1: %\documentclass[twocolumn,floatfix,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: % Some other (several out of many) possibilities
4: %\documentclass[preprint,aps]{revtex4}
5: %\documentclass[preprint,aps,draft]{revtex4}
6: \documentclass[aps,pra,twocolumn]{revtex4}% Physical Review A
7: \usepackage{graphicx}% Include figure files
8: \usepackage{dcolumn}% Align table columns on decimal point
9: \usepackage{bm}% bold math
10: \usepackage{amssymb}
11:
12: \begin{document}
13: \preprint{APS/123-QED}
14:
15: \title{Single atom detection of calcium isotopes by atom trap trace analysis}
16:
17: \author{S. Hoekstra, A. K. Mollema, R. Morgenstern, H. W. Wilschut and R. Hoekstra}
18: \affiliation{Atomic Physics, KVI, Rijksuniversiteit Groningen, Zernikelaan 25, 9747 AA, Groningen, The
19: Netherlands}
20:
21: \date{\today}% It is always \today, today,
22: % but any date may be explicitly specified
23:
24: \begin{abstract}
25: We demonstrate a combination of an isotopically purified atom beam and a magneto-optical trap which enables the
26: single atom detection of all stable isotopes of calcium (40, 42, 43, 44, 46 and 48). These isotopes range in
27: abundance from 96.9 \% ($^{40}$Ca) to 0.004 \% ($^{46}$Ca). The trap is loaded from an atomic beam which is
28: decelerated in a Zeeman slower and subsequently deflected over an angle of 30$^{\circ}$ by optical molasses. The
29: isotope selectivity of the Zeeman slower and the deflection stage is investigated experimentally and compared
30: with Monte Carlo simulations.
31: \end{abstract}
32:
33: \maketitle
34:
35: \section{\label{sec:level1}Introduction}
36: Trace analysis of long-lived isotopes has become an important tool in modern science. From medical science to
37: environmental research, from nuclear safety to archeology, the capability to detect low-abundance long-lived
38: isotopes has opened many research fields~\cite{60}. For the detection of trace elements Atom Trap Trace Analysis
39: (ATTA) is a promising and potentially very powerful new technique~\cite{223,213}. In a recent experiment one
40: million year old groundwater from the Nubian Aquifer (Egypt) has been dated using ATTA, by detecting very small
41: traces of $^{81}$Kr~\cite{215}. Because there are no interferences from other elements ATTA has the potential to
42: also detect the long-lived isotope $^{41}$Ca at its natural abundance level of $10^{-14}$~\cite{111}. This would
43: open the possibility to perform radio-calcium dating with $^{41}$Ca, which has a half-life of $10^{5}$
44: years~\cite{111,110}. Furthermore, $^{41}$Ca could be used as a tracer to directly monitor the bone loss and
45: retention rates of human subjects in both research and diagnosis of osteoporosis~\cite{109}. As an alternative
46: to Accelerator Mass Spectrometry (AMS), ATTA may provide a small-scale and cost-effective detection method of
47: rare isotopes, such as the example of $^{81}$Kr showed. Recently ATTA has been compared with the established
48: technique of Low Level Counting (LLC)\cite{217}. A competing compact laser-based method is Resonance Ionization
49: Mass Spectrometry (RIMS), which has been successfully developed in recent years \cite{312,313,315}. RIMS
50: combines the selectivity of laser spectrometry with the technique of mass separation. Recently it has been shown
51: that ATTA is already sufficiently sensitive to detect $^{41}$Ca in enriched ($10^{-8}$) calcium samples for
52: biomedical applications \cite{65}. We have set up an ATTA experiment with the ultimate goal of detecting
53: $^{41}$Ca at the natural abundance level.
54:
55: ATTA experiments combine various optical techniques, each of which is isotope selective. The mechanism of
56: isotope selection is the repeated excitation (by a laser) of an optically accessible electronic transition in
57: the neutral atom. Because of the isotope shift the scattering force induced by light of a fixed frequency is
58: different for the different isotopes. It is the ratio between the natural linewidth and the isotope shift of the
59: pumping transition that determines the selectivity that can be reached between neighboring isotopes. For the
60: calcium isotopes the isotope shift is about five times the natural linewidth; therefore isotope selectivity in
61: laser pumping of calcium isotopes is possible. Reducing the Doppler broadening of the transition is an essential
62: ingredient for the isotope selectivity. Therefore, cold atoms are an ideal and necessary tool for
63: ultra-sensitive isotope detection. Samples of cold atoms can be obtained by laser cooling and trapping in a
64: magneto-optical trap (MOT)~\cite{103}, which is the central element of an ATTA experiment.
