
\chapter{Design and construction of a new ion source}

\newcommand{\otwoplus}{O$_{2}^{+}$}

\section{Introduction}

This Chapter discusses the design and realization of a new ion source,~created to study new species and to probe other predissociated states than those found using conventional electron impact ion sources.

 A steady-state continuous cooled molecular beam is ionised, skimmed and focussed into an ion beam. A source was designed for this purpose and for ease of attachment to the existing vacuum apparatus, with the minimum disruption or modification.
 The experimental principles are similar to those outlined in Chapter 2: a collimated ion beam is formed, parent species mass selected and fragmented with collinear laser radiation. Fragments are separated from the parent ions using a second electromagnet and detected at an electron multiplier. \\
 This Chapter discusses results obtained with this new apparatus and outlines possible experimental studies.



\section{Jet cooling of ion beams}


\subsection{Overview}
Cooled molecular beams are created upon adiabatic expansion of high pressure gas through a `nozzle' into a low pressure background.
Upon supersonic expansion, random motion of a static gas is converted into directed mass flow in the expanding jet. If the system is isentropic during expansion, then the temperature of the gas is reduced.
 A `skimmer' is  positioned in front of the expansion to extract the centre of the beam (from the `zone of silence') and avoid shock waves which inevitably occur during supersonic expansion. 



\subsection{Background}
A given gas behind a nozzle has a negligible velocity, called the {\it stagnation state} with a pressure and temperature, P$_{0}$ and T$_{0}$ respectively. The region into which the gas will expand has a pressure of P$_{\rm b}$. The difference between the nozzle and background pressure causes gas to expand from the nozzle region. The molecules may reach speeds of greater than the speed of sound in the region directly in front of the nozzle. At a given distance from the nozzle (the Mach Disk), the gas will reach sub-sonic speeds. The distance (x$_{\rm M}$) of the Mach disk from the nozzle aperture (of diameter D) can be approximated to Equation~\ref{equation:machdisk}. \\
To minimise collisions in the beam to avoid re-heating, the skimmer has to lie inside this disk. 
\begin{equation}
{\rm \left(\frac{x_{M}}{D}\right) = 0.67 \left(\frac{P_0}{P_b}\right)^\frac{1}{2}}
\label{equation:machdisk}
\end{equation}


The temperature of the gas upon expansion is given by  Equation ~\ref{equation:jettemp}

\begin{equation}
{\rm \frac{T}{T_0}=\left(1+\left(\frac{\gamma-1}{2}\right)M^{2}\right)^{-1}}
\label{equation:jettemp}
\end{equation}

 where M is the Mach number (Equation~\ref{equation:machno}) and $\gamma$ is the ratio of the Heat Capacities.

\begin{equation}
{\rm M= \frac{speed~of~molecules}{local~speed~of~sound}}
\label{equation:machno}
\end{equation}

 For clustering to occur, and cooling observed, it is desirable to maximise the stagnation (nozzle) pressure and minimise the nozzle diameter. However, decreasing the nozzle diameter reduces gas throughput and hence beam strength. The pumping speed of the vacuum apparatus, the nozzle diameter, the position of the Mach disk from the source, the stagnation pressure and the desired temperature are all considered before such an experiment is designed.





\subsection{Related Literature}

Molecular beams are applied in a wide variety of Chemical and Physical problems ranging  from the smallest diatomic to large organic molecules. For a review see \cite{scoles1} \cite{scoles2}.\\
However, the  number of  studies using cooled molecular ions are considerably  less than those for the corresponding neutral molecules, due to low beam strength attainable for molecular ions.\\
 This study uses the technique of ionisation directly after expansion to create molecular ions. It has taken influences in it's apparatus design and the experimental techniques used to create molecular ions from several earlier studies, including:\\ 


  Larsson et al. ~\cite{larss1} have used a supersonic jet source in conjunction with a High Frequency Deflection technique \cite{hfd} to make lifetime measurements for various molecular ions. A molecular beam was crossed by an electron beam, skimmed and emission spectra observed for N$_{2}^{+}$, CO$^{+}$ and N$_{2}$. Temperatures of around 15 K were achievable, although this was less than the predicted temperature. This though to be due to ambient heating of the beam during the ionisation process. The jet cooled N$_{2}^{+}$  ${\rm B^{2}\Sigma~-~X^{2}\Sigma}$ (0-0) band was shown to be significantly less congested than the corresponding room temperature spectrum. A redistribution of state population occurred, where low lying states became more highly populated (and hence their transitions had a greater intensity). \\

Carrington, Shaw and Taylor \cite{carringtonjet} recorded microwave spectra of Ar$_{2}^{+}$ and Ne$_{2}^{+}$. The ions were created by crossing a neutral free jet from a liquid nitrogen cooled 20 $\mu$ nozzle with an electron beam. Stagnation pressures of up to 3.5 bar were used, giving pressures of 5 $\times 10^{-5}$  mbar in the source region. Ions were accelerated out of the source using potentials of up to 10 kV in a direction perpendicular to the molecular beam, then mass selected, and enter a waveguide whereupon transitions are pumped using microwave radiation. Daughter ions are separated from the parent beam using an electrostatic analyser and are detected at a Faraday cup or an electron multiplier. \\

