used to generate images, while others use large language models like the one powering ChatGPT and Claude, but both approaches struggle to ensure their material generations achieve chemical stability or follow fundamental principles about how chemicals interact and behave.

The solution has been to add another layer of computing on top of the generative process to filter out unstable materials.

"It's becoming easy to generate the material structure," Cheng says. "But the validation process, especially the part where you test the stability, has a huge computational cost. It's something like 90 percent of the computational cost for creating usable materials, and it can take weeks or months."

Big companies with huge computing budgets can afford to run those processes, but many small companies and research labs can't, potentially limiting innovation in the field.

"In academia, where we have fewer resources, I think we can still achieve strong performance with smarter designs and other approaches," Kulik explains. "Generating a model and then down-selecting for stability is inefficient. There's a high computational cost. But if we put a language model in the beginning of the process to constrain the generation, you can significantly enhance the ratio of stable materials generated."

The new study involved MIT researchers affiliated with the departments of Materials Science and Engineering, Chemistry, Chemical Engineering, Physics, and Nuclear Science and Engineering. Together the researchers combined AI diffusion models with a language model. In the first stage of their process, the language model produces chemically valid formulas. In the second stage, the diffusion model uses that formula to generate the corresponding atomic structure of the crystal material in coordination with the underlying material generation model.

"Diffusion for typical material generation is a slow process — you can think of it like 1,000 steps to create one material," Luo says.

"In contrast, when our model is used in the beginning, you can think of it like five steps. It allows you to screen out the unstable materials to generate higher quality materials. And it works with any models generating materials," Tang adds.

The researchers showed their approach created more stable materials an order of magnitude more efficiently than approaches that rely on screening materials after they're generated. When fine-tuned on stability metrics, their approach produced crystalline materials that achieved 68 percent mechanical stability and 85 percent metastability, which measures if a material stays in a stable state when undisturbed.

The researchers then used their approach to generate material candidates with high thermal conductivity and easy polarization in an electric field.

"These are materials useful for the semiconductor industry and high thermal conductivity materials relevant to data center cooling," Ju Li says. "In principle, you could also use this to create other properties, but thermal conductivity has become really important for cooling data centers. There's been a huge increase in energy use in that industry, and 30 percent of that energy goes to cooling. The industry needs materials with high thermal conductivity to more efficiently remove the heat."

Democratizing material design

The new approach doesn't work with every kind of material — it works best with solid structures with highly ordered internal arrangements. Still, the approach could be used to generate stable new crystalline materials with a host of important properties.

"We are not just generating stable materials, we're also prioritizing performance," Cheng says. "Any time you have two goals, achieving those goals with anything over 50 percent is hard in this field. In the past, people might have a goal for specific properties and not stability, or vice-versa, and get a single-digit percentage of materials that fit their goal."

Ultimately the approach will enable more researchers to develop novel materials for a range of next-generation applications.

"This will save huge computation costs and time by removing downstream selection requirements," Li says. "That will help not only large efforts that generate hundreds of millions of materials, but also smaller research groups with targeted applications."

The work was supported, in part, by the U.S. Department of Energy, a Mathworks Engineering Fellowship, the National Science Foundation, and the U.S. Defense Threat Reduction Agency.


