Abstract
Download Conference Presentation
In recent years, the cost of <em>de novo</em> DNA synthesis has significantly decreased due to advancements in multiplexing with dense multi-electrode arrays (MEAs). The current state-of-the-art technology allows for approximately 100 million electrochemically driven synthesis sites per 1 cm² synthesis area (~1 µm² cell size). Increasing this density further presents challenges, as the size of the classical, statically addressed switch beneath each synthesis site (SRAM-based, minimum of 6 transistors) cannot be easily reduced without compromising compatibility with the required electrochemical voltages. To address this limitation, we propose a dynamic addressing scheme, analogous to that used in dynamic random-access memory (DRAM), which requires only a single transistor per synthesis site. Our estimates suggest that this approach could enable up to 10 billion sites/cm² (~0.01 µm² cell size)—a 100-fold improvement over the current state-of-the-art. This can be manufactured using a 40 nm-class process node, ensuring compatibility with the high voltages required for electrochemical synthesis.
Our dynamic addressing architecture is based on a crossbar array configuration (Fig. 1a), where the rows are connected to the gates of the selector transistors, and the columns supply a driving voltage to each cell. Transient activation of a row, combined with synchronous activation of the driving voltage on the relevant columns, rapidly charges the intrinsic capacitance of the synthesis cell to a voltage required for electrochemical deblocking (ED). Deactivation of the row then traps the charge within the cell, enabling it to drive the ED reaction until its row is refreshed in the next addressing cycle or its overpotential becomes insufficient to sustain the ED reaction (Fig. 1b). The refreshing duty cycle must be optimized to address as many rows as possible without compromising the ED rate. We have experimentally demonstrated ED with dynamic addressing using a small-scale demonstrator MEA, driven by an external custom-built multiplexer/potentiostat. Additionally, we developed a lumped-element model, calibrated with experimental data, to predict theoretical performance and optimize synthesis cell geometry.
Our findings indicate that dynamic addressing can be a viable solution to achieve ultra-high-density DNA synthesis. This work contributes to the much-needed cost-per-base reduction essential for making DNA-based data storage viable.