Dr. William Phillips discusses time dilation, atomic timekeeping, and quantum computing
Dr. Phillips at NIST lab © Robert Rathe
“When you are courting a nice girl an hour seems like a second. When you sit on a red-hot cinder a second seems
like an hour. That's relativity.” Albert Einstein gave this humorous description to an inquiring journalist who
wanted to know more about the theory of Relativity. Certainly the human mind perceives time in a subjective way
– as Einstein alludes to – but what about clocks? Don’t they give us a universally stable reference we can depend
on to live our lives and make the trains run on time? Yes and no. Relativity tells us that time varies, but for most
practical purposes it’s negligible. You don’t need to consider it to be on time for a business meeting or sporting
event, but if you build global positioning satellite systems or atomic clocks it starts to matter.
Dr. William Phillips is a winner of the Nobel Prize in physics, professor at the University of Maryland, and a
physicist at NIST (the National Institute of Standards and Technology). Relativity and time dilation are a part
of his workday as he advances the science of atomic timekeeping. He’s also a very nice man and was willing to spend
some time on the phone with me discussing these very complex subjects. Following is a transcription of our
Q. Do you have a favorite definition of time?
Yes. The one I like best is the one given by Einstein which is “time is what a clock measures.” It sounds like
kind of a flip answer, as if it doesn’t answer the question, but in fact it does. By taking seriously the idea
that time is what a clock measures Einstein was able to come up with a deeper understanding of the nature of time
than had been the case before. In particular, he thought about a particular kind of clock and he imagined what
would happen if this clock were moving with respect to an observer. He came to the conclusion that the observer
would see that clock as ticking more slowly than if the same clock were standing still with respect to the
observer. Taking the idea seriously, that time is what a clock measures, you come to the conclusion that time is
running more slowly for the person who is moving from the observer’s point of view. We now know, from the basis
of experiments, that this is true.
Q. From my reading on the NIST website it seems there are a lot of different things
that you do, only some of which have to do with atomic clocks and time keeping.
My work is not so much directly working with atomic clocks and time keeping. Rather, the things that my
colleagues and I have done have made it possible for scientists in other laboratories to make better clocks.
So we are more of the providers of the technology that has allowed a new generation, so to speak, of atomic
clocks to be made by others.
Q. In the first question you are talking about how time is relative to what is
Time is relative to the reference frame in which you are measuring it. So if I am sitting at rest in my
laboratory and I see somebody who is moving with respect to me and I look at their clock I’m going to conclude
that their clock is running slower than my clock – even though the two clocks are completely identical in every
other respect. Now that observer, in his moving laboratory is looking at me – motion is just relative – he’s
going to come to the same conclusion, that my clock is running slower than his. This sounds paradoxical but
it is completely consistent with everything we know about the way the world works. Let’s say that you are in
your lab and your colleague who has a completely identical clock goes off on a trip. You look at what is
happening to your colleague and conclude that your colleague’s clock is running slow. Your colleague looks
back at you and concludes that your clock is running slow. Let’s say the colleague comes back from the trip.
Well, whose clock is slower? The person who goes and comes back is the one whose clock is slower. In order
to go away and come back you have to turn around, and there is no ambiguity about turning around. There is
ambiguity about who is moving and who is not moving, because motion is relative. This experiment has been
done in a number of different ways. One way is to get two atomic clocks and put one in an airplane and send
it on a trip. When it comes back you see what time it is registering and you find out it is a little bit slow.
There is no question that this actually happens. Einstein came to the conclusion that time is running at a
different rate depending upon who is looking at it and from what reference frame (whether you are moving with
respect to some other thing you are trying to observe).
Hafele-Keating experiment: a test of the theory of relativity in October of 1971. Joseph C. Hafele and
Richard E. Keating took four cesium-beam atomic clocks aboard commercial airliners and flew twice around
the world, first eastward, then westward, and compared the clocks against those of the United States
Naval Observatory. There was a difference.
Q. So that is the experiment that you were talking about with the atomic clock
and the plane?
Yes. That experiment has been done and it verifies what Einstein predicted.
