Henry Baltes

Solder-bonded micromachined capacitive pressure sensors

We report the first solder-bonding of low-cost silicon absolute pressure sensors. The goal of the work is to solder a pressure sensor wafer and a CMOS wafer containing the signal conditioning circuitry together face to face under vacuum. The result is a very flat capacitive absolute pressure sensor which can be used in harsh environments for automotive, medical, barometric, and other applications. We successfully demonstrated this approach using sensor wafers with micromachined silicon membranes and CMOS wafers without signal conditioning circuitry. Solder frames surrounding the membrane are electroplated on the sensor wafer. Different solder materials such as Au/Sn and SnPb were examined. Characterization of the hermetic prototypes in a pressure chamber showed a sensitivity of 0.8 fF/mbar, in good agreement with finite (FE) element simulations. With special regard to the sensitivity of the sensor a quadratic membrane, a rectangular membrane and a square membrane with a boss, all with an area of 2.25 mm2, were compared using FE analysis. The rectangular membrane has the largest sensitivity.

Interfaces for Microsensor Systems

We propose interface design strategies for CMOS sensor systems. The objective is to provide a microprocessor compatible sensor on a single chip. The interest of noise-shaping and oversampling techniques providing great flexibility and robustness are demonstrated. A new architecture for an oversampled noise-shaping A/D converter is also presented. It allows a significant simplification of the on-chip circuits, at the expense of high-speed signal processing at a remote data acquisition site.

Thin film front protection of CMOS wafers against KOH

KOH based silicon bulk micromachining of fully processed CMOS wafers is still a challenge since KOH heavily attacks the aluminum metallizations. At present, the protection of the front of such wafers is still accomplished using mechanical fixtures. These fixtures prevent batch processing. This paper reports a novel protection scheme for the front side of fully processed CMOS wafers against KOH solutions. Since no mechanical fixture is required the new scheme allows batch micromachining. The protection method uses standard CMOS equipment and materials only and is therefore fully CMOS compatible. Different protection schemes based on silicon nitride and oxynitride PECVD thin films are investigated. With a 4 hour etch in a 27 weight-% (wt%) KOH solution at 95 degree(s)C, membranes consisting of the CMOS dielectrics were successfully produced. After KOH etching the protection layers are removed in an reactive ion etcher (RIE). Two aspects of the protection schemes were investigated in detail. First, we analyzed the influence of the stress in the nitride layer on the fabrication yield. With an optimized recipe with a compressive stress of -150 MPa, more than 99% of all contact pads remain intact after the KOH etching. Secondly, potassium contamination of the RIE etcher is negligible if the wafers undergo an RCA cleaning procedure. Secondary ion mass spectroscopy showed that the total alkaline contaminations in thermal oxide, silicon, silicon nitride and silicon oxynitride after the RCA cleaning are not higher than those in reference samples not exposed to the KOH solution.

CMOS resonant sensors

This paper focuses on CMOS resonant sensors, i.e., resonant sensors fabricated with CMOS technology in combination with compatible micromachining steps. After reviewing resonant sensor principles, micromachining techniques applicable to CMOS resonant sensors are discussed. Subsequently, different excitation and detection mechanisms for silicon-based resonant sensors are compared. Finally, three examples of CMOS resonant microsensors, namely, an ultrasound proximity sensor, a chemical sensor and a vacuum sensor are discussed.

Radiation Effects on Carrier Transport

The model equations and boundary conditions reviewed in the preceding chapter describe electrical transport in semiconductor microtransducers in the absence of external fields. As summarized in Sect. 2.7, external fields appreciably alter carrier transport by introducing asymmetries in device operation. For example, radiation alters the generation-recombination rate, and hence, the electrical carrier transport in a semiconductor by virtue of its wavelength- and material-dependent absorption. In this chapter, we review the physical effects induced by radiant signals along with associated model equations relevant to simulation and subsequent optimization of optical microtransducers.

Magnetic Field Effects on Carrier Transport

The domain of magnetic signals ranges from the very weak biomagnetic fields (∼ 10 fT) to the very high fields associated with superconducting coils (∼ 10 T) (see [1–6]). As a measure of the field strength H, we use the related magnetic induction B whose unit is 1 tesla = 1 Vs/m2 and is related to the field strength as: B = μ0 H in vacuum, where μ0 is the free space permeability. In this very large span of over 15 orders of magnitude in field strength, the lower limit of field strengths (< 1 μT) requires relatively sophisticated detection devices and techniques [4], such as the flux-gate magnetometer, fiber optic magnetometer, nuclear magnetic resonance, and the superconducting quantum interference device, while the higher field strengths can be resolved by semiconductor magnetic sensors. Our discussion on the modeling issues will be restricted to the latter. Here, the signals are associated with geomagnetism (30–60 μT), magnetic storage media (∼ 1 mT), permanent magnets for contactless sensing (5–100 mT), and current carrying conductors (∼1 mT at 10 A) [6]. These signals lend themselves to two categories of direct and indirect applications [1–3]. Direct applications include measurement of the geomagnetic field, reading of magnetic storage media, identification of magnetic patterns in cards and banknotes, and control of magnetic apparatus. In indirect applications, a non-magnetic signal is detected via the magnetic field which is used as an intermediate carrier. Examples include voltage-free current detection and watt-hour meters, and contactless sensors, based on mechanical displacement of a permanent magnet, for detection of linear or angular displacement and velocity.

