Stable lasers

by user

Category: Documents





Stable lasers
Laser Stabilization
Kevin Cox
Advisor: Jun Ye
Over the past fifty years, lasers have become one of the most valuable tools for studying
physics. Monochromatic, phase coherent laser light can be used for experiments which were
impractical or impossible with traditional light sources. Of course, since the laser's
invention, fundamental limitations on a laser's freqency stability and phase coherence time
have been recognized [1]. Yet, most lasers are limited by technical noise sources such as
vibrations, temperature fluctuations, and thermal noise of the resonator mirrors. This
instability limits a laser's usefullness for applications such as precision spectroscopy,
measurements of physical constants, and atomic clocks. For this reason, significant research
continues in the field of laser stabilization. The Ye goup at JILA is currently endeavoring
to achieve a laser with linewidth below .5 hertz. This result will match the best laser
stability ever recorded and increase the precision of the group's strontium atomic clock. In
this paper, I will report on this experiment and the progress made during a ten week
period with the group. Furthermore, this paper will serve as a brief introduction to the
Pound-Drever-Hall method of laser stabilization which is commonly used in ultra-stable
laser systems.
Laser Stability
A common and useful way to characterize a laser's frequency noise is by its spectral
density. The spectral density of frequency variations is defined as
Sv ≡
(f )
[Hz 2 /Hz].
Where b is the bandwidth of the frequency noise, and νr.m.s.
is the \power" of the frequency
excursions at a given frequency (note the peculiar units Hz 2 /Hz). The frequency spectral
density of a laser is the cause of its linewidth, ∆ν, and lineshape. Of course, the lineshape
of a laser is the same as saying the spectral density of optical power, Se . Given a constant
frequency spectral density over a bandwidth b, if the noise bandwidth, b < ∆νr.m.s. , then
the lineshape is Gaussian. If b > ∆νr.m.s. , the lineshape will be Lorentzian. In general, a
laser's lineshape is a convolution of the two. [2] In either case, the linewidth is proportional
to the amplitude of the frequency modulation, νr.m.s
∆ν / Sv (f )
Is it accurate to assume the frequency variations of the laser are uniform? The answer is,
sometimes. The careful physicist must calculate Se from the laser's autocorrelation
function [3]. Yet, in general, common noise sources contribute to the noise over known
frequency ranges. Figure 1 shows a plot of limiting noise sources over a 100kHz bandwidth
for the current clock laser setup.
Figure 1: Spectral density of frequency fluctuations for a 698nm laser stabilized below 1
kHz. Thermal noise is the dominant noise source at low frequencies and dominates the
instability of the laser.
From Figure 1, we see that the limiting factor to laser stability is thermal noise. This
thermal noise is caused by fundamental fluctuations in the mirror substrate (Brownian
motion) at nonzero temperatures. The job of the physicist stabilizing a laser is to decrease
this noise spectral density as much as possible over a large bandwidth. To do this,
Fabry-Perot cavities and feedback control techniques such as the Pound-Drever-Hall (PDH)
method are used.
Fabry-Perot Cavities and PDH Stablization
Fabry-Perot (FP) cavities are useful for stabilizing a laser because they act as a constant
frequency reference. More precisely, FP cavities resonate at distinct frequencies governed
by the relation,
ω = nc/2L,
where n is an integer and L is the length of the cavity. This equation is easy to understand
using wave optics. If the laser frequency agrees with the condition of Equation 3, for each
pass through the cavity the electromagnetic field, the light will constructively interfere.
More complete descriptions can be found in a good optics book [4]. For a cavity with
perfect mirrors, the resonant condition is exact. However, if the mirrors have loss–and they
always do–then the cavity has resonant peaks with width, δν centered around the resonant
frequency in Equation 3. In fact, the linewidth of the cavity is best characterized by the
cavity's finesse. Finesse is the most useful quantity used to characterize an FP cavity.
F =
where ωF SR is known as the free spectral range of the cavity and is defined by the frequency
spacing between cavity resonances, c/2L. The finesse of a cavity is solely dependent upon
the optical losses as light travels between the mirrors. Cavities with a finesse of over
100,000 are used to construct ultra-stable lasers with linewidth below one hertz.
The goal of the physicist stabilizing lasers is to compare his laser's frequency with that
of the cavity and use a feedback system to correct for the error. The laser transmission
through an FP cavity is shown below in Figure 2.
Figure 2: Transmission coefficient from a Fabry-Perot cavity.