Bae et al. \cite{bae} demonstrated a cluster ion source capable of producing collimated, intense beams of positively or negatively charged clusters. Cluster ions were created using an electron gun focussed directly to the aperture of the pulsed nozzle. The ions were then skimmed and ion optics used to focus the ions, followed by mass selection. Laser radiation from a  Nd-YAG system placed anti-parallel to the beam direction created photofragments which were selected using an Electrostatic Analyser and detected on a Multichannel Plate. Through the study of the mass spectra it was shown possible to create Nitrogen dimer ion (N$_{2}^{+})_{\rm n}$ with n of up to 14. \\

Coe et al. \cite{coe} used a new technique for measuring molecular ions with sub-Doppler resolution. They used direct absorption techniques coupled with fast ion beams, to record a spectrum of HF$^{+}$. 3.5 torr of HF was expanded from a water cooled 0.5mm diameter nozzle, and directed towards an emitting filament. A current of 1$\times10^{-5}$ A was attained at 3 kV. This is (at least) an order of magnitude larger than normal ion beam experiments. The absorption pathlength of the laser system was increased using two 98 \% reflecting mirrors. Transitions were Doppler tuned into resonance with an I.R. laser by varying the acceleration voltage. Decreases in laser power transmitted through the ion beam occurred where a transition was located. A minimum linewidth of 40 MHz was found for the source and  the resolution proved  capable of  resolving hyperfine splittings in the X$^{2}\Pi$ ~state of HF$^{+}$.\\

Okumura, Yeh and Lee recorded  vibrational spectra for ${\rm H_{3}^{+}(H_{2})_{n}}$. Molecular ions were created using a Corona discharge source in which the stagnated gas was ionised before expansion using a corona discharge tip (a nickel plated sewing needle). It was ensured that the hydrogen gas was extremely pure, and all water vapour eliminated. The 75 $\mu$ platinum nozzle was cooled by liquid nitrogen or Freon- this decreases the temperature of gases leaving the nozzle, improving clustering and decreasing collisions.  The ion beam was skimmed 7mm from the nozzle, and ions focussed and accelerated after skimming. Cluster ions of up to H$_{15}^{+}$ were found to be created in the source.\\
necessary

Bieske and Maier have recorded spectra of a wide range of molecular ions using free jet sources.  An example of their work can be found in the study of the B$\leftarrow$X transition of ${\rm N_2^{+}..He}$ \cite{biesken2he}. A water cooled pulsed nozzle with a stagnation pressure of 3 bar (1:100 Nitrogen:Neon) was used as a free jet source. Ionisation takes place inside a shielding cage using a double filament arrangement (constructed from spark plugs). Ions are skimmed and ion optics focus the beam into a quadrupole mass spectrometer. A dye laser is scanned and the daughter ions are collected using a channeltron detector.  The entire assembly of the ion source (skimmer, filament and shielding cage are sprayed in graphite to reduce the accumulation of surface charges and stray electric fields which accelerate and warm the ions. This warming had a visible effect on the spectrum, and re-coating was  every few days. The photofragment spectra corresponded to transitions in the N$_{2}^{+}$ chromophore suggesting the molecule behaves as a free internal rotor.\\


\section{Experimental}

\subsection{Vacuum Apparatus}
The apparatus can be broken down into three specific areas: Ion source, laser and detection systems. The laser and detection apparatus are those outlined in Chapter 2.\\
The jet source consists of two differentially pumped chambers, one chamber housing the nozzle and skimmer and the second containing ion optics and a flight region. The source and flight chambers are pumped by Edwards EO6K  and Balzers BG540 diffusion pumps respectively, which are backed by an Edwards EM18 rotary pump. The source chamber is separated from the diffusion pump by an Edwards butterfly valve, allowing the source to be removed without the need for the diffusion pump to be cold. The flight chamber has an electric/gas baffle, controlled by in-house electronics, a nitrogen cylinder providing the constant gas supply to the valve.\\
A gas line is connected to the flight chamber to `roughen' both chambers using the rotary pump. This line is sealed when not in use using a VG isolation gate valve. Each diffusion pump can be individually sealed from the backing line using Speedivac valves.\\
Pressure in the backing lines and in the jet region are measured by a VG Pirani gauge (VGPV1) and a ZWW16 ion gauge respectively.  Both gauges are controlled and displayed using a VG IGPA3 ion gauge control unit. A second ion gauge (Kratos Analytical) is mounted in the flight region, the pressure reading displayed on a Kratos Vacuum Controller. \\
The diffusion pumps, ion gauge controllers, rotary pump and electric/gas baffle  are all powered through an in-house device, which cuts power to the diffusion pumps in the event of a power cut or loss of vacuum in either chamber.\\



\begin{figure}[!ht]
\centering
\resizebox{5in}{!}{\includegraphics{figures/jetbeamcon.eps}}
\caption{Construction of the premixture apparatus.}
\label{figure:hundb}
\end{figure}