二、生词与短语精讲

用法

先通读正文,再回来核对。建议先遮住中文释义自测。

单词/短语音标词性英文释义中文释义文中例句
framework/ˈfreɪmwɜːk/n.a set of ideas or rules used as a basis框架、体系Now, MIT researchers have developed a framework that can be applied at the beginning of the materials generation process to vastly improve the stability rate while achieving targeted material properties.
reliably/rɪˈlaɪəbli/adv.in a way that can be trusted to be correct可靠地、稳定地One reason for the translation gap is that current models don't reliably factor in the chemical stability of the materials they generate, and unstable materials aren't very useful in the real world.
factor in/ˈfæktər ɪn/phrasal v.to include something as a relevant element把……考虑在内One reason for the translation gap is that current models don't reliably factor in the chemical stability of the materials they generate, and unstable materials aren't very useful in the real world.
allocate/ˈæləkeɪt/v.to give out money, time or resources for a purpose拨出、分配That forces industries to allocate huge computational budgets to screening out all the unstable materials they generate, in some cases leaving behind a tiny fraction of usable options.
screen out/skriːn ˈaʊt/phrasal v.to remove unwanted items after checking筛除、淘汰It allows you to screen out the unstable materials to generate higher quality materials.
fraction/ˈfrækʃn/n.a very small part of something一小部分That forces industries to allocate huge computational budgets to screening out all the unstable materials they generate, in some cases leaving behind a tiny fraction of usable options.
vastly/ˈvɑːstli/adv.to a very great degree极大地Now, MIT researchers have developed a framework that can be applied at the beginning of the materials generation process to vastly improve the stability rate while achieving targeted material properties.
valence/ˈveɪləns/n.the combining power of an atom, set by its outer electrons价、化合价In a paper published today in Nature Computational Science, the researchers show how CrysVCD allowed several commonly used material models to meet those valence shell rules more often, and used it to achieve high lattice-dynamics stability — a stringent stability test — in nearly 70 percent of computational material generations. ...
constrain/kənˈstreɪn/v.to limit or restrict what can happen约束、限制But if we put a language model in the beginning of the process to constrain the generation, you can significantly enhance the ratio of stable materials generated. ...
lattice/ˈlætɪs/n.the regular arrangement of atoms in a crystal晶格The researchers call their approach "crystal generator with valence-constrained design, or CrysVCD." In a paper published today in Nature Computational Science, the researchers show how CrysVCD allowed several commonly used material models to meet those valence shell rules more often, and used it to achieve high lattice-dynamics stability — a stringent stability test — in nearly 70 percent of computational material generations.
stringent/ˈstrɪndʒənt/adj.very strict and demanding严格的、苛刻的The researchers call their approach "crystal generator with valence-constrained design, or CrysVCD." In a paper published today in Nature Computational Science, the researchers show how CrysVCD allowed several commonly used material models to meet those valence shell rules more often, and used it to achieve high lattice-dynamics stability — a stringent stability test — in nearly 70 percent of computational material generations.
thermal conductivity/ˈθɜːml ˌkɒndʌkˈtɪvəti/n.the ability of a material to carry heat导热率、热导率The researchers then used their approach to generate material candidates with high thermal conductivity and easy polarization in an electric field.
dielectric constant/ˌdaɪɪˈlektrɪk ˈkɒnstənt/n.a measure of how a material responds to an electric field介电常数They also showed the approach could support the creation of materials with specific desired properties, like high thermal conductivity or high dielectric constant, which is important for computer chips and data centers.
envision/ɪnˈvɪʒn/v.to imagine something as a future possibility设想、展望A hint of how the researchers envision people using their system is in the name.
diffusion model/dɪˈfjuːʒn ˈmɒdl/n.a generative model that builds data out of noise step by step扩散模型The diffusion model uses that formula to generate the corresponding atomic structure of the crystal material in coordination with the underlying material generation model. ...
respectively/rɪˈspektɪvli/adv.in the same order as the items just listed分别、各自Joining Li on the paper are Mouyang Cheng SM '26 and Weiliang Luo, MIT doctoral students in materials science and engineering and chemistry, respectively ...
undergraduate/ˌʌndəˈɡrædʒuət/n.a student studying for a first degree本科生... Hao Tang PhD '26, a recent graduate in materials science and engineering; Bowen Yu, a senior undergraduate in physics; Yongqiang Cheng, a staff scientist at the Oak Ridge National Laboratory ...
validation/ˌvælɪˈdeɪʃn/n.the act of checking that something is correct or acceptable验证、校验"But the validation process, especially the part where you test the stability, has a huge computational cost.
down-select/ˌdaʊn sɪˈlekt/v.to reduce a large list to the best few层层筛选、优中选优Generating a model and then down-selecting for stability is inefficient.
affiliated with/əˈfɪlieɪtɪd wɪð/adj.officially connected with an organization隶属于、就职于The new study involved MIT researchers affiliated with the departments of Materials Science and Engineering, Chemistry, Chemical Engineering, Physics, and Nuclear Science and Engineering. ...
valid/ˈvælɪd/adj.acceptable according to a set of rules合乎规则的、有效的In the first stage of their process, the language model produces chemically valid formulas.
atomic structure/əˈtɒmɪk ˈstrʌktʃə(r)/n.the arrangement of atoms in a substance原子结构In the second stage, the diffusion model uses that formula to generate the corresponding atomic structure of the crystal material in coordination with the underlying material generation model.
an order of magnitude/ən ˈɔːdə əv ˈmæɡnɪtjuːd/n.a factor of about ten一个数量级、约十倍The researchers showed their approach created more stable materials an order of magnitude more efficiently than approaches that rely on screening materials after they're generated.
fine-tune/ˌfaɪn ˈtjuːn/v.to make small adjustments to improve performance微调When fine-tuned on stability metrics, their approach produced crystalline materials that achieved 68 percent mechanical stability and 85 percent metastability, which measures if a material stays in a stable state when undisturbed.
metastability/ˌmetəstəˈbɪləti/n.the state of staying stable if left undisturbed亚稳性When fine-tuned on stability metrics, their approach produced crystalline materials that achieved 68 percent mechanical stability and 85 percent metastability, which measures if a material stays in a stable state when undisturbed.
crystalline/ˈkrɪstəlaɪn/adj.made of crystals with an ordered internal pattern结晶的、晶态的Still, the approach could be used to generate stable new crystalline materials with a host of important properties.
polarization/ˌpəʊləraɪˈzeɪʃn/n.the separation of electrical charge in a material极化The researchers then used their approach to generate material candidates with high thermal conductivity and easy polarization in an electric field.
in principle/ɪn ˈprɪnsəpl/adv.as far as the basic idea is concerned原则上、理论上In principle, you could also use this to create other properties, but thermal conductivity has become really important for cooling data centers.
prioritize/praɪˈɒrətaɪz/v.to treat something as more important than others优先考虑We are not just generating stable materials, we're also prioritizing performance," Cheng says.
vice-versa/ˌvaɪs ˈvɜːsə/adv.with the two things mentioned swapped round反之亦然In the past, people might have a goal for specific properties and not stability, or vice-versa, and get a single-digit percentage of materials that fit their goal." Ultimately the approach will enable more researchers to develop novel materials for a range of next-generation applications.
single-digit/ˌsɪŋɡl ˈdɪdʒɪt/adj.less than ten (used of a percentage)个位数的In the past, people might have a goal for specific properties and not stability, or vice-versa, and get a single-digit percentage of materials that fit their goal." Ultimately the approach will enable more researchers to develop novel materials for a range of next-generation applications.
novel/ˈnɒvl/adj.new and not existing before新颖的、全新的In the past, people might have a goal for specific properties and not stability, or vice-versa, and get a single-digit percentage of materials that fit their goal." Ultimately the approach will enable more researchers to develop novel materials for a range of next-generation applications.
downstream/ˌdaʊnˈstriːm/adj.happening later in a process下游的、后续的This will save huge computation costs and time by removing downstream selection requirements," Li says.

