A Piezoresistive Sensor for Pressure Monitoring at Inkjet Nozzle

Jia Wei and Pasqualina M. Sarro

Delft Institute of Microsystems and Nanoelectronics Delft University of Technology Delft, the Netherlands Email: {j.wei, p.m.sarro}@tudelft.nl
Abstract This paper presents a MEMS piezoresistive sensor for monitoring the fluidic pressure at the nozzle of an inkjet during droplet generation. The device consists of a sensing membrane (150 m wide and 1 m thick) with a nozzle orifice (20 m in diameter), and piezo-resistors placed around. The pressure information is useful in detecting missing droplets and estimating the size of the generated droplets. The device is fabricated on SOI wafers with an IC-compatible process. A resistance variation of 8.7% is measured with a 1105 Pa applied pressure. The sensitivity is 3.910-7V/Pa in a Wheatstone bridge configuration with 1 V supply voltage. The detected pressure signal can be used to implement a close-loop control to replace the open-loop control in most current commercial inkjet printheads, for better volume precision and system reliability.

Trinh Chu Duc

Faculty of Electronics and Telecommunications UET, Vietnam National University, Hanoi Hanoi, Vietnam Email: trinhcd@vnu.edu.vn

II.

CONCEPT AND DESIGN

Fig.1 shows the working principle and structure of the proposed sensor. During the formation of a droplet, a pressure wave is generated by the piezoelectric actuators attached on the sidewall of the inkjet channel and travels towards the nozzle. The sensor is placed at the nozzle part and consists of a thin membrane with an orifice in it. Piezo-resistive sensors are placed on the top surface of the membrane (Fig.1.a), allowing the detection of the membrane deformation caused by the fluidic pressure variation at the nozzle (see Fig.1.b).

I.

INTRODUCTION

The drop-on-demand (DOD) inkjet technology is one of the most successful applications realized with MEMS devices. This technology is relevant to many industrial fields besides document printing, such as production of flexible electronics, dispensing biomedical samples and maskless fabrication of microstructures [1]. Many applications require an accurate and reproducible droplet size [2]. A potential way to improve the precision and reproducibility of the droplet size is to replace the current open-loop inkjet system with a close-loop control. To do this, a device for in-situ monitoring of the fluidic condition inside the inkjet channel during the operation needs to be integrated into the printhead. The sensor can be designed to detect the fluidic motion [3], or to monitor the pressure variation inside the printhead. In this paper, a MEMS based piezo-resistive sensor is designed and fabricated to monitor the pressure variation at the nozzle orifice for inkjet application. The basic principle, mechanical design and fabrication process of the device are illustrated and discussed in the following sections. An experimental validation is carried out by measuring the resistance variation of the piezoresistor under different pressures.

Figure 1. Schematic view of the proposed device illustrating the sensing principle: a) the cross-section of the sensor; b) the sensor is placed at the end of an inkjet channel to detect the pressure at the nozzle.

By detecting the phase and the amplitude of the pressure wave at the nozzle and comparing them with the actuation signal, the required information about the fluidic condition inside the ink channel can be extracted. Such information can be potentially used to detect any clogging, air-bubbles and mechanical failures inside the printhead.

This work is supported by the IOP program from the Dutch government.

978-1-4244-8168-2/10/$26.00 2010 IEEE

2093

IEEE SENSORS 2010 Conference

To determine the place of the piezoresistive elements which operate as a mechanical sensor for detecting deformations of the nozzle membrane, the distribution of the stress on the top surface of the nozzle is investigated with finite element method (FEM) analysis. When a fluidic pressure is applied on the bottom surface of the nozzle membrane, there are two main regions containing large stresses, as shown in Fig.2. At the nozzle membrane, a highly compressive stress can be found. This is similar to the case of a conventional piezoresistive pressure sensor based on membrane deformation. The second region with high stress is along the nozzle orifice, where a tensile stress can be obtained. This is caused by the elongation of the orifice perimeter due to the deformation of the membrane in the out-of-plane direction.

Therefore, a membrane with smaller size and thickness is preferable. In this experiment, the minimum achievable thickness is about 1 m. The maximum fluidic pressure generated in DOD inkjet channel is typically around 1 to 2105 Pa [2]. The diameter of the nozzle orifice (d1) is chosen to be 20 m. To detect such a pressure range with a membrane thickness of 1 m, the size of the nozzle membrane should be around 150 m.

