The majority of experiments presented here were performed on an apparatus designed to produce clusters from solid materials. The technique produces a plasma using a pulse of laser light focused on a rod of material contained within a small channel and filled with a pulse of inert gas. The rapid cooling of the vaporized material within the carrier gas and its expansion into an enclosing vacuum chamber results in the formation of small clusters of material. A continuous flow of carrier gas is not feasible due to pump limitations, so that a gas valve pulsing synchronously with the laser is employed. The laser ablates the surface of the material on which it is focused requiring that a fresh surface be presented to it. This is done by vaporizing the surface of a rod that is slowly rotated and translated, effectively cutting shallow threads. The complete design is commonly referred to as a Smalley source.¹¹

The packet of neutral and charged clusters that exits the source consists of a distribution of sizes that depends sensitively on the local conditions. To obtain size specific information the clusters are detected on the basis of their mass to charge ratio in a time of flight (TOF) mass spectrometer. In the case of neutral clusters they are first ionized with a visible or UV laser. Before or during detection, optical properties of the clusters are probed primarily through their absorption of light. The mass distribution and its dependence on source conditions is also observed as a manifestation of cluster properties.

The apparatus described in the following section was built and designed over a number of years and its features have continued to advance as the applications have become more sophisticated. Sufficient effort has been put into its design that it is versatile for studying the gas phase properties of neutral or positively charged clusters of a wide range of solid materials. To date, clusters of Ag, Cu, Au, Pt, C, Ta, and Nb have been generated, as well as complexes with benzene or N2 and mixtures of these materials such as Ag_xCu_y. The apparatus has also been used to deposit clusters onto surfaces for electron microscopy by Joanna Hunter¹² and to study the catalytic properties of deposited clusters by George Alameddin.¹³ These projects employed a transfer arm for collection of deposited clusters that has been documented in those references. Although the author has been involved extensively with the design, setup and modifications of this equipment, some information on its operation is already available and has not been reproduced here.¹⁴

The apparatus is sufficiently complex that considerable care must be taken to ensure its operation. The failure of one component is not always easily separated from the rest of the apparatus. Alignment of the vaporization laser and the molecular beam are critical, as is the timing of the various valves, lasers and pulsed voltages. Additionally the optimal settings for many of these parameters are dependent on each other so that optimization is not straightforward. Successful operation typically requires many months of practice.

The experimental apparatus consists of a two stage differentially pumped molecular beam machine shown in Figure 2.1. This setup was used to determine silver cluster ionization potentials.

To produce clusters with a lower internal energy the vaporization block, nozzle and valve were cooled to about 120K by flowing liquid N2 through a copper sleeve surrounding the nozzle. Cooling the source assembly typically resulted in an order of magnitude more signal, although mass distributions and laser timing changed substantially in the process. Subsequent experiments with other source configurations suggest that an important feature of this design is the indirect cooling of the pulsed valve and enclosed carrier gas. This is essential to produce clusters cold enough to observe adducts such as AgnN2 or AgnArm. These adducts were produced under the same conditions that low temperature ionization potential measurements were performed. At room temperature these complexes were not observed. Other researchers have produced small atom or dimer transition metal rare gas adducts in room temperature sources either without an enclosing block or with very short expansion channels.¹⁶ However such designs produce only very low intensities of larger clusters, n>10.

The signal intensity is surprisingly sensitive to the power used for vaporization and varies significantly with the material being vaporized and the carrier gas pressure. Silver requires powers in the range of 20-30mJ/pulse. Higher or lower power levels often yield very weak signals for both neutrals and ions. At high powers more ions and fewer neutrals are produced. Copper requires high powers (yet even then gives a poorer signal than silver), and carbon requires low powers. The details of the vaporization process remain somewhat mysterious. Its operation is somewhat of an art. (Attempts to optimize it have consumed too many hours and met with only partial success).

Detection Chamber. The source and detection chambers were separated by a 1mm 30º cone angle skimmer located approximately 5cm downstream from the nozzle where the clusters exit the source block. When only neutral clusters were of interest, ionized clusters were deflected by a small metal plate at -50V just behind the skimmer.

Cluster ions were detected with a 1m drift tube Wiley-McLaren TOF mass spectrometer¹⁷ oriented collinear to the beam. Since neutral clusters were ionized with a short pulsed laser, voltages in the mass spectrometer were held constant. Perpendicular orientations were occasionally employed but suffered from requiring an addition deflecting field whose optimum voltage was mass dependent. Ion detection was accomplished with two chevron oriented microchannel plates (MCP). The resulting signal was amplified, discriminated, and then fed as pulses (ion counting mode) into a PC based multichannel scaler board with 5ns resolution. The start of data acquisition was set by the laser trigger, or, in situations where high resolution was required, by a photodiode that removed the excimer trigger jitter of approximately 150ns.

For future reference it should be noted that a setup for electron impact ionization can also be attached between the extraction plates but this has not been exploited extensively as it causes fragmentation.

