Dr. Matthew Walhout
Download the printer-friendly Word Document of this page.
The first project of the summer of 2007 was to replace the existing vacuum system with the one designed and built in 2006. The new system eliminated the need for oil diffusion pumps and liquid nitrogen cooling, making for a cleaner vacuum system, and allowing a magneto-optical trap (MOT) to be obtained much more quickly.
The vacuum system is divided into two major sections by a valve which is opened by compressed air when plugged in. This valve separates the source(s) from the slowing tube and trap area; it is located just upstream of the slowing tube and magnet. The upstream section contains a slow leak valve, the discharge source, the UV/IR source, the ICEMAKr (Inelastic Collisional Excitation for Metastable Atomic Krypton), a turbo pump connected just below the discharge source, and a cryopump connected by two hoses – one just downstream of the turbo pump, one just upstream of the slowing tube. The typical operating pressure in this section is of order 10-7 Torr by the sources and 10-8 Torr in the cryopump chamber.
Turbo pumps are basically high-tech fans that blow air out of the system; at full speed the fans rotate at 1500 Hz. The Cryogenics cryopump freezes everything out. It has very porous activated charcoal inside it, which has high surface area on which to condense contaminants. Its temperature can be monitored on its control box.
The pressure in the source area is monitored by a Pfeiffer Compact Full-Range Gauge with its own readout on the turbo pump control box. It can read all the way up to atmospheric pressure, unlike the ion gauges, which we generally don't turn on until pressure is down in the 10-4 Torr range. Just downstream of the turbo pump chamber is an ion vacuum gauge. It uses current to determine the pressure. As a filament releases electrons, they travel across to an anode. As the pressure in the chamber decreases, fewer particles obstruct the traveling electrons, and current increases. Another ion gauge is connected to the cryopump chamber.
As the krypton enters the system through the leak valve, the source(s) excite(s) the atoms into the desired metastable state. The DC discharge source runs current through the krypton gas much like is done in neon lights. One end of the glass discharge tube is grounded; the other end by the nozzle is held at high voltage by a wire wrapped around the tube with the wire’s end near the nozzle opening. Its current is supplied by a 1.5V battery amplified 1000 times by a Trek HV amplifier.
The UV/IR source uses a UV lamp made by an RF discharge in krypton (Young et al., 2002). The fast oscillation in high voltage between the two coils strikes and maintains the discharge. The UV light from this process goes through a magnesium fluoride window (it won't go through glass) to excite the atoms in the atomic beam to the 1s4 state. Immediately after that, an IR laser (819.223 nm) excites the atoms to the 2p6 state which can decay into the desired metastable state.
Finally, the ICEMAKr is an electron bombardment source. It works by sending a current through a filament, ejecting electrons; four grids can have independently controlled voltages to create an E-field to propel electrons toward the atomic beam. When the electrons hit atoms, they transfer their energy and can excite them into the metastable state.
The other side of valve – the high vacuum side around the low 10-8 Torr range – includes the Zeeman slower coils, a turbo pump, a MOT (Magneto-Optical Trap) chamber, a UV detector, an opposing coil and an ion pump. The Zeeman slower coils create a magnetic field which Zeeman shifts the energy levels in the metastable atoms in order to maintain resonance with the slowing laser (811.515 nm reading on our wavemeter) as the atoms change speed and Doppler shift. The atoms slow down as they scatter photons. The resonant photons can be absorbed by the atoms, yielding a small amount of momentum opposing the atom's forward motion such that mΔv = h/λ. The photon is re-emitted in a random direction; the vector sum of all the emissions gives no net momentum change. Thus the total change slows the atoms down gradually as they move through the slowing tube. By the time they reach the MOT chamber, they are slow enough to be caught.
The end of the slowing magnet closest to the MOT chamber is the high magnetic field end, which needs to be balanced out to keep the zero point of the magnetic field near the center of the chamber. This is accomplished by using an opposing coil on the opposite side of the chamber combined with fields in the x, y and z directions from the square coils. These also work to cancel out the earth's magnetic field. These magnetic fields are controlled with individual power supplies which are hooked up to DMMs to read the current.
The MOT has opposing Helmholtz coils to create a magnetic field gradient with a zero point at the center. We tried to create a gradient of between 5 and 10 Gauss/cm along the coil axis. This gradient makes for a position-dependent Zeeman shift in the atoms such that the lasers shining in along three axis will push atoms back to the center.
On the bottom of the trap chamber is the channel electron multiplyer (CEM) which detects the ions which resulting from collisions between metastable atoms in a dense trap. This, reads out to an EG&G Ortec MCS data acquisition board in a computer. The UV detector above the trap requires another magnesium fluoride window and an argon-flushed chamber to allow the UV photons from the atom trap up to the detector. A photo-multiplier tube transforms the UV photons into signal pulses that can be amplified and sent to another MCS counter in the computer. The UV signal is used to detect the presence of the long-range krypton dimers whose energy levels we want to study.
The ion pump is not used at this time but would contribute to lower the vacuum pressure. The pressure in this whole section is monitored by an ion gauge near the turbo pump between the slowing tube and MOT chamber.