Figure 2. Simulated sensor structure showing the distribution of the mechanical stress in x-direction under a pressure load on the bottom surface of the nozzle membrane.

Figure 3. FEA simulation results illustrating the relation between the expected stress concentration and the required nozzle membrane size. The thinner and larger the membrane is, the higher the stress concentrating on the piezo-resistors.

Theoretically, both stress regions can be used for piezoresistive sensing. However, the compressive stress is located at the edge of the membrane and this position is highly depended on the position accuracy of the backside cavity. Many effects during the etching, such as the under-etch and the misalignment of the crystal orientation, can create an unexpected shift of the membrane size and position up to 10~20 microns. This generates a challenge for the alignment of the piezoresistor to the edge of the membrane, especially when the size of the membrane scales down to the level of 100 microns. On the other hand, the tensile stress located at the edge of the orifice is less influenced by the position of the nozzle membrane. Taking into account the inaccuracy of the cavity etching process in this demonstrator fabrication, the second stress region is chosen. The sensitivity of the membrane can be modified by tuning its dimension. Fig.3 shows the relation between the expected stress concentration on the membrane surface and the required membrane size with different membrane thickness (1 and 2 m) under 1 kPa pressure load. A same stress concentration can be achieved with different combination of membrane size (d2) and thickness (t). To detect the pressure variation with a certain sensitivity, a design with a smaller membrane requires a thinner membrane thickness. This is a trade-off between the device size and the fabrication limit. In commercial inkjet systems, a large amount of nozzles are often arranged in 1D or 2D arrays for a higher value of pixels-per-square-inch (PSI).

Fig.4 illustrates the placement of the piezoresistors and their electrical connections. The piezoresistors are placed along the [110] orientation of the silicon crystal. In order to avoid as much as possible mechanical influences on the membrane thickness due to the electrical connections, the aluminum wires are kept far from the nozzle orifice. This means that the piezoresistors need to be connected to the aluminum wires through buried connections formed by doped silicon regions, as shown in Fig.4.

Figure 4. A schematic drawing of the placement of the piezoresistors together with their connections.

2094

To reduce the piezoresistive effect of this connection layer, which may degrade the sensitivity of the piezoresistor by introducing an opposite resistance variation, these connections are placed mainly along the [100] orientation of the silicon crystal, where the piezo-resistive coefficient is close to zero. A heavily doped p-type layer (up to ~1020 atoms/cm3) is used. This further reduces the piezo-resistive effect of the connection layer, as the piezo-resistive coefficient drops with the doping concentration. This high doping concentration also reduces the contribution of the connection layer to the total measured resistance, limiting its negative influence on the sensitivity. Thus, the detected resistance variation can be considered as a clean signal generated only by the piezoresistors placed along the edge of the orifice. III. FABRICATION

IV.

RESULTS AND DISCUSSIONS

Fig.6 shows the fabricated device where the nozzle membrane is 1 m thick and 160 m wide. An orifice of 20 m in diameter is located at the center of the membrane. Four piezoresistors (8 m long, 1 m wide and 300 nm thick) are symmetrically placed along the edge of the orifice.

An IC-compatible process is used to fabricate the proposed device. The starting material is an SOI wafer with a 400 nm thick buried oxide layer and a (100)-plane device layer with an initial thickness of 340 nm. A 600 nm thick n-type (As+, 1E16 atoms/cm3) epitaxy layer is grown on the wafer, and the piezoresistors are created by using a second epitaxy layer of 300 nm thick (B+, 1E18 atoms/cm3), as shown in Fig.5a. A reactive ion etching (RIE) is then used to define the dimension of the piezo-resistors. The resistors are placed along the edge of the nozzle orifice, mainly oriented along the [110] direction. The reference resistors are fabricated with the same dimension and orientation and placed close to the sensing resistors. All resistors are isolated from each other by the reversed biased ntype epitaxy layer and p-type isolation rings created by ion implantation (Fig.5b). The heavily doped p-type connection layer is fabricated in the same ion implantation step used for the isolation. After metallization and passivation, the wafer is etched in a 25% tetramethylammonium hydroxide (TMAH) solution at 85 C to create the backside cavity, landing on the buried oxide layer. Finally, the buried oxide layer is removed in a BHF solution with the front side of the wafer protected by a photoresist layer. Finally, a thin oxide layer is deposited at the bottom of the nozzle membrane to provide dielectric isolation (Fig.5c).

