On the operation of a source of cold SrF molecules
Jonathan Ellen 2214962
Supervisors: Dr. S. Hoekstra & Dr. T. Schlath¨ olter
Faculty of Mathematics and Natural sciences
by Jonathan Ellen 2214962
Supervisors: Dr. S. Hoekstra & Dr. T. Schlath¨olter
In this Bachelor thesis I describe the work I did on the instalment of a new source chamber for SrF molecules. The molecules are created by laser ablation of a rotating rod of SrF2, which are then entrained in argon. The new chamber should be better at producing molecules with a lower internal temperature because of an added housing to improve collisional cooling. No molecules were detected during my project. Efforts were made to align the setup, get the timing between valve and ablation right and modify the duration and strength of the carrier gas. Future improvements are discussed, including rebuilding the rod housing and the use of a skimmer.
1 Overview of our and other experiments 1
1.1 Introduction. . . 1
1.2 Deceleration and trapping of molecules . . . 3
1.2.1 Molecule source. . . 3
1.2.2 Stark decelerator . . . 3
1.2.3 Laser cooling . . . 4
1.3 Other experiments . . . 5
1.3.1 Introduction . . . 5
1.3.2 Fritz Haber Institute, Berlin. . . 5
1.3.3 Vrije Universiteit, Amsterdam. . . 6
1.3.4 Weizmann institute, Israel . . . 6
1.3.5 EPFL, Switzerland . . . 6
2 A new source chamber 7 2.1 Introduction. . . 7
2.2 Setup . . . 8
2.2.1 Ablation chamber . . . 8
2.2.2 Carrier gas . . . 9
2.2.3 Rotating rod and housing . . . 9
2.2.4 Measurement chamber . . . 10
2.3 Future improvements . . . 12
2.3.1 Dye laser . . . 12
2.3.2 Detection laser frequency . . . 13
3 Measurements and results 14 3.1 Introduction. . . 14
3.2 Measurements . . . 14
4 Conclusions and remarks 17
A Aligning a dye laser 18
Overview of our and other experiments
The standard model describes most physical interactions and their force carriers. As the pinnacle of modern science it has proven to be remarkably robust. In an effort to search for physics beyond the standard model researchers try various approaches. One of these approaches is to find parity violation.  So far molecules have never been used to search for parity violation despite their predicted sensitivity. The transition strength is governed by parity selection rules. Thus, if a deviation were to be found in these transitions, this might be used to prove that the standard model is not yet complete.
Experiments on extremely cold molecules will be precise enough to say something about parity violation.
Our experiment is aimed at decelerating molecules and ultimately bring them to stand- still, after which they are trapped in an optical dipole trap for measurement on a parity forbidden transition. These molecules will have a temperature of around 150 µK. In order to decelerate molecules to such low temperatures a special Stark decelerator is combined with laser cooling.  This is done in a setup that is described below. Obvi- ously the entire experiment is being conducted at high vacuum to minimize the amount of unwanted collisions and to allow for supersonic expansion.
My project has been to improve one of the components of this decelerator, the source chamber. This includes fixing any present problems and doing measurements to opti- mize the new chamber. Optimizing in this case means trying to make a lot of molecules and getting as many of these molecules in the X2Σ+ (v=0) electronic and vibrational ground state as possible. Having the molecules in the ground state is important for Stark
Chapter 1. Overview of our and other experiments 2
deceleration. As shown in figure1.1the energy of the (1,0) rotational state goes up with increasing field strength. This state is thus a ’low field seeker’ since it is attracted to an environment with a low electric field strength. Therefore these molecules will stay trapped in a potential minima. 
The laser used to detect molecules is sensitive only to the electronic level transitions that are also used to laser cool the SrF however. This gives no information on the dis- tribution between ground and exited states, which is a direct indicator of the rotational temperature of the sample. A second project will therefore be to align a dye laser, which can scan over the large wavenumber range necessary to measure this temperature.
