1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%%for ArXiV
3: \documentclass[pra,superscriptaddress,twocolumn]{revtex4}
4: \usepackage[dvips]{graphicx}
5:
6: \begin{document}
7: \title{Magnetic trapping of buffer-gas cooled chromium atoms and
8: prospects for the extension to paramagnetic molecules.}
9:
10:
11: \author{Joost M Bakker} \affiliation{Humboldt Universit\"at zu Berlin,
12: Institut f{\"u}r Physik, Hausvogteiplatz 5-7,10117 Berlin, Germany}
13: \email{bakker@physik.hu-berlin.de} \affiliation{Fritz-Haber-Institut
14: der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin,
15: Germany}
16:
17: \author{Michael Stoll} \affiliation{Fritz-Haber-Institut der
18: Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany}
19:
20: \author{Dennis R Weise}\affiliation{Universit\"at Konstanz,
21: Fachbereich Physik, 78457 Konstanz, Germany}
22:
23: \author{Oliver Vogelsang} \affiliation{Universit\"at Konstanz,
24: Fachbereich Physik, 78457 Konstanz, Germany}
25:
26: \author{Gerard Meijer}
27: \affiliation{Fritz-Haber-Institut der Max-Planck-Gesellschaft,
28: Faradayweg 4-6, D-14195 Berlin, Germany}
29:
30: \author{Achim
31: Peters}\affiliation{Humboldt Universit\"at zu Berlin, Institut
32: f{\"u}r Physik, Hausvogteiplatz 5-7,10117 Berlin, Germany}
33:
34: %%%for ArXiV
35: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
36:
37: %% \documentclass{iopart}
38: %% \usepackage{iopams}
39: %% \usepackage[pdftex]{graphicx}
40: %% \usepackage{a4wide}
41:
42:
43: %% \begin{document}
44: %% \title[]{Magnetic trapping of buffer-gas cooled chromium atoms and
45: %% prospects for the extension to paramagnetic molecules.}
46:
47: %% \author{Joost M Bakker$^{1,2}$, Michael Stoll$^{2}$, Dennis R
48: %% Weise$^{3}$\footnote{present address: EADS Astrium GmbH,
49: %% Claude-Dornier-Str., 88090 Immenstaad, Germany}, Oliver
50: %% Vogelsang$^{3}$, Gerard Meijer$^{2}$ and Achim Peters$^{1}$}
51:
52: %% \address{$^{1}$Humboldt Universit\"at zu Berlin, Institut f{\"u}r
53: %% Physik, Hausvogteiplatz 5-7,10117 Berlin, Germany}
54: %% \address{$^{2}$Fritz-Haber-Institut der Max-Planck-Gesellschaft,
55: %% Faradayweg 4-6, 14195 Berlin, Germany}
56: %% \address{$^{3}$Universit\"at Konstanz, Fachbereich Physik, 78457
57: %% Konstanz, Germany }
58:
59:
60: %% \ead{bakker@physik.hu-berlin.de}
61:
62: \date{\today}
63:
64: \begin{abstract}
65:
66: We report the successful buffer-gas cooling and magnetic trapping
67: of chromium atoms with densities exceeding $10^{12}$ atoms per
68: cm$^{3}$ at a temperature of 350 mK for the trapped sample. The
69: possibilities to extend the method to buffer-gas cool and
70: magnetically trap molecules are discussed. To minimize the most
71: important loss mechanism in magnetic trapping, molecules with a
72: small spin-spin interaction and a large rotational constant are
73: preferred. Both the CrH ($^6\Sigma^+$ ground state) and MnH
74: ($^7\Sigma^+$) radicals appear to be suitable systems for future
75: experiments.
76:
77: \end{abstract}
78:
79: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
80: %%%for ArXiV
81: %%\submitto{\jpb}
82: \maketitle
83: %%%for ArXiV
84: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
85:
86: \section{Introduction}
87: \label{sec:intro}
88: The buffer-gas cooling and magnetic trapping of CaH in 1998
89: constituted the first experiment where large samples of cold neutral
90: molecules were confined by externally applied electromagnetic fields
91: \cite{Weinstein:Nat395:148}. Since then, a number of alternative
92: methods has been developed to produce and capture cold molecules, such
93: as Stark deceleration of polar molecules in a molecular beam and
94: subsequent electric trapping \cite{Bethlem:Nat406:491},
95: photoassociation of laser-cooled atoms in a magneto-optical trap
96: \cite{Vanhaecke:PRL89:063001}, and molecule formation by tuning
97: through magnetic Feshbach resonances in an optical dipole trap
98: \cite{Jochim:PRL91:240402}. Among these approaches, the thermalization
99: with a cold buffer gas is the most general cooling method, as it is
100: applicable to virtually all atomic and molecular species. It also
101: offers the benefit of producing very large samples at sub-Kelvin
102: temperatures, which form a good starting point for further cooling by
103: other techniques, such as evaporative cooling.
104:
105: Doyle and co-workers have successfully applied the buffer gas method
106: \cite{DeCarvalho:EPJD7:289} to trap atomic chromium and europium
107: \cite{Weinstein:PRA57:R3173,Kim:PRL78:3665} as well as several other
108: atoms \cite{Hancox:Nat431:281}. Apart from the above mentioned
109: experiments on CaH, they also demonstrated buffer-gas cooling of VO
110: and PbO \cite{Weinstein:JCP109:2656,Egorov:PRA63:030501} molecules.
111: Although highly successful, buffer gas cooling in combination with
112: magnetic trapping of neutral species up to now had only been
113: implemented by this one group.
114:
115: Here we report on our buffer-gas loading experiment, aiming at the
116: production of large samples of trapped cold molecules. We have
117: validated the performance of the new system by buffer-gas cooling and
118: trapping of atomic chromium (Cr). Cr was chosen as it has a large
119: magnetic moment of 6 $\mu_B$ and strong, well-characterized electronic
120: transitions suitable for detection using absorption spectroscopy.
121: Moreover, the most abundant isotope $^{52}$Cr (86 \%) has zero nuclear
122: spin, which considerably simplifies the spectrum. As mentioned above,
123: Cr was also one of the first atoms to be magnetically trapped using
124: buffer-gas cooling by Doyle and co-workers, and it has been shown that
125: evaporative cooling to quantum degeneracy is feasible
126: \cite{Griesmaier:PRL94:160401}. In our current experiments we trap Cr
127: atoms and observe holding times exceeding one minute.