The system has two grating stabilized laser sources for the PA (photo association), and for the slowing and MOT lasers; there are also various other free-running sources, one of which is currently in use for the UV/IR source.
The wavelengths are measured by feeding a beam into a Wavemeter. Since the reset is not consistent, it is often necessary to turn the meter on and off to restart its measuring attempts. The wavelength for free-running lasers is adjusted using temperature and current controls. The difficulty with these controls is the likelihood of mode-hopping: where a small change in current or temperature causes a large jump in wavelength as the cavity is filled by a different integer number of wavelengths rather than the same number of a slightly different length.
For the grating stabilized lasers, an extra control is involved. These have a laser beam coming out by way of a diffraction grating. The reflected beam (zeroth order diffraction) comes out while the diffracted beam goes back in (order = 1), causing the interference to stabilize the wavelength. The diffraction grating can be moved in tiny increments in and out, essentially adjusting the length of the cavity and therefore wavelength, using the Piezo controls. These supply variable voltages to the Piezo on which the grating is mounted; the different voltages cause it to expand and contract by small amounts, thus moving the grating in and out. Higher voltage means more expansion, yielding a shorter cavity and therefore shorter wavelength.
The PA laser begins about at resonance with the metastable atoms (810.660 nm); at resonance, the atoms will absorb the photons and jump to a state from which they can cascade back to the ground state. Thus this destroys the trap. But as the PA laser is detuned from resonance it can excite atom pairs into a special long-range krypton dimer molecule. In this state the atoms behave as though oscillating on a long spring. From here, one of the atoms can decay down to the ground state, emitting a UV photon which is observed by the UV detector above the MOT.
The slowing and MOT lasers come from the same source, but acousto-optic modulators (AOMs) adjust the wavelengths so the slowing laser is about 400 Mhz below the MOT lasers to account for the Doppler shift of the moving atoms. The slowing master laser is monitored with an oscilloscope which observes the laser absorption through the glowing discharge on the table. The master feeds into the slave laser which is monitored with another oscilloscope. The detuning is controlled by AOM.
One system is set up to control the timing of all the things involved in the PA experimenting. This experiment requires the slowing laser and MOT lasers to be off while the PA laser flashes on for a short time. At the same time the ion and UV detectors need to start counting in a new channel (or bin), and the voltage on the piezo (controlling wavelength) for the PA laser needs to step to the next incremental setting. All these things happening together we'll call one event. This is accomplished through the use of triggerable pulse generators, and the pulses are monitored with a 4-channel oscilloscope.
There is a master driving pulse. Adjusting its period changes the total period from one event to the next. This pulse goes through the inverter/divider to divide its frequency down to our current running frequency around 10 Hz. From there it triggers the pulse which will turn the MOT and slowing laser off and the PA on. This pulse is quite short to prevent the complete destruction of the trap – around 20 ms. Its length is controlled by changing the duration on its pulse generator. Its period is set to “external trigger.”
The MOT, slowing, and PA lasers are turned on and off with a shutter. Every edge in the signal sent to the shutter reverses the shutter's current state. So if the shutter starts closed when it is switched to “remote” from “local,” the short signal will quickly open and close it for the PA flash.
The divided pulse also triggers a third pulse generator, again with period set to “external trigger” to send a short pulse to the MCS board “channel advance” to tell the computer to start counting in a new channel; an upward slope triggers channel advance to step forward when the 'dwell time' in 'pass control' is set to external for both boards. It is also possible to put this pulse through a division so that multiple events could be added up in one channel to increase statistical value. When the channel advance steps forward, it causes a ramp up in the voltage output to the piezo through a circuit to convert it correctly. This will scan the PA laser down in frequency from the resonant value in incremental steps synced with the channels.
The PA/MOT/slower pulse triggers the MCS gate function through a fourth pulse generator. The gate starts the counting of signals from the detectors when the MOT and slower turn off and the PA turns on with a high voltage; it stops at the end of the flash when the voltage goes back low. The gate has its own pulse generator so that the opening of the gate can be delayed slightly after the MOT is off to ensure it’s really off; this can be done by increasing the Delay on the generator. Additionally, the pulse to the gate can be slightly shorter (the duration setting) than the PA/MOT/slower pulse to ensure that it stops counting when the MOT is back on.
To tie the channels from the MCS output to absolute frequency measurements of the PA scan we’ve started using a signal from a Fabry-Perot. The output from the Fabry-Perot goes to a circuit which creates an oscillating signal. Each high voltage will be recognized by the MCS as a TTL signal to count. High output from the Fabry-Perot at specific wavelengths creates a high-frequency oscillation, which in turn is translated to a high peak in counts on the MCS readout. We also use a detuned sideband to create secondary, smaller peaks. The MCS channels for the Fabry-Perot are synchronized with the channels for the UV detection. So, by knowing the separation between the large peaks, between the small peaks, and between the large and small from the Fabry-Perot, and knowing one frequency – like that of the atomic resonance which can be seen in the UV signal, an absolute frequency scale can be assigned to the UV signal. This allows us to maintain approximately 10 MHz accuracy for frequency scans ranging over a few GHz.
On a completely different note, we watch the MOT with two cameras. One transmits directly to a TV monitor. The other one goes through an imaging board in another computer and uses the Scion Image program.