1.1 Solid-State Sensors

A sensor is a device which detects or measures a physical property and records, indicates, or responds to it. Solid-state sensors have no mobile parts and they must not be confused with transducers or actuators which react depending on the sensor response. Between sensor and actuator, a signal processing unit controls the whole system (Figure 1.1). Because the signal processing unit can be built as a semiconductor device, it is desirable to have the sensor device and the signal processing unit on the same chip. Integrated sensors can be built by taking advantage of semiconductor technology [15].

Figure 1.1: Block diagram of a sensor system.

Integrated sensors offer cheap solutions for controlling industrial processes and they are a good choice for medical applications. For any given application, the strength of a physical quantity should be measured. A solid-state sensor is designed in such a way that the measurand, the physical property to be sensed, exploits a physical phenomenon within the sensor structure. This physical phenomenon leads to an electrical response that can be detected and magnified with electronics. However, the design of such an integrated sensor must be accomplished under some restrictions.

From the process and cost viewpoint, the sensor structure must be manufactured without modifications to the process flow in a mass production batch. Device fabrication must be fully compatible with standard process conditions. Integrating the bias and control circuitry will be straightforward. Unless the packaging material properties affect the sensor behavior, the standard packing process can be used. This allows to build low-cost sensors with standard semiconductor processes.

Figure 1.2: Projection of the micro electro-mechanical sensor (MEMS) unit shipments in the car industry versus year.

From the design viewpoint, the sensor structure should take advantage of the process steps and of the various electronic devices that can be built. Full numerical simulations must be carried out in order to better analyze and predict the behavior of a given sensor. Simplifications will not work, because they do not take properly into account effects like temperature or doping profile variations. Besides, the measurand is not implicitly present in the partial differential equations that must be solved, making the design process complex.

All of the above steps impact the design cycle of any integrated sensor. Monolithic integration of sensors is of great interest for the microelectronics industry, because it will provide cheap and reliable sensors [15]. According to In-Stat/SMR ^1.1, despite the growing demand of the car industry (see Figure 1.2), to date, only a few sensor devices have been integrated into a high-volume production IC process.

There are several drawbacks that refrain the monolithic integration of sensor devices into IC chips. Packaging, thermal management, noise, cross-talk, are some of the drawbacks that must be overcome before getting integrated sensor devices into mass production. However, more fundamental research must be first done before attempting to transfer a sensor device into mass production. This is the case of magnetic sensors that might also find a wider opportunity in the car industry. Thus, this thesis attempts to fulfill that lack of fundamental research by providing a full three-dimensional numerical simulation tool with magnetic effects.