A silicon chip designed to eavesdrop on living neurons has been given an unexpected second career: writing DNA. In a study published in Nature Electronics, a Harvard-led team reports a semiconductor that builds 64 different DNA strands side by side on its surface, using nothing more aggressive than finely controlled electric currents and water.
The result matters because most synthetic DNA today is made with phosphoramidite chemistry, a mature, high-volume method that can produce millions of sequences at once but leans on hazardous organic solvents and centralized production facilities. Researchers have long eyed enzymatic synthesis, which works in mild water-based conditions much like a living cell, as a safer and potentially more accessible alternative. Its weakness has been scale: earlier enzymatic demonstrations managed only about a dozen sequences at a time. The Harvard chip's 64 parallel strands, each up to 39 nucleotides long, set a new benchmark for the approach.
How the chip writes
DNA is assembled one nucleotide at a time, and after each addition a temporary blocking group must be stripped away, a step called deprotection that is triggered by local acidity. The trick to making many different sequences at once is to lower the pH only at the specific sites due to receive their next building block in a given cycle.
The chip does this electrochemically. According to the journal paper, its surface carries an array of 256 ring-electrode pairs, each a programmable synthesis site. At any chosen spot, an inner electrode injects current to generate protons that acidify the immediate area, while an outer electrode draws current to mop up protons drifting outward, keeping the acidic pocket tightly confined. Repeating this pattern cycle by cycle lets the chip grow dozens of distinct strands independently.
The hardware was not built for this. It was originally developed in Donhee Ham's lab by former doctoral student Jeffrey Abbott to record electrical activity across large populations of neurons. "At a certain point, we wondered whether that same current control could be redirected from cells to molecules," said Ham, a professor at Harvard's John A. Paulson School of Engineering and Applied Sciences. "It worked."
To show the payoff, the team encoded a 169-byte text into the 64 sequences, a small proof that the method could someday contribute to DNA-based data storage, an idea attractive precisely because it would demand DNA at enormous scale, where cutting solvent use matters most.
The researchers are candid about the limits. When they packed synthesis sites closer together to push density higher, the experiment failed, but it revealed the true bottleneck. "The chip did what we asked it to do: it localized low pH at selected sites," said co-first author Han Sae Jung. The problem lay in the deprotection chemistry, whose reactive intermediates drift between neighboring sites, not in the silicon. The paper argues that switching from an indirect to a more direct acid chemistry could let throughput scale along with the chip.