A simplistic way to lock the laser frequency would be to measure the laser transmission
from the cavity in order to lock the frequency to some point detuned slightly from
resonance. One would not lock directly on resonance with this method since the
transmission signal is symmetric around this point. The feedback system would not know
which direction to apply a correction. For this reason the range of a suitable error signal is
limited to one side of the transmission signal. More importantly, this method is flawed
because the error signal would be sensitive to amplitude noise in the laser as well as
frequency noise. Pound-Drever-Hall stabilization addresses both of these problems.
The golden idea of PDH stabilization is to modulate the laser's frequency and monitor
the laser's reflection from the FP cavity in order to obtain a more useful error signal. This
error signal is obtained by measuring the amplitude and phase of the reflected light. There
is no device which can directly measure the phase of an oscillating electric field, but the
PDH method gives us a simple and clever way to make the measurement. For a FP cavity
with no losses, the reflection coefficient for the electric field is given by,
F (ω) =
r(exp(i ∆νωf sr ) − 1)
1 − r2 exp(i
∆νf sr
For a laser modulated with a modulation amplitude, β, the electric field can be expanded
as a series of Bessel functions and written [5],
E(ω) = E0 [J0 (β)eiω + J1 (β)ei(ω+Ω)t − J1 (β)ei(ω+Ω)t ]
Where ω is the laser frequency, Ω is the modulation frequency, and J0 and J1 are the
Bessel function coefficients of the series expansion. Equation six shows how modulating the
electric field creates an incoming wave which appears in frequency space as a large carrier
with two sidebands. How can we conceptualize this new modulated laser interacting with
the FP cavity? The system simply behaves as if three waves were interacting with the
cavity: The carrier with frequency, ω, and the two sidebands with frequencies ω + Ω and
ω − Ω. The reflected light is this electric field with each term multiplied by the reflection
coefficient at the corresponding frequency. The magnitude of the electric field squared is
the power of the reflection signal.
P =Pc |F (w)|2 + Ps [|F (ω + Ω)|2 + |F (ω − Ω)]|2
+ 2 Pc Ps Re[F (ω)F ∗ (ω + Ω)
− F (ω)F ∗ (ω − Ω)]cosΩt + Im[F (ω)F ∗ (ω + Ω)
− F (ω)F ∗ (ω + Ω)]sinΩt + (2Ωterms)
This is measured in a photodetector. The important terms in this measured signal are
those that oscillate at the frequency Ω. The goal is to demodulate this oscillating term to
measure the factor in front which contains the error signal. This can be done with either
the sine or cosine term. The solution is found by combining the reflection signal with a pure
sine oscillation from the local oscillator. We then have an electronic signal which contains,
Im[F (ω)F ∗ (ω + Ω)
− F (ω)F ∗ (ω + Ω)]sinΩt ∗ sinΩt
The constant term out in front is the error signal, E(ω), and the sine sqaured term becomes
sin2 (Ωt) = 1 − cos2 (ωt)
Aha! By mixing the two sines together, we have created a demodulated term with the error
signal. Experimentally, the rest of the measured signal is thrown away with a low pass
filter. Also, in real setups, the phase of the two sine terms is never necessarily equal, so a
phase shifter is used to create the correct error signal shown below in Figure 3.
Figure 3: Measured error signal from a cavity with F=1000. The red trace shows the
measured optical carrier with sidebands.
Now that a useful error signal has been produced, the laser frequency is controlled using
standard feedback methods.
Technical Considerations
The slope of the error signal around resonance, shown in Figure 3, is very nearly [5]
4 Pc Ps
where Pc and Ps are the power of the laser at the carrier frequency and sideband frequencies
respectively. ∆ν is the cavity linewidth. So, cavities with higher finesse have a larger
sloped error signal compared to the noise of the system. This allows for greater control.
Cavities with relatively lower finesse, perhaps 100 or 1000, are usually limited by quantum
shot noise at the photodetector in the PDH configuration. Since the slope of the error
signal is not as large as that of the ultra-high finesse cavities generally used, the shot noise
fluctuations limit the sensitivity of the setup to smaller frequency deviations of the laser.
Another technical consideration is mode matching to the FP cavity. Traditional FP
cavities with planar mirrors are rarely used in practice because they are quite sensitive to
misalignments. Cavities with spherical mirrors are preffered. The standing waves of this
type of cavity are actually Gaussian beams [4]. The lasers being used must be focused
correctly so that the beam divergence matches the radius of curvature of each mirror.
If a laser is locked to a cavity, fluctuations in the cavity length contribute to noise just
the same as inherent laser frequency fluctuations. Because of this, all FP cavities used for
laser stabilization must be precisely temperature controlled and constructed of materials
with low thermal expansion coefficients. Temperature control becomes an important
technical aspect to consider when constructing a laser stabilization setup.