高频搭配与短语动词

搭配中文例句
lead to导致、带来That hasn't led to a huge leap in the number of new materials being used to improve the performance of products like computer chips and rockets.
work backward反向推演、倒推Of particular interest are models that can start with a desired material property and work backward to deliver a material that achieves that goal.
be of particular interest特别值得关注Of particular interest are models that can start with a desired material property and work backward to deliver a material that achieves that goal.
filter out过滤掉、剔除The solution has been to add another layer of computing on top of the generative process to filter out unstable materials.
be affiliated with隶属于The new study involved MIT researchers affiliated with the departments of Materials Science and Engineering, Chemistry, Chemical Engineering, Physics, and Nuclear Science and Engineering.
in coordination with与……协同、配合In the second stage, the diffusion model uses that formula to generate the corresponding atomic structure of the crystal material in coordination with the underlying material generation model.
in contrast相比之下In contrast, when our model is used in the beginning, you can think of it like five steps.
rely on依赖、依靠The researchers showed their approach created more stable materials an order of magnitude more efficiently than approaches that rely on screening materials after they're generated.
a host of许多、大量Still, the approach could be used to generate stable new crystalline materials with a host of important properties.
in part部分地The work was supported, in part, by the U.S. Department of Energy, a Mathworks Engineering Fellowship, the National Science Foundation, and the U.S. Defense Threat Reduction Agency.

三、长难句解析

长难句 1

原句:Now, MIT researchers have developed a framework that can be applied at the beginning of the materials generation process to vastly improve the stability rate while achieving targeted material properties. It works by ensuring every design satisfies certain key rules of chemistry relating to the electrons around the materials' atoms before the expensive generation step begins.

结构拆解:第一句主干是 MIT researchers have developed a framework,用现在完成时强调成果已经取得;that can be applied at the beginning of the materials generation process 是定语从句修饰 framework,其中 can be applied 是"情态动词+被动语态";to vastly improve the stability rate 是不定式短语作目的状语;句末 while achieving targeted material properties 是 while + 动名词结构作伴随状语,while 后省略了主语和 be 动词。第二句主干是 It works,by ensuring... 是"by + 动名词"表示方式;ensuring 后接省略 that 的宾语从句 every design satisfies certain key rules of chemistry,relating to the electrons... 是现在分词短语作后置定语修饰 rules,句末 before the expensive generation step begins 是时间状语从句——"先验证、后生成"的先后关系全靠它点明。

参考翻译:如今,MIT 的研究者开发出一个框架,可以在材料生成流程的开端就投入使用,从而在达成目标材料性能的同时大幅提高稳定率。它的做法是:在昂贵的生成步骤开始之前,先确保每一个设计都满足若干关于材料原子外围电子的关键化学规则。

长难句 2

原句:They also showed the approach could support the creation of materials with specific desired properties, like high thermal conductivity or high dielectric constant, which is important for computer chips and data centers.