65:
66: The final sensitivity that can be reached in an ATTA experiment is limited either by the background of $^{40}$Ca
67: atoms or by the loading rate of the trap. The background of $^{40}$Ca atoms can be reduced by improving the
68: isotope selectivity, the loading rate can be increased by improving the efficiency. In the ATTA measurements
69: reported on in \cite{65} the isotope selectivity was the limiting factor for the sensitivity. In this article
70: the isotope selectivity has been investigated in detail. This results in a setup in which the search for
71: $^{41}$Ca atoms is no longer limited by the isotope selectivity but by the loading efficiency. The loading
72: efficiency is limited at the moment by the laser power available for the experiment.
73:
74: In section~\ref{sec:setup} of this article the various parts of our experimental setup will be presented. In
75: section~\ref{sec:results} three different loading schemes for the MOT are compared with the purpose of improving
76: the isotope selectivity. Monte Carlo simulations were made to help understand the isotope selectivity of the
77: Zeeman slower and the deflection stage. Finally, experimental data of the detection of single calcium atoms in
78: the MOT are presented.
79:
80: \section{Experimental setup}\label{sec:setup}
81: \begin{figure}[htb]
82: \includegraphics[width=\columnwidth]{figure1}
83: \caption{\label{fig:setup}A schematic overview of the experimental setup.}
84: \end{figure}
85: Our experimental method is schematically depicted in figure~\ref{fig:setup}. The metallic calcium sample is
86: placed in oven 1 from which atoms are evaporated to create an atomic beam. The divergence of the atomic beam of
87: a selected isotope can be reduced using optical compression with laser beams perpendicular to the atomic
88: beam~\cite{249}. The other isotopes are not affected or even pushed out of the atomic beam. The second step in
89: the selection of the isotopes, and essential for the efficient loading of the magneto-optical trap, is a Zeeman
90: slower. It slows the atoms down by the absorption of photons from a counter-propagating laser beam~\cite{246}.
91: The Doppler shift resulting from the longitudinal velocity of the atoms is continuously compensated by a Zeeman
92: shift, induced by the magnetic field along the axis. The resulting slow atom beam leaving from the oven is
93: deflected over 30$^{\circ}$ in the direction of the Magneto-Optical Trap (MOT) by a standing wave field tilted
94: at the proper angle. This deflection stage is an essential step in the isotope selectivity and the total
95: sensitivity that can be reached. As will be shown in this paper only the selected isotope is effectively
96: deflected and guided into the trap. This improves the signal-to-noise ratio in the detection of the rare
97: isotopes, and thereby enables the use of a higher atom flux. The detection of the rare isotopes takes place in
98: the MOT. Only atoms of one isotope can be trapped for a given laser frequency. By scanning the laser frequency
99: and observing the fluorescence of the trapped atoms we detected all stable isotopes of calcium down to the
100: single atom level. The trap can also be loaded from a second oven; it is connected to the deflection chamber
101: with a valve as indicated in figure~\ref{fig:setup}. This second atom beam enables us to make a direct
102: comparison between the different methods by which the trap can be loaded. This will be presented in
103: section~\ref{sec:results}. The various parts of the experiment will be presented here in some more detail.
104:
105: At 423 nm, calcium has a strong resonance transition from the ground $4s^{2}$ $^{1}$S$_{0}$ state to the $4s4p$
106: $^{1}$P$_{1}$ state which is rather well suited for laser cooling. The required 423 nm laser light for the
107: cooling is generated by frequency doubling of the output of an 846 nm diode laser (Toptica Photonics AG). By
108: amplification in a tapered amplifier up to 500 mW of 846 nm light is produced; after the frequency doubling with
109: an LBO crystal we typically have 55 mW of 423 nm light available for the locking of the laser (5 mW), the
110: compression stage (10 mW), the Zeeman slower (20 mW), the deflection stage (5 mW) and the MOT (15 mW). The
111: amount of laser power available is the limiting factor for the efficiency of the experiment, and could be
112: increased by investing in a more intense source of light at 846 nm. The main laser beam is split and frequency
113: shifted using beam-splitters and acousto-optical modulators (AOM). The laser frequency is locked to the cooling
114: transition of calcium by means of polarization spectroscopy~\cite{4}, which is done on an atomic beam from oven
115: 2. The average trapping time of calcium in a MOT is limited to $\sim$ 20 ms due to a weak leak (10$^{-5}$) from
116: the $4s4p$ $^{1}$P$_{1}$ to the $^{1}$D$_{2}$ state. From the $^{1}$D$_{2}$ state roughly 75 \% decays to the
117: ground state within 3 ms, and can be recaptured if the diameters of the trapping laser beams are large enough.
118: The rest of the atoms are lost from the trap, and limit the trapping time. The trapping time can be increased by
119: repumping the atoms from the $^{1}$D$_{2}$ to the $5s$ $^{1}$P$_{1}$ state, from which they quickly decay back
120: to the ground state \cite{54}. The laser light required for the repumping is generated by a home built diode
121: laser operating at 672 nm, of which typically 5 mW is available for the MOT. Only for the measurements with
122: single atoms presented here the repump laser has been used.
123:
124: The calcium atoms are placed in an oven from which the atoms are evaporated. The ovens have 10 exit channels
125: each with a diameter of 1 mm and a length of 10 mm. A ceramic tube around the oven holds tantalum wires which
126: are used to heat the oven to temperatures in the range of 400 to 600 $^{\circ}$C. The Maxwellian velocity
127: distribution for calcium atoms with a temperature of 600 $^{\circ}$C peaks at a velocity of 600 m/s.
128:
129: The transverse velocity is less than one tenth of the longitudinal velocity, provided the mean free path of the
130: atoms is larger than the length of the exit channel of the oven. Directly after the atoms leave the oven we have
131: the possibility to apply transverse compression of the atom beam. The Doppler shift $\omega_{D}$ of the moving
132: atom, the isotope shift $I_{s}$ and the natural linewidth $\Gamma$ of the transition together determine the
133: effective scattering rate of a specific isotope. For a specific laser detuning $\delta$ and power $s_{0}$ the
134: scattering rate $\gamma_{p}$~\cite{205} is given by
135: \begin{eqnarray}
136: \gamma_{p}=\frac{s_{0}\Gamma/2}{1+s_{0}+(2(\delta +\omega _{D}+I_s{})/\Gamma )^{2}} \label{eqn:scattering}
137: \end{eqnarray}
138: As can be seen from table~\ref{tab:isotopes} the typical isotope shift between two adjacent isotopes of calcium
139: is about 160 - 200 MHz. The natural linewidth of the cooling transition is 34 MHz, and the Doppler shift for the
140: calcium atoms is 2.1 MHz/(m/s).
141: \begin{table}[hb]
142: \caption{\label{tab:isotopes}The abundances of the stable isotopes of calcium and the isotope shifts~\cite{15}
143: of the $4s^{2}$ $^{1}$S$_{0}$ - $4s4p$ $^{1}$P$_{1}$ transition. Also indicated is $^{41}$Ca. The shifts given
144: for $^{41}$Ca and $^{43}$Ca are for the 9/2 hyperfine component.}
145: \begin{center}
146: \begin{tabular}{c|c|c}
147:
148: {\it Isotope} & {\it Natural abundance(\%)} & {\it Isotope Shift (MHz)}\\
149: \hline \hline
150: 40 & $96.94$ & $0$\\
151: 41 & $1 \cdot 10^{-12}$ & $166$\\
152: 42 & $0.65$ & $393$\\
153: 43 & $0.14$ & $554$\\
154: 44 & $2.09$ & $774$\\
155: 46 & $4 \cdot 10^{-3}$ & $1160$\\
156: 48 & 0.19 & $1513$\\
157: \hline
158: \end{tabular}
159: \end{center}
160: \end{table}
161:
162: Because the transverse Doppler broadening is limited the compression is isotope selective: when tuning the laser
163: frequency in between two adjacent isotopes there will be a force on both isotopes, opposite in sign. The result
164: is that the lighter isotope of the two will be pushed away from the beam axis while at the same time the
165: transverse velocity component of the heavier isotope can be reduced, resulting in an improved transmission
166: through the Zeeman slower. We have done Monte Carlo simulations to investigate this effect in more detail: a
167: detailed comparison with experimental results will be reported on in the near future.
168:
169: The Zeeman slower is designed to decelerate atoms with initial velocities up to 1000 m/s (corresponding to 85\%
170: of the atoms) down to 50 m/s. The total length is 0.86 meter. The magnetic field has a maximum of 0.16 Tesla at
171: the entrance of the Zeeman slower, and decreases towards the exit. The laser beam counter-propagating the atoms
172: has a $\sigma^{+}$ polarization and a typical power of 20 mW. To avoid an extra velocity spread of the atoms,
173: the slowing process has to be terminated quickly, cf. reference ~\cite{87}. Therefore, the magnetic field has a
174: negative offset (as a result the slowing laser is detuned by -320 MHz) and is terminated with a strong positive
175: magnetic field. The laser beam is focused on the exit channels of the oven. As a result there is a gradient of
176: the laser intensity along the Zeeman slower, for which the shape of the magnetic field slope has been adjusted.
177: The coil generating the magnetic field is divided in 8 independently adjustable sections so that the field can
178: be optimized.
179:
180: In the deflection chamber the desired isotope is deflected out of the slowed atom beam by a combination of two
181: counter-propagating red-detuned laser beams which cross the atom beam under an angle of 30$^{\circ}$. At this
182: point the Doppler broadening is sufficiently reduced so that direct isotope selection is possible, even when
183: probing the beam under 30 degrees. At this point the laser beam has a diameter of only 0.5 cm, the same detuning
184: as the trapping beams, and has a typical power of 5 mW. Only the selected isotope will be deflected so that it
185: can reach the trap, which is located 40 cm further downstream.