Q. And they also factor that into the GPS system?
Absolutely and even more subtle things have to be factored into GP Systems. The Global Positioning Systems
have a constellation of satellites and each has several atomic clocks onboard. Obviously these satellites are
moving and are relativistic facts. Furthermore, as the satellites are at a different altitude than we are,
another theory of Einstein’s, called the “General Theory of Relativity,” that has to do with the understanding
of gravity, also has an effect. Clocks closer to the earth run slower than those that are higher because of
gravitational influence. All these things are taken into account, as well, in order to understand the Global
Positioning Systems. You might say that the GPS is one of the best confirmations of Einstein’s theories.
If we didn’t take into account what Einstein told us about time then the GPS just wouldn’t work.
Q. So the fact that they can verify positioning by this system and verify that
it’s occurred is proof of the Relativity Theory?
Yes, of both the General and Special Relativity Theories.
Q. If you took out those adjustments that they make then your accuracy would
be way off?
That’s right. In fact, I heard a story of that when they first set up the GPS that they weren’t entirely
sure about some of the corrections. So, in order to cover themselves, the engineers were able to look at the
data both ways – both with and without the General Relativity corrections. It became clear after a very short
period of time of running they were able to tell which method was right. It was the one based on Einstein’s
Q. During his time were there any ways to experiment
& determine if he was
right or not?
There were certainly things at that time that could be checked. For example, when his theory of General
Relativity came out, one of the things that it predicted was that when starlight went by the edge of the sun
the starlight would bend by a certain amount because of “gravitational lensing”. It wasn’t long after that
an expedition went out to measure a total eclipse of the sun, when you could see the stars near the sun,
and the measurements they made confirmed what he had predicted.
Q. You’re group leader of the laser cooling and trapping group at NIST and you
experiment with cooling atoms with laser light. When I think of a laser I think of a beam of light imposing
kinetic energy on whatever the beam lands on, but you’re sort of doing the opposite?
That’s right. The usual thing is if you shine laser light on something the light is absorbed and that
something heats up. So what we do is quite counter-intuitive. So how does that work? Besides transferring
energy to stuff when light is absorbed, it also transfers momentum – that is to say light pushes on things.
So if you could arrange to have the light push on the atoms in such a way as to make them slow down then that’s
equivalent to cooling down the atoms.
Q. So it’s like a car rolling towards you, and if you start pushing against it
then your opposing force against the car slows it down?
A better analogy: lets say that a car is rolling toward you and you start firing ping-pong balls at it from
a repeating gun that can fire thousands and thousands of ping-pong balls. If you could fire enough ping-pong
balls at it you can slow it down. The atom is really a big massive thing with lots of momentum and the light
that you are shining on it can be thought of as a stream of photons … particles of light… and each of them
carries a tiny amount of momentum. But the action of a large number of them together allows you to slow the
Q. Are you slowing the entire atom or the electrons?
We’re slowing down the center of mass. The electrons are basically a separate thing. So what we do is
separate the center of mass motion of the atoms (in our minds) from the motion of the electrons. It is actually
a very good approximation to say that the atom as a whole is moving along and then the electrons are moving
with respect to the nucleus, and I don’t have to think about the two things together. I can separate out those
The NIST-F1 Cesium atomic clock. It is so accurate that it will neither gain nor lose one second in more
than 60 million years.
Q. You’re slowing it for what purpose?
One of the reasons is to reduce the Relativity effect that makes moving clocks run more slowly. Our atoms
are the tickers for the atomic clock, and if they are moving, they will appear to tick more slowly. So we want
to reduce that and eliminate, or nearly eliminate, one of the important sources of uncertainty in atomic clocks
that use faster atoms. There are other reasons also. When atoms are moving very fast you simply don’t have much
time to look at them. By slowing them down we give ourselves more time to look at them and therefore we can make
Q. So the atoms you are slowing down are being used to provide the time-keeping
pulses of the clock?
Exactly. We developed the techniques for slowing down atoms. In particular we developed techniques for the
ones that are used in primary atomic frequency standards. That atom is Cesium 133 – an isotope of the Cesium
atom – the one that is most common and radioactively stable. It had been agreed upon as being the atom that
one uses for the definition of time. The “standards” clocks all use Cesium atoms. We developed laser-cooling
procedures first with Sodium, then with Cesium. Then other people used these techniques to make atomic clocks
using laser-cooled Cesium atoms.
