IT之家 9 月 29 日消息，上海理工大学智能科技学院顾敏院士、张启明教授团队今日在《自然 · 光子学》在线发表研究成果，自主研发单光束多维光存储工艺。
该工艺使单个三维存储点可区分 1024 种状态，对应 10 bit 信息容量，并达成高速写入和宽场并行读取。DVD 尺寸介质的理论存储容量约 0.4 Pb，数据读取可靠性超过 99.99%。
传统二进制存储中，一个记录单元通常只有“0”和“1”两种状态。张启明表示，新工艺让一个记录单元具备 1024 种可区分状态。
过去需要多个记录单元承载的信息，现在可借助提高单个记录点信息容量完成。光存储提高容量不再只靠缩小记录点或增加层数。
以 8K 灰度图像为例，一个像素有 256 种灰度取值，需 8 个二进制单元保存。10 bit 存储点完成 8 bit 灰度后，多出 2 bit 可为原位光学 AI 训练给出编码空间。
传统超分辨光存储路线常需双束光协同写入和读取，带来精密配准、稳定控制等平台难题。团队此次以单束单色光分别完成写入和读取，打破高容量光存储对双光束读写架构的依赖。
在现有 0.4 Pb 基础上，若将层间距由 5 μm 压缩至 1 μm，灰阶编码拓展至万级，理论容量有望提高至约 20 Pb。
顾敏表示，面向将来海量数据存储，Pb 容量并不是唯一衡量指标。真正走向应用，还要兼顾读写速度、平台复杂度和设备集成能力。
博士生张慧和 Simone Lamon 副研究员为论文共同第一作者；Simone Lamon 副研究员、刘丽英教授、张启明教授与顾敏院士为共同通讯作者。上海理工大学为论文第一署名单位，复旦大学和东南大学为合作完成单位。
IT之家附论文链接：
https://www.nature.com/articles/s41566-026-02004-z
本文来源：https://www.ithome.com/1/008/488.htm
若有版权问题，请来信告知：13951431913#139.com 或在右方提交评论。
IT Home reported on September 29 that the team led by Academician Min Gu and Professor Qiming Zhang at the School of Intelligent Technology, University of Shanghai for Science and Technology, published research results online in Nature Photonics today on a self-developed single-beam multidimensional optical storage process.
The process enables a single three-dimensional storage point to distinguish 1024 states, corresponding to 10 bit information capacity, and achieves high-speed writing and wide-field parallel reading. The theoretical storage capacity of a DVD-sized medium is about 0.4 Pb, and data reading reliability exceeds 99.99%.
In conventional binary storage, a recording unit usually has only two states, “0” and “1.” Qiming Zhang said that the new process allows one recording unit to have 1024 distinguishable states.
Information that previously required multiple recording units can now be carried by increasing the information capacity of a single recording point. Increasing optical storage capacity no longer relies solely on shrinking recording points or adding layers.
Taking an 8K grayscale image as an example, one pixel has 256 grayscale values and needs 8 binary units for storage. After a 10 bit storage point handles 8 bit grayscale, the extra 2 bits can provide encoding space for in situ optical AI training.
Conventional super-resolution optical storage approaches often require two beams of light to cooperate for writing and reading, creating platform challenges such as precise alignment and stable control. This time, the team completed writing and reading separately with a single monochromatic beam, breaking high-capacity optical storage’s dependence on a two-beam read-write architecture.
Building on the existing 0.4 Pb, if the layer spacing is compressed from 5 μm to 1 μm and grayscale coding is expanded to the ten-thousand level, the theoretical capacity is expected to increase to about 20 Pb.
Min Gu said that for future massive data storage, Pb capacity is not the only metric. To truly move toward applications, one must also balance read-write speed, platform complexity, and device integration capability.
PhD student Hui Zhang and Associate Researcher Simone Lamon are co-first authors of the paper; Associate Researcher Simone Lamon, Professor Liying Liu, Professor Qiming Zhang, and Academician Min Gu are co-corresponding authors; the University of Shanghai for Science and Technology is the first affiliation of the paper, and Fudan University and Southeast University are collaborating institutions.
IT之家附论文链接： https://www.nature.com/articles/s41566-026-02004-z