Non-linear resonance enables a quartz crystal microbalance sensor to detect single nanoparticles

Apr 21, 2026

Researchers boosted quartz crystal microbalance sensitivity to 100 femtograms by exploiting non-linear resonance, enabling single particle and protein detection.

(Nanowerk News) Researchers have demonstrated that pushing a quartz crystal microbalance sensor into a non-linear vibration regime can detect individual micro- and nanoparticles, reaching a mass sensitivity of approximately 100 femtograms without any surface modification or complex nanofabrication. The work, published in Microsystems & Nanoengineering (“Precise detection of single particles and bio-sensing applications on quartz crystal microbalance using non-linear resonance behavior”), was carried out by a team at Ewha Womans University and Korea University.

Key Findings

Quartz crystal microbalance (QCM) sensors work by correlating shifts in the resonance frequency of a vibrating quartz crystal with changes in mass on its surface. They are robust, scalable, and widely used in thin-film deposition monitoring, air-quality measurement, and biomolecular interaction studies. However, conventional QCM systems operating in their linear mode typically offer mass sensitivity only in the nanogram range, limiting detection of very small analytes such as individual nanoparticles or trace proteins. Nanoelectromechanical system (NEMS) sensors based on graphene and carbon nanotubes have achieved far greater sensitivity, down to the attogram and yoctogram scale. But these devices suffer from practical drawbacks including position-dependent sensitivity, environmental instability, and poor reproducibility. The approach reported by Jaehyun Kim, Sang Wook Lee, and colleagues is conceptually simple. Rather than redesigning the quartz crystal or adding functional coatings, the team increased the driving voltage applied to a standard 6 MHz QCM. At low voltages, the crystal vibrated in the linear regime, producing symmetric Lorentzian resonance curves with a quality factor of approximately 39,000. Above about 3 V the resonance curve became asymmetric, and above 5 V an abrupt amplitude drop appeared near the resonance frequency. This behavior is characteristic of non-linear Duffing-type oscillation, where a cubic stiffness term distorts the resonance curve, creates a region of multiple coexisting amplitude solutions, and allows the system to jump suddenly between them. The researchers identified 6 V as the optimal operating point, producing a sharp, well-defined, and repeatable amplitude drop. Voltages below 5 V gave insufficient contrast, while voltages above 9 V introduced instability. All measurements were performed under controlled conditions of 22.5 ± 1 °C and 40 ± 5% relative humidity. The sensing strategy fixes the driving frequency just before the amplitude-drop point. Any added mass shifts the resonance curve to a lower frequency, pushing the system past its critical point and triggering a sudden amplitude collapse. This intrinsic amplification effect makes very small mass additions visible as large, unambiguous signal changes. This differs from other resonance-based enhancement strategies. High-Q QCM sensors require special quartz geometries or additional fabrication to boost the quality factor, and bifurcation-based mass sensing has mostly been explored at the theoretical level. The non-linear amplitude-drop method achieves its sensitivity gain simply by increasing the drive amplitude of an unmodified commercial QCM. To validate the concept, the team compared particle detection in both modes. Silica micro- and nanoparticles were dispersed in deionized water and deposited as 1 µl droplets at concentrations of 100 ng/ml, 200 ng/ml, 10 µg/ml, and 1 mg/ml. In the linear regime, small mass changes at low concentrations were difficult to resolve. In the non-linear regime, clear frequency shifts of 9 Hz, 10 Hz, 15 Hz, and 97 Hz appeared across the concentration range. Single-particle detection was then demonstrated. Individual 1 µm silica particles were transferred onto the QCM surface by dispensing a small droplet onto a glass slide to resuspend particles, aspirating the suspension, and depositing it onto the crystal. Each particle produced a reproducible 1 Hz frequency shift. Repeated sequential depositions confirmed this consistency, and the same QCM could be reused without replacement. Protein sensing experiments provided further validation. Bovine serum albumin (BSA) was diluted in T50 buffer and adsorbed onto the gold electrode, producing a 4 Hz shift for a 1 pg loading. Anti-BSA antibody diluted in PBS (pH 7.3) was then applied and incubated for 20 minutes. After five rinses with deionized water, a 1 Hz shift corresponding to approximately 100 fg of bound antibody was measured. The team notes that this value reflects the resolution limit of the measurement electronics rather than the intrinsic capability of the platform. Quantitative comparison showed that the linear mode struggled to resolve mass changes below about 10 pg, while the non-linear mode detected particles and protein binding events at masses three to four orders of magnitude smaller. Previous QCM studies have primarily reported detection limits in terms of solution concentration, typically around 20 ng/L. The non-linear approach instead enables direct detection of surface-bound analytes at the single-particle level. The team also tested whether the amplitude-drop phenomenon survives in liquid, where viscous damping typically degrades QCM performance. Even in water, where the quality factor dropped to about 1,600, the non-linear amplitude drop remained observable with sufficient driving force. A fixed-frequency monitoring experiment showed an immediate amplitude drop when a 1 µl droplet arrived on the surface, suggesting detection speed is governed by mechanical response time and mass transport rather than frequency scanning. Future integration with microfluidic or nanofluidic systems could further reduce damping by confining the sensing area and minimizing fluid volume. The researchers identify potential applications in real-time biomolecular diagnostics, protein characterization, nano-plastics and fine dust monitoring, and multiplexed QCM arrays for tracking multiple targets. The study was carried out by Jaehyun Kim, Yugyeong Je, Sung Hyun Kim, Dong Hoon Shin, and Sang Wook Lee. Shin is affiliated with Korea University, while the remaining authors are based at Ewha Womans University. Sung Hyun Kim also holds an affiliation with the Kavli Institute of Nanoscience Delft.