Q. If the atoms within the atomic clock are whizzing around too fast then
Relatively enters into it and affects the accuracy?
Exactly, relativity enters into it and is one of the features that limit the accuracy.
Q. What are the challenges of working at the atomic level? Obviously you can’t
see what you’re working with, correct?
That’s not entirely true. I know that in popular thought – the thought that I had for a long time – was
that atoms are too small to see. That’s not really true, however. It would be sort of like saying that stars
are too far away to see – but we do see stars. We see stars with the naked eye. We just don’t see how big they
are. As far as we can tell stars are just points of light. The same is true of atoms. We can see atoms by
shining light on them the same way you see most anything else. But because the wavelength of the light is bigger
than the size of the atom we can never tell how big the atom is by shining light on it and seeing the reflected
Q. You aren’t entirely working blind then?
No, measurements are never blind in the sense that we always have ways of making our measurements. It’s
not often that we care whether we can actually see something with our eyes as long we have instruments that
allow us to see what we are measuring. Except in this instance we can in fact see the atoms with our eyes by
using the reflected light.
Q. For experimenters, there’s often the issue of signal to noise ratio. When you’re
getting down to something as small as an atom there is certainly a very small signal to work with. When your
instruments are telling you certain things how do you trust that it’s a signal and not noise since you’re
dealing with such a minute quantity?
For most of the experiments we do in my laboratory we don’t look at single atoms. So when I say that we can
look at single atoms I’m referring to experiments that are done in other laboratories where they do look at
single atoms. I can tell you exactly how that’s done and how they can tell the difference between signal and
noise. We intend to be doing experiments with single atoms in our own laboratories and use the same techniques
that other people have successfully used. Let’s say that I got a single atom and I’m holding it in a “trap”.
How it works is a combination of magnetic fields and laser beams that hold an atom confined in a certain region
of space. The laser beams illuminate the atom; the light from the atom comes off and we can collect it with a
lens. Then take the light from the lens, put it onto a photo detector, and see how much light there is. As you
point out, there isn’t much light from a single atom. But it is enough for us to measure. To give you an idea,
we shine laser light on the atom; the atom may scatter 10 million photons per second – that sounds like a lot –
but we may get to collect only a tiny fraction of that – maybe one percent if we’re lucky. Of the ones we
collect we may not have a very high efficiency rate of detecting all of them. In the end, it’s typical that we
may be seeing hundreds or thousands – maybe tens of thousands of photons per second if we’re lucky from a single
atom. But it turns out that’s still pretty good. Because the noise comes from the fact that everything else in
our experiment will also scatter light from the laser. What we try to do is be very careful and not have the
laser hit anything else. It’s never perfect so there’s always a little extra light coming back. Our photo
detector has a certain level of what we call “dark counts”. This would be the noise in the detector itself.
Things that look like signal but are noise. Very often that can be made quite low - a few per second. Depending
upon the nature of the detector sometimes it’s higher than that, but a few per second is not impossible.
How do we tell when we are seeing this signal that it’s an atom? One way is just don’t put an atom in there
and then see what it looks like. Then when you put an atom in there if the light that comes out is a lot higher
then you know that you’ve detected an atom. A very common thing for people to do is to trap a few atoms, maybe
three, and then look at the amount of light coming out. After a certain amount of time you lose one because your
vacuum wasn’t perfect and some of the background – the residual air that was left in one of your vacuum systems -
hit one of the atoms and knocked it out. There can be any number of reasons why you might lose one. You wait a
bit and the light goes down a little bit. You wait a little bit and then the light goes down a little bit
further. Eventually you lose all your atoms. This might happen over a period of seconds or minutes. This way
you can see each individual atom as it leaves your trap, for whatever reason, and you know exactly how much
light comes from each individual atom. Then the next time you look you can tell how many atoms you got just
by how much light is coming out of there.