Ene rgy (cm-1
0.5 0 0.5 1.0 1.5 2.0
Electric ﬁeld (kV/cm)
0 20 40 60 80 100
Figure 1.1: The energy shift due to the Stark effect for SrF in the X2Σ+ ground state. The rotational number N=1 and projection on the electric field axis MN=0 level
is used for trapping the molecules, (N,MN)=(1,0).
Chapter 1. Overview of our and other experiments 3
1.2 Deceleration and trapping of molecules
1.2.1 Molecule source
The strontium fluoride (SrF) must be created in a so-called source chamber. So far laser ablation of a SrF2sample has been used to create SrF molecules which are then entrained in an inert carrier gas (Xenon or Argon) and transported to the Stark decelerator. In order to prevent the ablation laser from damaging the sample to much, it is mounted on piezoelectric mounts so that it can moved around a little. The created SrF has a high (around 10K) rotational temperature however. My project is to work on a new version of this chamber that has a housing around the SrF2 sample. The reduced space will improve the number of collisions, which helps cool the sample. In addition, a new way of moving the SrF sample is utilized, so that a fresh spot is ablated at each iteration.
The carrier gas expands into the vacuum at supersonic speeds. The collisions between the molecules and the carrier gas help cooling the sample. Because of these collisions all molecules travel approximately at the same speed and in the same direction giving rise to a ’directed flow’. In this directed flow molecules have transferred most of their internal energy to kinetic energy, leaving them internally cold. The rotationally coldest molecules are then selected by a skimmer, which is a conically shaped orifice which only lets the center of the beam, the so-called ’zone of silence’, pass. Figure 1.2, taken and modified from  shows a schematic overview of this shock wave and the zone of silence.
The skimmer also prevents turbulent gas flows from destroying the beam by blocking their path. The resulting velocities after the skimmer are too high (around 300 m/s for Xenon and 500 m/s for Argon) for laser cooling however. To bring the molecules to a more usable velocity, a Stark decelerator is used. 
1.2.2 Stark decelerator
The Stark decelerator used in our experiments utilizes moving potential minima to keep the molecules confined in space. If these wells are then decelerated the trapped molecules will decrease in speed with them. The principle is shown in figure1.3. The advantage of this method over traditional Stark decelerators is that molecules trapped in the electric fields in three dimensions and therefor cannot escape, resulting in a very low loss. 
The guiding fields are the result of an oscillating voltage applied to a series of rings mounted on rods. The rings form a tunnel for the molecules to fly through with the
Chapter 1. Overview of our and other experiments 4
Figure 1.2: Schematic overview of the shock wave created when supersonically ex- panding gas. The idea is that only the central zone of silence contains molecules with
the desired properties. A skimmer only allows this center to pass.
potential minimum in the center. The timing of the voltage on the rings is of utmost importance. When molecules arrive at the decelerator they should find a potential well there, which should then be moving along with them through the decelerator.
1.2.3 Laser cooling
After the molecules have been slowed enough and their occupation in phase space is redistributed in such a way that laser cooling is possible,we apply it to bring the SrF to near standstill. The principle of laser cooling is to have a laser beam counter-propagating the direction of the SrF beam. In our case the lasers will come from four sides, cooling the molecules trapped in the electric field wells. By tuning this laser to a transition the molecules can be excited, after which they fall back to their initial state, sending
Chapter 1. Overview of our and other experiments 5
the bulk energy in all directions causing them to decrease in energy.   One of the main reasons for using SrF in the first place is because of its favourable Franck-Condon factors. These determine the likelihood of transitions to the old state to occur after exciting them to another one. In this case the SrF will fall back to the ground state 98%
of the time, ready to be re-excited by the laser.
Longitudinal position (mm)
Transverse position (mm)
0 2 4 6 8 10 12
0 1 2 3
10 20 30 40kV/cm
Figure 1.3: The electric fields generated by applying a high voltage current to the metal rings (the blue dots and red triangles). The molecules are trapped in the blue potential minima. The entire field configuration moves to the right at decreasing speed.
