The materials from which this simple sensor is made have been carefully chosen. They have had to be compatible with one another during the manufacturing process – so that for example, etching the material in one layer did not affect another material laid down previously. It had to be possible to shape them into the desired form, though some compromises also had to be struck. For instance, the V-groove trench is not the ideal geometry for the pit behind the beam but it is very easily made in this way.
All the materials laid on top of the silicon have been made on the spot – they have been deposited there atom by atom and have assembled themselves into a continuous layer. Careful adjustment of process conditions and process gas composition has been necessary to obtain the right material.
The graph in Figure 4 suggests that it is possible to engineer the silicon nitride to have a tensile stress very close to zero. Suggest a reason why the designer of the Pirani sensor might want to ensure that the tensile stress is appreciable, rather than trying simply to make it as small as possible. Hint: consider the changes in the device when it is operating.
In normal use the beam will be heated by the metal track that it supports. This will tend to lessen any tensile stress inherent in the material. Since a tensile stress in the beam keeps it taut, it is reasonable to build in a tensile stress sufcient to enable it to remain tensile even when operating.
Even the platinum has arrived as individual atoms in a sputtering process, and the properties of this film can also be affected by the conditions under which the sputtering was carried out. Sputtering would have been done in a low-pressure argon atmosphere, and some of this gas inevitably gets incorporated into the metal film. The properties of the thin metal film differ from those quoted in reference books for the bulk material. This is for a variety of reasons, ranging from its not being fully dense, to its having an intrinsic stress, slightly distorting the crystal lattice. For very thin films, such a large proportion of the atoms present are at or near the surface that the material may be regarded as consisting entirely of surfaces – or at least as having no parts that are not influenced by a surface. This allows metals, opaque in their bulk form, to transmit light in the visible part of the spectrum.
This example is typical of the roundabout way in which microsystems are manufactured. In exchange for the benefits of highly parallel processing (of the order of 1000 devices on each wafer), combined with the high precision afforded by the essentially photographic techniques of pattern transfer, the price that has been paid is the necessity of working with thin layers, and building up the third dimension by stacking those layers.
Paradoxically, even though we can hold them in our hands, these micro and nanoscale devices remain somehow beyond our reach. I mean this in the sense that often their working parts are too small to be seen directly, even in some cases with optical or electron microscopes, and are too fragile to touch directly. So, although in one way they are strong (their small mass makes them resistant to damage through high acceleration), it's often the case that even allowing them to become wet will destroy them, as the surface tension forces arising as they dry again will make their moving parts stick together irretrievably. Contamination from contact with other objects can also damage them, because frequently the chemistry or shape of their surfaces plays some important role in their function.
Even though the processing involved in making the micro-Pirani sensor has taken us safely to the stage where there is a large number of finished items arranged together on the surface of a wafer, this ‘unreachability’ places an obstacle between this point and having the sensor installed and working in some vacuum system somewhere. The process of dicing, in which the wafer is sawn with a high-speed, diamond-impregnated abrasive wheel kept cool by jets of water, is so violent that serious thought has to be given to how to avoid carnage. Even if this is achieved, the sensor chips then have to be mounted and given electrical connections, all without touching the suspended bridge.
In microelectronics, keeping the devices intact is a little less tricky. There are no fragile moving parts, the important stuff is going on within the materials on the chip, and there is no need to communicate with the outside world, except electrically. This means that once the wafer has had its final coating of tough insulating passivation deposited, it can be cut up into chips that can be handled with suction cups, and (once the wires have been attached) covered in glue or encased in plastic to protect them.
This issue for micro and nano technologies, of having to communicate physically or chemically with the outside world yet having to be protected from it too, is all-pervasive, and you will see it determining the way things are designed time and time again. It is sure to appear at some point in the next topic: the atomic force microscope.
The MEMS Pirani sensor has a thin metal film resistor supported on an insulating beam spanning an open trench. Identify reasons for the following:
(a) the metal film is platinum
(b) the beam is made from silicon nitride
(c) the trench is cut out of silicon.
(a) An inert metal is needed that will operate stably as a resistor at high temperature in an oxidising gas; also one that is resistant to the etchants used to remove silicon during manufacture.
(b) It is an electrical and thermal insulator that can be formed by undercutting silicon, leaving a beam of material that can be inherently stretched by internal stresses built in as the material is formed.
(c) It is a robust mechanical substrate that can be chemically etched.