Abstract
High-density storage of big data in the information age calls for higher data transmission rates. Holographic data storage technology, based on the principles of three-dimensional volume holographic recording and two-dimensional planar transmission, delivers an extremely high data transmission rate. Given the need for encoding two-dimensional data pages, spatial light modulators (SLMs) are employed for information encoding, and the modulation region of SLMs has become a key limiting factor for the data transmission rate of holographic data storage systems. However, to overcome the limitation that the limited imaging area of SLMs in holographic data storage fails to fully exploit the potential of this technology, this paper proposes an ultra-high-speed holographic data storage system based on extending data page size. The signal beams modulated by two digital micromirror devices (DMDs) are stitched and aligned through an optical system to achieve an expanded data page size. The reference beam is modulated using a mask to conserve the limited pixel resources of DMDs. Coupled with the 5:16 amplitude data encoding rule, the proposed system achieves an ultra-high data transmission rate of 20.06 Gb/s.
Subject terms: Optics and photonics, Physics
Introduction
Holography was first proposed by British physicist Dennis Gabor to improve the resolution of electron microscopes1. Endowed with inherent characteristics including three-dimensional stereoscopic properties and high information capacity, it has been widely applied across diverse fields over decades of development, such as information display2, biomedical engineering3, and holographic data storage4. As the widespread application of artificial intelligence technology in human society has triggered explosive information growth, it has exerted enormous pressure on efficient data storage5. Current societal data applications have unleashed an increasing volume of cold data, which has become “hotter” in the data-driven AI era and imposed higher requirements on the transmission rate of data storage systems. Unlike traditional bit-by-bit storage methods, holographic data storage leverages volume holography principles to record two-dimensional encoded data pages in three-dimensional holographic storage media6–8. It achieves higher data transmission rates by virtue of two-dimensional data page transmission for data transfer. As stored data must be encoded into two-dimensional encoded data pages9,10, electro-optical conversion via light field modulation devices is essential. Common spatial light modulators (SLMs) currently include digital micromirror devices (DMDs)11 and liquid crystal on silicon (LCOS) spatial light modulators12.
The LCOS-SLM achieves multi-level signal modulation by applying different voltages to the liquid crystal, thus realizing high information modulation for each pixel13–16. However, limited by the response rate of liquid crystal molecules, which is usually on the millisecond order, it is not suitable for ultra-high-speed data page encoding. In contrast, DMDs—based on semiconductor technology—enable high-speed, high-efficiency, and reliable spatial light modulation. Boasting advantages of high resolution, high contrast, fast response time, and high reliability, they can easily achieve high refresh rates of tens of kilohertz, finding wide applications in digital projection17–19, optical communication20,21, and various imaging fields22–24. They further offer reliable hardware support for high-speed data modulation in holographic data storage, making DMDs the preferred choice for light field modulation in most current high-speed modulated holographic data storage systems25–27.
However, constrained by the fabrication process of DMDs, the size of individual micromirrors cannot be further scaled down. Further miniaturization will face inherent challenges associated with physical resolution, imaging quality and fabrication complexity. Consequently, the number of addressable micromirrors is inherently limited within the confined footprint of the display chip. This results in stringent constraints on the effective DMD area allocated for signal modulation. This limitation curtails the volume of data modifiable per frame and thereby impedes the data writing rate of the system. Furthermore, governed by its intrinsic modulation mechanism, a DMD attains its maximum modulation frame rate exclusively when encoding binary information. This necessitates the implementation of binary ON–OFF modulation in holographic data storage configurations, which in turn imposes an upper limit on the data capacity of a single DMD-modulated data page. Meanwhile, the finite imaging area of the DMD must be shared for the concurrent modulation of the reference beam. This further compresses the spatial resources dedicated to signal beam modulation. Collectively, these constraints form a critical bottleneck that hinders the enhancement of data writing speed in holographic data storage systems.