1.3 Other experiments
Cold molecules are often required for precision measurements and also the research on new deceleration techniques is very interesting. Reading up for my project I encountered some of the groups doing research in this field, some of which I will mention here.
1.3.2 Fritz Haber Institute, Berlin
At the Fritz Haber institute research is being done to a variety of fields. One group focusses on the manipulation of molecules using various techniques, which may then be applied to slow and cool them. Another traps molecules on chips. Recently they managed to improve a Stark decelerator by decelerating only one third of the time (a so called s=3 mode) and using the other electrodes to improve the focussing. After the head of the institute moved to Nijmegen a lot of research has spread out to other laboratories across Europe. 
Chapter 1. Overview of our and other experiments 6
1.3.3 Vrije Universiteit, Amsterdam
The VU Amsterdam is the home of the group of Hendrick Bethlem, with whom our group has united under the ’Broken mirrors and drifting constants’ banner. They are interested in accurately measuring a possible time variation of the electron to proton mass ratio. In order to do this they are building an ammonia fountain to increase the measurement time to half a second.  Other experiments include utilizing the internal rotations of methanol and a synchrotron to store and do collision experiments with molecules.  
1.3.4 Weizmann institute, Israel
Similar to our Stark decelerator, a Zeeman decelerator may be used as well. The Zeeman technique uses magnetic rather than electric fields to slow molecules. However, the principle of timing the field strength is almost the same. This group managed to use the Zeeman effect to bring Oxygen to near rest. 
1.3.5 EPFL, Switzerland
While working at the Fritz Haber Institute Andreas Osterwalder designed the first trav- elling wave Stark decelerator of the kind that we use today.  At the moment the main focus lies with reaction dynamics at the low temperatures at which quantum effects start to play an important role.
A new source chamber
As mentioned before, one of the best ways to improve the deceleration experiment is to improve the source chamber. If the source chamber can produce colder and slower molecules, and with a higher intensity, less deceleration in the Stark decelerator is needed. Since the acceptance of the decelerator, i.e. the amount of molecules actu- ally being trapped in the potential wells depends strongly on the deceleration it is of utmost importance to start off with slow molecules.
My project in its entirety will thus consist of first getting the new chamber to work. Af- ter this the chamber has to be optimized. In order to do this the created SrF molecules must be detected and their rotational and vibrational temperature must be minimized.
The temperature is calculated by measuring the fraction of molecules in the ground state. A wide range of wavelengths must be scanned to obtain this fraction. A dye laser can give us access to this wide range, but the available dye laser was not working at the start of this project. Therefore a third goal is to get this old dye laser operational to do the temperature measurements. In the mean time molecules may be detected using the diode laser that is also being used in the main setup. This laser is tuned to the transition between the ground- and first exited state of SrF and should induce decay which can then be measured. A sketch of the design of the source chamber is given in figure2.1.
The design for such a source chamber is similar to the work of the group of professor E. Smalley and is subsequently named a Smalley source.  Later this group designed a way to rotate the sample in front of the ablation laser after also encountering the problem with the ablation laser burning holes in the sample. The total measurement setup is more like the one used in .
Chapter 2. A new source chamber 8
Figure 2.1: Overview of the new source chamber. The carrier gas enters from the left. The SrF2 sample is located at the white circle. The ablation laser beam enters
from the top, through the triangular entrance.
2.2.1 Ablation chamber
The ablation chamber has four major components. First there is the carrier gas entrance, which consists of a solenoid valve that is opened in pulses to let gas pulses into the vacuum that last 100-200 microseconds. Secondly we have the rotating sample of SrF2
that we are ablating using an nD:YAG laser. The laser is the third part, although not technically inside the vacuum chamber. The rod is kept inside a specially designed chamber to direct the carrier gas over the sample and promote collisions in order to decrease the translational temperature of the SrF. The fourth part is a skimmer, to extract the desired cold molecules from the gas plume. This skimmer, though highly recommended, was only installed near the end of my project, since it was planned to try without one at first. 
