Development of methodology and Networked simulating tools with orientation on micro electromechanical systems (MEMS)

The goal of the project is the significant expansion of computer-aided designed micro electromechanical systems (MEMS) and large-scale integrated circuits, and increase of design efficiency thanks to new methodology usage and universal networked toolkit of MEMS modeling. The main objectives of this project are:

1. Development of methods of arbitrary non-linear functions that describe the MEMS functioning and behavior introducing to the system;
2. Improvement of design procedures and MEMS models by as classical so non-traditional approaches to a design;
3. Creation of a graphic library of the MEMS components;
4. Improvement of the optimization procedures due to an introduction of a previous search of a decision region based on genetic (evolutionary) methods;
5. Analysis of the micro electric mechanical structures (MEMS) for different applications and development of the models’ basic elements of such structures (say, s of ultrasound membrane, thermosensitive bolometer and so on);
6. Analysis of algorithms that perform correction of MEMS layout in submicron range for increasing the yield ratio and to use available lithographical equipment maximum effective;
7. Development of MEMS designing methodology based on the modified NetALLTED;
8. Analysis of protection ways of intellectual property (IP) as to MEMS and their components in the network designing;
9. The verification MEMS being designed and their parameters.

It is planned to realize all these objectives on the NetALLTED base (Networked ALL Technology Designer), a domestic system for complicated technical systems simulation, the components of which could be electronic modules, mechanical, hydraulic, air, electromagnetic subsystems. The system is based on the original numerical algorithms, which in their majority are unknown to the foreign users.

Today MEMS world market amounts to $5 billion, and if to put it into account the rapid growth for the last few years, it is expected it would be 10 billion till 2010. The current situation and forecasts for future of the main MEMS market segments are shown on fig.1.

• The production of ink jet heads. The main producer – Hewlett Packard. It is considered, that this tendency will not be developed any longer, given the way for the new print technology step-by-step.
• Pressure sensor. The most spread have in medicine and automobile branch, their part will be increasing thanks to the appearing of the new utilization, such as pressure control systems in wheels and others.
• Silicon microphones. They occupy not a large part, but for the last few years grow was more than 100%.
• Accelerometers and gyroscopes. The main use remains to be automobile branch, which is developing actively. Moreover accelerometers are going to apply in navigation systems and even in mobile phones.
• Optical MEMS. The main part remains to be DMD devices from Texas Instruments. It is expected increasing in others branches: infrared sensors for security systems, optical connection systems, various displays and others.
• Micro-liquid systems. The main application of these systems are: medicine, chemistry, biology.
• HF MEMS. This is perspectives and branch is developing recently, and the mass production of HF wireless connection, which are based on MEMS are not so high.

The success in MEMS production is directly connected with development if microprocessing technology, which allows mass production more and more complicated systems, and CAD tools. The main tendencies are shown in the follow table:

	Today	After 5 -10 years
New material	plastic, magnetic, bio-compatible	biodegradable, nano-structure (nanotubes, nanofibre and others)
Scaling	built-in sensors, multicomponent solutions	Work on cells level, alive organism, nano-MEMS
Integrated electronic	reducing of energy consumption, growing of calculating powers	Built-in power sources, increase of the circuit complication
Design and control tools	New modeling tools for new MEMS application, technological processes modeling	Increasing of calculation accuracy, creating more complicated models
Production	Including new not traditional materials to the technological processes	Progress in photolithography technology, in production of the complicated three-dimensional structures, improved equipments for packing and MEMS assembly.

Traditional technology of microcircuits production became the main for production electromechanical micro-devices, such as sensors and drives. During MEMS production today beside traditional mono- and polycrystalline silicon, oxide and silicon nitride, which were successfully taken from technology of production of integration circuits, new polymer materials are used widely too, which demands the develop technological processes and equipments .

The silicon microprocessing methods allow to product microsystems, which sizes are parts of millimeter. The microprocessing means forming three-dimensional micromechanics structures in silicon substrate or on its surface. These methods open for developments possibilities for building fundamentally new microdevices. Using such materials as mono and polycrystalline silicon, silicon nitride and others different mechanic microstructures are formed – the parts of MEMS: hanger bracket, diaphragms, hollow, holes, springs, cog-wheels and others. Today their sizes are about microns and hundreds of nanometers.