This paper proposes an ultra-high-speed holographic data storage system based on extending data page size. Without considering the response of the holographic storage media, two DMDs are used to modulate the signal beam. The upper and lower parts of the modulated signal light are processed by an optical system to stitch the modulated light fields, achieving a larger data modulation area. Modulation of the reference beam adopts mask modulation, which does not occupy the limited modulation area of the DMDs. It is modulated with the signal light part into collinear holography via an optical system. Meanwhile, in data encoding, the 3:16 encoding scheme9 is upgraded to a 5:16 encoding scheme. The amount of information encoded per encoding unit is increased from 8-bit to 12-bit, further enhancing the information modulation capacity per pixel. Experimental results show that the ultra-high-speed collinear holographic data storage system based on extending data page size can achieve ultra-high-speed data transmission rate at 20.06 Gb/s. This provides reliable support for holographic data storage systems in the current era of “hotter” data.
Methods
Without considering the response of the holographic storage media, the data writing speed of a holographic data storage system depends on two factors: the amount of modifiable information per frame of the SLM and the refresh rate of the SLM. The refresh rate of SLM is a fixed value limited by the device’s modulation principle, and it cannot be exceeded in practical storage applications. Thus, it is determined that the only way to improve the system’s data writing speed is to increase the amount of data modulated per frame. In a collinear holographic data storage system, the advantage of the high refresh rate of the DMD is fully leveraged. Therefore, the amount of modifiable information per frame of the DMD determines the data writing speed of the system.
According to the two-beam interference principle of holography, an encoded data page can be divided into two parts: the signal beam and the reference beam. Traditional collinear holographic data storage uses the same DMD for loading both beams. To ensure equal energy between the signal beam and the reference beam, it is required that the number of pixels used for reference beam modulation is equal to that for signal beam modulation. However, the reference beam is arranged in a radial pattern in the outer ring, with radial lines maintained at a 3° interval28,29. This results in a large number of pixels being required for reference beam loading, leaving only a small portion of the already limited DMD pixels for signal beam modulation. This greatly restricts the quantity of modulable information per frame. Therefore, this paper proposes a data encoding scheme as shown in Fig. 1.
Fig. 1.
Schematic diagram of dual-DMD modulation data encoding scheme.
To make full use of the limited imaging area, the reference beam is loaded using a mask and no longer occupies the limited modulable pixels of the DMD. The signal light part is split into two DMDs for loading. Each DMD has a resolution of 1024 × 768 pixels, so the area available for signal light modulation is 1024 × 1536 pixels. Considering that the effective aperture of optical devices is usually circular, the signal light area is thus encoded as an approximately circular region as much as possible. At this point, each DMD imaging area loads 1002 data subpages, so each encoded data page contains 2004 data subpages.
In terms of data encoding rules, a 5:16 amplitude encoding format is adopted. That is, in each encoding unit composed of 4 × 4 pixels, only 5 pixels are in the “ON” bright state, while the remaining pixels are in the “OFF” dark state. In this encoding mode, 4096 of these cases are selected to represent 12-bit data, meaning each encoding unit can carry 12-bit data. After user data is converted into a bitstream, it can be encoded into encoding units according to the encoding rules. Each data subpage contains 32 encoding units and includes a 4 × 4-pixel positioning mark at its center, which is used for positioning the data subpage and pattern correction. In the experiment, the bit error rate (BER) and signal-to-noise ratio (SNR) are used to evaluate the encoded data pages imaged by the optical storage system. The calculation formulas of the BER and SNR shown in Eqs. (1) and (2), where 
and 
represent the average value of ON and OFF pixels, respectively.
and 
denote the variances of ON and OFF pixels8.
 |
1 |
 |
2 |
The data capacity of a single encoded data page in the holographic data storage system can be given by Eq. (3), where 
denotes the amount of information contained in each encoded data page, 
denotes the number of data sub-pages contained in each encoded data page, 
denotes the number of encoding units in each data sub-page, and 
denotes the amount of information contained in each encoding unit.
 |
3 |
In the encoding rules proposed in this paper, each encoded data page contains 2004 data sub-pages, each data sub-page includes 32 encoding units, and each encoding unit can encode 12 bits of data. Through calculation, it can be obtained that the single-page data capacity of the extended encoded data page can reach 769,536 bits. In contrast, the traditional collinear holographic data storage system adopts a single DMD for modulation. It employs the 3:16 amplitude encoding scheme and requires the same DMD to modulate both the signal beam and the reference beam simultaneously8. Therefore, each encoded data page contains 51 data sub-pages, each data sub-page consists of 32 encoding units, and each encoding unit encodes 8 bits of data, with the single-page data capacity of each encoded data page thus reaching 13,056 bits.
