Porous silica films boost DNA data storage capacity with 11-fold yield enhancement

DNA has emerged as a compelling medium for archiving information over centuries, but turning this promise into practical reality requires solving a fundamental engineering problem: how to write DNA sequences onto silicon chips at high density without sacrificing the quality of what gets written. Researchers have now demonstrated a solution using porous silica films that dramatically increases the amount of usable genetic material produced per square centimeter of chip surface.

The challenge lies in a stubborn trade-off. Electrochemical synthesis—a method that uses electrical current to build DNA strands directly on microelectrode arrays—can pack synthesis sites densely onto a chip, much like semiconductor manufacturing does. But packing sites closer together typically means each site produces fewer DNA molecules, a problem that undermines the reliability of data recovery. A team working with microelectrode arrays found that a standard flat silica surface yielded only about 108 picomoles of oligonucleotides per square centimeter. That's not enough for practical data storage systems that need to reliably encode and retrieve information.

The researchers addressed this by coating their microelectrode chips with a porous silica film fabricated through a sacrificial templating approach. They mixed silica nanoparticles with polystyrene microspheres, deposited the mixture onto the chip, then heated it to 400 degrees Celsius. The polystyrene burned away, leaving behind an interconnected network of tiny pores—essentially creating an internal surface area far larger than the chip's flat exterior. The pore structure resembles controlled pore glass used in traditional column-based DNA synthesis, a proven platform for high-yield oligonucleotide production. When oligonucleotides were synthesized on this porous film using standard phosphoramidite chemistry, the yield jumped to 1,821 picomoles per square centimeter, an 11-fold enhancement over the planar baseline.

But yield alone doesn't guarantee utility. The researchers had to verify that the porous architecture didn't introduce errors or allow unwanted chemical reactions to spread across the chip. During electrochemical synthesis, acid is generated at specific electrode sites to remove protective chemical groups from growing DNA strands. In a porous film, acid could theoretically diffuse sideways and cause unintended reactions at neighboring sites. To prevent this, the team arranged electrodes in alternating patterns—anodes and cathodes side by side—so that basic species generated at cathodes would neutralize excess acid and confine the reaction to its intended location. Fluorescence imaging confirmed that this strategy worked: the chemical reaction stayed localized to the target electrodes even with the porous film in place.

The practical test came next. The researchers encoded a 60-byte text message—"Shanghai Institute of Microsystem and Information Technology"—into four DNA sequences, each 121 nucleotides long. These strands were synthesized in parallel on different electrode sites of the porous-film-coated chip. The resulting oligonucleotides were then sequenced using high-throughput methods. Despite some errors during synthesis—primarily deletions at a rate of 5.49 percent, with insertions and substitutions below 1.5 percent—the team was able to reconstruct the exact original sequences through clustering and consensus analysis. The four reconstructed strands were then converted back into bytes and decoded to recover the original 60-byte message without loss.

This demonstration establishes that porous silica films make electrochemical DNA synthesis practical for information storage. The error rates are comparable to or better than other custom-built on-chip synthesis systems, and the yield enhancement means more usable DNA molecules per unit chip area. The 121-nucleotide strands represent an improvement over previous electrochemical synthesis work, which had managed 104-nucleotide sequences. Longer synthesizable strands reduce the overhead of fixed primer-binding regions needed for library preparation and sequencing, improving the net information density. The path forward involves further refinement of the deblocking and coupling chemistry to extend synthesis length and lower error rates, but the fundamental viability of the approach is now established.