Delays with a microsecond precision were necessary between the opening of the pulsed valve, the firing of the vaporization laser, and the firing of the ionization laser. These were obtained with a Stanford Research Systems DG535 delay generator. Valve opening time was approximately 210m s; then 100-200m s later the vaporization laser was fired. For room temperature operation the pulsed beam had a temporal width of 400-550m s at the mass spectrometer but the greatest cluster intensity was located within approximately the first 50m s of this pulse. These two settings, vaporization and ionization laser delays, were critical to obtaining sufficient signal intensity. Typical mass spectra were acquired after summing the signals from 1K-10K pulse cycles.

The design here is an extensive modification of Phase 1 but only changes are described here. A schematic is shown and described in Figure 2.3. Primary modifications consist of the redesign of the source and pulsed valve assembly, and the addition of a reflectron time of flight mass spectrometer. Modifications were also made to the electronics to observe positively charged ions coming from the source. These modifications required a number of additional delay generators and voltage pulsers.

The pulsed valve was mounted external to the vacuum chamber allowing for convenient access. The source block was then mounted on thermally insulating spacers within the vacuum chamber just opposite the pulsed valve. This design permitted the source block and nozzle to be cooled independent of the pulsed valve that operated unreliably at low temperature. With this setup it was possible to operate the block and nozzle consistently at temperatures from 140 to 300K. The nozzle had an internal diameter of 2mm with a 40mm length channel terminating with a 10º conical extension. The excimer power used for vaporization was 50-80mJ per pulse.

A 1m reflectron time of flight mass spectrometer was installed perpendicular to the molecular beam. Detection of ions required that the deflecting field behind the skimmer described previously be removed. Additionally, the extractor and repeller voltages of the mass spectrometer were pulsed (<5ns rise time at 2000volts) as the intense portion of the cluster pulse passed through this region. The reflectron allowed for enhanced mass resolution of approximately 750. For photodissociation experiments the reflectron was used as a 'double' mass spectrometer by placing a mass gate approximately half-way along its length within the first stage. This mass gate deflected ions when activated and transmitted ions when grounded. It was designed so that the duration of the electrical grounding pulse (250volts and 5ns rise/fall time) did not have to be shorter than a few microseconds as was required by the pulsed electronics. This was accomplished by using the front end of the mass gate to deflect the faster low mass ions preceding the peak of interest, and the back end of the mass gate to deflect the slower high mass ions following the peak of interest.

Mass selected ions where photodissociated between the mass gate and reflecting fields at points determined by the delay of the dissociating laser. The laser was directed through both a quartz window and the reflecting field. With an appropriate choice of reflecting fields, charged fragments generated within the field free region could be mass analyzed based on their kinetic energy. Higher kinetic energy species follow longer trajectories through the reflecting field and hence arrive later at the detector. Species that fragment within the reflecting field produce broadened peaks in the mass spectrum.

The reflecting fields were also pulsed in order to reduce the background signal coming from pump oil ionized by the laser. Their voltages were turned on shortly after firing the laser. After passing through the mass gate the ion beam was collimated to ensure complete intersection of laser and ion beams. By varying the timing of the laser it was possible to induce fragmentation anywhere between the mass gate and reflecting fields. Under typical voltage settings this allowed for a 30m s time window. With additional effort the extraction voltages of the mass spectrometer could also be changed to increase this time window.

The experimental sequence executed by the delay generators consisted of the following events with approximate timings:

In an attempt to reduce an interfering signal from ionized pump oil within the source chamber, the oil based diffusion pump was replaced with two turbomolecular pumps. One pump was located below the chamber (500 l/sec) and another smaller pump was mounted to the side of the reflectron mass spectrometer. This was only partially successful in eliminating the pump oil signal. Further reduction can probably be obtained by installing a liquid N2 shroud.

Other modifications include the stationary mounting of the skimmer and block. This makes beam alignment relatively permanent and reduced the considerable amount of time spent ensuring correct alignment. Also, to simplify operations a stepper motor with automatic software control was added to rotate the vaporization rod, and the vaporization lens was mounted outside the vacuum chamber to reduce the amount of cleaning that was required. To monitor the temperature of the source, thermocouples are attached to the block and nozzle. These were interfaced to the PC for constant temperature recording. A resistive heater was attached to the pulsed valve and its temperature was maintained near room temperature with automatic control.

Although not used in the experiments documented here, the source also can generate bimetallic clusters. This was accomplished by splitting the vaporization laser and designing a block to hold two rods of material. Details of this design are given elsewhere.¹⁴ It has successfully produced mixed clusters such as AgxCuy.

Software control has been designed to scan the laser over the dye region and optimize the angle of the doubling crystal. This was accomplished through an analog to digital board attached to the PC. A complete description is available in the source code to the software.

13) N. G. Alameddin, Ph.D. Thesis, Northwestern University, Dec. 1992.