186:
187: The atoms are trapped by a standard Magneto-Optical Trap. Typically 15 mW of laser power is split over 3 laser
188: beams which are retro-reflected to make the 6 required trapping beams. The magnetic quadrupole field is
189: generated by two coils in anti-Helmholtz configuration placed outside the vacuum (902 windings each, typical
190: current 3 A) resulting in a magnetic field gradient of 8 mT/cm. The fluorescence of the atoms in the trap is
191: measured by a photo-multiplier tube (PMT) in photon-counting mode. A lens system of 2.54 cm diameter is placed
192: inside the chamber at a focal distance of 3 cm from the trapped atoms to efficiently collect the fluorescence.
193: This lens system is adapted from reference \cite{13}. Outside the vacuum trapping chamber the light from the
194: trap is focused through a pinhole before detection by the PMT to reduce background light. For a large total
195: dynamic range of the detection system also a CCD camera and a photo-diode are used. The windows of the MOT have
196: been mounted on extension tubes and are anti-reflection coated. In order to further reduce the amount of
197: scattered light the inside of the vacuum chamber is coated black. The typical pressure in the MOT is $10^{-9}$
198: mbar which can rise to $10^{-8}$ mbar depending on the oven temperature.
199:
200: The abundance of very rare isotopes can no longer be determined directly from the intensity of the fluorescence
201: from the trapped atoms if the average population of the trap falls below one atom. In this case we have to be
202: able to count the number of atoms arriving and leaving the trap over a certain period of time, and determine the
203: abundance from a comparison with the known abundance of another isotope. Experimental results on the detection
204: of the single atoms are presented and discussed in section \ref{sec:singleatoms}.
205:
206: The present sensitivity of $^{41}$Ca detection by ATTA, as reported by Moore et al \cite{65}, is limited by the
207: background of $^{40}$Ca atoms present in the trap, and by the trapping time of the atoms. Even though the
208: $^{40}$Ca atoms are not trapped while detecting $^{41}$Ca, they contribute to the fluorescence signal from the
209: trap, obscuring the signal from the less abundant isotopes. It is therefore a key issue to further increase the
210: isotope selectivity. Besides the fluorescence background from these $^{40}$Ca atoms, there is the fluorescence
211: background due to scattering of the trapping laser beams from the windows and walls of the trapping chamber. For
212: any measurement of single atoms in a MOT this source of background has to be made as small as possible.
213:
214: We have compared the isotope selectivity of three different trap loading schemes. As an indication of the
215: isotope selectivity of our system the amount of fluorescence due to the hot $^{40}$Ca atoms is measured while
216: trapping the calcium isotope $^{43}$Ca. The experimental results are presented in the next section.
217:
218: \section{Results}\label{sec:results}
219: \begin{figure}[htb]
220: \includegraphics[width=\columnwidth]{figure2}
221: \caption{\label{fig:figure2}The fluorescence of calcium isotopes in an atomic beam after a single excitation
222: step.}
223: \end{figure}
224: The starting point of the experiment is the atomic beam as it leaves the oven. The natural abundance of the
225: various isotopes in this beam can be seen in figure~\ref{fig:figure2}. In this figure the fluorescence of the
226: atomic beam coming from oven 2 is shown, excited only by the vertical trapping laser beam in the MOT. The
227: fluorescence is measured by a PMT under right angles with both the atom beam and the laser beam. The peaks have
228: been fitted with Lorentzian profiles, all having the same width (38 MHz) and an amplitude corresponding to the
229: natural abundance as given in table~\ref{tab:isotopes}. The fit curve is in good agreement with the measured
230: spectrum. The measured width of 38 MHz is slightly larger than the natural linewidth of 34 MHz: this corresponds
231: to a transverse velocity component of 1.9 m/s in the atomic beam. This figure gives an indication of the limited
232: maximum selectivity that can be reached in a single excitation step: the power of ATTA lies in the fact that the
233: same transition is excited very often.
234:
235: \subsection{Comparing different loading schemes}\label{sec:comparing}
236: \begin{figure}[tb]
237: \includegraphics[width=\columnwidth]{figure3}
238: \caption{\label{fig:figure3}Fluorescence in the trap loaded from oven 2. The background signal is caused by
239: $^{40}$Ca passing through the trap with its Maxwell-Boltzmann velocity distribution and excited by the
240: horizontal trapping laser beams.}
241: \end{figure}
242: In this section three different loading schemes are compared with respect to isotope selectivity. The simplest
243: configuration is to load the MOT directly from a thermal beam. Since the capture velocity of the trap is $\sim
244: 50$ m/s only the low velocity tail of the Boltzmann distribution is trapped. When the MOT is operated on one of
245: the heavier isotopes, the large amount of $^{40}$Ca limits the detection of these less abundant isotopes in the
246: trap. The horizontal trapping beams which intersect the atom beam at 45$^{\circ}$ can excite the fast $^{40}$Ca
247: atoms in the atom beam as the frequency of the trapping laser beams is scanned over the various isotopes. The
248: broad velocity distribution of these hot $^{40}$Ca atoms can be seen in figure~\ref{fig:figure3} as it dominates
249: the fluorescence at frequencies where for example trapped $^{46}$Ca atoms should be visible. To enable
250: comparison with figures \ref{fig:figure4} and \ref{fig:figure5}, the fluorescence is set to 1 at the resonance
251: frequency of $^{46}$Ca (1160 MHz). The hot atoms not only scatter the laser-light, thereby obscuring the signal
252: from the trapped atoms, but also shorten the trapping time of other isotopes by collisions with the trapped
253: atoms. The ratio between the fluorescence of the trapped $^{43}$Ca and the background $^{40}$Ca is $\sim 0.15$.