So that’s the sort of thing that’s done and it’s easy to tell the difference between the signal and noise
because atoms don’t behave the same way as the noise. If I change the color of my laser just a little bit,
then the atom won’t respond anymore because the atom only scatters light if the color is just right. But all
the rest of the stuff scatters in exactly the same way and the noise on the detector doesn’t change any, so if
I change things just a little bit and all of a sudden the signal changes then I know there must have been an
Q. So certain atoms only reflect laser light of a certain wavelength?
Every atom has what is known as its “resonance frequencies”, which is a certain frequency range that
corresponds to the resonance in that atom.
The first “universal” programmable quantum information processor able to run any program allowed by
quantum mechanics. -
Credit: J. Burrus/NIST
Q. My next question deals with quantum entanglement. Einstein had described
it as “spooky action at a distance.” He didn’t theorize quantum entanglement himself. Someone else presented
the idea to him and I understand he didn’t much care for it.
It’s perhaps a little bit more complicated than that in the way it worked out theoretically. It’s true that
Einstein did not come up with the idea of quantum entanglement. I believe that it was Erwin Schrödinger who
first described the idea of quantum entanglement. I think that he gave it its name in German, which translated
to “entanglement.” What Einstein did was think about this question of quantum entanglement more carefully than
most other people had. He came to the realization that if you believed in this theory of quantum mechanics,
with the idea of quantum entanglement, that it would imply some rather bazaar things. This was that “spooky
action at a distance” that he didn’t much like. He thought that the behavior that quantum entanglement implied
was just too weird. He just couldn’t believe that the world would work that way. It just didn’t seem
reasonable. In a sense he was right – it isn’t very reasonable. It leads you to conclude that things do not
have properties until you measure them. And that the properties of an object that you have in your laboratory
are not just dependent on the things in that laboratory, but those properties might also depend on things that
are going on in another laboratory. That seems really weird. But as it turns out it’s true. So Einstein's
intuition was correct – plus his careful thinking about it – in that entanglement implies really weird stuff.
What he was wrong about was that nature can in fact be that weird. He didn’t think nature could be that weird.
In fact, nature can be that weird.
Q. So then your opinion of quantum entanglement would be that it’s true.
Have there been experiments that have proven it?
Absolutely. I believe it is true is because the experiments have shown it without any question.
Q. What’s an example of one of those experiments?
The most famous set of experiments were done by a friend of mine named Alan Aspect (pronounced “as-pey”).
In the early 1980s he did a set of experiments to test whether nature is as strange as quantum mechanics
predicts. Effectively he was doing experiments to test whether Einstein and his colleagues (Einstein,
Podolsky, and Rosen were the authors of the famous paper that called this into question.) were correct about
the features that would exist if quantum mechanics were correct. The results convinced almost everybody who
had any doubts. Now, there are always people who have lingering doubts, and will wonder if maybe there is a
loophole, and think that we should do the experiments more carefully. Every time the experiments have been
done more carefully however, (many of those more careful experiments were done by Aspect himself) it always
turned out that quantum mechanics proved to be the right description.
Q. Is the premise correct that ‘physically distant particles will change in
some way in unison with out any apparent mechanism or any communication between them?’
That’s a reasonable description of what is happening, but I might take issue with the question of
“changing”. The result of a measurement on one particle is linked to the result of a measurement on another
particle even though there is no connection between the two particles; there’s no way of the two particles
communicating or interacting with each other. In other words, the fate of the two particles upon measurement
are entangled, or linked. Lets say that the measurement is whether the particle is spinning clockwise or
counter-clockwise. There is no way I can predict whether any particular one that I measure is going to be
spinning clockwise or counter-clockwise. If I measure one of them to be clockwise, then depending on how the
state was prepared, the other one is going to be counter-clockwise even though it is in a different laboratory
and there is no connection between the particles. That’s the nature of entanglement, and that is the thing
that Einstein couldn’t handle, because it goes against our intuitive idea of the nature of reality. But as
it turns out Einstein was wrong about this one.
Q. So it doesn’t necessarily mean that if the first particle you looked at was
spinning clockwise and you were able to grab it, stop it, and then spin it counter-clockwise, that its twin
particle somewhere else would then stop and spin in the opposite direction as well?
Certainly not, that’s not what it means at all. It just means that if you measure the thing then the
measurement on the other one will be correlated.