\subsection{Ion source}

The ion source is mounted in a stainless steel chamber above the Edwards E06K diffusion pump. The jet nozzle is mounted at the end of stainless steel tube, which in turn is mounted on bellows to allow axial translation of the jet assembly. Horizontal and vertical translation of the nozzle can be obtained by the adjustment of six translational screws.\\
The gas line passes through an O-ring seal and an electrically isolated Teflon Swagelock seal and enters the nozzle region. A pinhole (Ealing) (diameter 12.5-75 $\mu$) is mounted inside a threaded holder, screwed to end of the gas line. An O-ring seal prevents gas from escaping at the gas line/nozzle holder interface.\\
A Micromass MM6 filament and trap are mounted on either side of the nozzle, normal orientation being the filament mounted vertically above the nozzle, the trap diametrically opposite.  A copper cage surrounds the ion source to prevent penetration of stray fields into this region. \\
The ion source can be translated with respect to the nozzle so that the distance between filament and nozzle can be adjusted. Typical filament/nozzle horizontal separations are between 2mm and 10mm. \\
The filament is regulated in Electron Impact Ionisation (e.i.i.) mode (i.e. the electron current arriving at the trap was held constant by varying the current through the filament (See Chapter 2)). It was found that the jet source, could maintain emission currents (trap current) of 1000 $\mu$A for several weeks. The conventional ion source can obtain currents of this order for several days. This is due to the `open design' of the nozzle: the flowing gas is ionised in an open space, as opposed to the normal e.i.i source (which confines gas in a source block). The electron beam passes unhindered through to the trap in the jet source, whereas in a standard e.i.i. source it has to pass through an entrance hole and an exit slit in a source block. Filaments are usually destroyed when too large a filament current is needed to keep the emission current constant. In a jet source, the filament is under less strain due to the lower current needed to obtain the same emission current.\\
This open design of ion source  allows filament changes and source cleaning to be carried out quickly and simply due to the ease of access.\\
Magnets mounted above the filament and below the trap cause the electrons to spiral, increasing their flight path as occurs in a conventional ion source.\\ 
The source and copper cage, filament and trap are connected to a high voltage programmed power supply (VG M18A) using Teflon coated copper wires connected via electrical feedthroughs. Trap and filament currents are read/regulated using an emission control and source control VG M93 and M71 respectively. The emission control unit has a small modification which allows the trap current (normally held at +50V), to be varied between 24 and 55V.\\
A high vacuum viewport (68mm diameter) is mounted at the side of the chamber, centred on the nozzle/skimmer region, allows visual nozzle adjustments to be made without danger of touching the skimmer.\\ 

\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(4,7.8)(0.5,0)
\put(0,0){\special{psfile=figures/jetsource.eps}}
\end{picture}
\caption{Nozzle design}
\label{figure:jetsource}
\end{center}
\end{figure}






\subsection{Skimmer and ion optics}

A skimmer (Beam Dynamics nickel type 1, orifice diameter 0.4-0.8 mm) is mounted on a partition wall of the source chamber. The mounting design is  such that the skimmer can be mounted and removed quickly and easily, without damage to the thin nickel walls: it is placed into a recess (2.3 cm  with a 1 cm hole through the center) of a stainless plate, over which  a mounting ring is then  placed  and held securely by three stainless steel screws. A Teflon coated copper wire is screwed onto the plate, passing through the feedthroughs, and  is connected to the high voltage supply (see Figure ~\ref{figure:skimmer}). The plate is placed inside the vacuum chamber and screwed into place using six mounting bolts. Teflon isolates the plate from the chamber partition, to allow application of voltages to the skimmer. Typically voltages for the skimmer can vary up to $\pm 50$ V of the source voltage, controlled and read by a VG M71 source control unit.\\

Ion optics are mounted onto the back of the partition wall,positioned  axially to the nozzle and skimmer. These consist of several earthed plates and two plates in the Y plane and two in the Z plane. As the plates are earthed individually,  a  T.I.M signal to be detected after skimming the ion beam (if desired). The lens stack design is similar to the stack at the first magnetic sector, however, it is possible to vary the voltages over a greater range than the conventional lens stacks using a separate ion optics control box (in-house device). Voltage range for these lenses are 0-10$\%$ of the acceleration voltage and are applied to the ion optic stack via feedthrough port mounted vertically in the flight chamber. Teflon coated copper wires connect the lens stack plates to this feedthrough plate.\\



\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.5)(0.8,0)
\put(0,0){\special{psfile=figures/skimmer.eps}}
\end{picture}
\caption{Illustration of the nickel skimmer mounting and the lens stack construction inside the flight region chamber.}
\label{figure:skimmer}
\end{center}
\end{figure}



\subsubsection{Alignment of the nozzle to the skimmer}

For the creation of cold molecular ions, it is essential that the nozzle and skimmer are aligned axially. To achieve such an alignment, two techniques were used: a coarse alignment procedure undertaken  with the source and flight region chambers at atmospheric pressure and separated from each other.  Fine alignment was achieved with the apparatus under vacuum.\\
 
A Helium-Neon laser was mounted on a 30 $\times$30 cm optical bench consisting of two steering mirrors facing in perpendicular directions. The optics were adjusted to steer laser radiation through the gas line and exit from the nozzle.  With the two chambers separated, the nozzle was aligned to the skimmer using the nozzle translational controls. When the laser spot could be observed passing cleanly through the skimmer, the nozzle and skimmer are aligned. The chambers were then reconnected and pumped down to vacuum. \\
Gas was then passed through the nozzle and small translations made using the translational screws. Flight region pressure was then monitored, if the nozzle/skimmer alignment became better, the flight region pressure increases. If the alignment was worse, the pressure fell. In this manner, the nozzle was correctly aligned to the skimmer, until a maximum was reached for all translations of the nozzle.