128:
129: In the second part of the paper we describe the criteria we have used
130: to select the molecules to be studied with this new experimental
131: setup. We discuss the technological constraints as well as the
132: relevant intrinsic properties of molecules that determine the chances
133: of success of buffer-gas cooling experiments in conjunction with
134: magnetic trapping.
135:
136: \section{Method and experimental setup}
137: \label{sec:expt}
138: Buffer-gas cooling of atoms and molecules is done by thermalizing them
139: with a cold background gas. Collisions with the background gas
140: particles take out small quantities of energy of the particles to be
141: cooled, and after many collisions the particles are thermalized with
142: the background gas. The choice of background gas is restricted to
143: those species that have at the (low) temperature of interest a large
144: enough density to facilitate a large number of collisions while not
145: reacting with the collision partner to form new species. Also, if
146: buffer gas cooling is combined with trapping using external fields,
147: the buffer gas should be insensitive to these fields. In practice, the
148: only gas that fulfills these conditions is helium.
149:
150: At the start of experiments, a dense helium vapor is created in an
151: experimental cell. The buffer gas density is chosen such that hot
152: particles injected into it are cooled sufficiently to be trapped by an
153: inhomogeneous magnetic field before they can diffuse to the cell wall
154: and adsorb to it. On the other hand, the density must be low enough to
155: allow for the particles to reach the trap center. Optimal buffer gas
156: densities are typically in the range of 10$^{16}$-10$^{17}$ atoms per
157: cm$^{3}$, which corresponds to the vapor pressure of $^4$He at $\sim$
158: 700 mK, or $^3$He at $\sim$ 300 mK \cite{DeCarvalho:EPJD7:289}.
159:
160: After the creation of the required buffer gas density, the species to
161: be cooled is injected into the cell. This can be done through
162: capillary injection \cite{Messer:PRL53:2555}, or by ablation from a
163: solid state precursor material placed within that cell. In another
164: current project, efforts are made to use a molecular beam to inject
165: molecules into the buffer-gas \cite{Egorov:EJPD31:307}. Upon
166: injection into the cell, the particles collide with the helium atoms
167: in the vapor. By thermalizing with the helium vapor, species in
168: low-field seeking (LFS) states are drawn into the local magnetic field
169: minimum; those in high-field seeking (HFS) states are repelled and
170: lost at the cell wall. To fully isolate the trapped sample, it is
171: necessary to remove the helium, as it forms a thermal link to the cell
172: wall. This can be done by cooling the cell down, thereby effectively
173: cryopumping the helium out.
174:
175: For the cooled species to be magnetically confined, a (strong)
176: magnetic field around the experimental cell, with a field minimum
177: coinciding with the cell center, must be applied. This can be done
178: using a magnetic quadrupole trap consisting of two parallel coils
179: where counter-propagating currents create a zero magnetic field in the
180: center between the coils. Typically, the trap depths required for
181: buffer gas cooling should be on the order of a few Kelvin. As the trap
182: depth $E$ (in units of temperature) is given by $E=\mu H/k_{B}$, where
183: $\mu$ is the magnetic moment of the species $H$ the magnetic field
184: magnitude, and $k_{B}$ the Boltzmann constant, this leads to typically
185: required field strengths of $\sim$ 2 T. These fields can be generated
186: by coils of superconducting material, for which a cooling to liquid
187: helium temperatures is required.
188:
189: %%%%%%%%%%%%%%%%%%%%%%% FIG 1 %%%%%%%%%%%%%%%%%%%%%%%%
190: \begin{figure}[t]
191: \centering \includegraphics[width=1.0\columnwidth]{cryostat-half}
192: \caption{Scheme of the experimental setup. Figure \ref{fig:setup}a
193: shows the experimental cell within the custom-made cryostat,
194: positioned within the magnet bore. Figure \ref{fig:setup}b depicts
195: a cut through the copper cell itself.}
196: \label{fig:setup}
197: \end{figure}
198: %%%%%%%%%%%%%%%%%%%%%%% FIG 1 %%%%%%%%%%%%%%%%%%%%%%%
199:
200:
201: The refrigerator for cooling the experimental cell and the liquid
202: helium bath that holds the magnet are implemented in a custom-made
203: cryostat which is schematically depicted in Figure \ref{fig:setup}.
204: The cryostat is built around a commercial $^3$He-$^4$He dilution
205: refrigerator system (Leiden Cryogenics Minikelvin 126-700), with a
206: specified cooling power of 500 $\mu$W at 120 mK. The magnet consists
207: of two coils of NbTi/Cu wire spun onto a titanium bobbin that is
208: designed to withstand the repulsive forces of the two coils. The
209: magnet design is similar to the Mark-3 design by Doyle and co-workers
210: \cite{Harris:RSI75:17}, but allows for three axes along which the cell
211: can be optically interrogated: one vertical and two horizontal. From
212: the bottom of the cryostat, three apertures at the different
213: temperature stages of the cryostat (room temperature, 77 K and 4 K)
214: allow for the implementation of 75 mm diameter windows. From the top
215: of the cryostat and from the four lateral sides there are apertures
216: for 25 mm diameter windows. To be able to use the four side windows,
217: there are 10 mm diameter clearances through the magnet support
218: structure at the height of the trap center.
219:
220: The experimental cell, made of oxygen-free high conductivity copper
221: (OFHC), is attached to the mixing chamber of the dilution refrigerator
222: and positioned in the center of the magnet bore. The cell is linked to
223: the cryostat by a 10 cm long copper rod of 15 mm diameter forming a
224: well-defined thermal impedance. Optical access into the cell is
225: provided by a 50 mm diameter window sealing the bottom of the cell. A
226: buffer gas fill line is implemented using a CuNi capillary. Finally,
227: the cell is equipped with RuO thermistors and a resistive heater for
228: temperature monitoring and control.