Q. That’s a good clarification, because a lot of things that I’ve read
suggested that whatever happens to one particle will happen to the other particle in unison or at the
Yes, but not because you did something to it. For example, this cannot be used to communicate faster
than the speed of light. A lot of people misunderstood about that, and that will not work. You couldn’t
do something here and have it affect what goes on over there in such a way as to send a message to somebody
at a distant place faster than the speed of light. The correlations will develop faster than the speed of
light. But no information could be transmitted because you have no way of knowing, nor of affecting, what
the result of the measurement will be.
Q. Is there an experiment that you would like to perform but can’t because
of physical or resource limitations?
There is always the experiment that one would like to do and the daily work of an experimental physicist
is getting around limitations. There are certainly things we dream about. One of those things is having a
quantum computer; something that could do computations using bits of information that were represented by
quantum objects. For example, atoms; we were talking about this hypothetical particle that was spinning
clockwise or counter-clockwise. You could say for clockwise I’m going to call “one” and counter-clockwise
I’m going to call “zero” as binary digits. The interesting thing about a quantum mechanical object is that
you could have it in the state where you don’t know if it was zero or one. In a certain very real sense it
is both zero and one at the same time until you measure it, and that makes it possible to do computations
that you can’t do with ordinary computers. The computations that you can do with such a quantum computer
would allow us to solve problems in quantum mechanics that for the moment are too difficult to solve on
ordinary computers. Having such a device is something that I would very much like. In fact, being able
to build such a device is what we are in the very early stages of trying to do right now. Having such a
device would allow us to answer some very interesting questions in quantum mechanics. It might allow us to
address/test some new ideas that people are coming up with that suggest different ideas from the standard
ideas of quantum mechanics.
Q. You said that a particle could have two states until you measure it?
Let’s say that I have a single particle and there are only two possibilities for this particle. It can
spin clockwise or counter-clockwise. I can construct a state – and I know exactly how to do it repeatedly
and very accurately – called the super positioning of it being clockwise or counter-clockwise. This is
something we do all the time in the laboratory. We make a particle that is spinning both clockwise and
clockwise at the same time. It literally has both states. This itself is essentially impossible to
imagine for a macroscopic object. You couldn’t imagine a top that was spinning in both directions at the
same time. But when the top is an electron then we can and do it all the time. If one direction of spin
represents the digit ‘one’ and the other represents the digit ‘zero’ then that means I’ve got both digits
represented at the same time. One way of thinking about how a quantum computer gains its power is by having
it do a calculation on both numbers at the same time. If we have lots of these then it turns out that a
quantum computer can be doing a calculation on lots different numbers at the same time. That allows it to
have a speed-up of the calculation and that makes all the difference.
Q. So you get an exponential speed advantage vs. a transistor or logic gate
that is simply either on or off?
You get an exponential speed up of the computation time. Let’s say that I want to do a computation on
a certain number that has a certain number of digits. The amount of time it takes to do the computation
grows exponentially with the number of digits if you’re using a standard computer. There are a lot of
computations that are like that and they’re hard problems to solve. A quantum computer can do that
calculation but it does not grow exponentially with the number of digits – it only grows like a polynomial
with the number of digits. That makes all the difference in the world for whether you can calculate the
answer or not in a limited time.
Q. Do you think this can be made reliable enough for critical, real-world
Dr. Phillips speaks at the World Science Festival about atomic clocks
Yes, but it certainly is a challenge. Making it reliable enough is something we’re working on.
Fortunately, it turns out that it doesn’t have to be perfect – that’s the wonderful thing. There are
procedures that have been developed theoretically for correcting any errors. So if we can make it reliable
enough then we are home free – we don’t have to make it perfect.
Q. It doesn’t have to be perfect? I once heard a NASA scientist say,
“we’re always one transistor away from failure.”
Well that’s true. Certainly in those kinds of situations you generally build in redundancy, so that if
something fails you’ve got a backup. In a certain sense, that’s the way error correction works.
You have redundancy, so if something bad happens you can tell that it happened and you can fix it.
Einstein once said, “Things should be made as simple as possible – but no simpler.” □