It's been a long time coming: more than 40 years have passed since holographic data storage was first put forth by a researcher at the now-defunct Polaroid Corporation. But with a new product on the market, holographic data storage seems to have finally arrived. InPhase Technologies (Longmont, CO) is now taking orders for their holographic drives: 300 GB per disk initially, with data transfer rates of 20 MB/s.
The product, called Tapestry, is aimed initially at customers who have large amounts of data that need to be permanently archived. According to Kevin Curtis, InPhase's CTO and founder, the fact that the technology will finally be sold and used is key. "It is critical to get units into customer hands to prove the viability of the technology. After that the focus shifts to improvements in transfer rates and capacity, and rewritable formats."
The basic idea for holographic storage hasn't changed: coherent reference and signal beams interfere inside a photosensitive medium (see Figure 1). Once the interference pattern is recorded, reversing the reference through it will reconstruct the signal beam (see Figure 2), which is acquired by a camera. The technology allows particularly dense storage because the holographic pattern is both very high resolution (with subwavelength feature sizes) and three dimensional: the data is stored in the volume of a material, not just on its surface or a few layers like a conventional optical disk.
What is new is that InPhase has fit what used to take an entire optical table into a package a bit longer than a shoe box. More crucially, its researchers have made significant advances in the technology's most critical area: materials and getting the most out of them.
Until recently, materials were the single greatest obstacle to progress. From the time the first holograms were recorded in the 1960s and through the late 1980s, silver halide materials (high-resolution photographic emulsions) were the medium of choice for most holographers. Though sensitive, they had the disadvantage of being thin, making them unsuitable for storage.
Lithium niobate -- a photorefractive crystal that has the advantage of being a bulk material -- became the main choice for holographic storage experimentation. However, researchers recognized quite early on that this was essentially a placeholder: it was not sensitive enough to be a commercial material, nor did it have a high enough capacity. It became increasingly clear that photopolymers (light-reactive plastics) would be the media of the future.
The main manufacturers of photopolymers for holography were Polaroid in the 1980s and then DuPont in the 1990s. Bell Labs, however, was the first corporation to focus on materials for storage. In 1999 the company announced it had developed a material with an M/# (pronounced "em number," a measure of storage capacity) of 42 in a 1-mm-thick layer: lithium niobate has an M/# of 1 to 1.5 in a 1-cm-thick crystal.
The new photopolymer consisted of two components: a high-index photopolymerizable monomer and a low-index matrix precursor. After deposition, the precursor would be polymerized, creating cross-linked structure within which the monomers could remain suspended, unreacted, until activated by light. When exposed they would link together, forming high-refractive-index features within the low-refractive-index host. The combined material had the advantage not only of high index modulation, but also of high clarity, low scattering, and high stability: the material (and therefore the recorded interference fringes) would not shrink or swell significantly after recording.
According to Curtis, the two-part chemistry approach was a crucial Bell Labs invention, "[It] allows us to optimize the material and the manufacturing of the media independently." But that would not have been enough to make commercialization possible. "The second major invention was the Zerowave manufacturing process that allows us to use inexpensive plastic and yield extremely flat media [about a quarter wavelength] over the entire disk."
Two other companies currently trying to develop holographic storage products, Daewoo (Korea) and Optware (Japan), both use the material.
Engineering for Capacity
Bell Labs -- by then part of Lucent -- launched InPhase Technologies to exploit the technology in 2000, and researchers started the work of optimization. In particular, they had to decide the best way of multiplexing holograms -- placing several holograms within the same volume of material -- to avoid several practical problems of holography. For instance, angular multiplexing, where each hologram is recorded with the reference beam at a different angle, was an efficient way to store data in theory. The interference fringes for each recording would form on differently angled planes, so many holograms could exist in the same volume.
However, the technique had serious problems in practice. For example, each hologram recorded in the same volume had a lower diffraction efficiency (efficiency of diverting a reconstruction beam into the desired image) than the last because some of the monomers that polymerized for the first would not be available for the second, and so on. This meant that the exposure time had to be different for each hologram to compensate for recording order.
Another problem was that the reference/reconstruction beam for an angled hologram at one location tended to partly overwrite/read-out the corresponding hologram at the neighboring location: this required either large separation between locations, lowering the information density, or caused severe reconstruction problems.
Figure 1: Data is displayed on a screen and picked up by the holographic signal beam. This interferes with the reference beam and the hologram is recorded.
Figure 2: The reconstructing beam comes into a given location and at the specified angle, retrieving the required data page.
The Bell Labs team developed a new technique called polytopic multiplexing to get over this problem. Here, both the recording and reconstruction of the hologram is restricted to the width of the object beam waist using a blocking filter. This prevents "leakage" into adjacent regions. Other techniques developed included a moving phase mask to make the structure of the multiple-exposure recording more uniform, and using slightly different reference angles for neighboring books as another means of preventing interference between the two.
Finally, the team developed a technique invented at the IBM Almaden Research Center in California -- the main focus of holographic data storage research through the 1990s -- to solve a huge problem: aligning emerging data pages with the detector. With a camera less than twice the size (in terms of pixels) as the data pages, plus optical markings for rough alignment, they are now able to interpret the complex data images despite rotation, warping, and non-uniform intensity distributions. This is crucial for getting information out reliably.
According to Curtis, Tapestry is just the first of several products they have in mind. "The first generation drive is a write once, read many (or WORM) format. Therefore, we are targeting digital archive applications that place a high value on the inalterability and security of the data for decades-long archive storage," says Curtis.
These include video and other rich media applications and business and medical data archiving. "In addition," he says, "with industry partners we have development programs going for very small holographic read only memory (HROM) for content distribution, and a consumer version of our recordable drive for mainstream storage and distribution. The HROM is a unique format -- think 50 GB on a postage stamp -- and the recordable is in a more standard disk format."
More feature articles for optics and photonics professionals are free to SPIE members, after log-in, in SPIE Professional magazine.