\subsection{Mixtures of Gases}

A mixture of precursor and carrier gases were used to create colder molecular ions than could be acheived without a buffer.  For jet sources, normal carrier gases are Argon or Helium. An apparatus was constructed which enabled two gases to be mixed in a third cylinder, which could be fully evacuated (via an independent rotary pump), to allow the use of spontaneously flammable gases. A 0-5 bar pressure gauge monitored the filling and nozzle pressures.  See Figure \ref{figure:jetbeamcon}. Gas flow into the mixing cylinder could be carefully controlled through the use of a VGMD7R leak valve.\\
Typical carrier gas to precursor gas ratios were 5:1, 10:1 and 20:1, and a typical nozzle (stagnation) pressure was 3 bar. 




\section{Review of \otwoplus}



\subsection{Introduction}
This section discusses a series of experiments which recorded a spectrum of  O$_{2}^{+}$~ using the new ion source. There are several advantages of using \otwoplus~  to characterise the apparatus:  large ion currents are achievable, relatively low cost and high purity available of the precursor gas O$_{2}$ and the ion has a well understood photofragment spectrum. The aim of this series of experiments was to prove that the apparatus could be seriously considered as an alternative to the e.i.i. source and also offer advantages in terms of rotational cooling of any spectra.  


\subsection{States and Spectroscopy of \otwoplus}

The spectroscopy of \otwoplus is well understood and has been studied at high resolution using laser photofragment apparatus several times \cite{carringtonsarre} \cite{cosby2} \cite{cosby4} \cite{mosreview} \cite{bae}, \cite{tad}. It is therefore considered to be a `benchmark molecule', from which experimental parameters can be deduced. 
The electronic transition of interest for these experiments is the ~${\rm b^{4}\Sigma_{g}^{-}\leftarrow a^{4}\Pi_{u}}$. 
All electronic transitions thus far observed in photofragment spectroscopy correlate to the ${\rm O(^{3}P_{2}) + O^{+}({^4}S^{0})}$~ system.
Rotational levels of the lower electronic state are split into four fine structure components through the spin-orbit interaction, these are  $^{4}\Pi_{\frac{5}{2},\frac{3}{2},\frac{1}{2},-\frac{1}{2}}$, and are labelled F$_{1}$, F$_{2}$, F$_{3}$ and F$_{4}$ respectively.
 Each rotational component of the ground state is Lambda doubled, with only negative parity levels present in the homonuclear ${\rm ^{16}O_{2}^{+}}$.
Each component is  separated by approximately 50 \waveno~ from the others (the exact value depending upon the vibrational and rotational states involved).\\

The upper (b$^{4}\Sigma_{g}^{-}$) electronic state has four spin components,  labelled: \\~${\rm F_{1}' ~(J'=N'+\frac{3}{2})}$,~ ${\rm F_{2}' ~(J'=N'+\frac{1}{2})}$,~ ${\rm F_{3}' ~(J'=N'-\frac{1}{2})}$~and~ ${\rm F_{4}' ~(J'=N'-\frac{3}{2})}$.
Due to zero spin of oxygen nuclei and nuclear spin statistics, only odd valued N' are present.
 Transitions between the two electronic states are labelled using the following nomenclature: N'${\rm \Delta J_{F{n}'F{m}''}(J'')}$\\
Where $\Delta J$ is P, Q or R for -1, 0, and +1 respectively. The majority of lines observed in the visible region predissociation spectra of \otwoplus~ are due to to transitions from v$^{\prime\prime}$=3,4,5 to v$^{\prime}$=3,4,5  vibrational states.





\subsection{Experimental Setup}



The chamber was connected to the ion beam apparatus using a CST gate valve. The first magnetic sector was rotated through 90$^{\circ}$ and current polarity reversed. Three sets of ion optics were used: immediately after the skimmer, after the first magnetic sector and before the second magnetic sector. Ion currents could be monitored at these positions using T.I.M. plates. Voltages for the ion optics mounted at the magnetic sectors are controlled by the high voltage supply for the conventional e.i.i. source. The supply was set to 2 kV, giving voltages of $\approx$200 V to the ion optics.  However, this value could be adjusted to optimise ion beam strength. In practice, for a jet source voltage of 2 kV, the e.i.i. high voltage supply was set to 1.8 kV (for O$_{2}^{+}$). Individual plate voltages were set using the ion optics control unit. The precursor gas enters the jet source through a stainless steel gas line, regulated by a needle valve. The precursor gas in the following experiments was oxygen or oxygen mixed with helium or argon.\\

