Physics 726: JQI rotation 2
Optogalvanic effect in Ytterbium
The Optogavanic effect(OGE) is observed in gaseous discharges. When a discharge is illuminated with light that is resonant with atomic transitions then there is a change in the electrical properties of the gas such as the conductivity. This change is registered in the variation of the current passing through the discharge [1,2]. In our experiment we shine a 399nm laser through a hollow cathode discharge of Ytterbuim and observe the variation in discharge current as a signature of Optogalvanic effect in Ytterbium. This wavelength is resonant with the 1S0-1P1 transition in neutral ytterbium as shown in figure 1.
The basic experimental setup (fig. 2) consists of a Yb-HCL (Hollow Cathode Lamp) which is operated at a discharge voltage of 0.3kV (5.7mA). The discharge current is passed through a ballast resistor of 30kW. When the discharge is not excited by the laser there is a finite potential drop across this resistor since the discharge current is steady. Therefore in order to produce OGE we modulate the intensity of the laser using a chopper wheel that operates at ~1kHz. This produces a small ac component in the discharge current and hence a voltage across the resistor due to OGE in the Yb discharge. The d.c. component of the voltage is now decoupled using a capacitor (100nF) and the remaining signal is fed into a lock-in amplifier. Assuming that any effect thereby produced due to the laser (including OGE) should have the same periodic modulation as the laser intensity, it is possible to use the reference from the chopper to lock the OGE signal. For this purpose we use a Sharp GP1A52HRJ00F photointerrupter to obtain the reference frequency from the chopper. There are additional components such as the oscillocope and the spectrum analyzer which are used for other investigations that will be discussed later.
Figure 1. Ytterbium transition lines Figure 2. Yb-OGE setup
Stability Measurements and Noise reduction
The long term goal is to use OGE for laser locking at 399nm. If we can scan the frequency within a narrow window about the resonant frequency then we expect to observe a peak in the OGE effect and hence the OGE signal. It is therefore possible to use this electronic signal as a feedback that would regulate the frequency at which the laser cavity resonates. Because of Doppler broadening and isotope shifts of the resonant line it is not possible to obtain a sharp peak for OGE in the spectrum. Therefore, we expect to see a non-zero OGE signal for frequencies close to the resonant frequency within the Doppler window (~2GHz) with a peak occurring at the resonant frequency. It is important that we obtain a low noise signal (preferabely Gaussian). Moreover it is necessary that the signal is stable at a given laser frequency , only fluctuating within a tolerance limit but not drifting with time. During the first stage of the Yb-OGE experiment  a drift was observed in the OGE signal over long enough timescales (~600 sec) for a fixed frequency (fig. 3). Therefore we have started the second stage trying to understand the cause of drift and studying the noise in the signal.
Figure 3. The OGE drift observed close to resonance. Figure 4. The signal stability at 1kHz lockin frequency. Laser beam is switched on at t=0sec.
Ground loop Harmonics
In order to study the noise present in the signal in the form of periodic perturbations (sinusoidal) we feed the signal parallely into a spectrum analyzer as illustrated by figure 2. It is observed that the signal along with the OGE component also carries noise occurring at multiples of 60Hz frequency . Since the power line operates at 60Hz this noise might occur due to the presence of ground loop harmonics. Figure 5 shows a narrow window of the analyzer (920-1120 Hz) which shows noise peaks at 960,1020,1080 Hz. The peak at 1000Hz is the component that oscillates at chopper frequency thereby confirming the signal produced as a result of the laser hitting the discharge (mainly OGE). It can be noticed that noise strength is of the same order or more than the signal strength. Therefore, it is important that the chopper is operated at a frequency different from the harmonics so that we are able separate OGE signal from noise.
Estimating Current due to other effects
The Yb-Hollow Cathode Lamp has the two discharge electrodes shaped one in the form of a disk surrounded by an annular ring . Since we shoot the laser along the axis of the tube most of it hits the electrodes after passing through the discharge. Therefore, there is a possibility that we also observe photoelectric effect along with the OGE. To examine this we shine the laser on the HCL before discharge and record the current across the ballast resistor (voltage) for varying potential drops across the discharge electrodes. Figure 6 shows the observations for three different cases. The highest photoelectric current is observed (red line) when the laser is let to fall on the annular ring electrode rather than at the center. Aligning it at the center such that it hits the central electrode reduces the current (blue line) thus suggesting that it is the anode. If we further focus the beam to converge at the center the current is further reduced (green line). However just around discharge voltage the signal behaves erratically due to more complicated irregular phenomena coming into play at the start of the discharge. However, when the discharge starts operating at more than 0.26KV the signal becomes stable. This also justifies the operating discharge voltage of 0.3KV that we use for observing OGE. Due to the HCL design it is not possible to prevent photoelectric effect. Therefore the signal observed would be a superposition of the OGE and the photoelectric effect provided that they are independent under practical circumstances. This should not create a problem since we would observe the OGE signal riding on a steady offset photoelectric voltage.
Fig 5. Presence of ground loop hermonics in the signal Fig 6. Observing photoelectric effect before and after discharge.
Noise reduction and Response time
For practical purposes of laser locking we require a real-time response of the OGE signal as the laser frequency is sweeped. Therefore it is important to minimize the integration timescale of the lock-in amplifier. We have studied the typical response time for the lock-in signal to get stabilized after the laser is switched on. This has been carried out for several integration time-scales of the lock-in amplifier. Figure 7 shows the signal response for different timescales and the associated noise with it. It is observed that the typical time scale for the signal to reach a stable value (with associated noise riding on it) is of the order of 4 x timescale. A reduction in the response time causes the signal to stabilize faster but at the same time it does not reduce noise riding the signal. Therefore there is a trade-off between the noise reduction and signal response of the lock-in.
Fig 7. The response times and signal-noise ratio for different timescales of lock-in operation
Drift in the signal might occur due to the finite time taken by the discharge to warm up and perform steadily. We did several runs by turning on the discharge gradually (fig. 6)and then observing the lock-in signal over long enough time scales(fig. 4). We observe that the discharge characteristics does not change much over time as far as the OGE signal is concerned. We observed that there is an occasional unlinking of the reference signal from the chopper wheel as detected by the lock-in amplifier. Because of the presence of a slight wobble in the chopper wheel the reference frequency detected by the lock-in is not a constant value over time. In fact while operating the chopper at 1.1 KHz (as was used previously) the occasional unlocking is observed. Due to unlocking there are sudden jumps in the value of the locked signal and this causes overloading. The time taken by this overloaded signal to reach its normal value could be estimated to be ~4 x timescale of integration in the lock-in. In the first stage the timescale used was 100sec pre-amplification and 1 sec post-amplification. Therefore the signal after overloading would take atleast 400sec to reach the normal value (fig. 3). We tried to reproduce the drift setting the timescale to these values and overloading the signal in the beginning (Fig. 8). However we kept the chopper frequency at 1Kz which is consistent and does not cause signal unlocking. Operating the laser at near resonant frequency of 751526GHz we were able to produce a drift in the signal which started at a very high value and started exponentially decaying to a normal value. The drift was over a reasonable 200 second time period. If the chopper is operated at 1.1 KHz then the occasional unlinking might produce several overloading instances and this would explain the near 600 seconds drift that was observed before.
Figure 8. Drift in the signal due to overloading
To observe OGE in Yb discharge we chose the chopper frequency to be 1Khz. This frequency is sufficiently far from ground loop harmonics and also produces a stable locked signal throughout the experiment. We attempt to manually scan the laser frequency within a 13Ghz range including the resonance frequency (751520-751533 Ghz). The timescale for the lock-in is fixed at 10 seconds which produces a very high signal to noise ratio. However, the response time being of the order of 40 seconds every frequency step is observed after at least 1 min duration. Since the intensity of the laser itself varies over the frequency window the OGE signal is rescaled assuming that it is directly proportional to the intensity of the incident light on the discharge. The final OGE signal obtained is shown in figure 9. After fitting the data with a Gaussian the resonant frequency estimated is 751526.8±.03 Ghz (398.911nm) and the approximate width is observed to be 2Ghz. It is also observed that the OGE signal rides over a voltage offset of the order of 67uV which might be due to photoelectric effect as discussed earlier.
Figure 9. Observed Ytterbium Optogalvanic effect
á The signal drift at a fixed value of laser frequency is produced due to un-locking of the amplifier at 1.1 kHz chopper frequency. This does not occur if the chopper frequency is 1kHz. The problem could be fixed if we replace the chopper wheel.
á The observed photoelectric offset on which the OGE rides could be removed if one could use a lamp with different electrode shape and orientation such that laser does not strike any of the electrodes. This could produce a sharper OGE response in the electronic signal detected by the amplifier.
á A 10 sec timescale gives a much higher signal to noise ratio compared to a 300ms timescale of the lock-in (fig. 7). However, it has a slower response time and hence cannot but used while real-time scan of the laser frequency is executed when OGE is applied in laser locking. Where as for a faster response the noise becomes an issue. Therefore as an alternative solution for noise reduction an oscilloscope might be used to make averages over several frequency sweeps and obtain a low noise OGE profile
á The manual scan of the frequency window (fig. 9) shows the accurate OGE profile and confirms that we are observing OGE in Yb discharge.
á The Gaussian profile for the OGE effect indicates a width of the order of 2Ghz. The Doppler broadening of the discharge is given by where the discharge temperature is ~600K . That gives a broadening of . Our observed broadedning of 2Ghz is because of isotope shifts. Naturally abundant isotopes of Yb (Yb 168,170,171,172,173,174,176) have spectral separations of the order of 1-1.5Ghz . This explains the observed brodening of 2GHz.
á As the frequency sweeps there modes inside the laser cavity and the optical fiber that carries the beam might change. This gives rise to an intensity variation with the frequency of the laser. To make sure that we observe OGE as a function of the frequency and not the intensity of the laser it is important to operate the whole operation at laser intensities above a threshold value such that saturation in the atomic transition inside the discharge is reached. In our experiment we have not detected or operated above this threshold value. This is indicated by the fact that the original OGE signal has to be rescaled according to the laser intensity to obtain a much regular profile (Fig. 9).
 Goldsmith, J. E. M., and Lawler, J. E., 1981, Contemp. Phys., 22, 235.
 Barbieri, B., and Beverini, N., 1990, Rev. Mod. Phys., 62, 603.
 Michael Jarret and Jiehang ZhangÕs rotation webpage: http://terpconnect.umd.edu/~mjarret/rot/
 AndrŽs David Cimmarusti's Rotation Page: http://www.terpconnect.umd.edu/~candres/rotation1.html
 Jae I. Kim. Et al. Optics Letters, Vol. 28, Issue 4, pp. 245-247 (2003)
Thanks to Prof. Chris Monroe for providing this project and also work space in his lab.
Thanks to Dr. Wes Campbell for his immense help and constant supervision under which this project was completed.
Thanks to Dr. Qudsia Quraishi for invaluable advise on the experiment.
Thanks to Dr. Emily Edwards for her help with the 399nm laser.
Thanks to Dr. Kihwan Kim for his help with the experiment and the 399nm laser.
Thanks to Prof. Luis Orozco for organizing this course.