| Next | Top of Page | Interestingly, most telephone companies have been investing heavily in the global conversion from copper wire to optical fiber because light does a better job of carrying information than does electricity. This is because photons (the basic unit of light) go faster, and have a higher
bandwidth than do electrons. Thus, photons are inherently more valuable
than electrons. If we can just get them to accomplish the logic tasks that
make computing work, they will become the next logical computing upgrade.
Over 65 major companies have invested heavily in the search for an inexpensive "nonlinear" crystal able to make one light beam turn another light beam on and off, which is a prerequisite for the production of a completely photonic (optical) computer. As explained below, their quest has yet to provide a practical elementary logic component.
Photonic logic, based on a different physical principle, is proving to be
the key to the production of a completely optical computing system. Such a
system would completely replace the start-and-stop surges of electrons
with tiny light beams that simply blink on and off, in order to carry
information and perform the logic of computing in light-speed photonic
computers... without slowing down the photons in some crystal or enslaving
them to some electro-optical process.
| FAQ | Photonics Menu | Home | Previous | Next |
Every since the 1930s, the advantages of light were recognized for carrying
information within the newly emerging computer science. The problem was that, back then, they lacked to tools needed to make light compute. As a result, the task fell to
electrons, and the electronic computer age was born. Since then, three major
events have laid the groundwork for the present effort at producing fully
photonic (optical) digital computers. The first was the invention of the laser.
Without the laser it was easy to see why researchers rejected light as a viable
computing medium. Ordinary optical signals are a noisy mish-mash of
difficult-to-control electromagnetic energy rather akin to
the noisy old spark-gap radio transmitters from the beginning of the 20th
century. Lasers, on-the-other-hand, produce much cleaner continuous wave
signals that are more like modern radio transmitters that can be used to
convey complex information.
The next discovery was the computer-generated hologram. Unlike lenses,
mirrors and so on, holograms can be calculated into existence using
the laws of optical physics. We can input any known optical signal, and calculate
the hologram(s) needed to direct and manipulate it in any way that the laws of
physics permit. When laser light is directed through the resulting hologram,
the light does exactly as we have calculated it to do. In a photonic computer
there will be many such optical signals. Each one will have its own
characteristics and will need to be redirected into some other configuration
in concert with many other such signals. Holograms allow us to do just that.
They allow us to interconnect optical components and even create optical
components. What's more, they can be mass-manufactured of inexpensive
glass, plastic and aluminum, just like the holograms on credit cards.
The third background element that has brought forth the photonic transistor
at this time rather than 20 some years ago when lasers first became readily
available has a more human element. Over the years a considerable
investment was made by those 65 companies, and a number of universities,
into the optical computing effort. This effort began long before lasers and
holograms. Work concentrated on electro-optical and nonlinear optical
methods of optical computing. Problems soon arose, problems that have
proven insurmountable.
First of all, any system that uses electrons in the process cannot be any faster
than the electrons themselves. Electronic signals are able to traverse a
microscopic electronic transistor and accomplish their assigned task in a little
less than a nanosecond. Light on-the-other-hand, travels 30 cm (about 11 3/4
inches) in that same nanosecond (a billionth of a second.)
A whole lot of photonic transistors can be placed in than same nanosecond
accomplishing entire computing tasks in the same time it takes a single electronic transistor to
even switch from off to on. Therefore, every electro-optical device will
always be hamstrung by the 'electro' part. If so, then there's not much reason to use light if it's
hung up on electrons.
Nonlinear optics exploits the optical properties of certain (rather expensive) materials that slow
light down to two different speeds at the same time. Two problems occur: the first is
obvious...the light is slowed down. The second is that in order to get light to
respond within such crystals so as to perform computing tasks, the lasers have to be so powerful
that the components get fried whenever one puts enough of them together to
do anything useful. The effort to produce inexpensive nonlinear crystals that
switch fast and at reasonable power levels has not been successful. In fact
such a substance has been dubbed "Unobtainium" by those in the
field. Their multi-million dollar effort has failed to produce marketable
optical computers. After all, how would you like to be the head of research at some
big company or university that has to go to the board of directors and tell
them that things just didn't work out, and all the money's gone? Worse yet, that some little outfit
in Montana found the secret right under their noses by examining optical effects
that they had rejected decades ago...(under entirely different technological
circumstances.) It's no wonder that both research and thinking gets
entrenched when budgets are at stake. So over the years, when ever anyone
suggested using optical interference, they were rejected out-of-hand based on
antiquated technology, and a strong desire to maintain the status quo.