254:
255: \begin{figure}[tb]
256: \includegraphics[width=\columnwidth]{figure4}
257: \caption{\label{fig:figure4}Fluorescence in the trap loaded from a slowed but not deflected atom beam. This
258: spectrum was taken before the deflection stage was installed (cf. fig. \ref{fig:setup})}
259: \end{figure}
260: When the trap is loaded from a slowed atom beam the amount of trapped atoms increases because a larger fraction
261: of the atoms falls within the capture range of the trap. Figure~\ref{fig:figure4} shows the fluorescence
262: detected in the MOT, while simultaneously scanning the frequency of the trapping laser and the Zeeman slower
263: laser beams over the range of isotopes. This measurement was done before the deflection stage was installed. It
264: can be seen that the relative contribution of the hot $^{40}$Ca atoms is greatly reduced. Comparing
265: figure~\ref{fig:figure3} to figure~\ref{fig:figure4}, the ratio of for example the trapped $^{44}$Ca to the
266: $^{40}$Ca-background increases from $2.5$ to $5000$. The reduction of the $^{40}$Ca-background enables the
267: detection of trapped $^{46}$Ca atoms with a natural abundance of only 0.004 \%, which was not possible when
268: loading the trap directly from the thermal beam. The figure is normalized to the intensity of the $^{46}$Ca
269: peak. The ratio between the trapped $^{46}$Ca and the background $^{40}$Ca is $\sim 3$, for $^{43}$Ca the ratio
270: to the $^{40}$Ca background is $\sim 15$. The increase of isotope selectivity is the result of both the
271: increased fraction of slow atoms and the isotope selectivity of the Zeeman slower itself. In section
272: \ref{sec:simulations} we will look at this in some more detail.
273:
274: \begin{figure}[tb]
275: \includegraphics[width=\columnwidth]{figure5}
276: \caption{\label{fig:figure5}Fluorescence in the trap loaded from a slowed and deflected beam.}
277: \end{figure}
278: We find that the background signal from fast $^{40}$Ca atoms disappears almost completely when loading the trap
279: from a deflected, slow beam. In the measurement presented in figure~\ref{fig:figure5} the background in between
280: all the trapped isotopes is found to be mostly dependent on the intensity of the laser trapping beams, and only
281: slightly on the oven temperature (atom flux). This is further illustrated by figure~\ref{fig:peakback}. Here the
282: fluorescence from the trap is shown while scanning over the $^{43}$Ca trapping frequency for two different oven
283: temperatures. While the $^{43}$Ca peak intensity increases by a factor of 6.6 (from $\sim 500$ to $\sim 3300$
284: counts/10 ms), the average background level changes only from $475.9 \pm 0.7$ to $492.5 \pm 0.5$ counts/10 ms.
285: This insignificant background increase of less then 4 \%, corresponding to $16.6 \pm 0.6$, is due to extra
286: $^{40}$Ca atoms. The purely $^{40}$Ca dependent background component should also have increased by a factor of
287: 6.6: therefore we can conclude that at 450 $^{\circ}$C the background contribution due to $^{40}$Ca is $2.5 \pm
288: 0.1$ counts/10 ms. The $^{43}$Ca signal relative to the $^{40}$Ca background is therefore $\sim 200$. When
289: comparing this to a directly trapped thermal beam (figure \ref{fig:figure3}) the ratio of the $^{43}$Ca signal
290: to the $^{40}$Ca background is increased by a factor of $\sim 1300$. The rest of the background is due to laser
291: light scattered from the walls and the windows of the trapping chamber, and can be further reduced to 125 counts
292: / 10 ms as shown by the data on single atoms, presented in section \ref{sec:singleatoms}. It is noted that the
293: measured $^{43}$Ca to $^{46}$Ca ratio changes, when comparing figure \ref{fig:figure4} to figure
294: \ref{fig:figure5}. This is discussed in section \ref{sec:fractionation}.