Nozzle pressures of 2-3 bar were used, resulting in a pressure of 10$^{-6}$ and 10$^{-7}$ torr in the source and flight regions respectively. 
Electrical breakdown occurred in the source chamber at approximately 200 V. Experiments isolated this problem to the Swagelock fitting. However breakdown was avoided when gas flowed through the nozzle. Breakdown at 2-3 bar occurred at voltages $>$2.5 kV (for oxygen).\\

The emission current of the filament was set to 1000 $\mu$A for the majority of experiments, the jet source programmed power supply set to 2 kV.
Optimal ion current arriving at the first magnetic sector was achieved when a small negative voltage was applied to the skimmer (typically -12 V).
As mentioned previously, the trap potential could be varied between 55 and 24 volts. This voltage was optimised to maximise ion current. Laser radiation from the CR699-29 tunable dye laser entered the vacuum apparatus through the Brewster window mounted on the cross piece at the  first magnetic sector.  
Due to the photofragmentation process (O$_{2}^{+} ~\rightarrow$ O$^{+}$ + O), O$^{+}$ ions  could be directed towards an electron multiplier when the laser radiation was collinear with the ion beam.\\

After optimisation of ion optics, the laser was scanned and an \otwoplus~ spectrum recorded. The following parameters were used: 30 Mhz sampling rate, 40 s/10 GHz segment, lock-in time constant 1 s, 500 - 10000 $\mu$V sensitivity (depending on signal to noise, ion current and laser power) and 1000 $\mu$A emission current. \\
The laser was scanned in regions between 17200 and 17300 \waveno~ to find the strongest lines of the (4-4) band of the ${\rm b^{4}\Sigma_{g}^{-} \leftarrow a^{4}\Pi_{u}}$ transition.
These recorded regions have line positions recorded to a precision greater than 0.003 \waveno, and a large density of lines \cite{cosby}. A recording taken using a conventional e.i.i. source is shown in Figure \ref{figure:o2normsource}. This spectrum has a dominant transition at 17261.3 \waveno~ made up of three components, a shoulder at 17261.338\waveno~ 9 P$_{31}$ (9.5) and a larger peak, at 17261.387\waveno~. This peak is a blend of two lines, the dominant 9 Q$_{21}$(9.5) and the much weaker 17 Q$_{11}$(18.5). This line has been deconvoluted into individual contributions using an energy analyser \cite{cosby} which samples different ranges of rotational levels above the dissociation limit. Several other transitions are found in this region, including an important transition arising from and arriving in the same spin states (the 17 P$_{21}$(18.5) transition) as the dominant 9 Q$_{21}$(9.5) transition. 

\newpage

\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.7)(2.8,0.8)
\put(0,0){\special{psfile=figures/o2s1901b.eps}}
\end{picture}
\caption{\otwoplus~ spectrum recorded in a conventional ion source. Labelled transitions taken from the study by Cosby et al. \cite{cosby}.}
\label{figure:o2normsource}
\end{center}
\end{figure}

\newpage
\section{Results}
\subsection{Jet rig results}

Initial results from the jet rig are comparable to those found from a conventional ion source. A spectrum of \otwoplus~ was recorded with a satisfactory signal to noise ratio, with only a small decrease in ion current of one order of magnitude (jet source achieved 1$\times10^{-8}$ A at the first magnetic sector lens stack T.I.M). The distance between the source and magnetic sector is significantly larger ($\approx$ 65 cm) compared to the conventional ion source ($\approx$ 15 cm), and hence a decrease in ion current was expected due to collisional and unimolecular decomposition of the molecular ions in the beam. 
Extra pumping capacity in the jet source compensates for flight region losses, as higher source pressures can be used. The pressure in the source chamber can reach 1$\times10^{-4}$ torr without adverse effects to the pump which is two orders of magnitude greater than that used in a conventional ion source. An initial spectrum of \otwoplus~ is shown in Figure~\ref{figure:o2s1612f}. Linewidths in this region are approximately  500 MHz, identical to those found in a conventional e.i.i. source (500 MHz) with an acceleration voltage of 2 kV. Both these values are higher than those reported by  Cosby et al. where all transitions have widths between 275 and 450 MHz \cite{cosby}. This is probably due to the low acceleration potential, in comparison with that used by Cosby (3.6 kV).  The signal to noise ratio is poorer in comparison to the conventional source spectrum, probably due to the lower beam currents from the jet source.\\ Comparison  between  intensities of  9 Q$_{21}$(9.5) and 17 P$_{21}$(18.5) in Figures ~\ref{figure:o2normsource} and \ref{figure:o2s1612f} show that virtually no intensity redistribution occurs between the jet source and conventional source with a 25 $\mu$ nozzle and no carrier gas.  \\


\subsubsection{Estimation of temperature}

The temperature of the ion beam from can be estimated through comparison of spectroscopic lines of \otwoplus.
Three assumptions were made to quantify the jet source temperatures:
\begin{itemize}

