Frequency stabilization of a diode laser by means of temperature
stabilization and passive resonant optical feedback
Kevin Henderson*
Department of Physics
and
JILA, University of Colorado, Boulder, CO 80309-0440
(August 5, 1999)
A single mode 635 nm 15 mW diode laser is built into a small module that is
temperature stabilized to within 0.05 ºC via a Hytek™ HY-5600 TEC controller.
This stability, however, only corresponds to a 100 MHz linewidth, or one part in
106. To improve this performance, a special holographic transmission grating is
used in an external optical feedback cavity that forces the semiconductor to
automatically lock its frequency optically to the cavity resonance. This method
can, in theory, be used to stabilize the laser frequency and reduce the linewidth
by at least a factor of 102 so that the overall stability corresponds to a 1 MHz
linewidth. Lastly, criteria are given for a small, hand held diode laser module
design, and a discussion is given for various applications.
Under most circumstances, the high sensitivity of diode lasers can generally lead to
disruptive effects in optical output, like mode hoping or intensity fluctuations. Under
well-monitored conditions (i.e., highly sensitive servo-systems), this sensitivity to
feedback can be put to advantageous use. Optical feedback techniques that are either
active or passive can narrow semiconductor-laser linewidths substantially and also
automatically stabilize the laser’s oscillation frequency.
However, these optical
techniques generally require that the diode laser be both temperature and current
controlled in order to be effective at locking in on one specific frequency. Optical
techniques, either active or passive, also generally require tuning (i.e. specific
*
REU student from the University of Arizona, Tucson, AZ 85715. Advisor: Dr. Dana Anderson,
Department of Physics and JILA, University of Colorado, Boulder, CO 80309-0440.
attenuation, reflectivity, geometry, etc.) for each diode laser. Fortunately, given the
intended purposes of the proposed diode laser module, a particular frequency is not
important. Rather, it is only crucial to maintain a single frequency with short-term
stability (< 1 hr or 1 day depending on use) and a narrow linewidth. In other words, the
frequency of the diode laser can be allowed to be different from day to day, however,
once it is turned on and is temperature stabilized, its short-term stability will not be
compromised.†
In order for the features described above to be actuated effectively, each part of the
diode laser module is tested for stability. The testing includes isolating any errant
features of the temperature servo-systems as well as the optical feedback loop and then
bringing the two systems together in order to optimize the stability of the diode laser
output. Demonstrating temperature stability of the diode laser, however, turns out to be a
difficult task. Initial examinations of the temperature stability of a diode laser frequency
output usually do not reveal a marked difference from standard output of a diode laser
without any temperature stabilization. There are essentially two main reasons why there
is a failure to detect a difference. First, the Hytek™ TEC controller can have poor
sensitivity in its feedback or be “tuned” improperly. Thermally faulty connections of the
thermistor to the diode laser mount can cause the Hytek™ TEC controller to become
ineffective as a driving servo-system. Likewise, poorly located thermistors can cause
thermal lag in the servo loop. Current limit and set resistors must also be properly
selected otherwise the circuit will not work as expected. Secondly, the resolution of the
†
Commercial diode lasers are normally wavelength rated as (nm) ± , where  is usually 5 to 10 nm.
spectrum analyzer (in this case, an Anritsu MS9030A, with 4 pm resolution, or one part
in 106) can be less than what is needed in order to measure the linewidth precisely.
Of course, interferometry techniques are also used to identify the stability and/or
resolution of spectral linewidths.
Interferometry techniques, in comparison to
spectroscopic techniques, have much higher resolution when measuring, in particular, the
difference between two linewidths. For testing purposes, two interferometers are set up
side by side so that numerous comparisons can be made between diode laser output with
or without temperature stabilization and diode laser output with or without passive optical
resonant feedback. To isolate the effects, tests for diode laser stability are, initially, done
separately, but non-quantitatively, i.e., interference fringes are simply “visibly” detected
and noted as sharper, distinct, etc. Invariably, sharper lines result when there is adequate
optical feedback present. However, there is a great deal of freedom associated with
“tuning” the optical feedback for the diode laser and little is known (so far) how efficient
the present techniques for optical feedback are. At present, only 4% reflection from a flat
glass surface has been used. It is not yet known whether surfaces with greater reflectivity
can appreciably improve the stability and/or lock in behavior of the diode laser.
Similarly, it is not entirely clear if there are certain “precise” geometrical angles or
locations within the diode cavity that can cause the diode laser to be more susceptible to
optical resonance.
To complicate matters, the final diode laser module not only relies on a reflecting
surface, but also, a transmission grating to select a single frequency. The transmission
holographic gratings used, which are lined to nearly 1500 lines/mm, are geometry
sensitive (see Fig. 1). The diffraction angles must be written precisely in the holography
process, otherwise direct optical feedback is impossible to accomplish. Because there are
inherent difficulties in writing consistent holographic gratings, it is also difficult to obtain
high efficiencies along with the correct geometry (a discussion of holographic grating
writing is beyond the scope of this paper, but, nevertheless, plays a significant part in
developing an optically stable diode laser). With the best efficiencies running usually as
high as 65%, a 4% reflection (with the correct geometry) would only produce roughly a
2.5% reflectivity to the diode laser with other losses occurring at each zero order (as the
beam passes through the grating twice).
Ideally, the problem of selecting an appropriate optical feedback technique and
identifying its success to stabilize the diode laser output is separate from the issue of
judiciously finding a suitable physical apparatus to optimize the temperature stabilization
of the diode laser. This is true in the case of this particular diode laser design. However,
as detailed above, it is clearly seen that the most of the front end optics must be built
around the passive resonant optical feedback loop. To date, after several designs have
been sketched, constructed or tested, one has been selected and its preliminary
construction has begun (see Fig. 2). After initial tests on two preliminary modules, it has
been concluded that there were both optical problems as well as thermal lag. As of now,
the final design has been completed, modulus one part, but has not yet been constructed
completely.
The efficacy of this device is of prime importance if it is to be used for a special hand
held application of the photorefractive effect.‡
This particular effect uses a
photorefractive crystal as a novelty filter. A “nose” is constructed on the total internal
‡
The photorefractive effect is essentially a grating writing effect and, hence, requires or, at the very least is
strongly dependent on, a narrow linewidth laser source in order for both beams to write the grating.
reflection (TIR) side of a prism by precisely placing small circular dots of specific
polymers on the TIR face. In the absence of the change the detector (a CCD camera with
ND filter) detects nothing. However, when a vapor, like water or ethyl alcohol, is
“breathed” across the surface of the polymer, it adsorbs onto the surface and a sudden
change in index of refraction as well as optical path length is detected. On the screen, a
bright spot appears in the location of the dot. Knowing, of course, which polymers are
most suitable for this process is yet to be determined. Nevertheless, it is certainly clear
for demonstration purposes, that a hand held device can be an effective tool for use of the
“nose”.
I would like to acknowledge Dana Anderson for all of his help and direction. I gladly
appreciate the quick-but-comprehensive shop class given by Bill Bickel. I would like to
thank the following for their help and guidance: Leslie Czaia, Dirk Mueller, Hongke Ye,
John Jost, Edeline Fotheringham, Valerie Damiao, and Scot Christiansen. I would also
like to thank the NSF, without whom this project and summer research experience would
not have been possible.