295: \begin{figure}[tb]
296: \includegraphics[width=\columnwidth]{figure6}
297: \caption{\label{fig:peakback}Fluorescence intensity of $^{43}$Ca and background level intensity for two
298: different oven temperatures. The $^{43}$Ca peak intensity increases by a factor of 6.6, while the fluorescence
299: background level remains almost constant.}
300: \end{figure}
301:
302: \subsection{Monte Carlo simulations}\label{sec:simulations}
303: \begin{figure}[tb]
304: \includegraphics[width=\columnwidth]{figure7}
305: \caption{\label{fig:zeemansl}A calculation of the final longitudinal velocity distribution (1000 atoms each) of
306: the isotopes $^{40}$Ca, $^{41}$Ca and $^{42}$Ca travelling through the Zeeman slower. The laser detuning is set
307: for $^{41}$Ca. An example of a Monte Carlo sampling of the initial velocity distribution upon entering the
308: Zeeman slower is shown in the top panel.}
309: \end{figure}
310: The isotope selectivity of the Zeeman slower can be understood on basis of the results of Monte Carlo
311: simulations. In the simulation results depicted in figure~\ref{fig:zeemansl}, the final velocity distributions
312: for $^{40}$Ca, $^{41}$Ca and $^{42}$Ca atoms were calculated for our Zeeman slower with the laser frequency
313: optimized for $^{41}$Ca. Most of the $^{41}$Ca atoms are decelerated to the desired velocity of about 50 m/s;
314: for both the other isotopes this is different. This can be understood as follows: for a given laser detuning all
315: the isotopes of calcium will be resonant in the Zeeman slower, but at different locations because of the isotope
316: shifts. These isotope shifts translate into different scattering rates in the Zeeman slower, according to
317: equation~\ref{eqn:scattering}. Therefore the final velocity of the different isotopes leaving the Zeeman slower
318: will be different. The magnetic field slope of the Zeeman slower is steeper at the end, where the average
319: velocity of the atoms is lower. If the laser is tuned to $^{41}$Ca atoms, then the $^{42}$Ca atoms will
320: initially also be decelerated, but the slope of the magnetic field is too steep: the change in velocity is not
321: enough to keep up with the change in magnetic field. Therefore this isotope will be lost from the slowing
322: process. On the other hand, a $^{40}$Ca atom is nicely slowed down until the end, but the slowing process
323: continues too long. In the simulation the far majority of these atoms undergo a reversal of their longitudinal
324: velocity. They will thus be lost even due to their small transversal velocity component and do not reach the
325: MOT.
326:
327: As can be seen from figure~\ref{fig:figure4} a significantly lower background from $^{40}$Ca is detected in the
328: MOT for a trapping detuning smaller than about 450 MHz. With this detuning we probe $^{40}$Ca atoms with a
329: velocity range below 200 m/s. Since there is at the moment no collimation of the atom beam at the exit of the
330: Zeeman slower these atoms probably do not reach the trap due to the divergence of the atomic beam. The Zeeman
331: slower is designed to slow the desired isotope to 50 m/s; from the fact that we trap these isotopes we conclude
332: that for these isotopes the gain due to the larger number of slow atoms outweighs the loss due to the transverse
333: velocity spread. However, many more of these slow atoms could reach the trap if we implement a second
334: transversal cooling stage at the exit of the Zeeman slower.
335: \begin{figure}[tb]
336: \includegraphics[width=\columnwidth]{figure8}
337: \caption{\label{fig:simulation}A simulation of the large-angle atomic beam deflection for the different
338: isotopes. Shown is the deflection angle as a function of the initial velocity for three different isotopes:
339: $^{40}$Ca, $^{41}$Ca and $^{42}$Ca. The inset shows the geometry of the atomic beam and the deflection laser
340: beams. The laser detuning is 40 MHz red detuned with respect to the $^{41}$Ca resonance frequency. Further
341: details are given in the text.}
342: \end{figure}
343:
344: To assess the functionality of the large-angle deflection by optical molasses we have performed Monte-Carlo
345: simulations. $^{40}$Ca, $^{41}$Ca and $^{42}$Ca atoms with a range of longitudinal velocities have been traced
346: through the deflection molasses. Absorption and emission probabilities are calculated along the way, and every
347: time a photon is absorbed and emitted the velocity of the atom is adjusted correspondingly. Stimulated emission
348: is also accounted for. Plotted in figure~\ref{fig:simulation} are the deflection angles as a function of the
349: initial velocity for $^{40}$Ca, $^{41}$Ca, and $^{42}$Ca atoms. In the inset the geometry of the atomic beam and
350: the deflection laser beams is shown. As in the experiment, the laser power is 5 mW distributed uniformly over a
351: beam diameter of 0.5 cm. The deflection laser is tuned 40 MHz below the resonance frequency of $^{41}$Ca. While
352: the $^{41}$Ca atoms are deflected well for velocities up to 90 m/s, the $^{40}$Ca are pushed away from the
353: atomic beam axis. This is the main reason that the combination of a Zeeman slower and the large-angle deflection
354: is so effective in selectively deflecting only one desired isotope out of the atom beam. Due to the large
355: additional detuning caused by the isotope shift, $^{42}$Ca is not affected at all. Similar simulations have been
356: done for different laser power, and it follows that both the isotope selectivity and the efficiency of the
357: deflection stage increase with increasing laser power.