229:
230: In the current experiment $^4$He is used as a buffer gas and the
231: appropriate density is created by bringing the entire cell to a
232: temperature of choice. The density is limited by the pre-defined
233: amount of helium let into the cell. Cr atoms are injected via
234: laser-ablation from a solid chip of Cr, positioned at the top of the
235: cell, using a $\sim$ 10 ns duration frequency-doubled Nd:YAG laser
236: pulse at 532 nm with a typical pulse energy of $\sim$ 15 mJ. We detect
237: Cr atoms in their $m_J=+3$ sublevel of the ground state by monitoring
238: absorption of light resonant with the $a^7S_3(m_J=+3) \leftarrow
239: z^7P_4(m_J=+4)$ transition around 425 nm. This is a nearly closed
240: transition where almost all excited Cr atoms decay into the initial
241: state (leakage $\sim 10^{-6}$). The light is generated by frequency
242: doubling the output of a grating-stabilized diode laser (Toptica
243: DL100) at 850 nm in a KNbO$_{3}$ crystal within an external bow-tie
244: cavity. The cavity is actively locked to the diode laser using a
245: Pound-Drever-Hall scheme \cite{Pound-Drever-Hall}. On long timescales
246: the cavity length is adjusted piezo-electrically to match the laser
247: frequency; on short time scales, the frequency of the diode laser is
248: adjusted to match the cavity length by modulating the diode current.
249: It must be stressed that the laser is not frequency-locked to an
250: external reference, so that long-term frequency drifts can occur. The
251: frequency-doubled light is amplitude stabilized by an acousto-optic
252: modulator (AOM), spatially filtered, expanded and (optionally) sent
253: through an aperture defining a beam diameter of a few mm.
254:
255: This beam is then sent into the cell using a beam splitter and
256: retroreflected by a mirror mounted at the top of the cell. For the
257: current experiments only the windows that give access from the bottom
258: of the cell are implemented, and to minimize optical etalon effects
259: all windows are wedged by $\sim$ 1$^o$ and anti-reflection (AR) coated
260: for the wavelengths used in the experiment. Light exiting the cell is
261: directed onto a photomultiplier tube (PMT, Electron Tubes 9813QB). A
262: fraction of the light that is not entering the cryostat is directed
263: onto a second PMT (Electron Tubes 9814QB) for normalization and
264: correction of residual intensity fluctuations. Typical light powers
265: used are less than 1 $\mu$W, to avoid both heating of the cell and
266: loss of atoms due to optical pumping effects. The measured
267: transmission is thus obtained by normalizing the detection signal on
268: the reference signal. When the detection laser is scanned in
269: frequency, the signal is corrected for etalon effects by recording a
270: background transmission signal on which the signal is normalized. In
271: an alternative detection scheme, the transmitted beam is directly
272: steered onto the light-sensitive chip of a CCD camera (PCO Imaging,
273: Sensicam SuperVGA).
274:
275:
276: \section{Magnetic trapping of Chromium}
277: \label{sec:results}
278:
279:
280:
281: %%%%%%%%%%%%%%%%%%%%%%% FIG 2 %%%%%%%%%%%%%%%%%%%%%%%%
282: \begin{figure}[t]
283: \centering \includegraphics[width=1.0\columnwidth]{cr-trapped}
284: \caption{Transmission through a sample of Cr atoms in the buffer
285: gas cell (top graph). The probe laser frequency is scanned
286: repeatedly over the atomic transition with a repetition rate of
287: 1 Hz. The inset shows a typical absorption profile (triangles)
288: after $\sim$ 14 s, together with a simulation (dotted line). The
289: other two graphs show the evolution of the particle density
290: (middle) and temperature (bottom graph) of the trapped atom
291: ensemble over time.}
292: \label{fig:raw}
293: \end{figure}
294: %%%%%%%%%%%%%%%%%%%%%%% FIG 2 %%%%%%%%%%%%%%%%%%%%%%%
295:
296: Figure \ref{fig:raw}a depicts a typical recorded transmission trace
297: spanning a time period of 1 minute. In this experiment, starting at
298: t=-3.5 s, the cell is heated for a period of 3 s with a heating power
299: of $\sim$ 20 mW. Then, at t=0 s, the ablation laser ( 15 mJ) is fired
300: to inject Cr atoms into the helium vapor. The optimal loading
301: conditions are found in a procedure where the heating power and
302: ablation are optimized to yield the best capture efficiencies. A field
303: free measurement under the same conditions suggests that the heating
304: over a few seconds ensures a full evaporation of the helium off the
305: cell wall, whereas the short waiting period allows for the buffer gas
306: to be cooled somewhat without noticeable cryopumping effects taking
307: place. The 15 mJ used for the ablation is a compromise between signal
308: intensity and the temperature at which the atoms are captured in the
309: trap.
310:
311: The trace shown in Figure \ref{fig:raw}a is recorded by repeatedly
312: scanning the probe laser frequency over a range of $\sim$ 10 GHz to
313: obtain a magnetic-field broadened absorption line while minimizing the
314: effect of long term drifts of the laser wavelength. In the scanned
315: signal, some interesting dynamics can be observed. In the initial few
316: seconds after ablation, the strong absorption signal first decreases,
317: then shortly grows again and finally decays slowly. Additionally, a
318: rapid narrowing of the line profile can be seen during the first 5
319: seconds following ablation. These different regimes will be discussed
320: in detail below.
321:
322: More detailed information about the density and temperature evolution
323: of the trapped atom ensemble can be obtained by simulating the spectra
324: and comparing them to the recorded ones. First we note that at
325: temperatures below 1K the expected Doppler broadening is less than 100
326: MHz, which is negligible in comparison with the observed Zeeman
327: broadened linewidths of 1 GHz and larger. For the modeling, it is
328: assumed that the cloud of atoms is in thermal equilibrium and can be
329: characterized by a Boltzmann temperature \cite{Luiten:PRA53:381}. For
330: any given point in the trapping field, the Zeeman shifted transition
331: frequency and a (relative) population density and a resulting
332: absorption cross-section are calculated. A transmission spectrum is
333: obtained by numerically integrating the absorption over the volume
334: defined by the probe beam. From the inset in Figure \ref{fig:raw}a it
335: can be seen that the simulated spectrum is in excellent agreement with
336: the recorded data.
337:
338: The simulations show that the spectral shape strongly depends on the
339: probe beam parameters and we have thus carefully measured beam
340: diameter and intensity profile. However, the relative position of the
341: probe beam with respect to the center of the magnetic quadrupole trap
342: is less amenable to direct measurements and must therefore be included
343: in the model. This relative position is especially important when the
344: probe beam is linearly polarized, as the projection of the driving
345: $E$-field onto the quantization axis of the atoms, the inhomogeneous
346: magnetic field, has a strong angular dependence. It is rather
347: straightforward to derive that the angular dependence of the
348: absorption coefficient is $(\cos^2\theta + \sin^2\theta \sin^2\phi)$,
349: where $\theta$ is the angle between magnetic field and light
350: propagation axis, and $\phi$ that between polarization direction and
351: magnetic field, respectively. To the best of our knowledge, this full
352: angular dependence has not been implemented in modeling of data in
353: previous buffer-gas loading experiments. Best agreement is obtained
354: when it is assumed that our 2 mm diameter beam is offset by 0.3 mm
355: from the trap center, which is well within the limits of optical
356: alignment. The remaining free parameters used in the simulations are
357: the particle density at the trap center and the Boltzmann temperature,
358: which are found using a least-squares fitting procedure.