\item{A conventional e.i.i. source creates (\otwoplus) molecular ions at a temperature of 500 K  \cite{carr4}.}
\item{Populations of rotational levels in the jet source can be modelled using a single Boltzmann distribution and temperature. This can be generally accepted as the case for molecular beam studies despite non-equilibrium conditions of a jet source.}

\end{itemize} 


 A redistribution of rotational populations results in an increase for low lying rotational states and a corresponding decrease in high lying states.
Assuming that the temperature of the conventional ion source and jet source can be modelled by a Boltzmann distribution,  it can be shown that the intensity of a given transition to N$^{\prime\prime}$ can be estimated from \cite{larss1}:

\begin{equation}
I_{N',N''}~\alpha~~ S_{N',N''} exp\left(\frac{-B^{\prime\prime}}{kT}(N''(N''+1)\right)
\end{equation}

Where ${\rm I_{N',N''}}$ is the intensity of a transition, ~${\rm S_{N',N''}}$ is the Honl-London factor for that transition, N$^{\prime\prime}$ is the total angular momentum excluding nuclear spin, T is the temperature and k is the Boltzmann constant.

A spectrum is recorded in the normal ion source and the intensity ratio of two lines in the same spin states taken.
The same spectrum is then recorded at a lower temperature T$_{red}$, and the ratio of the two lines again taken, the ratio {\em of the ratios} of the intensities are used to determine a temperature for the cooler spectrum, assuming that the temperature of an e.i.i. source is 500 K. As the ratio of the two Honl-London factors remaining constant with temperature, this assumption can be made.

Over 50 scans were recorded, using different source conditions and the following subsections outline these different series of experiments used to characterise and optimise the jet source.




\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.7)(2.8,0.8)
\put(0,0){\special{psfile=figures/o2s1612f.epsi}}
\end{picture}
\caption{\otwoplus~ spectrum using Jet source. Labelled transitions taken from study of Cosby et al. \cite{cosby}.}
\label{figure:o2s1612f}
\end{center}
\end{figure}



\subsubsection{Nozzle Diameters}

Spectra for \otwoplus~ were obtained using two  different nozzle diameters, made from  25$\mu$ and 12.5$\mu$ pinholes (Ealing). No discernible loss of ion current occurred with the smaller nozzle, as the stagnation pressure was increased to compensate for reduced throughput. See Figure~\ref{figure:plotnoz}.\\
 Analysis of the recorded spectra using several different mixes of precursor gases show that a significant change in the temperature occurs when the smaller nozzle is used. The intensity ratio between the 9 Q$_{21}$(9.5) and 17 P$_{21}$(18.5) transitions recorded using the conventional ion source and for the two different nozzle diameters are measured and temperatures deduced (see Table ~\ref{table:plotnoz}). A significant decrease in intensity of the 17 P$_{21}$(18.5) line relative to the  9 Q$_{21}$(9.5) line is found to occur with the smallest nozzle, corresponding to a significant decrease in the temperature of the ion beam. The temperature of the ion beam is found to be $\approx$ 125 K, a reduction of approximately 400 K from a conventional e.i.i. source. This clearly demonstrates that the ion source can create a significantly cooled molecular ion beam, with a reasonable ion current and without discernible loss to the signal to noise ratios and linewidth.




\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.7)(2.8,0.8)
\put(0,0){\special{psfile=figures/plotnoz.epsi}}
\end{picture}
\caption{Comparisons of \otwoplus~spectra with nozzle diameter. The lowest trace shows normal ion source spectrum. Middle trace shows the jet source spectrum with a 25$\mu$ nozzle. The upper trace shows the spectrum with a 12.5$\mu$ nozzle. Note shifted mode hop due to change in laser segment scan length.}
\label{figure:plotnoz}
\end{center}
\end{figure}




\begin{table}[!ht]
\centering
\begin{tabular}{|c|r|r|}
\hline
&&\\
{\bf Nozzle diameter}&{\bf Intensity ratio}&{\bf Temp (K)} \\
&&\\
\hline
\hline
N/A (Electron Impact source)&9.41&       500 (assumed)\\
&&\\
25$\mu$&17.51&       264\\
&&\\
12.5$\mu$& 75.06&       126\\
\hline
\end{tabular}
\caption{Comparison of the intensities of the  9 Q$_{21}$(9.5) and 17 P$_{21}$(18.5) transitions for different nozzle sizes. Pure \otwoplus~ is the precursor gas.}
\label{table:plotnoz}
\end{table}