358:
359: For $^{40}$Ca we can analyze the deflection behavior in some more detail. Only the velocity component along the
360: axis of the molasses laser beams (called $v_{mol})$ is affected by the deflection molasses. The $^{40}$Ca atoms
361: are much closer to resonance to the molasses laser beam pushing them away from the trap (laser 2, cf. fig
362: \ref{fig:simulation}). The number of scattered photons depends on the scattering rate as given in
363: formula~\ref{eqn:scattering}, and on the time the atoms spend in the optical molasses. For the $^{40}$Ca atoms
364: at the given detuning this scattering rate is a Lorentzian with its maximum at $v_{mol} = 50$ m/s. This
365: corresponds to an atom with a longitudinal velocity (called $v_{long}$) of 100 m/s. During the time that the
366: atoms are in the optical molasses $v_{mol}$ will change due to the scattering. The atoms can be accelerated to a
367: final $v_{mol}$ which depends on the time spent in the molasses and the linewidth of the transition. The
368: deflection angle is determined by the ratio between $v_{mol}$ and $v_{long}$. The minimum in the deflection
369: angle for $^{40}$Ca atoms around 20 m/s is the region where $v_{mol}$ is limited by the time the atom spent in
370: the molasses. For $v_{long}<5$ m/s the deflection reaches the maximum possible deflection angle of 60$^{\circ}$,
371: i.e. parallel to the molasses laser beams. For $v_{long}>30$ m/s $v_{mol}$ reaches a maximum value limited by
372: the linewidth of the transition. The deflection angle is then just determined by the ratio of this maximum value
373: of $v_{mol}$ and the initial longitudinal velocity.
374:
375: \subsection{Single atoms}\label{sec:singleatoms}
376: \begin{figure}[bt]
377: \includegraphics[width=\columnwidth]{figure9}
378: \caption{\label{fig:singleatoms}Fluorescence of individual atoms detected in the MOT. The dotted lines indicate
379: the fluorescence level for 1, 2 and 3 atoms in the trap. The integration time of the photon-detector is 4 ms.}
380: \end{figure}
381: Single atoms can be detected in a magneto-optical trap because the trapped atoms are continuously scattering
382: photons from the trapping laser beams. Provided the amount of detected photons not originating from the trapped
383: atom is low enough, a single trapped atom can be detected as a temporal burst of light. In order to test the
384: sensitivity of the detection system, we have detected the fluorescence of trapped calcium atoms for low oven
385: temperatures. With decreasing oven temperature the atomic beam becomes less and less intense, down to a loading
386: rate of only a few atoms per minute. At some point the fluorescence from the trap displays discrete steps: we
387: can count the number of atoms in the trap. Shown in figure \ref{fig:singleatoms} is the fluorescence of single
388: $^{40}$Ca atoms detected in the trap for a period of 10 seconds. We have also detected single atoms from all
389: other stable calcium isotopes. During the time that an atom is trapped it can scatter photons at an estimated
390: rate (depending on laser detuning and intensity) of $1.3 \cdot 10{^6}$ s$^{-1}$. We detect $35 \pm 3$ photons
391: per 4 ms per atom, which results in a total photon detection efficiency of 0.7 \%. The background level is 50
392: photons per 4 ms. This should not be confused with the noise level, which is limited by statistics. Trapping the
393: atoms longer enables us to use a larger integration time which reduces the statistical error on the signal.
394: During the measurement of figure \ref{fig:singleatoms} the repump laser was used: we concluded from trap decay
395: time measurements that the average trapping time improved from $\sim$ 20 ms to $\sim$ 200 ms. It is important to
396: note that the background contribution due to hot $^{40}$Ca atoms such as reported in section
397: \ref{sec:comparing}, even at high oven temperatures, is much lower then the fluorescence signal of a single
398: trapped atom.
399:
400: \subsection{Isotope fractionation}\label{sec:fractionation}
401: The final issue we will address in this article is that of isotope fractionation effects in the experiment. The
402: maximum fluorescence of the various trapped isotopes in figure~\ref{fig:figure5} is compared to literature
403: values~\cite{248} in table~\ref{tab:abundance}. The absolute amount of $^{40}$Ca could not be measured
404: accurately in the same experiment due to the high intensity of the fluorescence: therefore the relative
405: abundance of the heavier isotopes was compared to $^{44}$Ca. The abundance of $^{44}$Ca was set to the
406: literature value of $2.09$.
407: \begin{table}[h]
408: \caption{\label{tab:abundance}The measured abundance of the stable calcium isotopes compared to literature
409: values from~\cite{248}. $^{44}$Ca is taken as a reference point, and for $^{43}$Ca two values are shown for two
410: different laser powers (50 mW / 70 mW).}
411: \begin{center}
412: \begin{tabular}{l|l|l}
413: {\it Isotope} & {\it Measured(\%)} & {\it Literature(\%)}\\
414: \hline \hline
415: 40 & $-$ & $96.94(16)$\\
416: 42 & $0.68$ & $0.65(2)$\\
417: 43 & $0.006 / 0.06$ & $0.14(1)$\\
418: 44 & $[2.09]$ & $2.09(11)$\\
419: 46 & $0.005$ & $0.004(3)$\\
420: 48 & $0.15$ & $0.19(2)$\\
421: \hline
422: \end{tabular}
423: \end{center}
424: \end{table}
425:
426: For all isotopes except $^{43}$Ca the agreement is reasonable. In the case of $^{43}$Ca we detect an amount of
427: fluorescence which would indicate an abundance of only $0.006 \%$ for the typically used laser power of $\sim$
428: 40 mW. This corresponds to only $\sim$ 4 \% of the literature value. This large discrepancy is due to
429: differences in the cooling and trapping efficiency of the odd isotopes of calcium: the odd isotopes $^{41}$Ca
430: and $^{43}$Ca have a nuclear spin of I=7/2. The resulting magnetic substructure of the ground and the excited
431: state influences the Doppler cooling force. Comparable observations have been reported for the case if the odd
432: strontium isotopes~\cite{236}. We have adapted the theoretical model developed for strontium and have solved the
433: generalized optical Bloch equations for the odd calcium isotopes taking into account all the hyperfine states
434: and their magnetic sub-states. We find that the maximum deceleration force on the odd isotopes for a
435: one-dimensional $\sigma^{+}$-$\sigma^{-}$ optical molasses configuration is only half as strong as that on the
436: even isotopes. This indicates that the number of photons emitted per trapped atom in a certain period of time is
437: only half that of the even isotopes. This alone cannot explain the reduction of a factor of 25 in the detection
438: of $^{43}$Ca. Since the cooling force is weaker this effect could also limit the efficiency of the deflection
439: and slowing of the odd isotopes. In order to investigate this issue, we have done a recent experiment where we
440: find that the measured $^{43}$Ca abundance depends significantly on the laser-power. Increasing the total
441: laser-power from 50 mW to 70 mW the ratio between $^{43}$Ca and $^{46}$Ca increases from 1.2 to 15. This ratio
442: can be more accurately measured than the relative abundances, due to non-linear effects in the detection system
443: for the very high count rates when measuring $^{44}$Ca. A ratio of 15 corresponds to a measured abundance for
444: $^{43}$Ca of 0.06 \%: this is roughly half of the literature value. This observation indicates clearly that the
445: efficiency of the deflection and the Zeeman slower is different for the odd and even isotopes, and depends
446: critically on the laser power available. Thus for the planned investigations on $^{41}$Ca, the isotope $^{43}$Ca
447: should be chosen as reference, due to its expected similar behavior in the experimental scheme.
448:
449: \section{Conclusion}
450: We have demonstrated the trapping of all stable isotopes of calcium involving cooling in a Zeeman slower and
451: deflection of the cooled beam by 30$^{\circ}$ in a tilted optical molasses. A corresponding experimental system
452: has been built with a sensitivity that enables us to detect even single trapped atoms of all these isotopes. To
453: suppress the interfering $^{40}$Ca fluorescence background when detecting the less abundant isotopes the
454: deflection of the slowed loading beam over 30$^{\circ}$ was found to be very effective. The background due to
455: $^{40}$Ca decreased by three orders of magnitude compared to direct loading from a thermal atomic beam. We find
456: that finally the background signal for single atom detection of the less abundant isotopes is no longer limited
457: by the $^{40}$Ca atoms but dominated by laser stray light. This enables single atom detection of all stable
458: isotopes. Improvements will be made to the setup: increasing the laser power will increase the efficiency and
459: further improve the performance and the isotope selectivity of the individual components: transversal beam
460: compression, Zeeman slower and deflection stage. Adding a second isotope selective compression stage directly
461: after the Zeeman slower is another potential improvement. With these changes the system will be ready to start
462: searching for $^{41}$Ca atoms at the natural abundance level.
463:
464: \begin{acknowledgments} The authors would like to thank the KVI technical staff (J. Mulder and J. Sa)
465: for their support, C. van Ditzhuijzen for her work on the Monte-Carlo simulations, J. W. Dunn for providing the
466: code to take into account the multilevel structure of $^{43}$Ca and Toptica Photonics for prompt assistance.
467: This project (00PR1887) is part of the research program of the Stichting voor Fundamenteel Onderzoek der Materie
468: (FOM) which is supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). Additional
469: support is from the EU contract NIPNET (HPRI-CT-2001-50034).
470: \end{acknowledgments}
471: \bibliographystyle{apsrev}
472: \bibliography{references}
473: \end{document}
474: %
475: % ****** End of file article.tex ******
476: