A laboratory study reports that changing the effective precursor dose used in atomic layer deposition, or ALD, was associated with which kind of MoS2 film emerged after sulfurization. The higher-dose path corresponded to a continuous monolayer, and that path showed stronger measured piezoelectric response, meaning electrical output recorded when the film is strained, than the porous-monolayer path: 141.38 pA per 1% lateral strain at 0.5 Hz, compared with 59.69 pA. The reported effective piezoelectric coefficient was 4.3 pm per volt for the continuous film and 0.8 pm per volt for the porous film.
The study asked whether the state of a molybdenum oxide film before sulfurization was associated with the final MoS2 monolayer outcome and device performance. It deposited MoOx by ALD, then sulfurized it to form MoS2. The sulfurization process was fixed for all samples while the work examined ALD kinetic flux, keeping the comparison focused on the deposition paths.
The film paths diverged before sulfurization
Across the tested paths, low, intermediate and higher effective ALD doses were associated with isolated flakes, porous monolayers and continuous monolayers, respectively. The porous path was also reported to have relatively more sulfur vacancies than the continuous path. That pattern links the process setting to the later comparisons, but the study supports an association between the paths and the outcomes rather than a definitive causal claim.
The difference was not simply a matter of how much oxide had been deposited. Before sulfurization, the oxide thicknesses for Path 2 and Path 3 were comparable: 1.082 nm plus or minus 0.016 nm and 1.064 nm plus or minus 0.028 nm, respectively. Path 3 was smoother, with reported root-mean-square roughness of 0.12 nm versus 0.36 nm for Path 2, and it had the more stoichiometric composition.
An intermediate sulfurization check showed different reported chemical states in the two paths. After sulfurization at 600°C for 10 minutes, the Mo4+ fraction was 0.75 in C-ML versus 0.37 in P-ML. The residual Mo6+ fraction decreased from 0.40 to 0.05 in the same comparison. The reported values were consistent with more advanced sulfurization in C-ML, but this measurement alone does not identify the mechanism behind the difference.
The cleaner film carried a stronger signal
The continuous film covered approximately 99.97% of the surface, compared with approximately 92.9% for P-ML, and it was reported to have significantly lower defect density. Its sulfur-to-molybdenum ratio was 2.014, compared with 1.945 for P-ML. The reported work function was 5.652 electronvolts for C-ML and 5.318 electronvolts for P-ML. The measurements were consistent with reduced electron accumulation and weaker free-charge screening, meaning less free charge was available to mask the measured response.
That pattern carried into electrical testing. Under lateral strain at 0.5 Hz, C-ML devices registered 141.38 pA per 1% strain, or 2.36 times the P-ML value of 59.69 pA per 1% strain. In piezoresponse-force microscopy, a nanoscale test of electromechanical response, the reported effective coefficient was 4.3 pm per volt for C-ML and 0.8 pm per volt for P-ML. The paper described the C-ML value as the highest among the compared binary TMD systems.
The voltage test showed the same direction of difference. At a bending speed of 6 centimeters per second and 0.88% strain, open-circuit voltage was 1.74 volts for C-ML versus 1.32 volts for P-ML. Stable C-ML signal output was also reported after 2,500 seconds of cycling at 0.37% strain and 0.5 Hz.
The advantage narrowed when the materials were compared at device scale. The C-ML-to-P-ML piezoelectric ratio fell from approximately 5 at the microscale to approximately 2.3 at the macroscale. The text notes that strain gradients around pinholes in P-ML may partly compensate for its weaker response in a larger device. Because the coefficient is reported as an effective value, the device-scale result should not be read as a direct measure of intrinsic piezoelectricity alone.
Promising prototypes, limited reach
The application tests were demonstrations in cell culture and prototype devices. In cell-culture assays, viability remained above 98.8% in both tested cell types, with no significant difference versus control. Each assay plated 40,000 cells in 12-well plates and cocultured them without or with MoS2 films for 1, 3 and 7 days. C-ML devices detected finger, knee and elbow bending and enabled fruit classification by weight.
In a soft-gripper event, pressure rose from 20 to 37 kPa within approximately 250 milliseconds after slippage was detected, and the object was not dropped. Lifting attempts without feedback failed. The report demonstrates an adaptive response in that test, not a general success rate for robotic handling.
Taken together, the results support a process-to-structure-to-device association in the tested films and prototypes. They do not establish human safety, clinical utility or long-term field performance, and the cell work remains an in vitro test. The study also does not show that ALD dose alone determines every final MoS2 property, because the comparisons follow laboratory process paths rather than a randomized intervention.
The paper's disclosures
The work was supported by Singapore's National Research Foundation, Prime Minister's Office, under the CREATE Smart Grippers for Soft Robotics programme. The authors declared no conflicts of interest and said that all data needed to evaluate the conclusions are present in the paper and/or Supplementary Information.
Paper data and sources
Original title: Monolayer MoSWith Ultrahigh Piezoelectricity: From ALD Dose Control to Device Performance.
Authors: Yun Li, Ming Rui Joel Tan, Subhasis Das et al.
Journal/Repository: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1002/advs.77323
Original paper