\subsubsection{Gas mixtures}

To increase the cooling of the molecular beam from the jet source, the precursor gas was mixed with various ratios of Argon and Helium buffer gases. Results using Oxygen and Argon mixtures  on the spectrum of ~\otwoplus are shown in Figure~\ref{figure:armix}. A marked effect on the spectrum is observed, in that the  19 Q$_{43}$(17.5) and the 17 P$_{21}$ (18.5) are observed to decrease to the extent where they are virtually indiscernible from the background signal. A similar but greater effect occurs when Helium is used as the buffer gas, however interestingly the signal to noise ratios remain unaffected using this gas. The Helium seeded beam has a greater ion current than that using pure Oxygen alone, and yields signal to noise ratios comparable to those found in a normal ion source. See Table~\ref{table:bufgas} for comparison between the different precursor gas ratios.\\
 As expected, Helium is shown to be the better buffer gas than Argon, and has an added benefit of increased ion currents and better signal to noise ratio. Helium mixed with Oxygen in the ratio of 5:1, gave an estimated temperature of 85 K. This is a reasonable temperature, which hopefully could be decreased with an increase of pressure and/or a higher ratio mixture, but is still significantly higher than that calculated from equations Equation~\ref{equation:jettemp} and \cite{drmiller} which yield 7 K. This is probably due to heating from the environment due to the proximity of the filament.  




\begin{table}[!ht]
\centering
\begin{tabular}{|c|r|r|r|}
\hline
&&&\\
{\bf Precursor gas}&{\bf Intensity ratio}&{\bf Temp }&{\bf Relative beam}\\
{\bf (ratio)}&&{\bf (K)}&{\bf strength}\\ 
&&&\\
\hline
\hline
&&&\\
O$_{2}$ (N/A)& 75       &  125     &77.3   \\
&&&\\
Ar (5:1) O$_{2}$  &    83      & 121       &5.7    \\
&&&\\
Ar (10:1) O$_{2}$  &    87       & 119       &21.6  \\
&&&\\
He (5:1)  O$_{2}$  &     289       &   85      &100 \\
&&&\\
&&&\\
\hline
\end{tabular}

\caption{Comparison of the intensities of the  9 Q$_{21}$(9.5) and 17 P$_{21}$(18.5) transitions for different ratios of carrier gas mixtures/precursor mixtures. Normalised daughter ion current for each precursor is also shown.}
\label{table:bufgas}
\end{table}





\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.7)(2.8,0.8)
\put(0,0){\special{psfile=figures/armix.epsi}}
\end{picture}
\caption{Effect of a using Argon as a carrier gas. Notice reduction in intensity  of the n=19 and 17  transitions relative to the smaller N. With a 10:1 mixture, these transition are impossible to distinguish from the background signal.}
\label{figure:armix}
\end{center}
\end{figure}


\subsubsection{Shielding the source}

To examine the effect of the source region having an open plan design, a series of experiments were undertaken using a modified ion source. This involved electrically `shielding' the region directly in front of the jet expansion. The filament is floated at approximately -70 V, the trap at +55 V, making it possible that the ion source deflects the ions towards the trap upon creation. It was reasoned that this would apply an artificial electric field gradient, deflecting the ions upon their creation, possibly leading to lower beam strengths and warmer beams.\\

 A small U shaped stainless steel plate was constructed, with holes on either side to allow the electron beam to pass cleanly in front of the nozzle and arrive at the trap. The U shaped `shield' was welded onto the nozzle mounting plate and floated at the same voltage as the source.\\

 An ion beam was created and spectra recorded for comparison. These are shown in Figure~\ref{figure:shield}. A small heating effect occurs in these spectra, corresponding to an increase in the intensity of transitions for high N, relative to those for low N. Typical temperature increases are of the order of 50 K, possibly due to heating of the shield by the filament, the heat then conducted to the region close to the nozzle. Table ~\ref{table:shieldgas} lists the temperatures found for Pure O$_{2}$, Ar and O$_{2}$ and He and O$_{2}$. A greater ion current is recorded is found in all the experiments when the shield is used, whilst the signal to noise ratio is seen to remain approximately constant. This increase of ion current is up to 5 orders of magnitude greater than that found without the shield in position. \\

The reason for this apparent increase of ion current, and apparent increase in temperature is uncertain. It is suggested that more ions are indeed now arriving at the skimmer, and not deflected by the electric field gradient. However, a warming effect occurs due to the extra nozzle heating, which is a distinct disadvantage if a colder beam is required.\\





\begin{table}[!ht]
\centering
\begin{tabular}{|c|r|r|r|}
\hline
&&&\\
{\bf Precursor gas}&{\bf Intensity ratio}&{\bf Temp }&{\bf Relative beam}\\
{\bf (ratio)}&&{\bf (K)}&{\bf Strength}\\ 
&&&\\
\hline
\hline
&&&\\
O$_{2}$ (N/A)& 33      &  178 &    474 \\
&&&\\
Ar (5:1) O$_{2}$ &    53      & 144  & 31.4 \\
&&&\\
He (5:1) O$_{2}$ &     68       &   130 &471   \\
&&&\\
&&&\\
\hline
\end{tabular}

\caption{Comparison of the intensities of the  9 Q$_{21}$(9.5) and 17 P$_{21}$(18.5) transitions using the shield and different ratios of carrier gas mixtures/precursor mixtures. Relative beam strengths are normalised for direct comparison with Table~\ref{table:bufgas}}
\label{table:shieldgas}
\end{table}







\begin{figure}[!ht]
\begin{center}
\setlength{\unitlength}{1in}
\begin{picture}(5,7.7)(2.8,0.8)
\put(0,0){\special{psfile=figures/shield.epsi}}
\end{picture}
\caption{Effect of a using a shield. Note the increase in the intensity of the high N transitions with the shield, than with the shield not in place. This occurs for pure oxygen precursors as well as mixtures of gases.}
\label{figure:shield}
\end{center}
\end{figure}