Figure 6. SEM image of the fabricated device. The inset image shows the thickness of the sensing membrane.

To validate the principle, preliminary tests are done with water pushed through the nozzle at different flow rates, while the corresponding resistance change of the piezo-resistors is measured. The measurement setup is schematically shown in Fig.7. The fabricated device is glued and wire-bonded on a standard ceramic housing with a hole drilled at the center of the ceramic substrate. A step motor system (KDS Model 200 Series) is used to provide a constant flow rate by pushing with a syringe through a plastic pipe, which is connected to the backside of the nozzle. A fine stream of water being jetted from the nozzle orifice is observed during the experiment, once the flow rate is above a threshold value (~100 L/min).

Figure 5. The main steps of the piezoresistive sensor fabrication.

Figure 7. Drawing of the measurement setup showing a bonded device during the measurement.

2095

Figure 10. Resistance variation versus fluidic pressure applied on the membrane. Figure 8. Measured resistance variation versus water flow rates through the nozzle.

Measured resistance change (%)

The resistance value of the piezoresistor is measured with an Agilent 4156C precision semiconductor parameter analyzer. Fig.8 shows the measured resistance variation with the applied flow rate. The nominal resistance is 9.97 k. With a maximum flow rate of 200 L/min, a resistance variation of 8.7% is measured.

variation is determined by the electrical noise. By taking into account the flicker noise and thermal noise at room temperature with a 50 kHz measurement bandwidth, the minimal detectable pressure is estimated to be around 40 Pa.

V.

CONCLUSIONS

Flow rate (L/min)

The device presented is suitable for detecting the pressure variation at the nozzle orifice for inkjet application. The data can be used to monitor the fluidic behavior of the printhead during the droplet generation, such as detecting missing droplets and estimating the size of the generated droplets. The device is successfully fabricated on SOI wafers with an ICcompatible process. A resistance variation of 8.7% is measured with a 1.04105 Pa applied pressure. The sensitivity is 3.910-7 V/Pa in a Wheatstone bridge configuration with 1 V supply voltage. ACKNOWLEDGMENT The authors would like to thank S. Mokkapati from the Electronic Instrumentation Laboratory of TUDelft/DIMES for many helpful discussions on realizing the measurement setup and the DIMES IC-process group for technical support. REFERENCES
[1] D.B. Wallace and D.J. Hayes, Solder Jet Optics Jet AromaJet Reagent Jet Tooth Jet and other Applications of Ink-Jet Printing Technology, Proc. of IS&Ts NIP18, San Diego, pp. 228-235, 2002. H. Ren, R.B. Fair, M.G. Pollack, Automated on-chip droplet dispensing with volume control by electro-wetting actuation and capacitance metering, Sens. Act. B, Vol. 98, pp. 319-327, 2004. J. Wei, C. Yue, M. van der Velden, Z.L. Chen, Z.W. Liu, K.A.A. Makinwa & P.M. Sarro, Design, fabrication and characterization of a femto-farad capacitive sensor for pico-liter liquid monitoring, Sensors and Actuators A: Physical, doi:10.1016/j.sna.2010.03.021, 2010. H. Wijshoff, Structure- and fluid-dynamics in piezo inkjet printheads, PhD thesis, University of Twente, 2008

Figure 9. FEM smulated relation between the flow rate and the applied fluidic pressure across the nozzle membrane.

The relation between the flow rate and the applied fluidic pressure across the nozzle membrane is obtained with FEM analysis (Fig.9). By using this relation, the resistance variation as a function of applied pressure across the nozzle membrane is extracted as shown in Fig.10. The results show an almost linear detection of fluidic pressure, especially at the high pressure region. This is consistent with the expected linear behavior of solid mechanical systems. The error in the measurement is mainly caused by the vibration in the fluidic system, since the liquid is driven by a step motor. When connecting two sensing resistors and two reference resistors to form a Wheatstone bridge with 1 V supply voltage, the resistance variation measured above provides a pressure sensitivity of 3.910-7 V/Pa. The minimal detectable pressure

[2]

[3]

[4]

2096