In contrast, compared with the traditional collinear holographic data storage system using single-DMD modulation, the collinear holographic data storage system based on dual-DMD modulation adopts 5:16 amplitude encoding. It not only retains the advantage of the high refresh rate of DMD but also gives full play to the advantages of holographic data storage, thereby improving the data storage transfer rate. The combination of dual-DMD modulation enables ultra-high-speed data transmission. The amount of data that can be modulated per frame by the proposed system in this paper has increased by approximately 59 times compared with the traditional system.
Signal beam modulation results
Based on the above-mentioned data page encoding scheme, a dual-DMD modulated holographic data storage optical path diagram as shown in Fig. 2 is designed in the experiment. First, after beam expansion, the light beam is split into two beams: one is used for modulating the signal beam, and the other for modulating the reference beam. The modulation of the signal beam is split into two parts, which are incident on the two DMDs respectively and loaded with the corresponding encoded data pages. After modulation, the upper and lower parts of the encoded data pages are transmitted through the 4F system and combined via beam splitter (BS) 2 to achieve imaging plane stitching of the dual DMDs. For the modulation of the reference beam, the custom-made mask is directly transmitted through the 4F system and combined with the stitched signal beam via BS3, thus realizing modulation of the entire data page. Finally, the light beam is scaled by the 4F system composed of lenses L11 and L12, ensuring that the entire modulated data page can pass through the entrance pupil of the objective lens for recording and reading in the storage media.
Fig. 2.
Schematic diagram of the ultra-high-speed collinear holographic data storage device based on dual-DMD modulation.
Typically, DMD display chips have two mounting methods: 0° and inclined 45°. As shown in Fig. 3 are the different mounting angles of DMD chips and light incidence angles. In the experiment, to facilitate observing the stitching of information pages between the two DMDs during optical path adjustment, a DMD chip mounted at 0° is selected. However, as shown in Fig. 3b, when the chip is mounted at 0°, there is a height difference between the incident light direction and the emergent light direction. Thus, the optical axis of the incident beam and that of the emergent beam are not at the same height. Moreover, affected by the manufacturing process, the tilt angles of micromirrors on the DMD are not all the specific ± 12°. This means that after determining the height of the emergent optical axis, the angle of the mirror and the height of the incident light cannot be directly inferred based on the principle of optical path reversibility. This makes the collimation of each DMD require extremely high precision, and the stitching alignment of the two DMDs poses great challenges.
Fig. 3.
Different mounting angles of DMD chips and light incidence angles. (a) shows the mounting angles of the chip. (b) shows the directions of the incident beam and emergent beam when mounted at 0°.
In the experiment, for the collimation of a single DMD, the dual-beam alignment method is adopted. Figure 4 shows the schematic diagram of the dual-beam alignment method. Light beam 1 passes through aperture 3 and aperture 4, which is used to calibrate the DMD display chip to be perpendicular to the optical axis. The DMD uses an x, y, z three-axis translation stage and α, β two-axis goniometers to ensure that the beam is incident on and emergent from the exact center of the chip. Light beam 2 is used to calibrate the angle of mirror 2. Since the tilt angles of micromirrors on the DMD are unknown, after light beam 1 is incident, according to the principle of optical path reversibility, when the reflected beam reflected to mirror 2 coincides with light beam 2, it can be considered that the incident angle of light beam 2 at this time is the deflection angle of the micromirrors. At this point, adjust the angle of mirror 2 to ensure that the light spots of the two beams passing through the four apertures all go through the exact center. Then, the DMD can be considered collimated in the optical path.