Mechanical and Fluidic Signals

Simulation of the static and dynamic behavior of solid structural and fluid mechanical variables, e.g., stress, strain, strain-rate, displacement, force, and velocity, is critical to the design and analysis of microsensors and microactuators in the mechanical domain. For example, in Chapt. 6, we saw how electrical transport is modified by piezoresistance. This, in addition to deflection-induced capacitance change, can be effectively utilized for conversion of signals from the mechanical to the electrical domain. Alternatively, as we will see in Chapt. 8, a micromechanical structure subject to an electrical, thermal, magnetic, or mechanical excitation signal, gives rise to micro-actuation in the mechanical domain. In this chapter, we deal with model equations and constitutive relations relevant to: simulation of mechanical (e.g., pressure) microsensors; computation of velocity profiles relevant to flow microsensors (needed in Chapt. 5) and selected microfluidic systems; computation of mechanical stresses induced by packaging or encapsulation of microtransducers or integrated circuits (needed in Chapt. 6); and simulation of mechanical microactuators, including fluidic damping effects (needed in Chapt. 8).

Thermal Non-Uniformity Effects on Carrier Transport

Physical properties of semiconductor materials and devices are sensitive to variations in temperature, whether generated from the ambient or internally in a device or integrated circuit (IC). While the variations in temperature associated with the ambient can be treated as uniform (isothermal) relative to device dimensions, internal heat generation is highly localized giving rise to a temperature gradient, which constitutes a non-isothermal signal. Various methods can be employed for detection of thermal signals. For measurement of ambient temperature, we can employ the highly predictable and stable temperature dependence of the base-emitter voltage V BE of a bipolar junction transistor. Together with co-integrated biasing, signal correction, and amplification circuitry, they provide an output voltage or current that is proportional to absolute temperature (PTAT) [1, 2]. On-chip temperature gradients or non-isothermal signals transduced by physical signals, not necessarily from the thermal domain (see [3, 4]), can be detected using thermoelectric or thermoresistive effects. Our discussion of modeling issues will be restricted to non-isothermal signals and related microtransducers; models pertinent to isothermal signals are reviewed in Chapt. 2.

Basic Electronic Transport

In contrast to very large scale integrated (VLSI) devices, microtransducers have relatively large dimensions and are not in the race to push the limits of feature size into the submicron regime. Thus with microtransducers, it is reasonable to assume a static picture for electrical transport in the device, whereby the mobile charge carriers are in equilibrium with the host lattice. This permits the use of the classical model comprising Poisson’s equation, which relates the electrostatic potential and space charge in the device, and the electron and hole continuity equations, which account for charge conservation, with current density relations based on the drift-diffusion formulation. Effects of non-static transport have become very important in VLSI devices where the active device dimensions are reaching scales (nm) where the carrier transit time becomes comparable to the collision time.

Mechanical Effects on Carrier Transport

Effects of mechanical stress on electrical properties of semiconductor materials and devices have been known since the invention of the transistor. The pioneering work of Bardeen and co-workers [1], on effects of pressure on the p-n junction, and subsequently, by Smith on stress modulation of resistivity [2], have led to what we presently term as the piezojunction and piezoresistance effects, respectively. Today, these effects are routinely exploited for sensing of mechanical signals. Integrated circuit (IC) mechanical microsensors and microactuators, including micro-electro-mechanical systems in closed loop operation, rely on modulation of the inherent electrical carrier transport induced by an external mechanical signal such as pressure, force, or acceleration. Here, the transduction mechanism from the mechanical to the electrical domain is by virtue of piezoresistance. A large number of applications and associated devices have been developed, along with a corresponding amount of literature. An extensive early review of the interaction of the mechanical signal with electrical transport in semiconductors is given in [3]. Specific references to recent developments are given in the sections that follow. In addition to piezoresistive and piezojunction effects, there is the piezoelectric effect, which however, is present in ferroelectric materials. Here, mechanical stress gives rise to electric polarization. The piezoelectric effect and associated applications are described in Sect. 8.6. Recent progress in mechanical sensors and micro-electro-mechanical systems is reviewed in [4]. The latter along with other mechanical microactuators is discussed in Chapt. 8.