Martin Teichmann Atomes de lithium-6 ultra froids dans la ... - TEL

CHAPTER 3. EXPERIMENTAL SETUP
specification, they should get to around 668 nm when heated to 60 ◦ C.
In the experiment, we usually heat them up to 80 ◦ C. Near a current of
280 mA the wavelength reaches 670 nm, and we can inject the diodes,
giving us more than 120 mW of power. Both temperature and current
are beyond the specifications of the diode.
The high current in combination with a high power is not easy to
handle. Thermal isolation is crucial, as at high temperatures the losses by
convection will be large, leading to an unstable diode. We glued rubber
isolations on the inside of the laser housing, and closed it hermetically.
Additionally, we used commercial diode mounts Newport Model 700
Series Temperature Controlled Mounts. Only the high price prevented
us from using them for all diodes.
In case the temperature controller fails, the temperature drops rapidly
and the diode will die, as the current is by far too high at ambient
temperature. After having lost some diodes this way, we installed a
temperature security system. We installed a temperature sensor LM35
into our own diode mounts and used the sensors AD592 in the Newport
mounts to monitor the temperature. Once it leaves a predefined range,
a relay installed in the current drivers will cut the supply to the diode.
During the evaporation of the atoms, it is imperative that we do not
allow any light that resonates with the atoms, as even at low powers this
leads to non-negligible heating. This is not easy to achieve if one has to
block some 100 mW. In the old setup, we used to switch off the AOMs
of the master lasers in order to disinject the lasers, such that they were
not resonant with the atoms anymore. Unfortunately, some of the new
lasers show an annoying behavior: while they are stable once injected,
they will not re-inject after the injection beam has been switched off for
a while. The diodes show a hysteresis behavior: one has to increase
the diode current above the injection point and decrease it again for the
laser to injection lock. As a first solution, we tried to constantly modulate
the laser diode current. This made re-injection possible, but not to an
extent that we could guarantee it. As this also affected long term stability
on normal conditions, we abandoned this technique. Instead, we had to
block the resonant light completely. This turned out to be harder than
expected, as even after a switched off AOM and a shutter there was still
enough light to decrease the lifetime of the atoms below a second. A
shutter in front of a small aperture solved this problem.
Still, the tapered amplifier gave us about 500 mW of optical power, far
from what the diodes are capable of. As luck would have it, this power
is equally distributed over the four frequencies that we use. This means
that we can use four lasers to generate the six beams needed for the
50