• Chinese Journal of Lasers
  • Vol. 50, Issue 18, 1813003 (2023)
Dejiao Hu1、2、**, Zhen Wang2, Tiewei Luo2, Qifan Xia1, Diqin Zhang3, Yuxia Zhao3, Mu Zheng2, Xiangping Li1, and Yaoyu Cao、*
Author Affiliations
  • 1Guangdong Provincial Key Laboratory of Optical Fiber Sensing and Communications, Institute of Photonics Technology, Jinan University, Guangzhou 510632, Guangdong, China
  • 2Amethystum Storage Technology Co., Ltd., Meizhou 514000, Guangdong, China
  • 3Technical Institute of Physics and Chemistry, Chinese Academy of Science, Beijing 100190, China
  • show less
    DOI: 10.3788/CJL230870 Cite this Article Set citation alerts
    Dejiao Hu, Zhen Wang, Tiewei Luo, Qifan Xia, Diqin Zhang, Yuxia Zhao, Mu Zheng, Xiangping Li, Yaoyu Cao. A Review of High-Capacity Hologram Data Storage for Big Data Applications[J]. Chinese Journal of Lasers, 2023, 50(18): 1813003 Copy Citation Text show less

    Abstract

    Significance

    With the explosive growth of global data in the information age, the amount of data to be stored is increasing exponentially. The secure, stable, and long-term storage of vast amount of data generated by the internet in various scenarios has become the foundation for the sustainable development of future information technology. The capacity of a single hard disk drive (HDD) can be as large as 20 TB, which plays an important role in many applications. An HDD is a typical magnetic data storage technique. However, the rapidly developing technology of solid-state disks (SSD), which employ electronic circuits to store and retrieve data, is superior in terms of a high transfer rate and favors applications requiring frequent data exchange. However, both magnetic- and electronic-circuit-based data storage have the disadvantages of high energy consumption, short lifetime, low safety, and poor stability. These issues become increasingly obvious for the storage requirement of “cold data,” which have large size and needs to be accessed less frequently. Optical data storage (ODS) has the advantages of low energy consumption, long lifetime (as long as 50 years), high safety, and high stability. Consequently, the utilization of the ODS technology in the field of large data storage has attracted considerable attention.

    The conventional ODS technique employs a focused laser spot to record and read data, and the data density depends on the lateral size of the laser spot. Based on this concept, the ODS industry developed compact discs (CD), digital video discs (DVD), blue-ray discs (BD), and archival discs (AD). In 2019, Sony released a second-generation AD with a capacity of 500 GB per disc. The minimum size of the data bit on the disc was approximately 68 nm, which is close to the physical limit of this technique. Further increase in the capacity requires an increase in the number of recording layers, which causes serious interlayer interference and limits the maximum number of layers. Inorganic glass was used as the recording medium to further increase the archive lifetime. Microsoft reported their silica-based compact discs in 2019, where data was recorded inside silica disc in the format of multi-layered 3-dimensional (3D) data bit called “voxel.” This technique is based on the interaction between a focused femtosecond laser and quartz that generates micro-gratings for data encoding. However, it exhibits drawbacks, such as a relatively low transfer rate, large physical size, and high cost.

    Holographic data storage (HDS) records data in photosensitive media in the form of holograms by utilizing the coherence of the laser beam, where the recorded data are represented by the amplitude, phase, or polarization of the light wave. In a parallel storage scheme, the data are scribed into 3D volume space, which makes full use of the thickness of the media and increases the storage capacity. The data is recorded and read as a 2-dimensional (2D) coded data page, which enables a high transfer rate of up to several hundred million bits per second (GBPS). Its predominant performance potential makes HDS the best candidate for next-generation optical data storage technology.

    Progress

    The foundation of the HDS technology lies in its multiplexing recording method, which guarantees a large storage capacity. HDS techniques can be categorized into two classes according to the relationship between the reference and signal beams: coaxial and off-axis. Coaxial HDS has an axially symmetric record and read beam, where the reference and signal beams share the same optical axis. Whereas, in an off-axis HDS system, the reference and signal beams are directed onto the media at a nonzero angle of incidence. Multiplexing techniques, such as orthogonal coding, phase-coding, and shift, can be employed in coaxial HDS systems. Commonly used multiplexing techniques for off-axis HDS systems include angular, peristrophic, and spherical reference wave-shift multiplexing. The HDS encompasses the practical implementation of various techniques, including optical, servo, signal processing, and coding systems. In 2000, Stanford University demonstrated a coaxial HDS prototype with a transfer rate of 1 GBPS using a pulsed laser as the light source. Subsequently, Sony designed an image-stabilizing technique for a coaxial holographic disk system to increase the transfer rate using a low-peak-power continuous laser. Around 2006, Optware and Sony demonstrated their prototype with a capacity of about 200 GB to 1 TB. During the same period, InPhase released its HDS drive, Tapestry, which employed an angular multiplexing method to increase data density. It has a capacity of 300 GB and a transfer rate of 20 MBPS. After 2017, Tokyo University of Science and Amethystum Storage Technology Co,. Ltd. developed an HDS system based on spherical reference wave cross-shift multiplexing, which combines shift and peristrophic multiplexing techniques. The shift interval can be as small as 5 μm, which enables a high multiplexing density (Fig. 3). Furthermore, the prototype was provided with a single-arm architecture, in which the reference and signal beams pass through the same relay lens system and split into two beams immediately before the recording media. This design reduces the path along which the reference and signal beams are separated, leading to a compact size and stable performance (Fig. 4).

    Another core foundation for HDS is the recording media. Photopolymer materials are regarded as the best candidates to date. Dynamic range, photosensitivity, shrinkage rate, scattering, and lifetime are used to characterize the performance of the media. The dynamic range is the largest amplitude of refractive index modulation that the media can achieve, restricting the maximum multiplexing density from the media aspect. A larger dynamic range allows for a higher multiplexing density and capacity. However, a large dynamic range is always accompanied by a large shrinkage rate, which is induced by the polymerization of monomers and can result in the complete deterioration of the hologram. To guarantee a high signal-to-noise ratio (SNR) for the readout data page, the shrinkage rate is typically required to be less than 0.1%. Photosensitivity is the main factor influencing the recording rate. Larger photosensitivity reduces the requirement of high laser power and complexity of the mechanism; however, it may decrease the lifetime owing to high chemical activity. Materials with superior comprehensive performance have been developed by Aprilis, InPhase, Akonia Holographics, and Mitsubishi Chemical (Table 1).

    Conclusions and Prospects

    HDS is a prominent candidate for next-generation optical data storage owing to its high storage density and potentially high transfer rate. The main obstacle for HDS to enter practical applications is the recording medium. Intensive studies on photochemical kinetics, particularly on the interaction between two multiplexed holograms during the recording process, are required. In addition to its applications in the field of big data, HDS has great potential in the field of artificial intelligence. Specifically, it offers promising prospects for storing the weight matrices of neural networks in holographic media, thereby enabling fast parallel calculations.

    Dejiao Hu, Zhen Wang, Tiewei Luo, Qifan Xia, Diqin Zhang, Yuxia Zhao, Mu Zheng, Xiangping Li, Yaoyu Cao. A Review of High-Capacity Hologram Data Storage for Big Data Applications[J]. Chinese Journal of Lasers, 2023, 50(18): 1813003
    Download Citation