Screen-Printed Microcantilevers for Environmental Sensing ^†

Abstract

This work investigates different geometries for fully screen-printed self-actuated self-readout microcantilevers made using a sacrificial layer process. The microcantilevers are made of PZT sandwiched between two gold electrodes. Q-factors as high as 1300 are reported. Promising theoretical sensitivities are reported (1000 Hz/µg at 620 kHz) and applications as humidity, temperature and particles sensor are explored in this work.

1. Introduction

More and more regions of the world are facing the challenges of air pollution due to gases and fine particulate matters (PM). This is especially true in fast-developing countries with rampant industrialisation releasing massive amount of dusts in the atmosphere. Recent reports of the WHO recommend for PM of 10 µm and 2.5 µm diameters a maximal exposure over 24 h of 50 µg/m³ and 25 µg/m³ respectively [1], which gives guidelines regarding the sensitivities to be achieved with PM sensors. Silicon microresonators have been widely used as sensors, including for particles detection [2,3]. In this work we propose self-actuated self-readout piezoelectric microcantilevers used as a gravimetric sensor in the scope of particulate matter detection for automotive application. It has already been demonstrated that such microcantilevers can be used as specific gas sensors in previous work [4]. Applications as a temperature and humidity sensor are explored. A test bench is mounted for particulate matter sensing. First results showed the proof of concept and further experiments are currently running with promising perspectives.

2. Materials and Methods

The microcantilevers are screen-printed using a sacrificial layer process [5], further described in the Appendix A. Made of a PbZr0.52Ti0.48O3 (PZT) ceramic beam sandwiched between two gold electrodes (Figure 1), the clamped-free microcantilevers are entirely screen-printed on alumina substrates. They are sintered at 900 °C where the sacrificial layer burns, releasing the beam during the sintering of the ceramics. High densification (88% of the bulk PZT) and high Q-factors (1300) are achieved. The unusual 31-longitudinal resonance mode is here privileged to obtain high resonant frequencies. Different geometries are investigated from 8 × 2 × 0.1 mm³ down to 1 × 1 × 0.1 mm³. The resonant frequency of the microcantilever is measured using a network analyser. The resonance frequency f₀ is followed through the maximum of the conductance G, image of the mechanical resonance. A polynomial model (Equation (1)) allows precise determination of the resonant frequency through the conductance measurement (Figure 2). Details about this equation are given in Appendix B. This is consistent with the low noise (σ) reported along the whole range of geometries investigated. Theoretical sensitivities are calculated and showed up to 1000 Hz/µg at 620 kHz (

Table 1. Theoretical (f₀ th.) and experimental (f₀ exp.) resonant frequencies, theoretical sensitivities (S), noise σ and quality factor Q exp.. The noise σ and the quality factor Q exp. are measured on one microcantilever for each dimension. f₀ exp. is the mean value obtained for each geometry. S calculations are made using f₀ exp. values. Refer to Appendix B for more details.

L/w (mm)	f0 th. (kHz)	f0 exp. (kHz)	σ(Hz)	Q exp.	S (Hz/μg)
8/2	90	90	1.5	100	−5
8/1	90	100	3.6	150	−10
4/2	170	180	0.9	250	−20
4/1	170	180	0.9	300	−40
2/2	370	340	0.8	250	−100
2/1	360	340	0.7	600	−200
1/2	810	780	1.2	800	−400
1/1	790	620	0.6	1300	−1000

).

$$G\approx 2\pi CQ{f}_0\left(1-4Q^2+8Q^2\frac{f}{f_0}-4Q^2\frac{f^2}{f_0^2}\right)+\dots )$$

(1)

This expression is used to fit the raw data and allows a more precise determination of the resonant frequency as illustrated in Figure 2.

A test bench is mounted for dust detection experiments, represented Figure 3. An optical particle analyser (Fidas^® Frog, Palas Gmbh, Karlsruhe, Germany) put at the end of the gas line is used as reference. The microcantilever is actuated beforehand for 20 min under a clean nitrogen flow (1 L/min) for stabilisation of humidity and temperature. For temperature and humidity measurements, the microcantilevers are put in a climatic chamber in which humidity and temperature can be controlled. An additional temperature and humidity sensor is added in the chamber as close as possible from the microcantilever. Twenty minutes-plateaus of humidity are carried out from 50% to 90% by steps of 10% of relative humidity (%RH), followed by temperature plateaus ranging from 25 °C to 45 °C by steps of 5 °C.

3. Results and Discussion

Humidity and temperature are steady during the dust experiments. The measurements are carried out on a 1 × 1 × 0.1 mm3 microcantilever. On the Figure 4, the first part of the curves shows a slowly increasing of f₀, corresponding to a mechanical stabilisation of the resonance. Then as particles are released, f₀ decreased constantly, resulting in a total shift of 23 Hz. Considering the theoretical sensitivity for this specific device, the mass of deposited particles can be estimated around 23 ng, giving collection efficiency around 12 ppm at 1 L/min. Experiments in the climatic chamber show that the resonant frequency of the microcantilever varies negatively with both temperature and humidity, with respectively −47 Hz/°C and −41 Hz/%RH for a 1 × 1 × 0.1 mm3 microcantilever (Figure 5). For the temperature, this is most likely due to changes in the stiffness of the beam, and for the humidity to the mass effect of the adsorbed water on the microcantilever. The wavy shape of the temperature profile is due to small spikes of humidity during temperature steps, both not being perfectly independent in the chamber.

4. Conclusions

This work showed proof of concept for screen-printed microcantilevers to be used as fine particles sensors. This sensor can also be used either as a temperature or humidity sensor with high sensitivities. Other experiments are currently running toward improving the particles collection efficiency using thermophoresis and decreasing the dust carrying flow as already explored by Liu et al. for SAW devices [6]. As new materials are being investigated worldwide, other improvements could also be made in using more ecological alternatives to PZT such as BNT or KNN piezoceramics [7].