All triggers controlling the separate actions taken in the creation process are sent from
Chapter 2. A new source chamber 9
a single trigger and delay box. This allows for accurate timing between the different stages. For example the ablation laser is Q-switched with a 190 microsecond delay between the flash and the Q-switch. On top of that the opening of the valve and the travel time between the valve and the SrF2sample all add additional delays which should be accounted for. The trigger box operates at 10 Hz, with each cycle of creating and measuring taking approximately 700 microseconds.
2.2.2 Carrier gas
In order to transport any created molecules from the ablation zone to the detection area, an inert carrier gas is used. In our case this gas was argon with a backing pressure of 3 bar. From earlier experiments we knew the maximum of the gas pulse has to pass the SrF2 sample approximately 400 microseconds after the valve is triggered.  Such a pulse is approximately 100-200 microseconds wide, so there is some room for error.
However, this gave an early indication of the timing between the ablation laser and the carrier gas valve.
In our efforts to find SrF molecules a wide range of pulse lengths, ablation laser delays and valve opening speeds have been tried. Since no SrF was detected, we cannot con- clude anything about the ’correct’ settings however.
2.2.3 Rotating rod and housing
Our rod of SrF2 is mounted on a threaded cylinder, which passes through the screw plate. A schematic of this system is given in figure 2.2, which has been taken from .
The rod itself is made by pressing SrF2 (mixed with a little boron for added stability) onto a metal rod. Now turning the cylinder will cause the rod to move up and down in addition to rotating. The connection to the stepping motor used to turn the rod is formed by a cylinder that had it’s centre bored out and with a guiding groove on the side. The cylinder holding the SrF2 fits in this hole and is rotated by a pin that fits into the guiding slit. As the large cylinder is rotated the guiding pin rotates the inner cylinder, which then pulls itself upwards through the screw plate. Downward motion is easily achieved by reversing the motor. The rotating rod system is similar to the one used by the group of Smalley.
Initially the threaded cylinder would get stuck inside the screw plate. Initially we thought the problem was that the threaded cylinder was too small. A slight misalign- ment might then cause the cylinder to shift a little and actually clamp itself inside the screw plate rather than rotate through it. A new, better fitting cylinder failed to solve
Chapter 2. A new source chamber 10
the problem, however. We then turned our attention to the general alignment of the screw plate and the valve. All parts are in fact connected to the screw plate, including the valve. However, the valve location is also fixed by the gas tube entering the vac- uum chamber. Since the alignment isn’t optimal, the valve pulled on the rod housing, causing friction. Having a loose cylinder actually proved helpful in mitigating this mis- alignment. Ultimately, installing a flexible gas tube solved the problem by having the valve rest solely on the screw plate.
More problems arose after the rod movement was fixed. The design of the housing is such that the sample blocks a large portion of the channel transporting the carrier gas.
Also the hole cut to let the ablation laser in is rather large. We believe that a lot of ab- lated material is swept with the deflected with the carrier gas into this hole. This might explain our inability to measure molecules. In an effort to solve this problem plans have been made for a new housing. This new version will have the ablation zone at the end of the housing, so molecule creation will take place in the middle of the supersonically expanding carrier gas. A downside to this design might be that, since there no longer is a channel after ablation, less translational cooling may occur.
2.2.4 Measurement chamber
After being entrained in the supersonically expanding carrier gas, the SrF molecules should arrive at the measurement chamber. Mounted on top of this chamber is a PMT.
Different band pass filters can be installed to reduce the background photon count. Usu- ally a 663 nm bandpass filter is used, since this corresponds to the transition by which 98 percent of the electrons decay. Alternatively a 685 nm bandpass filter may be used.
This corresponds to second most used transition. The advantage to this filter is that it does not allow the detection laser light through, reducing the background photon count significantly. The downside is obviously that only a small percentage of electrons actu- ally decay via this transition. 
Since the detection laser beam is reflected of a mirror on the end of the measurement chamber and hence passes through the detection zone twice, its alignment is critical. At first, the detection laser beam passed through an aperture mounted directly inside the vacuum. This aperture reduced the stray light count rate. However, by forcing us to pass the aperture and still be orthogonal to the end mirror it reduced our options when setting up the detection laser. Because of a slight misalignment of the parts holding the mirrors in place this became a large problem. In the end the front aperture was removed.
