Conductive MOF Gate Electrodes Offer a Tunable Route to Lower-Voltage Transistors

A KAIST-led team has patterned a conductive metal–organic framework (MOF) into transistor gate electrodes with ordinary photolithography. The gates switched three different semiconductors, and in one comparison they matched a platinum gate. That is a fabrication result. The headline aim, lower operating voltage, is a design direction the work enables. It is not a measured outcome of this paper.

According to the paper in Advanced Materials (published online 15 September 2026) and a KAIST release dated 26 October, the group is led by Sarah S. Park of KAIST's chemistry department. Collaborators come from Pukyong National University, POSTECH and Yonsei University.

01Why the gate material matters

A field-effect transistor turns current on and off through the voltage on its gate. Applying a voltage to the gate electrode controls the flow of current through the transistor. The voltage at which the channel turns on, the threshold voltage, depends partly on the work function of the gate. The work function is the energy an electron needs to escape from the surface of a material. The KAIST release says it affects the voltage at which a semiconductor turns on.

In practice, a gate's work function is mismatched to the channel in many device stacks. The threshold is then shifted away from where a designer wants it, and the usual fixes are extra doping, thicker or different dielectrics, or an applied bias. The conventional alternative is to pick a different metal. Gate electrodes are mostly metals, and each metal has fixed electronic properties. The designer's menu is therefore limited to the metals that exist and that a fab can process.

Conductive MOFs offer a different knob. They link metal ions and organic molecules in a regular structure, and changing the metal–molecule combination changes the electronic properties. If the work function can be set by chemistry, the gate can in principle be chosen to place the threshold voltage where a given channel material needs it. A lower threshold lets a circuit run at a lower supply voltage. Dynamic switching energy scales with the square of that voltage, which is why researchers care.

02The material and the process

The film in this study is Cu₃(C₆O₆)₂, which the authors describe as a highly conductive two-dimensional MOF grown by chemical vapor deposition. Its measured work function is 4.51 eV, between typical n-type and p-type semiconductors, so one film can control both channel types. This is the first limit on the "tunable" label. The paper demonstrates one composition with one work function. Tuning across compositions is the stated rationale, not something these devices show.

The harder problem was integration. A MOF has to be uniform, patterned precisely, and kept undamaged, because processing damage can reduce its ability to conduct electricity. The team's solution is a two-step wet etch:

• Ammonium persulfate (APS), a slow-reacting etchant, defines the initial pattern.
• A fast chromium etchant is then applied briefly to remove residue.

The paper says this sequence gives high-conformity patterning fully compatible with standard photolithographic processes, while preserving the underlying device layers. According to the release, the structure, chemical state and conductivity of the film were largely preserved after processing. The authors also present this as the first demonstration, in their words "for the first time," of conductive MOF films serving as directly integrable electrodes in transistor arrays.

03What the devices showed

The team built transistors on three channel materials:

The authors say this shows the generality of the MOF as a gate material. The release says the electrodes switched current on and off in all three.

The most informative test used MoS₂. A MOF gate and a platinum gate were fabricated on the same channel. The release reports similar turn-on voltage and charge-transport characteristics. The paper's abstract likewise reports transfer characteristics and mobilities comparable to conventional metal gates. The release gives no numerical threshold or mobility values, and the full-text tables were not available to this reporter. "Similar" therefore means parity with platinum. It does not show a lower threshold. An electrode that merely matches a metal gate is a useful baseline, and the benefit comes only if the work function is then shifted deliberately.

The reproducibility tests are modest but relevant:

• Array yield: 23 IGZO transistors were placed on one substrate, all gates were formed in a single patterning step, and all 23 operated normally.
• Cycling: An IGZO transistor kept stable performance after 400 consecutive gate-voltage operations.
• Storage: A MoS₂ device retained its MOF-gate characteristics after about a year in ambient air, according to the release.

Twenty-three devices and 400 cycles are proof-of-concept numbers. Commercial gate stacks are judged on billions of switching events and on statistics across millions of devices.

04What this could mean for chip power, and what remains open

The power argument is plausible but indirect. Gates are not the main place where a chip loses energy. They are one lever on the threshold voltage, and the threshold sets how low the supply can go. Supply voltage, subthreshold behaviour and leakage also depend on the dielectric, the channel, doping and geometry. A tunable gate adds a design parameter, which is most useful in material systems where threshold control is difficult. Oxide thin-film transistors like IGZO and emerging 2D channels such as MoS₂ and MoTe₂ are examples. The demonstrations here are on exactly those platforms, not on silicon CMOS logic.

Several constraints remain:

• Scale. The team's own list of next steps includes making the electrodes smaller and more precise, producing them uniformly over large areas, and assessing process reproducibility and long-term reliability. Wet-etch patterning of micropatterned films is far from the feature sizes of advanced logic.
• Power. The researchers say they plan to test whether the approach lowers power consumption in practice. That test has not been done.
• Process fit. CVD growth and a chromium etchant must still fit into a production flow. The release does not address growth temperature, substrate limits or contamination rules.
• Chemistry. Only one MOF composition is shown. The case for tunability rests on other MOFs also being patterned with this APS–Cr method.

Park said the goal is for researchers to design electrodes with the properties each semiconductor needs, instead of selecting from existing metals. The next result to watch is a pair of MOF compositions with different, measured work functions. They would need to show correspondingly shifted threshold voltages on the same channel, with mobility and stability intact. That would turn a patterning demonstration into a tuning demonstration.