Fig. 4.
Schematic diagram of the dual-beam alignment method.
To ensure that the tilt angles of micromirrors on the DMD meet the requirements for holographic data storage, according to the requirements of references20, after collimating the DMD, checkerboard patterns and all-on patterns are loaded onto it to test the brightness uniformity of light spots in the diffraction pattern. Figure 5 shows the diffraction spots under different patterns. It can be seen from the figure that three DMDs are tested respectively, distinguished by three dashed boxes in blue, orange and green. When loading the all-on pattern, only the diffraction spots in the blue box are four light spots with uniform brightness. Similarly, when loading the checkerboard pattern, the bright spots displayed by the detector become five bright points. Except for the large central spot, the four surrounding bright spots have uniform brightness and are symmetrically distributed. Thus, it can be considered that only the deflection angles of micromirrors on the DMD in the blue box meet the requirements of holographic data storage. However, it can be seen that the remaining two DMDs exhibit uneven bright spots when loading the two types of images. This situation will lead to uneven brightness during the modulation of encoded patterns in holographic data storage. Some encoded “ON” pixels will diffuse to surrounding “OFF” pixels or be submerged by noise, making it impossible to distinguish between noise and information, which affects the accuracy of data recording and reading.
Fig. 5.
Different diffraction images when DMD load all-on and checkerboard patterns. (a) shows the conditions of different DMDs when loading the all-on pattern. (b) shows the conditions of different DMDs when loading the checkerboard pattern.
After repeating the above steps to collimate the DMDs, select those with suitable micromirror angles, and choose two such DMDs for the construction of the overall optical path. The key consideration in system construction is the alignment and stitching of imaging regions. According to the encoding scheme shown in Fig. 1, the imaging regions of both DMDs are 1008 × 600 pixels, with each DMD pixel size being 13.68 μm. Thus, the size of each DMD’s imaging region is 13,789.44 × 8208 μm. The stitching direction is vertical, so the height difference between the emergent optical axes of the two DMDs should be greater than 8208 μm. Considering that it is difficult to control micron-precision positioning during system construction, the height difference between the two optical axes is set to 8.3 mm, allowing a certain gap at the stitching position, which does not affect subsequent calculations. For the two DMD channels, 4F imaging systems are constructed using lenses (L) L5, L6 and L7, L8 respectively for object-image transmission, and BS2 is used for beam combining and stitching.
Figure 6 shows the schematic diagram of signal beam stitching effect detection. After BS2, the combined signal beam is detected using a charge coupled device (CCD) to observe the confocal status and stitching alignment status of the two signal beams.
Fig. 6.
Schematic diagram of signal beam stitching effect detection.
After repeated adjustments of the heights of the two DMDs and system construction, the imaging experiment results of the dual-DMD modulated signal beam region as shown in Fig. 7 is obtained. It can be seen that the brightness uniformity of the image and the clarity of the encoded patterns within the overall imaging area are relatively satisfactory. From the local enlarged view, it can be observed that there is still a separation of approximately 8 CCD pixels in the vertical direction between the encoded regions of the two DMDs (with the CCD pixel size being 3.45 μm). This is inevitable, and it is difficult to achieve “seamless stitching”. However, the existing gap does not have any impact on subsequent image decoding, nor does it cause waste of the limited imaging area, so it is acceptable. Through calculation, the BER of the signal beam region at this time is 1.56 × 10− 4, and the SNR is 6.58.
Fig. 7.
Imaging experiment results of the dual-DMD modulated signal beam region.
On this basis, we used the system to load different encoded data pages and collected them with a detector, obtaining 100 images of the dual-DMD modulated signal beam region. The BER and SNR of each image were calculated respectively. As shown in Fig. 8 is the distribution of BER and SNR of the 100 dual-DMD modulated signal beam images. It can be seen from Fig. 8a that the distribution of BER is random, with an average BER of 0.014%, which is much lower than the 1% requirement for practical applications5. From Fig. 8b, it can be observed that the SNR is distributed between 4 and 7, with an average SNR of 5.43, indicating that the reliability of the data meets the requirements. From the data, it can be concluded that the modulation of the signal beam meets the requirements.
Fig. 8.
Distribution of BER and SNR for 100 dual-DMD modulated signal beam images. (a) shows the BER distribution of 100 images. (b) shows the SNR distribution of 100 images.
Reference beam modulation results
In holographic data storage, in addition to the modulation of the signal beam, modulation of the reference beam is also required. Figure 9 shows the schematic diagram of the structural parameters of the custom reference beam mask. According to Texas Instrument’s product manual, as shown in Fig. 9a, the size of each micromirror is 13.68 μm. This size includes the reflective areas of the micromirror and the gaps between the two reflective areas. The DMD surface has hundreds of thousands of micromirrors, so it can be regarded as a grating in holographic data storage. At this point, the grating constant can be considered as P = 13.68 μm. Therefore, to keep the grating structures of the signal beam and the reference beam consistent, the influence of this structure is also considered in the custom reference beam mask.
Fig. 9.
Schematic diagram of the structural parameters of the custom reference beam mask. (a) shows the micro-mirror structure of the DMD. (b) shows the overall size structure of the mask. (c) shows the partial enlarged view of the mask with a ratio of 1000:1.
In practice, when customizing the mask, parameters of the coating process need to be considered. The reference beam pattern consists of “bright pixels”, so a negative mask is adopted in mask design. The modulated pattern area is transmissive, while the remaining area is chromium-plated and opaque. Under this condition, parallel light can be loaded with the designed reference beam pattern when passing through the reference beam mask. Figure 9b presents the overall structure. The inner ring with a diameter of 26.4 mm is the coated area, which is opaque and used for subsequent beam combining with the signal beam. The width of the outer ring of the reference beam is 17.6 mm, and the total width of the entire reference beam is 4.4 mm. The angle of each radial line is 3°. Figure 9c is a partial enlarged view at a ratio of 1000:1.
Considering that the minimum line width of the coating process is 1 μm and the processing error is 0.1 μm, the overall grating constant needs to be consistent with that of the DMD. Therefore, the transmissive area of the coating is designed as a rectangular region of 12.68 μm, with a spacing of 1 μm between two transmissive regions. Regarding the coating thickness, affected by the line width precision, higher line width precision results in a thinner chromium layer. This leads to the optical density (OD) value of the coated part of the entire device being only 3, meaning that the intensity of the transmitted light is 1/1000 of the incident light intensity.
Considering that the pixel size of the CCD in the actual optical path cannot detect a precision of 0.1 μm, the custom-fabricated mask was inspected using a Nikon-E100 microscope. Figure 10 shows the microscopic inspection images of the custom mask. Figure 10a shows the transmissive region of the reference beam. It can be seen that the width of a single transmissive region marked in the figure is 12.635 μm, and the total with the spacing is 13.613 μm, indicating that the fabrication precision meets the design requirements. Figure 10b shows the angle between two radial lines. Test point 1 is 3.04° and test point 2 is 3.046°, which are consistent with the design requirement of 3°. The OD value is defined as the common logarithm of the ratio of incident light intensity to transmitted light intensity. To test the OD value of the coating, when the incident light intensity is 4 mW, the measured transmitted light intensity is 2.12 μW. The OD value of the coating is calculated to be 3.17, which meets the design requirement of OD value ≥ 3.
Fig. 10.
Microscopic inspection images of the custom mask. (a) shows the dimension marked view of the partial light-transmitting area. (b) shows the angle marked view of the overall radiation line.
Experimental measurements show that the various parameters of the custom reference beam mask meet the design requirements and can satisfy the experimental requirements. In the experiment, according to the dual-DMD modulated holographic data storage optical path diagram shown in Fig. 2, the reference beam mask performs object-image transmission through the 4F system constructed by lenses L3 and L4. The reference beam modulated by the reference beam mask and the signal beam modulated by the dual-DMD are combined via beam splitter BS3, and the optical axes of the two beams are adjusted to coincide, thus realizing the ultra-high-speed collinear holographic data storage system based on dual-DMD modulation.
System test results
To test the data read–write rate of the system, the CCD in Fig. 2 is replaced with a high-speed camera. As shown in Table 1 are the relevant parameters of the high-speed camera used in the experiment. The Revealer -XJ1200 high-speed camera is used in the experiment, which has a resolution of 1280 × 1024, a pixel size of 14.6 μm, and a frame rate of 13,500 FPS at full frame. Since the frame rate of the DMD is 27,995 Hz, a resolution of 128 × 128 pixels are selected for the test, and the frame rate of the high-speed camera in this case is 100,000 FPS.
Table 1.
Relevant parameters of the high-speed camera used in the experiment.
| Model | Revealer -XJ1200 | |
|---|---|---|
| Pixel size | 14.6μmx14.6 μm | |
| Resolution | 1280*1024 | |
| Frame rate | Resolution | Frame rate |
| 1280 × 1024 | 13,500 FPS | |
| 1280 × 512 | 27,000 FPS | |
| 640 × 480 | 28,500 FPS | |
| 128 × 128 | 100,000 FPS | |
| Shutter | Global Shutter | |
| Minimum exposure time | 1us | |
| Maximum memory | 256 GB | |
To avoid the interference of the holographic storage media response on the data transmission rate test of the optical system, a mirror was used instead of the holographic storage media in the test experiment, so as to obtain the data transmission rate of the collinear holographic data storage system during the data writing process. The DMD-modulated signal beam is directly returned to the high-speed camera via the mirror for detection to test the data writing rate of the system. The test images loaded on the DMD are as shown in Fig. 11. The image size is 1024 × 768 pixels, and the central 128 × 128 pixels are marked with numbers 1–30,000 respectively, which are used to test the actual frame rate of the system.
Fig. 11.
Actual frame rate test images loaded on the DMD.
In the test, the signal output of the DMD is connected to the external trigger input of the high-speed camera. Each time the DMD switches one image, the high-speed camera receives a trigger signal and captures one frame of image. Figure 12 shows the experimental results of the tested system frame rate. This interface is the operation interface of the high-speed camera. As can be seen from the exported data, the operating frame rate of the high-speed camera is 30,303 FPS, and the actual detection duration is 923.8 ms. Within the 923.8 ms detection duration, the total number of frames captured by the detector is 27,995. This means that the DMD of the system can work accurately at 27,995 Hz, which is consistent with the original intention of the overall system design.
Fig. 12.
Experimental results of the tested system frame rate.
Based on the test results of the system frame rate and the aforementioned 5:16 amplitude encoding rule, the single-frame data encoding capacity of the dual-DMD modulated collinear holographic data storage system is 769,536 bits. The refresh rate of the DMD is 27,995 Hz, that is, the DMD can display 27,995 frames of data per second. This indicates that the ultra-high-speed collinear holographic data storage system based on data page extension proposed in this paper can achieve a data transmission rate of 20.06 Gb/s during data writing.
Ultra-high-speed data writing can be achieved. However, the data reading rate of holographic data storage systems is currently largely limited by the frame rate of high-speed cameras and the conversion rate of photoelectric conversion devices. The conversion rate of the photoelectric conversion device. Taking the high-speed camera used in the test system as an example, its frame rate is 13,500 FPS at a full-frame resolution of 1280 × 1024 pixels. Each detection pixel contains 8-bit data. Thus, the data reading speed per second is given by Eq. (4):
 |
4 |
In the formula, Bps denotes the data transmission rate, FPS represents the frame rate of the high-speed camera, R stands for the total number of pixels of the information image, and BD indicates the bit depth per pixel. This means that when the frame rates of all components in the system can be matched, using this high-speed camera for data reading can achieve an ultra-high-speed data transmission rate of 131.84 Gb/s, which is much higher than that of current traditional optical storage.
Discussion
Holographic data storage stores two-dimensional encoded data pages in three-dimensional holographic storage media, featuring advantages of high transfer rate and high storage density. However, limited by the size of a single DMD pixel, the imaging area available for data modulation is limited. Meanwhile, traditional collinear holographic data storage uses the same DMD to load both the signal beam and reference beam. As a result, a large number of pixels cannot be used for information modulation, leading to a significant waste of limited imaging pixels and preventing the high-rate advantage of holographic data storage from being fully utilized. This paper proposes an ultra-high-speed holographic data storage system based on extending data page size. In signal beam modulation, precise stitching of the dual-DMD imaging planes is achieved optically, expanding the imaging area for signal beam modulation. Additionally, a custom-made reference beam mask is used to realize the loading of the reference beam, avoiding the waste of limited imaging area. On this basis, combined with the 5:16 amplitude encoding method, the collinear holographic data storage achieves an ultra-high data transmission rate of 20.06 Gb/s during data writing. Compared with existing data storage technologies on the current market, the data read and write rates of this holographic data storage technology are much higher than those of common technologies currently available.
Currently, the advantages of holographic data storage technology are constrained by the performance of photoelectric conversion devices. With the continuous advancement of photoelectric conversion devices in the future, the performance of holographic data storage is expected to be further improved. While maintaining high-density storage, it will be capable of achieving ultra-high-speed data read and write rates, thereby meeting the demands of the big data era where cold data is gradually “turning hot.” Although the ultra-high-speed collinear holographic data storage system based on data page extension is proposed in this paper, realizing ultra-high-speed data storage in holographic storage media still requires considering the response of holographic storage media, as well as the influence of multiple factors such as laser power for data recording, recording exposure methods, servo spatial addressing, crosstalk, and shift multiplexing. Further in-depth research is needed in the future to achieve ultra-high-speed data storage in holographic storage media.
Author contributions
Y.L. and S.K. conceived the idea; H.L. generated the data used in the experiments; Y.L. and S.K. set up the experimental device and conducted the experiments under the discussion and guidance of D.L. and X.T.; D.L., X.X., and Y.L. provided suggestions during the experiment; Y.L., R.Y., and X.T. analyzed the results. All authors reviewed the manuscript.
Funding
This study was funded by National Natural Science Foundation of China (grant number U22A2080).
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the corresponding author upon reasonable request.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yongkun Lin and Shenghui Ke contributed equally to this work.
Contributor Information
Dakui Lin, Email: lindk@fjnu.edu.cn.
Xiaodi Tan, Email: xtan@fjnu.edu.cn.
References
- 1.Gabor, D. A new microscopic principle. Nature161, 777–778 (1948). [DOI] [PubMed] [Google Scholar]
- 2.Liu, Y. C. et al. Dynamic interactive bitwise meta-holography with ultra-high computational and display frame rates. Opto-Electron. Adv.7, 230108 (2024). [Google Scholar]
- 3.Li, Z. S. et al. Deep learning assisted variational Hilbert quantitative phase imaging. Opto-Electron. Sci.2, 220023 (2023). [Google Scholar]
- 4.Lin, X. et al. Optical holographic data storage-the time for new development. Opto-Electron. Eng.46, 180642 (2019). [Google Scholar]
- 5.Zhao, M. et al. A 3D nanoscale optical disk memory with petabit capacity. Nature626, 772–778 (2024). [DOI] [PubMed] [Google Scholar]
- 6.Horimai, H. & Tan, X. Holographic information storage system: Today and future. IEEE43, 943–947 (2007). [Google Scholar]
- 7.Lin, X. et al. Collinear holographic data storage technologies. Opto-Electron. Adv.3, 19000401–19000408 (2020). [Google Scholar]
- 8.Horimai, H., Tan, X. & Li, J. Collinear holography. Appl. Opt.44, 2575–2579 (2005). [DOI] [PubMed] [Google Scholar]
- 9.Gu, H., Cao, L., He, Q. & Jin, G. A two-dimensional constant-weight sparse modulation code for volume holographic data storage. Zhejiang. Univ. Sci. C.05, 430–435 (2005). [Google Scholar]
- 10.Liu, H. et al. Modulation code optimization for holographic data storage. Opt. Lett.50, 4834–4837 (2025). [DOI] [PubMed] [Google Scholar]
- 11.Zhou, J. & Rong-Zhi, W. U. Digital Micromirror Device. Chin. J. Liq. Cryst. Disp.25, 313–314 (2003). [Google Scholar]
- 12.Yang, Y., Forbes, A. & Cao, L. A review of liquid crystal spatial light modulators: Devices and applications. Opto-Electron. Sci.2, 230026 (2023). [Google Scholar]
- 13.Hao, J. et al. Lensless phase retrieval based on deep learning used in holographic data storage. Opt. Lett.46, 4168–4171 (2021). [DOI] [PubMed] [Google Scholar]
- 14.Fan, R. et al. Phase retrieval based on deep learning with bandpass filtering in holographic data storage. Opt. Express.32, 4498–4510 (2024). [DOI] [PubMed] [Google Scholar]
- 15.Zheng, S. et al. Orthogonal matrix of polarization combinations: Concept and application to multichannel holographic recording. Opto-Electron. Adv.7, 230180 (2024). [Google Scholar]
- 16.Hao, J. et al. Lensless complex amplitude demodulation based on deep learning in holographic data storage. Opto-Electron. Adv.6, 220157 (2023). [Google Scholar]
- 17.Zhang, C., Zhang, D. & Bian, Z. Dynamic full-color digital holographic 3D display on single DMD. Opto-Electron. Adv.4, 200049 (2021). [Google Scholar]
- 18.Velez-Zea, A., Hoyos-Peláez, C. A. & Barrera-Ramírez, J. F. High-quality binary amplitude hologram generation for digital micromirror device based holographic display. Opt. Lasers Eng.10.1016/j.optlaseng.2025.108994 (2025). [Google Scholar]
- 19.Liang, Q. et al. High-fidelity tissue super-resolution imaging achieved with confocal2 spinning-disk image scanning microscopy. Light Sci. Appl.14, 260 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Li, N. et al. Photonic terahertz phased array via selective excitation of nonlinear Pancharatnam-Berry elements. Nat. Commun.16, 8159 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhao, S. et al. Generation of structured beams with optional turbulence levels using a Digital Micromirror Device. Opt. Commun.529, 129084 (2023). [Google Scholar]
- 22.Dou, K. et al. Off-axis multi-wavelength dispersion controlling metalens for multi-color imaging. Opto-Electron. Adv.3, 190005 (2020). [Google Scholar]
- 23.Meng, W. et al. 100 Hertz frame-rate switching three-dimensional orbital angular momentum multiplexing holography via cross convolution. Opto-Electron. Sci.1, 220004 (2022). [Google Scholar]
- 24.He, C. et al. Pluggable multitask diffractive neural networks based on cascaded metasurfaces. Opto-Electron. Adv.7, 230005 (2024). [Google Scholar]
- 25.Lin, Y. et al. Anti-noise performance analysis in amplitude-modulated collinear holographic data storage using deep learning. Opt. Express32, 29666–29677 (2024). [DOI] [PubMed] [Google Scholar]
- 26.Liu, J. et al. Optimal micro-mirror tilt angle and sync mark design for digital micro-mirror device based collinear holographic data storage system. Appl. Opt.56, 4779–4784 (2017). [DOI] [PubMed] [Google Scholar]
- 27.Xu, J. et al. Matching for realizing high-speed reading in holographic data storage system. Opt. Express32, 46259–46272 (2024). [Google Scholar]
- 28.Shimura, T., Ichimura, S., Fujimura, R., Kuroda, K., & Horimai, H. Calculation of the Pixel Spread Function with a Simple Numerical Model for the Collinear Holographic Storage System. Optical Society of America, (2005).
- 29.Shimura, T. et al. Analysis of a collinear holographic storage system: Introduction of pixel spread function. Opt. Lett.31, 1208–1210 (2006). [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the corresponding author upon reasonable request.