\newpage

\subsubsection{Other experimental considerations}

Several other details and  experimental considerations about this construction of a  jet source ion beam, which are summarised below:

\begin{itemize}
\item{A negative skimmer voltage of 11.5 V with respect to the source is essential to create ion beams of a reasonable strength.}
\item{No heating effect occurs when the emission current of the filament is raised from 1000 to 2000 $\mu$A. No extra cooling is observed with the emission current at 500 $\mu$A.}
\item{Changes in the trap voltage had no observable heating effect, but greater currents achieved with the trap kept at low voltage ($\approx$25 V)}
\item{Nozzle-filament distances of approximately 6 mm yield the strongest ion beams.}
\item{The distance between the nozzle and skimmer is also critical. For a 12.5 $\mu$ nozzle, it is difficult to detect any photofragmentation of \otwoplus at 10 mm. 5 mm gave reasonable ion currents and sufficient cooling to record the spectra below.}\\

\end{itemize}

An increase in ion beam current and cooling could be obtained if the nozzle pressure was increased to greater pressures than 3 bar. The diffusion pump in the source region showed no signs of stalling when operating with Helium at 3 bar, (the backing line pressure was 0.05 torr,  well below the maximum allowed 0.5 torr). However, any effect of higher pressures on the pinhole surface needs to be considered, although little damage of the nozzle surface was observed during the experiments documented in the previous sections.\\
 A secondary method for increasing the beam current would be increasing the high voltage applied to the source region. However, this probably involves a re-design of the nozzle/gas line assembly, to prevent electrical breakdown which inevitably occurs above 2.5 kV.\\
 Finally, water (see Bieske et al. \cite{biesken2ne}) or liquid nitrogen  (see Carrington et al. \cite{carringtonjet}) cooling of the nozzle may yield further reduction in  the beam temperature with only minor modifications (for water at least) to the existing apparatus. 




\newpage

\section{Conclusions}

It has been demonstrated that a jet ion source can be created, capable of creating molecular ions with temperatures typically around  100 K.  It has been shown that a continuous jet can be formed, ionised immediately after supersonic expansion, skimmed and a photofragment spectrum recorded. The spectroscopic linewidths and signal to noise ratios are seen to be comparable to those obtained using a conventional e.i.i. source. \\

\subsection{Future studies}

Many alternatives for experiments exist which are suitable for the new ion source. A re-recording of many spectra for predissociated species which exist in the literature could be undertaken, in an attempt to probe low lying rotational states. This idea is  similar in concept to recent studies by Hechtfischer et al., \cite{ulrich1} where vibrationally cooled \chplus~ spectra were recorded using an ion storage ring as a source. This probed the lowest lying rotational and (due to the long lifetime in the ion storage ring) vibrational state using a photofragmentation  detection techniques similar to those outlined above. New predissociated transitions in the region between the spin-orbit limits were found. It is not expected for vibrational cooling to occur on the timescale of a jet cooling experiment, although it is hoped that previously unobserved rotational states of predissociated states could be found.\\ 
 One molecule possibly considered is \sihplus~. In previous photofragment studies, only the highest lying rotational states (J $\geq$ 16)  could be observed \cite{sarresih}, which have a complex spectrum.\\
This technique may prove invaluable for the spectroscopy of many molecules, for example a simplification of the complex photofragment spectrum of NO$^{+}$ could be achieved. Previously, the high density of lines have prevented any significant rotational assignment of the photofragment spectrum \cite{cosbyno} \cite{geersbatey}.\\

Finally, a series of experiments for recording photfragment spectra of  molecular ions with rare gas ligands could be devised. This is similar in concept to the work of Bieske et al., in which the electronic spectra of N$_{2}^{+}$..Rg, where Rg is Helium \cite{biesken2he} \cite{biesken2hen} or Neon \cite{biesken2ne} \cite{biesken2nen}. In this series of experiments, the mass selected molecules from a jet source, were irradiated and photofragmented. N$_{2}^{+}$ was collected as a function of laser wavelength. Where transitions occurred between electronic states of the chromophore, large numbers of N$_{2}^{+}$ fragments arrived at the detector. Therefore, the spectra closely resembled those of the cold chromophore, suggesting a low barrier to internal rotation of the N$_{2}^{+}$ ion. A `breaking off' in rotational progressions enabled a dissociation energy for these molecules to be estimated. \\
Interesting possibilities exist for such molecules in the new jet experiment. An series of experiments can be planned to study molecules such as CH$^{+}$..He, to ascertain the role of the Helium ligand when the chromophore spectrum is dominated by quantum mechanical tunneling \cite{sarrech1}. Similarly a photofragment spectrum of SiH$^{+}$..Rg could prove interesting due to the majority of transitions arising due to predissociation via Feshbach resonances \cite{sarresih}.






