Microelectromechanic systems development depends a lot on process automation of new microdevices design. MEMS specifics lays in combining of electric and mechanic part, so then developer face not only traditional IC cases of analysis of additional tasks of controlling work dynamic of mechanic system, but solid control and others. The modern MEMS design systems, such as ANSYS, CoventorWare, IntelliSense and others come to tens of modules for analysis of electrostatic, electromagnetic, thermo mechanic, hydrodynamic and others effects. They have rich models libraries for elements with different physical nature. Application of such systems allows decreasing the design process of different MEMS, till weeks or months, which depends on project novelty or complication. Wide popularity among MEMS developers has MEMSCAP system. The system includes design tools starting with simulation tool and finishing with MEMS chip topology designer. Its distinctive feature is 2D and 3D graphic supporting and industrial formats description of the projects GDS II, CIF, EDIF and DXF.

In MEMS top-down design approach the first thing to build is general conception of and its circuit, which based on the behavioral subsystems model. Then specification of solution for realization of used systems is provided, their optimization is made, after these by their results the iteration cycle of specification of all system parameters are made. Then the whole electromagnetic and electromechanic design of the system are provided and the project go to production for the technical processing. The average design time in this case is about one week for project processing from its very beginning and then several minutes for refinement iteration.

MEMS based experiments showed, that NetALLTED systems, developed in this project, allows to analyze and design the separate elements of such type as well as built from them systems. It is possible to design and research complicated systems which have hydraulic, pneumatic, electric, electromagnetic and others elements, which have influents on mechanic elements.

It is become possible due to the one approach for introduction physically different systems. The last one is based on choosing the basic variables according to the subsystem physical nature and on choosing in systems with different physical nature of some set of basic components. So then it is become possible the presentation of physically inhomogeneous system like an equivalent scheme. Mathematical MEMS model may be represented as equivalent RLC circuit. The equivalent circuit of physically inhomogeneous system has high dimensionality, so then for increasing MEMS design efficiency there is a need to reduce its dimensionality. The problem of reducing the MEMS model dimensionality is bringing to the task of reducing the dimensionality of the equivalent RLC circuit.

For now there is two ways to solve this problem with the minimal lose in accuracy:

• to form passive macromodels (AWE, RICE, Arnoldi, PRIMA algorithms);
• to use Y-∆ transformation (star-triangle and back) for the consequence switching off internal circuit nodes (algorithm TICER and others).

Modern MEMS design methodology has been formulated which includes the following:

• During the top-down MEMS design process the planned application of the device is first investigated in order to determine the most critical parameters of the system. This is done using simple MEMS behavioral models which reflect the relation between device's parameters, its geometrical dimensions and the underlying physical phenomenon's characteristics, ignoring MEMS manufacturing technology at this stage. When the critical parameters of the MEMS have been determined the user develops photomasks for manufacturing moveable parts of the device and models the corresponding technological manufacturing process.
• MEMS are characterized by significant nonlinear relationships between different physical phenomena which are being used in the device. It is not sufficient to take into account the single underlying physical phenomenon while designing the system. Physical phenomena can be described using partial differential equations which can be numerically solved using finite element method or finite differences method. Such an approach leads to creating high order models which are used for determining mechanical properties of flexible microstructures and electrostatic field distribution between their electrodes and thus provide the ability to design different MEMS components. Such models, however, are too expensive, in terms of computational resources, to be used for modeling the microsystem in the whole.
• Complete MEMS can only be investigated at the higher abstraction levels such as schematic and system levels for which still accurate macromodels can be used. Creating such models by hand takes a lot of time and can lead to numerous mistakes and usually employs significant simplifications (only first and second degree of freedom). More valuable solution is automatic generation of such macromodels by extracting the necessary information from the detailed finite element models which have been built at the earlier design stages. This can be done by using model order reduction techniques which were developed by Linn Hobbs, Jan Mehner and others. Then complete behavioral model of the whole MEMS can be compiled in VHDL-AMS language.
• There are several integrated development systems in the industry today for MEMS design and modeling which have the necessary tools to perform all the design stages – MEMSCAP and CoventorWave. These systems include topology editor, three-dimensional (solid) editor, finite-element modeling package and model order reduction and exporting tools.
• A few kinds of simple behavioral models of MEMS have been reviewed in the scope of the work in order to select several MEMS types for subsequent use in design routines of the modified version of NetALLTED package.