As expected the background count rate increased, but we believe the alignment space we
Chapter 2. A new source chamber 11
Figure 2.2: Schematic overview of a system, similar to ours, implemented to rotate a sample of SrF2 inside the vacuum. Our version is mounted directly on the flange
without positioning blocks and its rod is directed upwards.
gained be the removal outweighs this. If the background proves to be a limiting factor it might be wise to install an aperture with a rather large diameter to block some of it while still easily allowing the detection laser through.
Chapter 2. A new source chamber 12
Trot=30 K Trot=10 K
Trot=20 K Trot=5 K Trot=2 K N=0
transition intensity (arb.units)
−1 0 1 2 3 4 5
transition frequency (cm-1)
15073.5 15074.0 15074.5 15075.0 15075.5
Figure 2.3: Simulation of the intensity of absorption in SrF. The dependence on temperature is clearly visible. The lower laying states are more populated at lower temperatures. By comparing this simulation with actual measurement something may
be said of the temperature of the sample.
2.3 Future improvements
2.3.1 Dye laser
In order to optimize the new source chamber a dye laser is needed to scan over the frequency range the states of SrF occupy and make a plot of the intensity. As shown in the simulation shown in figure2.3, the intensity of the different states are dependant on temperature. By comparing the results with the simulation the temperature of the sample may be determined. The only available dye laser had not lased for over five years. As part of my project I worked on getting it back to lase. The details of doing this are included in appendix A.
As can be concluded from the inclusion of this part in the ’Future improvements’ section, we were unable to use the dye laser for measuring, even though we did manage to get it to lase. This is due to the fact that no molecules were detected in the new source chamber, rendering the temperature measuring capabilities of the dye laser useless.
While trying to get the dye laser power to the expected values and installing the required optical components lasing was lost. Getting it back to lase should not be very hard
Chapter 2. A new source chamber 13
however. After it lases again the last optical components should be installed and the output power should be maximized. Also a laser locking system still should be installed before it can be used for measurements. The final location of this entire setup is still uncertain too. The distance between the source chamber and the laser setup is almost 50 meters at the moment and moving it or using an optical fiber are both mediocre options at best. Getting the dye laser fully operational will take at least two weeks of work, but might easily be delayed to a month.
2.3.2 Detection laser frequency
The exact alignment of the detection laser in order to excite the SrF molecules has not been obtained yet. Although the detection laser is locked to an iodine state which should have the same frequency, no molecules have been measured as of late. Since the same detection laser is also used for the main deceleration experiment being able to reliably excite the SrF is very important. Finding molecules in either experiment fixes the frequency for the other. Since we are so close already it should be a matter of weeks to find it.
Measurements and results
In this chapter I will present the data gathered when working on the new source cham- ber. Unfortunately no clearly positive results were obtained. However, new methods of detecting SrF molecules or optimizing the source chamber may be extracted from these results.
Measurements are done using an MCA visualization program. This displays the counts of the PMT in a time versus time graph. By having one scale being a lot larger than the other many shots can be analysed at the same time.
In figure3.1 I display the most interesting and promising result we had, the figure is a screen shot of the program we use to display our data and is thus of rather low quality.
From the simple comparison between the speed of the carrier gas (500 m/s) and the dis- tance to the detection zone (30cm) we expect the molecules arrive after approximately 500µs. The highest count rates of our data arrive after 1000µs. A second feature is the regular ’beating’ visible in the count rate. These beats are spaced 30 seconds apart.
To investigate these phenomena we first tried to block the detection laser. If the signal then disappears, it originates from the desired molecules. Shutting down the detection laser had no effect. So if the large signal is not coming from SrF molecules, then what is creating this light?