359:
360: A possible error in the parameters obtained this way arises from a
361: breakdown of the assumption of thermal equilibrium. This is likely the
362: case at times shortly after the ablation pulse, where the changes in
363: temperature are on comparable time scales as the time it takes to
364: record one spectral profile. For this reason, the temperatures and
365: densities obtained before t=5 seconds are given without error bars and
366: should be considered as indicative only. Additionally, long-term
367: drifts in the laser frequency could amount to errors of a few \%. The
368: densities and temperatures of the sample as obtained from the spectral
369: simulations are depicted in the lower two graphs in Figure
370: \ref{fig:raw}. The size of the error bars reflects the reduced
371: signal-to-noise level at later times, when absorption signals become
372: less pronounced.
373:
374: Data at short times after ablation have not been published by Doyle
375: and co-workers previously and it is interesting to speculate on the
376: nature of the observed dynamics. The time evolution can roughly be
377: divided into three periods which are indicated by I, II and III in
378: Figure \ref{fig:raw}b. During period I, the temperature falls rapidly
379: from $\sim$ 1.5 K at t=2 s to some 500 mK after 5 s. This is
380: attributed to so-called spilling effects, where during trap loading
381: the warmest particles are lost over the trap edge. As a result, the
382: overall temperature drops.
383:
384: In Period II, the sample temperature does not change as dramatically
385: as in Period I. The sample appears to be in thermal equilibrium as it
386: has a temperature similar to that of the cell wall. The obvious {\it
387: recurrence} in the transmission spectrum after 5 s is also observed
388: in the density. The increase in the density is interpreted as a net
389: diffusion of cooled atoms into the trap center.
390:
391: After $\sim$ 20 s, the temperature of the cell wall (not shown) has
392: dropped to around 350 mK and the gas-phase helium within the cell is
393: now cryopumped to the cell wall. As a consequence the sample now
394: becomes thermally isolated and the temperature of the trapped cloud is
395: now no longer decreasing. Its final temperature thus corresponds to
396: the cell temperature at the time of thermal disconnect.
397:
398: At times later than 20 s, a clear continuous loss of Cr is observed.
399: It is not expected that evaporation plays a large role at this point,
400: as the effective trap depth at these temperatures is very large and
401: evaporation rates are expected to decrease exponentially with
402: effective trap depth \cite{Ketterle:AAMOP37:181}. A further indication
403: for this is that the observed temperatures do not change, which rules
404: out effective evaporative cooling. As additionally the buffer gas
405: temperature is similar to or lower than the Cr temperature, we can
406: thus rule out elastic collisions as a loss mechanism.
407:
408: As a consequence, the observed trap loss must be due to inelastic
409: collisions, which change the spin of the atom. Several such processes
410: are known and they can be distinguished by the number of Cr atoms
411: involved. The dominant loss mechanism may be found by fitting
412: functional forms to the number density evolution. In an earlier set of
413: data recorded over 180 s (where the signal was noisier and thus less
414: suitable for quantitative data analysis) a clear signature is found of
415: two-body loss processes, i.e. collisions in which two Cr atoms are
416: involved. For the data presented in Figure \ref{fig:raw}, the
417: signature of two body loss processes cannot be easily discerned from
418: that due to collisions with background gas. If we leave one-body decay
419: out of consideration, we obtain a maximum value for the two-body
420: inelastic collision cross-section of 6$\cdot$ 10$^{-13} cm^2$ at T=350
421: mK , consistent with a previous measurement
422: \cite{deCarvalho:JOSAB20:1131}.
423:
424:
425: %%%%%%%%%%%%%%%%%%%%%%% FIG 3 %%%%%%%%%%%%%%%%%%%%%%%%
426: \begin{figure}[t]
427: \centering
428: \includegraphics[width=1.0\columnwidth]{cr-absorption-imaging}
429: \caption{Absorption images of a trapped cloud of Cr atoms, at three
430: different times (columns) and at three different laser
431: frequencies (rows). The size of the image is 8.6 mm x 6.9 mm. }
432: \label{fig:ccd}
433: \end{figure}
434: %%%%%%%%%%%%%%%%%%%%%%% FIG 3 %%%%%%%%%%%%%%%%%%%%%%%
435:
436:
437: In the measurements presented above, the probe laser diameter is
438: limited to $\sim$ 2 mm and is aligned close to the trap center. This
439: way not the whole trapping volume is addressed. In an alternative
440: experimental configuration, the full waist of the collimated probe
441: beam is sent through the trap center, retroreflected and steered
442: directly onto the light-sensitive element of a CCD camera. For these
443: imaging experiments the laser frequency is kept at a fixed value.
444: Since the atomic resonance frequency is determined by the
445: field-dependent Zeeman shift, a fixed frequency experiment selectively
446: probes for atoms in specific 'shells' within the trapping volume. The
447: results of these measurements are depicted in Figure \ref{fig:ccd} for
448: three probe laser frequencies and three different times at the maximum
449: trap depth of 6.8 Kelvin. The images shown (1 second gate time) are
450: obtained as the ratio of an image recorded with Cr in the trap and one
451: where no Cr is in the system. The size of the images is 8.6 mm
452: (horizontal axis) by 6.9 mm (vertical). Since the beam is collimated,
453: the spatial extent of the absorption signal on the CCD element has a
454: 1:1 correspondence to the trapped cloud size.
455:
456: The first set of images is recorded at the frequency of the field-free
457: atomic transition. A strong absorption signal is clearly visible at
458: the trap center. One can see that the absorption is anisotropic,
459: which is a manifestation of the use of linearly polarized probe light.
460: When the probe laser is tuned towards higher frequencies, a ring
461: structure becomes apparent in the images. Here, it is obvious that the
462: angular intensity distribution is proportional to $\sin^2\phi$, as
463: discussed above. At a frequency-offset of +1.1 GHz, the strongest
464: absorption is observed in a ring with a radius of $\sim$ 1.5 mm. The
465: reason for such a pronounced ring is that the probe beam line-of-sight
466: here coincides with a 'shell' of Cr atoms trapped at the same field
467: magnitude. Again, the asymmetry attributed to the laser polarization
468: is clearly visible. The ring structure allows for a consistency check
469: of our trapping field calculations. If we assume that the probe beam
470: is collimated well, this implies that at a radius of 1.5 mm the
471: magnetic field calculated from the Zeeman shift $\Delta \nu=H\mu_B
472: (g_{f}m_{J,f}-g_{i}m_{J,i})/h$ is 76 mT. This is indeed consistent
473: with our trapping field calculations (69 mT at 1.5 mm), used in the
474: spectral simulations described earlier.
475:
476: Apart from this strong outer ring of absorption, some weaker but still
477: pronounced rings are also visible. The origin of these rings is
478: unknown to us. It is unlikely that they are due to population in lower
479: $m_J$ states, as they persist for tens of seconds and it is expected
480: that population in lower $m_J$ states is lost more quickly than in the
481: $m_J=+3$ state due to a combination of lower trap depth and
482: spin-exchange collisions. It can be speculated that they are caused by
483: diffraction of the beam by the dense atomic cloud in the trap center.
484: It is, however, unfortunately not straightforward to simulate this
485: diffraction pattern due to our probe geometry where the probe beam
486: passes the trapped cloud twice. Without diffractive effects included,
487: a semi-quantitative agreement between our simulated data and the
488: images is found. We therefore do not extract temperature and density
489: information out of these images. For the absorption measurements where
490: only a small probe beam is steered through the center of the cloud,
491: such diffractive effects are not expected to play an important role as
492: the cloud size is larger than the beamsize, and the sensitive area of
493: the PMT is significantly larger than the probe beam diameter.
494:
495:
496: %%%%%%%%%%%%%%%%%%%%%%% FIG 4 %%%%%%%%%%%%%%%%%%%%%%%%
497: \begin{figure}[t]
498: \centering
499: \includegraphics[width=1.0\columnwidth]{cr-trapdepth-variation}
500: \caption{Density (upper graph) and temperature (lower graph)
501: of a trapped cloud of Cr atoms as a function of trap depth,
502: measured at three different times; at t=7 s (squares), t=14 s
503: (circles) and 35 s (triangles) after ablation. The lines in the
504: upper graph are guides to the eye, whereas in the bottom graph they
505: represent the average temperature.}
506: \label{fig:trapdepth}
507: \end{figure}
508: %%%%%%%%%%%%%%%%%%%%%%% FIG 4 %%%%%%%%%%%%%%%%%%%%%%%
509:
510:
511: To investigate the potential of this system for future experiments on
512: the buffer gas cooling and trapping of molecular species, it is of
513: interest to evaluate the trapping efficiencies at lower trap depths.
514: Such a measurement is performed by lowering the current through the
515: two anti-Helmholtz coils. The results of these tests are depicted in
516: Figure \ref{fig:trapdepth} where the observed number densities and
517: temperatures of a trapped Cr cloud, evaluated at three different
518: times, are shown as a function of trap depth. Each value is obtained
519: by averaging over several measurements. The trends that can be seen
520: from these graphs are clear: a lower trap depth leads to a substantial
521: drop in the number of captured atoms. The temperature of the trapped
522: atoms is independent of trap depth. It seems thus obvious that a
523: shallower trap captures less particles, while the temperature behavior
524: gives us supporting evidence that no evaporative cooling, and thus no
525: evaporation takes place. It is obvious from this Figure that if a
526: molecule were to be trapped, it should possess a large enough magnetic
527: moment to allow for a trap depth of at least 5 K. Molecules that could
528: fulfill this condition are discussed in Section \ref{sec:molecule}.
529:
530: \section{Search for a molecular candidate for buffer gas cooling}
531: \label{sec:molecule}
532:
533: As He atoms are highly nonreactive and do not easily form new species,
534: we can consider the helium buffer gas cooling a universal cooling
535: method. As such, it has already been shown that one can use this
536: technique as a source of cold molecules \cite{Maxwell:PRL95:173201}.
537: However, if the cooling is to be combined with magnetic trapping of
538: the (paramagnetic) species, the universality is severely limited due
539: to several constraints. At the heart of these constraints lie
540: inelastic collisions, in which molecules in a low-field seeking (LFS)
541: state undergo a transition into a high-field seeking (HFS) state, a
542: process often referred to as Zeeman relaxation. The HFS molecules are
543: then rapidly ejected from the trapping region. The ratio of inelastic
544: to elastic collisions is thus crucial for the efficient transfer of
545: energy while maintaining the molecules in a trappable state. The
546: collision partners can be a molecule and a helium atom, but also two
547: molecules. The former case is crucial for the feasibility of
548: buffer-gas loading, while the latter governs a possible evaporative
549: cooling process to further lower the temperature and increase the
550: phase-space density of molecular samples.
551:
552: Theoretical studies on the mechanisms of inelastic collisions with
553: helium atoms in external fields are few, and have been limited to
554: molecules in $^2\Sigma$ and $^3\Sigma$ ground states \cite{
555: Volpi:PRA65:052712, Krems:PRA67:060703,Krems:PRA68:051401,
556: Krems:JCP117:118,Cybulski:JCP122:094307}, although a study on
557: collisions between He and the OH radical (in a $^2\Pi$ state) is
558: forthcoming \cite{Groenenboom:privcom}. For $^2\Sigma$ molecules the
559: Zeeman relaxation occurs through spin-rotational coupling between the
560: rotational ground state and rotationally excited states
561: \cite{Krems:JCP120:2296}. Calculations for CaF and CaH molecules have
562: shown that Zeeman relaxation rates are strongly dependent on the
563: spin-rotation interaction constant $\gamma_{sr}$ and decrease with the
564: energy difference between rotational states. The experimental data
565: available for these molecules confirm this
566: \cite{Weinstein:Nat395:148,Maussang:PRL94:123002}.
567:
568: For higher-spin states, no experimental data is available. It is,
569: however, predicted that the Zeeman relaxation rates are considerably
570: higher. In $^3\Sigma$ states, the spin-spin interaction is the
571: dominant factor and again a larger energy difference between the
572: rotational states decreases the couplings between LFS and HFS states.
573: As a consequence, the spin-spin interaction parameter $\lambda_{ss}$
574: should be relatively small and the rotational constant large to
575: successfully trap molecules \cite{Krems:JCP120:2296}. For the NH
576: ($^3\Sigma^-$) radical, the calculated ratio of elastic to inelastic
577: collisions has been shown to be favorable for trapping experiments
578: \cite{Cybulski:JCP122:094307}.
579:
580: About the collisional properties of molecules in ground states other
581: than $^2\Sigma$ or $^3\Sigma$ little is known. An unsuccessful attempt
582: was made to buffer gas cool and trap VO molecules ($^4\Sigma^-$ ground
583: state), but it is unclear whether Zeeman relaxation was a problem
584: \cite{Weinstein:JCP109:2656}. Theoretical calculations that predict
585: collisional properties of such high-spin molecules are unknown to us.
586: Based on the lack of such data, we here propose to buffer gas cool
587: molecules with a high-spin ground state to investigate their
588: suitability for magnetic trapping and possible evaporative cooling
589: experiments. By extrapolating the cases built for $^2\Sigma$ and
590: $^3\Sigma$ molecules, we assume that the candidate molecule should
591: possess a large rotational constant B and a small spin-spin
592: interaction constant $\lambda_{ss}$ to minimize spin-spin couplings.
593: As diatomics containing hydrogen atoms have the largest rotational
594: constants, metal hydrides are a natural choice.
595:
596: Apart from such general criteria more practical considerations also
597: play a role. As is observed above, the trap depth plays a crucial role
598: in the possibility to confine species over longer time periods. In the
599: present experiment, trap depths of 5.3 K and above have been employed
600: with success. A useful quantity to describe trapping is the capture
601: efficiency parameter $\eta$, defined as the ratio of trap depth to
602: sample temperature; $\eta = \mu H_{max}/k_{B}T_{mol} $, where $\mu$ is the
603: molecular magnetic moment, $H_{max}$ the magnetic field strength at
604: the trap edge and $T_{mol}$ the Boltzmann temperature of the molecular
605: sample.
606: In the current experiment the trap edge is formed by the cell wall,
607: where the highest achievable magnetic field is 1.7 T.
608: Thus, at the observed initial temperatures of 1.4 K (evaluated
609: at t=2 s), $\eta$ should be larger than 3.8 for a successful (initial)
610: confinement of a substantial sample of molecules. This yields a minimum
611: magnetic moment of about 4.5 $\mu_B$.
612:
613: \begin{table}[tb]
614: \centering
615:
616: \caption{Diatomic molecules with high-spin electronic ground states
617: and their rotational constant (B), spin-spin interaction constant
618: ($\lambda_{ss}$) and spin-rotation interaction constant ($\gamma_{sr}$). All
619: numbers are given in cm$^{-1}$. The ratio of $\lambda_{ss}$ over B is an
620: indicator for the strength of Zeeman relaxation.}
621: \begin{tabular}{lc|rrlrr}
622: \hline
623: & & $B$ & $\lambda_{ss}$ & $|\lambda_{ss}/B|$&$\gamma_{sr}$&ref.\\
624: & &&&&$\times 10^3$&\\
625: \hline
626: \hline
627: GdO &$^9\Sigma^- $& 0.355& -0.104& 0.29 & 0.1&\cite{Kaledin:JMS165:323}\\
628: MnH &$^7\Sigma^+ $& 5.606& -0.004& 7.2 10$^{-4}$ & 31.3&\cite{Gordon:JMS229:145}\\
629: MnF &$^7\Sigma^+ $& 0.353& -0.005& 1.3 10$^{-2}$ & 0.6&\cite{Sheridan:CPL380:632}\\
630: MnCl&$^7\Sigma^+ $& 0.158& 0.037& 0.24 & 0.4&\cite{Halfen:JCP122:054309}\\
631: CrH &$^6\Sigma^+ $& 6.132& 0.233& 3.8 10$^{-2}$ & 50.3&\cite{Corkery:JMS149:257}\\
632: MnO &$^6\Sigma^+ $& 0.501& 0.574& 1.1 & -2.4&\cite{Namiki:JCP107:8848}\\
633: CrF &$^6\Sigma^+ $& 0.379& 0.539& 1.4 & 13.6 &\cite{Katoh:JCP120:7927}\\
634: MnS &$^6\Sigma^+ $& 0.195& 0.350& 1.8 & -2.4 &\cite{Thompsen:JCP116:10212}\\
635: CrCl&$^6\Sigma^+ $& 0.167& 0.266& 1.6 & 2.2&\cite{Katoh:JCP120:7927}\\
636: CrO &$^5\Pi_r $& 0.524& 1.148& 2.2 & 10.5&\cite{Sheridan:ApJ576:1108}\\
637: CrN &$^4\Sigma^- $& 0.624& 2.611& 4.2 & 7.0&\cite{Sheridan:ApJ576:1108}\\
638: VO &$^4\Sigma^- $& 0.546& 2.031& 3.7 & 22.5&\cite{Cheung:JMS91:165}\\
639: NbO &$^4\Sigma^- $& 0.432& 15.58& 36 & 334.0 &\cite{Adam:JCP100:6240}\\
640: TiH &$^4\Phi $& 5.362& & & 182.3&\cite{Launila:JCP104:6418}\\
641: \hline
642: NH &$^3\Sigma^- $&16.343& 0.920& 5.6 10$^{-2}$ & 54.7& \cite{Brazier:jms120:381}\\
643: O$_2$&$^3\Sigma^- $& 1.438& 1.985& 1.4 & -8.4&\cite{Berden:PRA58:3114}\\
644: CaH &$^2\Sigma^+ $& 4.229& & & 46.7& \cite{Barclay:ApJ408:L65}\\
645: CaF &$^2\Sigma^+ $& 0.339& & &1.3&\cite{Maussang:PRL94:123002}\\
646: \hline
647: \end{tabular}
648: \label{tab:molecules}
649: \end{table}
650:
651:
652: Based on the above mentioned criteria, we have performed an extensive
653: literature search to find an interesting and suitable molecular
654: candidate for buffer gas cooling and magnetic trapping. The results
655: are collected in Table \ref{tab:molecules}. To compile this table,
656: only molecules have been listed that possess an electronic ground
657: state with at least a fourfold spin multiplicity. For comparison, the
658: molecules studied so far using buffer-gas cooling have been included:
659: CaH, NH, CaF as well as the oxygen molecule ($^{16}$O$^{18}$O), a long
660: time prime candidate. As a further restriction, only molecules that
661: have been observed in experimental work and for which an experimental
662: value of the rotational constant is available are included. A column
663: is added with the ratio of the spin-spin interaction and rotational
664: constants, $|\lambda_{ss}/B|$, to guide the selection process in
665: minimizing the expected effects of Zeeman relaxation.