However, there is one over-riding thing to remember. Interference based
Photonic Transistors work! They have been tested in the lab,
and we are continuing to make rapid progress toward the goal of replacing
nearly all electronics with photonics on a global scale.
What is Photonic
Computing?
Well, electronic computing uses electrons to perform the logic that makes computing work. Photonic computing uses photons of laser light to do the same job, only thousands of times faster. Electronic transistors are whittled into silicon wafers to make modern computer chips. Today's technology, however, is pushing the electron to its physical limits. As a result, the manufacturing processes are becoming increasingly expensive for producing even minor improvements.
However, photons are manipulated using inexpensive computer-generated
holograms made of plastic or glass. Photonic computers, therefore, will be far more valuable than their slower electronic counterparts, and far less expensive to manufacture.
Short History of
the Photonic Transistor:
In 1989 the Photonic Transistor was invented at the Rocky Mountain Research Center, and then tested in the laboratories of the University of Montana, and Montana State. In 1992, U.S. Patent 5,093,802 was issued to the Rocky Mountain Research Center. Since that time, the entire basis of
interference-based photonic computing has been growing and growing. Even
that crude first example was able to accomplish what the 'experts' said was
impossible, making one light beam turn another light beam on and off
without the use of some electronic gizmo in the middle.
How the first
Photonic Transistor works:
A "photonic" computer should use photons. Photons are the basic unit
of electromagnetic energy just as electrons are the basic unit of electricity.
Unlike the nonlinear optical materials that require a large supply of photons
to bias them up to some switching level, Photonic Transistors need only signal
levels of photons to work. Just as we can see certain things at night that emit
only a handful of photons per second, so too, the Photonic Transistor must be
able to operate using small amounts of energy.
The next desirable attribute is that they should always operate at the full speed of light. Unlike the so called SEEDs (Self Electro-optic Effect Devices) that are popular with the big budget people, Photonic Transistors do not use electricity in any way shape or form. The fundamental physical control and manipulation processes used do not slow down the light. The only retardation occurs during the very short time that the energy must pass through a dense medium such as a thin hologram. The Photonic Transistor is vacuum compatible, meaning that they can be operated in air or even in a vacuum where there light moves at the universal speed limit.
Optical Interference is a process of energy rearrangement that occurs when two laser beams pass through the same point at the same time. (Not that this is the only such circumstance, but it is the one under discussion at the moment.) The energy pattern forms an interference image that depends upon the wavefront pattern, input energy levels, and phase components of each the two input beams, along with the geometry of the encounter.
Interference has another very important property. If we accurately know all of the input parameters of all of the inputs, the output interference image formed can be calculated by a process called the "Linear addition of amplitudes" or the "Vector addition of amplitudes."
Since digital computers operate at discrete energy levels, (two levels in the case of binary logic,) Each two-input photonic logic gate will have 4 possible combinations of its inputs being either high or low...on or off. As a result, 4 different images need to be calculated, one for each input combination. During high speed computing, the interference image will switch continually among this set of images. Taken together, they form a "Dynamic Image".
Therefore. At any given location within the dynamic image, the amplitude, and thus the energy level (which is proportional to the square of the amplitude,) will change among 4 different static states as determined by the optical arrangement of the transistor. If we place an image component separator, such as a mask with a hole in it, at any location within the dynamic image, then any energy that shows up at the hole goes through the mask into the output. Any energy that does not show up at the hole is prevented by the mask from contributing to the output of the transistor.
If we select a location for the hole that switches through a set of energy patterns that match what we want this particular transistor to do, then the transistor will do what we want. The output will have a modulated waveform that is dictated by the optical arrangement and controlled by the states of the input beams. The transistor has one other feature of over-riding value. The calculated image set produces definite outputs through the hole that define the output energy in all of its states. This precise description of the output signal can then be used to accurately calculate the next transistor, lens, hologram or what ever. By mathematically stepping from one device to the next, we will be calculating into existence entire photonic computers.
Usually there are multiple sites in the Dynamic Image that have compatible energy-state series. Holes can be placed at these locations too, and its energy combined with that from the other holes to produce composite transistors with all different kinds of attributes.
Now that we have a method of calculating and manipulating known light signals so as to make them do all sorts of things: