Macro-molecular data storage with petabyte/cm^3 density, highly parallel read/write operations, and genuine 3D storage capability

Macro-molecular data storage with petabyte/cm3 density, highly parallel read/write operations, and genuine 3D storage capability Masud Mansuripur and Pramod Khulbe Optical Sciences Center, University of Arizona, Tucson, Arizona 85721, masud@u.arizona.edu [Published in Optical Data Storage 2004, edited by B.V.K. Vijaya Kumar and Hiromichi Kobori; Proceedings of SPIE 5380, pp272-282 (2004). doi: 10.1117/12.562434] Abstract. Digital information can be encoded in the building-block sequence of macro- molecules, such as RNA and single-stranded DNA. Methods of “writing” and “reading” macromolecular strands are currently available, but they are slow and expensive. In an ideal molecular data storage system, routine operations such as write, read, erase, store, and transfer must be done reliably and at high speed within an integrated chip. As a first step toward demonstrating the feasibility of this concept, we report preliminary results of DNA readout experiments conducted in miniaturized chambers that are scalable to even smaller dimensions. We show that translocation of a single-stranded DNA molecule (consisting of 50 adenosine bases followed by 100 cytosine bases) through an ion-channel yields a characteristic signal that is attributable to the 2-segment structure of the molecule. We also examine the dependence of the rate and speed of molecular translocation on the adjustable parameters of the experiment.

1. Introduction. Many of the traditional problems in disk and tape data storage can be overcome if data-blocks were to be released from the confines of a disk (or tape) and allowed to float freely between read/write stations (i.e., heads) and permanent “parking spots.” The heads and parking spots thus become fixed structures within an integrated chip, while the macro-molecular data blocks themselves become the (mobile) storage media. In this scheme, a large number of read/write heads could operate in parallel, the heads and parking spots would be constructed (layer upon layer) in a truly 3-dimensional fashion, and individual nanometer-sized molecules – strung together in a flexible macromolecular chain – would be used to represent the 0’s and 1’s of binary information. In this paper we discuss the potential advantages of this alternative scheme for (secondary) data storage and, to demonstrate the feasibility of the concept, we present results of experiments based on DNA molecules that travel within micro-fluidic chambers.

In principle, the four nucleotides of DNA can be used to represent a 2-bit sequence (A = 00, C = 01, G = 10, T = 11), although practical considerations may impose certain restrictions on the specific sequences that can be used to encode the information. A data storage device built around this concept must have the ability to (i) create macromolecules with any desired sequence of building blocks, i.e., write or encode the digital information into macromolecular strands; (ii) analyze and decode the sequence of a previously created macromolecule, i.e., read the recorded information; (iii) provide an automated and reliable mechanism for transferring the macromolecules between the read station, the write station, and designated locations (parking spots) for storing each such macro-molecule. Although methods of writing and reading macromolecular strands are currently available (e.g., arbitrary sequences of oligonucleotides can be synthesized, and DNA sequences can be deciphered), these methods require large machines and are slow and expensive. In an ideal molecular data storage system, routine operations such as write, read, erase, store, and transfer must be carried out within an integrated chip, reliably and at high speed. 2

2. Device architecture. As a possible alternative to present-day mass data storage devices (e.g., magnetic and optical disks and tapes), we envision a system in which data blocks are encoded into macromolecules constructed from two or more distinct bases, say, x and y; the bases can be strung together in arbitrary order such as xxyxyyxy… xyx to represent binary sequences of user- data (x = 0, y = 1).1,2 The macromolecular data blocks must be created in a write station, transferred to parking spots for temporary storage, and brought to a read station for decoding and readout. The erase operation is as simple as discarding a data block and allocating its parking spot to another macromolecule. (In principle, discarded molecules can be recycled after being broken down to their constituent elements.) The parking spots and read/write stations depicted in Fig. 1, for example, are µ-fluidic chambers connected via µ-channels and µ-valves (not shown) that enable automatic access through an electronic addressing scheme.2 With the dimensions of the various chambers indicated in Fig. 1, one can readily incorporate, on a 1.0 cm2 surface area, a total of 106 parking spots (~ 0.25 cm2), 1000 read/write stations (~ 0.1cm2), and

…(Full text truncated)…