Lastly, PDH stabilization suffers from a sensitivity to so-called \Residual Amplitude
Modulation" (RAM). When our laser is frequency modulated at frequency Ω, amplitude
modulation is introduced at the same frequency. This spurious RAM will be recieved on
the photodetector as a term with sin(Ωt) oscillation and will be demodulated into the error
signal. As the RAM changes amplitude over time, possibly with variations in temperature,
the baseline of the error signal can fluctuate. For lasers locked to high finesse cavities, this
can be a serious problem. The best solution currently known is to prevent RAM from
leaking onto the signal.
Experimental Setup: 40cm Cavity
The Ye labs strontium atomic clock has achieved an overall uncertainty of 1 ∗ 10−16 . The
linewidth of the clock laser is .5 hertz. Despite this fact, laser stability is still the limiting
factor in the precision of the atomic clock. Earlier in this paper, Figure 1 showed us that
the limiting factor for this stability is currently thermal noise. One way to reduce the
contribution of this noise is to design a longer cavity. Indeed, this is one of the ways that a
new 40cm cavity stabilization system can improve the current clock laser stabilization.
This cavity, which is currently being setup for integration into the strontium clock, is
shown below in Figure 4. The new 40 cm cavity is expected to have a factor of 10 increased
insensitivity to thermal noise. This is in part due to the increased length of the cavity.
Also, the mirrors of the 40cm cavity are coated with fused silica substrate instead of the
ultra low expansion (ULE) coating, the coating on the current fabry perot cavity. Fused
silica exhibits lower thermal noise than ULE material.
Figure 4: 40cm fabry-perot cavity made of ULE material with fused-silica mirrors.
The Pound-Drever-Hall locking scheme discussed above will be used to create a 698nm
stable laser system with this optical cavity. However, before the diode laser can address the
super-cavity, it must be \pre-stabilized" to a linewidth around 1kHz. This was
accomplished with a smaller optical cavity with a linewidth of 2.8MHz and a finesse around
1000. A diagram of the experimental setup for PDH locking is shown below in Figure 5 [5].
Figure 5: Schematic for a PDH setup. The reflection signal from the optical cavity is
detected and combined with the local oscillator signal in order to demodulate the error
From this setup, the spectral density was measured using a Fourier Transform machine.
The extrapolated linewidth of the laser was 550 200Hz. The stabilization lock was seen
to have a bandwidth of nearly 2.5MHz which is typical for a PDH locked laser.
The current atomic frequency standard created by the strontium clock in Ye labs has
reached a total uncertainty of 1 ∗ 10−16 . This uncertainty is less than the national
standard–the Cs ion clock housed at the National Institute of Standards and Technology
(NIST). By creating a more stable laser system, lasers will be able to probe the narrow
transitions in strontium even more precisely. The current laser stabilization system uses a
vertically mounted 7cm Fabry-Perot cavity with a finesse around 200,000 [6]. A new
system, consisting of a 40cm horizontal cavity with comparible finesse is currently being
integrated into the atomic clock system. The 40cm cavity is expected to have a factor of
ten less sensitivity to thermal noise, or thermal fluctuations in the mirror substrate. My 10
week project project was to setup the prestabilization of the 698nm laser to a
prestabilization cavity. After the prestabilization, the laser's stability will be comparable
with the 40cm cavity's linewidth. The prestabilization achieved a linewidth near 550hz, a
very suitable stability which will allow further stabilization to the 40cm cavity.
[1] A. L. Schawlow and C. H. Townes, \Infrared and Optical Masers," Phys. Rev. 112,
19401949 (1958).
[2] D. S. Elliott, R. Roy and S. J. Smith, \Extracavity laser band-shape and bandwidth
modification," Phys. Rev. A, 26, 12 (1982).
[3] John L. Hall and Miao Zhu, \An Introduction to Phase-Stable Optical Sources,"JILA
Reprint No. 4738.
[4] B. E. A. Saleh and M. C. Teich, \Fundamentals of Photonics", 2nd ed.(John Wiley &
Sons, Hoboken, NJ, 2007).
[5] Eric D. Black, \An introduction to Pound-Drever-Hall laser frequency stabilization,"
Am. J. Phys. 69, 79-87 (2000).
[6] Mark Notcutt, Long-Sheng Ma, Jun Ye, and John L. Hall, \Simple and Compact 1-Hz
laser system via an improved mounting configuration of a reference cavity," Optics
Letters 30, 12, 1815-1817 (2005).
Fly UP