Chapter 3. Measurements and results 15
0 500 1000 1500 200
Figure 3.1: Photon counts (by color) at certain times after the trigger (horizontal axis). The trigger is sent at the same time as the ablation laser, the light of which can be easily seen at the same time as the trigger (t=0). The vertical axis is time in seconds. Large peaks are seen at t=1000 µs. The repeating strength increase is due to
the phase of the rod in it’s vertical motion.
We suspect that clusters of SrF are being created which are so hot that they glow. The heavier clusters might also explain the longer travel time measured. Reducing the ab- lation laser power did not seem to reduce the problem. At a certain power all signal disappeared, indicating that the ablation process is not happening. Changing the focus of the ablation laser light so that the not all power is concentrated to one spot might help in the future, although none of our attempts worked.
As mentioned before, the rod housing might not be optimal either. The large entrance hole for the ablation laser has to be filled by the carrier gas before it escapes towards the detection zone. In addition the channel the gas has to move through improves not only the rotational temperature, but also stimulates cluster growth.
Another possibility is that the valve settings are off. When too little gas is released from the valve, i.e. it is not open long enough, a lot of photons reached the detection PMT. This resulted in a high peak in the count rate around 100 microseconds after the ablation laser fired. The most likely explanation seems to be that a lack of carrier gas force will keep the SrF trapped in the ablation zone. The trapped plasma is hot enough to emit its own light, which we then detect after it finally reaches the detection zone.
Chapter 3. Measurements and results 16
Opening the valve for longer times immediately got rid of the second peak.
We checked the delay settings between the ablation and the valve opening too. First of the delay was set to a large number (700 microseconds or so), causing the valve to open way before or after the ablation took place. As expected, the signal dropped. Then the delay was slowly changed to the value taken from . At the maximum signal strength we assume the timings to be optimized, although again no molecules were detected.
Whenever modifications are made to the position of the rod housing (or it is replaced altogether) I recommend that this procedure is repeated.
Conclusions and remarks
Although no definitive results were obtained, it was very usefull to go through the experiments we did. It helps to get to understand the way experiments and theory are linked, which is very important. Also a lot of small yet necessary improvements were made to the new source chamber setup, like installing a flexible gas tube to get rid of the alignment problems. The thought and work put in the housing of the SrF2 pill I believe will especially prove to be of value.
It was unfortunate however that most if not all planned tasks proved to be much harder than expected, or at least took much longer than expected. This was notably the case in getting the dye laser operational, This of course is just the way it goes but did remove the opportunity to work on the data analysis part of the project.
I expect the new source chamber to be operational fairly soon, especially if molecules are detected in the main setup again and the detection laser frequency can be checked.
Hopefully it will be installed in the main setup soon!
Aligning a dye laser
The dye laser of interest to our experiment is a 20 year old Model 380D Frequency stabilized Ring Dye Laser. Since it was inactive for over a decade the alignment of the mirrors was way of. Following the manual literally we managed to get it to lase again, although installing the optical components necessary for stabilization and power amplification did not work out yet.
The basics of this particular laser are as follows. A dye (a liquid) rather than a solid is used as a lasing medium. Light makes multiple trips through the laser following a ring shaped path (hence the name) which is illustrated in figureA.1. This means that lasing may occur in both directions. One of the optical components, called the unidirectional device, is specially designed to prevent lasing in one direction. This should make getting the initial lasing easier.
Aligning this laser is so difficult because all mirrors effect the light beam in two directions, and aligning it in one way may easily misalign it in another. We found the spot shapes mentioned in the manual to be very important, even more so than the output power.
In fact, at one stage we maximized the output power (mostly ignoring spot shapes) on what later appeared to be a reflection rather than the light beam. The mirror that has the most influence on the spot shapes seemed to be the Pump mirror, which directs the incoming pump laser light onto the dye. Spot shapes are most easily measured at M3 and M4. When lasing, three bright spots coming from the birefringent filter were visible on the lower set screw used to manipulate M2.
Appendix A. Aligning a dye laser 19
Figure A.1: Optical schematic overview of the dye laser, taken from its manual.
The part within the blocked line is meant for stabilization and is not installed yet.
Optical components missing are: Astigmatism compensator, Dual galvoplates and the thin etalon.
 C. J. Hawthorn, K. P. Weber, and R. E. Scholten. Littrow configuration tunable external cavity diode laser with fixed direction output beam. Review of Scien- tific Instruments, 72(12):4477–4479, December 2001. URLhttp://link.aip.org/
 J.E. van den Berg, S.C. Mathavan, C. Meinema, J. Nauta, T.H. Nijbroek, K. Jung- mann, H.L. Bethlem, and S. Hoekstra. Traveling-wave deceleration of srf molecules.
Journal of Molecular Spectroscopy, 300(0):22 – 25, 2014. ISSN 0022-2852. doi:
http://dx.doi.org/10.1016/j.jms.2014.02.004. URL http://www.sciencedirect.
com/science/article/pii/S0022285214000484. Spectroscopic Tests of Funda- mental Physics.
 M. A. Duncan. Laser vaporization cluster sources. Review of Scientific Instru- ments, 83:041101/1–19, April 2012. URL http://scitation.aip.org/content/
 M Barr, K M O’Donnell, A Fahy, W Allison, and P C Dastoor. A desktop supersonic free-jet beam source for a scanning helium microscope (shem). Measurement Science and Technology, 23(10):105901, 2012. URLhttp://stacks.iop.org/0957-0233/
 Y. Sebastiaan, T. v.d. Meerakker, H. L. Bethlem, and G. Meijer. Taming molecular beams. Nature Physics, 4:595–602, 2008. doi: 10.1038/nphys1031. URL http:
 C. J. Foot. Atomic Physics. Oxford universtity press, 2005. ISBN 978 0 19 850695 9.
 H. L. Bethlem, M. Kajita, B. Sartakov, G. Meijer, and W. Ubachs. Prospects for precision measurements on ammonia molecules in a fountain. The European Physical Journal Special Topics, 163(1):55–69, 2008. ISSN 1951-6355. doi: 10.1140/epjst/
 Ludwig Scharfenberg, Henrik Haak, Gerard Meijer, and Sebastiaan Y. T. van de Meerakker. Operation of a stark decelerator with optimum acceptance. Phys.
Rev. A, 79:023410, Feb 2009. doi: 10.1103/PhysRevA.79.023410. URL http:
 Paul Jansen, Li-Hong Xu, Isabelle Kleiner, Wim Ubachs, and Hendrick L.
Bethlem. Methanol as a sensitive probe for spatial and temporal variations of the proton-to-electron mass ratio. Phys. Rev. Lett., 106:100801, Mar 2011.
doi: 10.1103/PhysRevLett.106.100801. URL http://link.aps.org/doi/10.
 Cynthia E. Heiner, Hendrick L. Bethlem, and Gerard Meijer. A synchrotron for neutral molecules. Chemical Physics Letters, 473(13):1 – 9, 2009. ISSN 0009- 2614. doi: http://dx.doi.org/10.1016/j.cplett.2009.02.069. URL http://www.
 E. Narevicius, A. Libson, C. Parthey, I. Chavez, Narevicius J., U. Even, and M. Raizen. Stopping supersonic oxygen with a series of pulsed electromagnetic coils: A molecular coilgun. 2008.
 A. Osterwalder, Samuel A. Meek, G. Hammer, H. Haak, and G. Meijer. Deceleration of neutral molecules in macroscopic traveling traps. Phys. Rev. A, 81:051401, May 2010. doi: 10.1103/PhysRevA.81.051401. URL http://link.aps.org/doi/10.
 M R Tarbutt, J J Hudson, B E Sauer, E A Hinds, V A Ryzhov, V L Ryabov, and V F Ezhov. A jet beam source of cold ybf radicals. Journal of Physics B: Atomic, Molecular and Optical Physics, 35(24):5013, 2002. URLhttp://stacks.iop.org/
 M. F´arn´ık. Molecular beams. 2010. URLhttp://www.jh-inst.cas.cz/~farnik/