48 HOLOGRAPHY
means of adjusting the power level of the readout beam may
be needed. '
The specific characteristics required of these light sources depend on other system parameters, such as the sensitivities anq efficiencies of the data composer, the beam deflecting devices, the hologram material and the output detector. If wavelength sensitivity is used for holographic storage on a thick medium, an electrically tunable laser would be desir- able so that the laser lines be sufficiently spaced in wave- length. The degree of temporal and spatial coherence required to permit full use of the space bandwidth product and dynamic range available from any storage material must also be considered. Several other factors involved in deter- mining a source suitable for the optical system, include the electrical (or chemical) to optical conversion efficiency, and wavelength and intensity stability.
DATA STORAGE 49
How a holographic erasable read/write memory works
Materials for Data Storage
One of the major considerations in developing any type of holographic memory is the recording or data storage materi- als. These should have the following desired features: high resolution; fast recording speed; minimum processing; high stability and reversibility; easy erasure and/or overwrite;
broad spectral response; high angular sensitivity and uni- formity.
Resolution governs the packing density and in turn the physical size of the memory. Recording speed and processing requirements determine storage rate. Stability is needed for any extended data stora'ge; this encompasses not only dimen- sional factors but also sensitivity to light exposure and envi- ronmental conditions. Reversibility governs the number of times the material may be recycled. The capability of selective erasure and overwrite governs applicability to read-write memories (as opposed to read-only memories on which no
50 HOLOGRAPHY
addition or alteration is possible). In some cases, only des- tructive readout may be feasible, thus requiring a rewrite cycle after readout. The spectral response should be suffi- ciently wide to allow the use of various frequency laser units for possible spectral multiplexing. A high angular sensitivity enables the storage of multiple sub arrays over the same location. Finally, high uniformity of the material's sensitivity is required for reliability and lower error rates.
Many materials have been studied for the purpose. Photo- graphic emulsions can obviously be used only for read-only memories, although the new arsenic trisulphide can be erased. For uses that require frequent recall of extensive tabulated data, this may offer the best system. Photopia sties or manganese-bismuth films have also been recommended for read and write purposes.
Some' alternative materials are: lithium niobate crystals for storage of upto 1000 holograms per crystal; alkali halide crystals for 1018bit storage; strontium-barium niobate crys- tals; iron-doped lithium niobate crystals and arsenic trisul- phide and other inorganic photochromics. All these materials exhibit refractive index changes in response to strong optical fields.
The required laser output power levels and wavelengths for read-in and read-out will be determined partly by the sensitivities of these materials and partly by their holo- graphic diffraction efficiencies. The phvsical dimensions of the storage medium will be determined primarily by the resolution of the material (space-bandwidth product) and geometrical factors, like thickness. The recording linearity and the light scattering property of the material will affect the signal-to-noise ratio of the readout data.
The useful life of the entire system will be determined partly by the available number of write, read and erase cycles.
The length of time data can be stored in the material will affect
DATA STORAGE 51
the rate at which it must be retrieved and relayed to more permanent ground-based data storage terminals.
Two-dimensional recording materials are simpler than three-dimensional materials as regards analysis of their be- haviour and their utilization in a system. For both reasons, a system based on storage in only two dimensions can be realised more readily.
The storage format in a two-dimensional medium is straight-forward. Each data block is likely to be stored in a Fresnel or Fourier transform hologram. The factors that should be determined for any particular system are the spa- tial extent of each hologram, the spatial-frequency range to be accommodated, and the number of overlapping holo- grams present on any region of the material. Also, the manner of division of the material into blocks corresponding to vari- ous beam deflection positions must be examined. The format is determined by such factors as the dynamic range of the material and the capabilities of the record and deflection systems.
Albert Friesem and Howard Roberts (1970) have investi- gated holographic data storage in a series of experiments in which digital data was recorded (regularly and randomly spaced array of transparent dots in a black background) in both Fourier transform and Fresnel holograms. The holo- grams were recorded on conventional photosensitive emul- sions and dichromated gelatin materials. The emphasis in the experiments was on determining the best format of the digital data, the effects of non linearities due to the recording media and the noise level in the reconstructed image.
They recorded Fresnel holograms with various reference- to-signal beam ratios to determine the effects of non-lineari- ties in the recording material on the reconstructed image. The noise level as well as its form, depends on the number of active elements in the digital data, the duty cycle and the extent and severity of non-linearity. It was observed that the
52 HOLOGRAPHY
noise due to the non-linearity of the recording medium gen- erally decreased with an increase in the number of active elements. When the Fresnel holograms of the digital data are recorded on dichromated gelatin plate, the diffraction effi- ciency of the hologram increases. For the Fourier transform type holograms an improvement in the signal-to-noise ratio was noted for the jittered array. The result is expected because such an array produces a more uniform spectrum, thus re- ducing the effects of the non-linearity of the recording film.
The effect of hologram aperture (a measure of opening to let in the light beams) on the signal-to-noise ratio (SIN) for Fresnel holograms was investigated and it was found that the reconstructed image of a binary signal of 104bits, was best with a large aperture, but acceptable results were achieved with an aperture of only 0.5 mm. Even this aperture is-almost twice the theoretical value found by calculations.
The holograms recorded in three-dimensional media have distinctive properties when compared with conventional pla- nar holograms. These are: The inherent high angular and wavelength sensitivities provide high data storage capabili- ties; and appropriate choice of thickness allows high diffrac- tion efficiencies.
The interference effects are recorded as surfaces within the recording medium, forming, in effect, a three-dimensional grating. The diffraction process from such a structure is analogous to X-ray diffraction from crystals and in this con- text, studies have been made to determine the response of simple hologr~m gratings. These studies provide informa- tion about diffraction efficiency, as well as angular lorienta- tion and wavelength sensitivities of thick holograms.
A variety of three-dimensional recording media have been identified, including thick photographic emulsions, lithh,lm niobate crystals, and photochromic materials. The last two offer particular advantages in data storage because of their relatively high thickness and excellent resolution capabilities.
DATA STORAGE 53
For example, a one em cube of photochroffiic glass or lithium niobate crystal with resolution elements of 1/3000 rom con- tains 27 x1012resolution cells. But, the major problem is how to effectively utilize all or most of these cells.
The wavelength or the direction of incidence of the readout beam may be varied with a corresponding change in diffrac- tion intensity from a hologram. Both these parameters, there- fore, can be effectively used to construct a hologram that stores a great multiplicity of images, each stored uniformly thr<mghout the recording medium and separately recon- structable .