结构拆解:主句主干是 They also showed (that)...,showed 后是省略了 that 的宾语从句;宾语从句的主干为 the approach could support the creation of materials。with specific desired properties 是介词短语作后置定语修饰 materials;like high thermal conductivity or high dielectric constant 是举例;句末 which is important for... 是非限制性定语从句,which 指代前面整件事(能造出这类材料),而不是某一个名词;computer chips and data centers 用 and 并列,说明产业落点。

参考翻译:他们还证明,这一方法能够支撑生成具备特定目标性能的材料,例如高导热率或高介电常数,而这对计算机芯片和数据中心都很重要。

长难句 3

原句:When fine-tuned on stability metrics, their approach produced crystalline materials that achieved 68 percent mechanical stability and 85 percent metastability, which measures if a material stays in a stable state when undisturbed.

结构拆解:句首 When fine-tuned on stability metrics 是时间状语从句的省略形式,完整说法是 When it was fine-tuned...。主句主干是 their approach produced crystalline materials;that achieved 68 percent... 是定语从句修饰 materials;句末 which measures... 是非限制性定语从句,which 指代前面的 metastability,从句中又套了一个 if 引导的宾语从句和一个 when undisturbed 的省略状语从句,形成三层嵌套。

参考翻译:在按稳定性指标进行微调后,他们的方法生成出晶体材料,达到 68% 的力学稳定性和 85% 的亚稳性——亚稳性衡量的是材料在不受外界干扰时能否保持在稳定状态。

四、专业词汇标注

专业词汇

这些词只需"认识",不必会拼写。

专业词汇中文解释所属领域
generative model生成模型:能从噪声或条件出发造出新样本的 AI 模型机器学习
diffusion model扩散模型:逐步去噪生成图像或结构的生成模型机器学习
large language model (LLM)大语言模型自然语言处理
valence shell / valence-constrained design价电子层/价约束设计:先保证外层电子满足成键规则化学
chemical stability化学稳定性:材料在环境中不易分解或变质材料化学
lattice-dynamics stability晶格动力学稳定性:原子振动层面也稳定的严格判据固体物理
metastability亚稳性:不受扰动时保持稳定,受扰动则可能转变材料科学
thermal conductivity热导率:材料导热的能力,数据中心散热的关键材料科学
dielectric constant介电常数:材料在电场中极化程度的量度电子材料
polarization极化:电场作用下材料内部电荷分布发生偏移电磁学
crystalline material晶体材料:内部原子长程有序排列的固体材料科学
semiconductor industry半导体产业:制造芯片的行业微电子
data center cooling数据中心散热:把服务器产生的热量带走计算机工程
atomic structure原子结构:材料中原子种类与排列方式材料科学
high-throughput screening高通量筛选:批量生成再逐个检验候选方案计算材料学

五、阅读训练建议

  • [ ] 第一遍:不看注释通读全文,记录用时 ___ 分钟
  • [ ] 第二遍:查生词表,回读卡住的句子
  • [ ] 第三遍:只看英文,口头复述每段大意
  • [ ] 输出:用 5 句英文写出本文摘要

关联阅读

- 01-04-大语言模型对医疗意味着什么 —— 同样是"生成模型进入专业领域":一边是诊室里的准确性与问责,一边是实验室里的稳定性与成本。

- 01-03-人工智能规模扩张之路 —— 本文反复出现的算力账单,正是那篇讲的三条指数曲线在科研场景里的样子。

- 01-05-自动驾驶人工智能的可解释性 —— 两种"驯服模型"的思路:一边把化学规则前置进生成过程,一边把决策过程翻译给人看。

- 01-01-算法单一化对招聘的影响 —— 从材料设计跳到招聘:算法走出实验室进入社会决策后,代价往往由谁承担。

六、来源与版权说明

  • 原文标题:AI helps design new materials that work in the real world
  • 来源:MIT News(麻省理工学院新闻办公室),2026 年 8 月 26 日,作者 Zach Winn
  • 原文链接:https://news.mit.edu/2026/ai-helps-design-new-materials-that-work-in-real-world-0826
  • 说明:本篇英文正文为原文节选,逐字摘录自报道开头至结尾的连续英文段落(仅略去图片说明、联系方式、文末分享按钮与相关链接等非正文部分,故正文内未使用 [...] 标记),仅用于个人英语学习;版权归原作者与麻省理工学院所有。