666:
667: An inspection of this table indicates that there are three molecules
668: that clearly stand out in terms of expected stability against Zeeman
669: relaxation: CrH in its $^6\Sigma^+$ ground state, MnH ($^7\Sigma^+$)
670: and MnF ($^7\Sigma^+$). All three molecules possess a large magnetic
671: moment, which is nominally 5 and 6 $\mu_B$, respectively. We
672: disregard the MnF molecule, as it has a substantially smaller
673: rotational constant, which would result in level crossings between
674: different rotational manifolds at fields lower than 0.25 T
675: \footnote{The magnetic field at which the first rotational level
676: crossings can occur can be approximated by $H_{cross}=2B/\mu$}. One
677: could point out that the TiH molecule also possesses a large magnetic
678: moment and rotational constant that should avoid these level
679: crossings; however, it is believed that non-$\Sigma$ state molecules
680: are highly susceptible to Zeeman relaxation, as direct couplings
681: between Zeeman states tend to become large \cite{Groenenboom:privcom}.
682: Since in addition no value of the spin-spin interaction constant could
683: be found for TiH, it is left out of consideration.
684:
685: Both CrH and MnH fulfill the requirements of a large rotational
686: constant, a small spin-spin interaction constant while at the same
687: time having a high-spin ground state which should be amenable to
688: magnetic trapping. Both CrH and MnH were first observed in the 1930`s
689: by Gaydon and Pearse
690: \cite{Gaydon:Nat140:110,Pearse:ProcPhysSoc50:201}. CrH has by far
691: attracted the most experimental attention as a prototype of a
692: high-spin diatomic molecule, especially since it was discovered in the
693: emission spectra of L-type brown dwarfs \cite{Kirkpatrick:APJ519:802};
694: for an overview, see \cite{Shin:ApJ619:407} and references therein.
695: Because of its high-spin ground state, CrH is employed as a probe for
696: magnetic field strengths in sunspots \cite{Engvold:AAsuplser42:209}.
697: The magnetic parameters of CrH are well-established in the
698: X$^6\Sigma^+$ electronic ground state by FIR laser-magnetic resonance
699: spectroscopy \cite{Corkery:JMS149:257}. Recently, we also measured the
700: Zeeman spectrum of the A$^6\Sigma^+ \leftarrow$X$^6\Sigma^+$ (1-0)
701: transition around 767 nm, thereby determining the spectroscopic
702: parameters of the excited state that are needed to simulate spectra
703: recorded in an inhomogeneous magnetic field \cite{Chen-inprep}. For
704: MnH, the experimental studies are more sparse, which is likely due to
705: the highly complex structure of the spectroscopic transitions which
706: are split into numerous hyperfine components. The A$^7\Pi^+
707: \leftarrow$X$^7\Sigma^+$ (0-0) and (1-0) transitions near 624 and 568
708: nm, respectively, are well characterized
709: \cite{Varberg:jcp92:7123,Varberg:JCP95:1563,Varberg:JMS156:296} and
710: the ground state has been studied using ESR spectroscopy
711: \cite{Vanzee:JCP84:5968,Vanzee:JCP85:3237}.
712:
713: Based on the extensive experimental information available, it seems
714: that the CrH radical is currently the most promising high-spin
715: candidate for studying the mechanisms involved in buffer gas cooling,
716: although it can not be ruled out that the significantly lower
717: spin-spin interaction of MnH will, in the end, favour this molecule.
718:
719: \section{Conclusions}
720: \label{sec:conc}
721: We have demonstrated buffer-gas loading of paramagnetic species into a
722: magnetic trap. In first experiments we confined Cr atoms with storage
723: times of more than one minute. The use of an absorption based
724: detection scheme combined with a simulation of the absorption
725: properties of the cloud in the presence of the inhomogeneous magnetic
726: trapping field allows us to extract relevant parameters. Given storage
727: times of more than one minute, thermal contact between the trapped
728: cloud and the walls of the experimental cell is reliably broken after
729: a period of 20 s. The final temperatures of $\sim$ 350 mK reached are
730: determined by the temperature of the buffer gas at thermal disconnect.
731: Observed initial number densities exceed 10$^{12}$ per cm$^3$. We have
732: also succeeded in implementing continuous non-destructive absorption
733: imaging of the trapped cloud, which can provide complementary
734: information on temperature and internal state distribution over the
735: whole trapping volume.
736:
737: Progressing towards the preparation of cold molecules, we have
738: outlined the criteria for selecting a molecular candidate to study in
739: buffer gas cooling experiments. Such a molecule should have, apart
740: from a high magnetic moment, a low cross-section for spin-changing
741: collisions which would result in trap loss. Due to their large
742: rotational constants and small spin-spin interaction coefficients, the
743: MnH and CrH molecules have been identified as potentially suitable
744: species. We will thus proceed by experimentally investigating the
745: relevant collisional properties of these molecules. Measurements of
746: the effect of trap depth on the capture efficiency show that, if the
747: spin-changing collision rates are sufficiently low, the current
748: experimental setup should allow for trapping and thermally isolating
749: CrH and MnH.
750:
751: %%\ack
752: {\bf Acknowledgment}
753:
754:
755: We gratefully acknowledge the support by the Deutsche Forschungs
756: Gemeinschaft (Forschergruppe Quantengase FOR282/2-1TP8). JMB
757: acknowledges funding through the European Network for Cold Molecules
758: (RTN2-2001-00498). We thank Bretislav Friedrich and John M. Doyle for
759: stimulating discussions and their help in constructing this
760: experiment. Last but not least we want to thank Tim Steimle for
761: bringing MnH to our attention, and the referee for helpful comments
762: that considerably improved the manuscript.
763:
764: \vspace{1cm}
765:
766:
767: %%\bibliographystyle{unsrt}
768: \bibliographystyle{apsrev}
769: \bibliography{string,mp,atoms,XH-radicals,buffergas,bas,general,misc}
770:
771: %% \newpage
772: %% \begin{table}[tb]
773: %% \centering
774:
775: %% \caption{Diatomic molecules with high-spin electronic ground states
776: %% and their rotational constant (B), spin-spin interaction constant
777: %% ($\lambda_{ss}$) and spin-rotation interaction constant ($\gamma_{sr}$). All
778: %% numbers are given in cm$^{-1}$. The ratio of $\lambda_{ss}$ over B is an
779: %% indicator for the strength of Zeeman relaxation.}
780: %% \begin{tabular}{lc|rrlrr}
781: %% \hline
782: %% & & $B$ & $\lambda_{ss}$ & $|\lambda_{ss}/B|$&$\gamma_{sr}$&ref.\\
783: %% & &&&&$\times 10^3$&\\
784: %% \hline
785: %% \hline
786: %% GdO &$^9\Sigma^- $& 0.355& -0.104& 0.29 & 0.1&\cite{Kaledin:JMS165:323}\\
787: %% MnH &$^7\Sigma^+ $& 5.606& -0.004& 7.2 10$^{-4}$ & 31.3&\cite{Gordon:JMS229:145}\\
788: %% MnF &$^7\Sigma^+ $& 0.353& -0.005& 1.3 10$^{-2}$ & 0.6&\cite{Sheridan:CPL380:632}\\
789: %% MnCl&$^7\Sigma^+ $& 0.158& 0.037& 0.24 & 0.4&\cite{Halfen:JCP122:054309}\\
790: %% CrH &$^6\Sigma^+ $& 6.132& 0.233& 3.8 10$^{-2}$ & 50.3&\cite{Corkery:JMS149:257}\\
791: %% MnO &$^6\Sigma^+ $& 0.501& 0.574& 1.1 & -2.4&\cite{Namiki:JCP107:8848}\\
792: %% CrF &$^6\Sigma^+ $& 0.379& 0.539& 1.4 & 13.6 &\cite{Katoh:JCP120:7927}\\
793: %% MnS &$^6\Sigma^+ $& 0.195& 0.350& 1.8 & -2.4 &\cite{Thompsen:JCP116:10212}\\
794: %% CrCl&$^6\Sigma^+ $& 0.167& 0.266& 1.6 & 2.2&\cite{Katoh:JCP120:7927}\\
795: %% CrO &$^5\Pi_r $& 0.524& 1.148& 2.2 & 10.5&\cite{Sheridan:ApJ576:1108}\\
796: %% CrN &$^4\Sigma^- $& 0.624& 2.611& 4.2 & 7.0&\cite{Sheridan:ApJ576:1108}\\
797: %% VO &$^4\Sigma^- $& 0.546& 2.031& 3.7 & 22.5&\cite{Cheung:JMS91:165}\\
798: %% NbO &$^4\Sigma^- $& 0.432& 15.58& 36 & 334.0 &\cite{Adam:JCP100:6240}\\
799: %% TiH &$^4\Phi $& 5.362& & & 182.3&\cite{Launila:JCP104:6418}\\
800: %% \hline
801: %% NH &$^3\Sigma^- $&16.343& 0.920& 5.6 10$^{-2}$ & 54.7& \cite{Brazier:jms120:381}\\
802: %% O$_2$&$^3\Sigma^- $& 1.438& 1.985& 1.4 & -8.4&\cite{Berden:PRA58:3114}\\
803: %% CaH &$^2\Sigma^+ $& 4.229& & & 46.7& \cite{Barclay:ApJ408:L65}\\
804: %% CaF &$^2\Sigma^+ $& 0.339& & &1.3&\cite{Maussang:PRL94:123002}\\
805: %% \hline
806: %% \end{tabular}
807: %% \label{tab:molecules}
808: %% \end{table}
809:
810:
811: \newpage
812: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 1 %%%%%%%%%%%%%%%%%%%%%%%%
813: %% \begin{figure}[t]
814: %% \centering \includegraphics[width=1.0\columnwidth]{cryostat-half}
815: %% \caption{Scheme of the experimental setup. Figure \ref{fig:setup}a
816: %% shows the experimental cell within the custom-made cryostat,
817: %% positioned within the magnet bore. Figure \ref{fig:setup}b depicts
818: %% a cut through the copper cell itself.}
819: %% \label{fig:setup}
820: %% \end{figure}
821: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 1 %%%%%%%%%%%%%%%%%%%%%%%
822: %% \newpage
823:
824: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 2 %%%%%%%%%%%%%%%%%%%%%%%%
825: %% \begin{figure}[t]
826: %% \centering \includegraphics[width=1.0\columnwidth]{cr-trapped}
827: %% \caption{Transmission through a sample of Cr atoms in the buffer
828: %% gas cell (top graph). The probe laser frequency is scanned
829: %% repeatedly over the atomic transition with a repetition rate of
830: %% 1 Hz. The inset shows a typical absorption profile (triangles)
831: %% after $\sim$ 14 s, together with a simulation (dotted line). The
832: %% other two graphs show the evolution of the particle density
833: %% (middle) and temperature (bottom graph) of the trapped atom
834: %% ensemble over time.}
835: %% \label{fig:raw}
836: %% \end{figure}
837: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 2 %%%%%%%%%%%%%%%%%%%%%%%
838: %% \newpage
839:
840: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 3 %%%%%%%%%%%%%%%%%%%%%%%%
841: %% \begin{figure}[t]
842: %% \centering
843: %% \includegraphics[width=1.0\columnwidth]{cr-absorption-imaging}
844: %% \caption{Absorption images of a trapped cloud of Cr atoms, at three
845: %% different times (columns) and at three different laser
846: %% frequencies (rows). The size of the image is 8.6 mm x 6.9 mm. }
847: %% \label{fig:ccd}
848: %% \end{figure}
849: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 3 %%%%%%%%%%%%%%%%%%%%%%%
850: %% \newpage
851:
852:
853:
854:
855: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 4 %%%%%%%%%%%%%%%%%%%%%%%%
856: %% \begin{figure}[t]
857: %% \centering
858: %% \includegraphics[width=1.0\columnwidth]{cr-trapdepth-variation}
859: %% \caption{Density (upper graph) and temperature (lower graph)
860: %% of a trapped cloud of Cr atoms as a function of trap depth,
861: %% measured at three different times; at t=7 s (squares), t=14 s
862: %% (circles) and 35 s (triangles) after ablation. The lines in the
863: %% upper graph are guides to the eye, whereas in the bottom graph they
864: %% represent the average temperature.}
865: %% \label{fig:trapdepth}
866: %% \end{figure}
867: %% %%%%%%%%%%%%%%%%%%%%%%% FIG 4 %%%%%%%%%%%%%%%%%%%%%%%
868:
869: \end{document}
870: