字幕列表 影片播放
-
Good morning to all of you. Welcome to the course on Microsystems and MEMS. So, MEMS
-
the full name is microelectromechanical systems and this is a very emerging area today and
-
lot of work is going on around the globe on MEMS and Microsystems and it has got enormous
-
applications. Today, we will highlight the basic applications and some of the introduction
-
on MEMS and Microsystems.
-
Now, the MEMS topic or microsystems is an offshoot of the microelectronics and so to
-
start the course on MEMS and Microsystems, we will start from microelectronics and let
-
us look back little bit into the back of the present MEMS which is microelectronics and
-
we will look into the history on microelectronics.
-
So, this particular figure you can see the three scientists who are the Bardeen, Brattain
-
and Schokly from the Bell Laboratories and you know these three scientists first discovered
-
the Point Contact Transistor which is in 1947. For that discovery they got Nobel Prize in
-
1956 and that was the first time Nobel Prize was awarded for an engineering device. You
-
know Nobel Prize is not given for any engineering branch. It is given on basic science, physics,
-
chemistry, mathematics, and etcetera. But here the Nobel Prize was given in physics
-
for engineering device.
-
Later on these three scientists developed a technique by which silicon can be oxidized
-
and oxidation demonstrated by them in 1953 at Bell Lab. Because of that demonstration
-
of oxidation from silicon, basically the people started thinking why we cannot make the transistor
-
on silicon monolithically. The first invention on transistor was a point contact that is
-
a diskette. Basically, they combined three pieces of silicon N P N and something like
-
that and then they took contact on each point and they got the transistor action.
-
This figure you can see is Jack Kilby and he invented integrated circuit in 1958.You can see here so that is the US patent and
-
which was submitted in February 6, 1959 and this was in September 1958 the patent was
-
written and that was the first monolithic integrated circuit you can see the picture
-
here and this gentleman is Jack Kilby and he got again Nobel prize for this particular
-
device in the year 2000.
-
So, from that time onwards people are not seeing back and they are proceeding forward,
-
particularly in the context of miniaturization of different components and making the integrated
-
circuits. Now this picture you see is the MOS transistor which contains over one million
-
MOS transistors and that is in early 1990s. Now if you look back to compare the revolution
-
in the integration of the components, we have to go back to the first transistor which is
-
also monolithic and that was made in 1960s, where four BJTs and several resistors are
-
connected together with some metal line. You can see this is a metal contact, this is a
-
metal born pad, here is another metal born pad and the four transistors and several resistances
-
were integrated to get some functions off to get some circuit.
-
Now, after that if we see the what is the status and the what are the trends of silicon
-
ICs, we have to see this particular figure where this has been obtained from CR route
-
map and here in one side you can see the minimum feature size of the transistor starting from
-
1970 to 2020 and on the other side you can see the number of transistors for DRAM chip.
-
So now look at this figure, here in 1980 where these minimum feature sizes was nearly say
-
2000 nanometer and then if you go, at present this is the present line 2004 and there you
-
can see the feature size is nearly 100 nanometer.
-
So, on the other side, the DRAM chip which is basically always cited the integration
-
level and that integration level starts from the 4 Kilobyte this is the nineteen four kilobit
-
DRAM which has come in 1970 and later on you can see from 4 Kilobyte to 64 Kilobyte then
-
1 Megabyte to16 Megabyte and so on. At present 256 Megabyte DRAM, where this is nearly in
-
1998 or 99 it came and there transistor count you can see here is one 10 to the power 9
-
so that means is nearly people are thinking at the moment so transistor level has come
-
of the order of 10 to the power 10 transistors for DRAM chip.
-
Now, if you here is one limit by 2010 and 2012, we can see the number of neurons in
-
the human brain in 15 centimeters cube is nearly 10 to the power 11. 11 is the total
-
number of the neuron cells and people are speculating that, by that time the number
-
of transistors for IC or DRAM chip will achieve in that level.
-
So now, this is another picture you can see which is basically Moore's Law and that gives
-
you the scaling of the CMOS. That means year to year how the integration is taking place,
-
you can see from here the model CMOS in 1980 which is nearly one micron technology and
-
then at present in 2004, the transistor size has come down to 19 nanometer only. Here basically
-
a route map is given by an international organization and that is the semiconductor route map they
-
call it, and in the last 34 years the scaling history is such that every generation feature
-
size sinks by 70 percent.
-
What do you mean by the generation? The generation basically, on an average of every 2.9 years
-
they speculate something, they foresee something and after that on that particular year again
-
depending on the progress of the design and progress of the technology they again just
-
to control something so the next two years what will happen, so that is basically the
-
route map. Earlier this generation is nearly 2.9 years but later on the progress is so
-
fast that has been changed to 2 years recently every two year they are speculating something.
-
Before that time target achieved, they got something else, much more progress has been
-
done.
-
Now, basically the scaling of the CMOS if you can see that the beginning of the some
-
micron era is started nearly say mid of 85, a mid of 80s that is 1985 nearly in that after
-
that deep UV lithography then 19 nanometer in 2004. So if you continue in this fashion
-
so reasonably by 2020, they will come to a limit and that limit of scaling, that means
-
there the problem is the lithography alignment or which will be the source of lithography.
-
So, optical lithography is a commercial technique, you know after they highlight, they proceed
-
to UV, deep UV and then after deep UV there are certain other lithography techniques which
-
are basically x-ray or lithography but those are not commercially viable.
-
The people want to stick on optical lithography itself so that's why optical lithography create
-
some problem after the 90s nanometer and below. Lot of intervention and techniques people
-
are using in the optical lithography so that they can get the feature size below these
-
90 or now at present 60 or 65 nanometer people are working on.
-
But now the silicon microelectronics basically if you see, that is the silicon wafer is 1
-
0 0 crystal orientation wafer. Now, at present the standard size of the silicon wafer in
-
industry is nearly 12 inches. Normally in some cases the 4 inch or 6 inch and 8 inch
-
wafers are also used in some of the small fabs but in big fabs they are working on 8
-
to 12 inches. Now, you know the lots of circuits are made on silicon wafer and individual single
-
chip and that is of the order of 2 centimeter square.
-
I am talking about a larger chip size which you can make now days, is of the order of
-
2 centimeter square that does not mean that feature size is also very large. Feature size
-
is very small and if you can get the larger die side that is another achievement. Feature
-
size goes down and this is known as the die and the die size goes up so that is the challenge
-
of the technology. Now currently what I just mentioned few minutes back, that the number
-
of transistors per chip has exceeded the 1000 million. So this is heading towards the billion
-
and projection in 2014 is 20 billion transistors per chip and that is
-
the projection at the moment.
-
Now, here is another diagram, where you can see the feature size goes down, so if you
-
look into the integrated circuit history, below this is a 0.1 micrometer and this is
-
10 nanometer range, this is the one transition 0.1 to 10 nanometer and then another transition
-
level is 10 nanometer to 1 nanometer. So 1 nanometer means this transition is very important
-
to people and is basically the quantum devices. Now you know the lattice, the constant of
-
silicon is nearly 3 or 4 or 5 nanometer. Now if you go into 10 nanometer that means few
-
2 or 3 the atomic diameter is like that. So, that means the few monolayers of silicon within
-
that if you go into this region then you have to make the devices.
-
So there, the normal physics of the transistor may not be valid and you have to go into the
-
quantum mechanical analysis and those are basically the quantum devices. Below that
-
are atomic dimension and the automatic atomic dimension level, when you make the transistor
-
lot of other problems will come into the picture. Then here is another table which shows that
-
the trend of the feature size as well as the wiring levels the mask count and supply voltage
-
that all this figure tells that is in 1997 where the supply voltage is nearly 2 volts
-
and in 2003 it is 1.2 volts and people are looking for the circuits and devices which
-
will work in the 1 volt and at 2012 it's a 0.5 to 0.6 volt.
-
So, you see how the feature size goes down and at the same time the chip size is going
-
up as just now I mentioned to 80 millimeter square so they are speculating in 2012 it
-
will be 1580 millimeter square. The number of transistors in that regime is 1.4 billion
-
and that is wiring level. That means in the present all the microprocessor chips, the
-
seven or eight level of metallization is being done there. You know if the number of levels
-
is more the technology is getting much more complicated and obviously the yield is another
-
important point which is another yardstick. So, how your technology is good, how it is
-
commercially viable etcetera, so yield should be very good.
-
So, for that if number of levels of metallization increases, so then automatically you have
-
to compromise with some yield but you have to make a compromise on a good yield. At the
-
same time, number of wiring level increases that is the motto and target. And mask count
-
is going to increase if the levels are more. So that the interlevel metal you have to have
-
a dielectric layer then some metal pattern then again another dielectric layer so automatically
-
the mask level will also increase. So that was the present scenario.
-
Now the size what i am discussing does matters how you can see this figure here. So, this
-
is the the area of the micromachining and this Nano machining. So another word you are
-
coming across now is machining. So machining, the terminology initially it was in mechanical
-
engineering, people they used to point machining, that means from a huge bulk steel or any of
-
the metal beam they used to machine it to get small miniature structures. Well later
-
on the machining is used in microelectronics laboratory also to fabricate the MEMS devices.
-
Now here you can see the 1meter, that is that elephant and then is gone down to 0.1 so that
-
is the chip size is you say this is the area where 0.01 meter to 1 centimeter in that area
-
you can get the IC chip.
-
Then this is the size of 1 millimeter the grain of sand and then biological cell comes
-
nearly 0.01 millimeter to 10 micrometer and then
-
comes 1 micrometer which is the smallest feature size 0.35 micrometer is sometimes back may
-
be in 1996,97 where that level is micro. Now, if you go beyond that 0.1 micrometer below,
-
that is entering to the nano area and there some of the examples are shown that the atomic
-
littering using the scanning telling microscope. This is
-
1 nanometer, they are obtained from the feature size then DNA is of the 2 nanometer wide that
-
is the size.
-
On the other hand if you see the machining, this is the micro machine gear which is nearly
-
100 micrometer, then this is dust particle side you can compare with that which is a
-
1 to 5 micrometer and these are some quantum electronic structures, the width of these
-
structures in nearly 200 Angstrom and the atomic level 1 to 4 Angstrom. So that means,
-
the size does matters because the with the reduction of the size and minimum feature
-
size all the technology is going to change and at the same time the equipments are going
-
to change and if you go below say 1 nanometer level, then physics of devices is going to
-
change.
-
So now, if you look here, the size also on the scale that is bottoms up and here it stops
-
down. So both are shown, here the few areas are is basically plant and animal cell and
-
this is the bacteria size is the 1 micrometer to 10 micrometer and 10 to 100 micrometers
-
are plant and animal cell. This is the area where basically the MEMS
-
are made and now if you go beyond that, which is not microelectronics that is basically
-
coming nano scales or nano electronic their size comes below the 100 nanometer to say
-
here is say 1 nanometer. And below 1 nanometer the different area so that 100 nanometer to
-
1 nanometer, that the size of the virus, proteins and the helical turn of the DNA and this is
-
the minimum feature size of a MOS transistor in 2004 you can see here. So people are working
-
so 10 nanometer or 15 nanometer level, obviously, that is the nano scale for the nano and these
-
are 100 nanometers MOS transistor you can see the gate dielectric is from here to here.
-
So now, the science of miniaturization with that you can see in case of machine, how the
-
size is going to reduce. And here the size is basically an accelerometer and here is
-
a MEMS accelerometer. Now, side by side if you compare you see initially, the conventional,
-
the accelerometer, the mass was nearly 1.5 Kilo more than that and its size is 15 centimeters
-
by 8 by 5 centimeter and it requires power of the order of 35 Watts. And on the other
-
side you can see that, if a MEMS accelerometer if you use so that mass is only 10 grams so
-
instead of 1587, it has reduced to 10 gram and size is very small compared to these.
-
Power instead of 35 Watts, is only 1 Milli watt and cost instead of 20,000 Dollars, it
-
has come down to only 500 Dollars. So, at present the price is further down and you
-
can get nearly 10 or 50 or 100 Dollars to have MEMS devices.
-
So, although the microfabrication was originally limited to silicon, now the field is open
-
people have started using some other materials than silicon particularly the polymers and
-
ceramics and composite materials and also coming into the picture and people are using
-
those materials for the MEMS devices and microsystem fabrication. Why people are sticking at the
-
beginning only on silicon? There are certain reasons. Because silicon technology is fully
-
mature and there is no new research to be done for the processing of silicon. That is
-
why they do not want to deviate from the silicon. But later, on the other material particularly
-
polymer and organic material has some other advantage in certain areas, so they are going
-
to switch from the silicon to other nonconventional materials also non-silicon materials also
-
for making MEMS.
-
So now, the science of miniaturization, if you look into that, although lithography has
-
been the current method of defining patterns, many forms of direct right schemes are also
-
being used now a days. Traditional electroplating molding is being used in micro-domain. This
-
is not used in case of this normal microelectronics. But in case of MEMS we are going to use the
-
electroplating micro molding and in this liga, all these are coming into the picture and
-
all the techniques, I will discuss in detail in the future classes.
-
Now, this is a Micro or Nano world. If you look here, you can see various application
-
areas. Those application areas are basically, one is the say the physical MEMS or physical
-
sensors which deals with pressure sensor, force, inertial, sound means that these are
-
the area where MEMS devices are made. Other area is micro-optics, optical areas which
-
we call it as MOEMS Micro-Optical Electromechanical Systems. So there is the optical domain, there
-
are lots of devices made using mircomachining technology. Then another is micro-probing,
-
the STM and AFM. The AFM components are being made using the MEMS.
-
Others are the micro-fluidics. Micro-fluidics is a major application in the biology and
-
not only the biological application, there are others in flow of gas and flow of fluid
-
in a micro channel, that dynamics is a is a very interesting and there lot of devices
-
are being made now. This is another area of micro-fluidics. In Bio-MEMS, lot of work is
-
going and one emerging area of research at present Bio-MEMS area. And by the actuation
-
and motion that is basically the actuators are also very important. If you want to make
-
a microsystem, what do you need? You need the sensor, you need actuator and you need
-
the signal conditioning circuits or processing circuits.
-
So actuator is one part of the microsystem. So if you look into all the microsensor development
-
that will not complete the microsystem development. If you want to have some microsystem, you
-
have to see how the actuators can be made. And it cannot only make how it can be integrated
-
with the sensor and also the signal condition circuit. So that is why this is another area
-
of Micro Nano world, how can you make the actuator a precisely working actuator and
-
is very important which is integrable with the micro sensor.
-
Now let us look into the MEMS history and if you look there, the gentleman is a fine
-
man you know, and he basically got a Nobel Prize in 1959. In physics, again he had a
-
vision and at that time he declared that there is plenty of room at the bottom. That means
-
on silicon inside you can make lot of small miniature devices and that means there are
-
a lot of spaces available which people are not utilizing. That vision has come through
-
in 1959 that has been basically announced. In 1967, if you look that transistor basically
-
Resonant Gate Transistors came from Wastinghouse, and then in 1989 is another thing that is
-
basically MEMS device started. This is the picture which is basically 1 micromotor and
-
has been developed in UC Berkeley.
-
And then in 1991, you can see the MEMS device which is one digital light processing chip
-
that is used in DLP projector machine. And in 1993, these are the accelerometer, if you
-
see commercial MEMS accelerometers coming from a