University of Southern California Dissertations and Theses

A technique for assessing water cluster sizes by pickup momentum transfer and a new source with loading system for lithium clusters

(USC Thesis Other)

A technique for assessing water cluster sizes by pickup momentum transfer and a new source with loading system for lithium clusters

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A TECHNIQUE FOR ASSESSING WATER CLUSTER SIZES BY PICKUP MOMENTUM TRANSFER AND A NEW SOURCE WITH LOADING SYSTEM FOR LITHIUM CLUSTERS by Chuanfu Huang A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (PHYSICS) August 2016 ii Acknowledgements In the six years of my graduate life, I would like to thank all people who helped me fulfill the project. First of all, I would like to thank my research advisor, Prof. Vitaly V. Kresin, whose expertise, passion, understanding and guidance make it possible for me to finish the project. No matter scientific problems or not, whenever I asked him for assistance or advices, he always can give me a satisfied support. I want to say he is a truly responsible and reasonable professor, and it is a great experience discussing problems with him, learning how to solve problems from him. I would like to thank other lab members, especially Dr. Nicholas G. Guggemos and Dr. Avik Halder, who directly helped me with learning cluster beam machines. Additionally, I am also thankful to other group members including Daniel Merthe, Malak Khojasteh and Patrick Edwards. I would like to thank my committee professors taking time on my thesis and defense. Finally, I would like to thank our machine shop staff members who are really nice: Don, Michael and Ramon. iii Contents Acknowledgements ....................................................................................................................................... ii Contents ....................................................................................................................................................... iii Abstract ......................................................................................................................................................... v Chapter 1. Introduction ................................................................................................................................ 1 1.1 What are clusters? .............................................................................................................................. 1 1.2 Why study water clusters? .................................................................................................................. 2 1.3 Ionization properties of water clusters ............................................................................................... 4 1.4 The work of this dissertation .............................................................................................................. 7 Chapter 2. Mixed Water Clusters .................................................................................................................. 9 2.1 Methods of doping clusters ................................................................................................................ 9 2.1.1 Producing mixed clusters by coexpansion ................................................................................. 10 2.1.2 Pick up by capillary ..................................................................................................................... 11 2.1.3 Mixed cluster mass .................................................................................................................... 17 2.2 Mixed clusters with other gases ....................................................................................................... 17 2.2.1 Methanol with water clusters .................................................................................................... 18 2.2.2 Ammonia with water clusters .................................................................................................... 19 Chapter 3. Water Cluster Equipment .......................................................................................................... 21 3.1 Water cluster source ......................................................................................................................... 23 3.2 Quadrupole mass spectrometer ....................................................................................................... 25 3.3 Detection system .............................................................................................................................. 26 3.3.1 Synchronous detector ................................................................................................................ 27 3.3.2 Chopper phase ........................................................................................................................... 29 iv 3.3.3 Synchronous mass spectrum ..................................................................................................... 29 Chapter 4. Measurement of Cluster Velocity ............................................................................................. 34 4.1 MCS ................................................................................................................................................... 35 4.2 Vollmer method ................................................................................................................................ 37 4.3 Fitting procedure .............................................................................................................................. 38 Chapter 5. Measurements and Discussion ................................................................................................. 43 5.1 Results ............................................................................................................................................... 43 5.2 Discussions ........................................................................................................................................ 46 5.3 Correction of speeds. ........................................................................................................................ 52 5.4 Cluster mean size .............................................................................................................................. 55 5.5 Future work ....................................................................................................................................... 57 Chapter 6. Design of Source with Loading System for Lithium Clusters ..................................................... 59 6.1 Recent theory .................................................................................................................................... 59 6.2 Loading system for lithium clusters .................................................................................................. 61 6.3 The different sources designed for lithium clusters ......................................................................... 66 6.3.1 Typical seeded supersonic nozzle source .................................................................................. 66 6.3.2 Upgrade to a TZM source ........................................................................................................... 69 6.3.3 Modified French source ............................................................................................................. 74 6.3.4 A newly designed stainless steel source .................................................................................... 76 6.4 Results ............................................................................................................................................... 80 6.5 Future work ....................................................................................................................................... 84 References .............................................................................................................................................. 86 v Abstract We have developed a technique for assessing water cluster sizes by pickup momentum transfer, revealing abundant fragmentation of water clusters is due to electron impact-ionization (70eV). Neat water clusters (H 20) n were produced by a supersonic nozzle source, while neutral acid-water clusters were produced with the same method by attaching a deuterium chloride molecule to neat water clusters. Other gases including CO 2, CH 3OH and NH 3 were also explored, but the mass spectra did not show efficient attachments. The speeds of the water cations (H 2O) n-1H + covering the size from n = 3 to 12 were investigated. By applying the momentum conservation law, we show that most of the detected water cluster ions originate from an original size of around n = 30-40. To check the validity of this conclusion, we also carried out the similar experiment for neat deuterium water clusters and corresponding mixed clusters with hydrogen vi chloride. Here, we found that the original deuterium water cluster sizes are slightly smaller than the original hydrogen water cluster sizes. In another part of the project, we designed a system for transferring a load of high purity lithium metal into a molecular or cluster beam source. A hot loading vessel was thoroughly baked out while empty and overpressured with argon. A clean Li rod was then dropped in through a long narrow tube. The thoroughly degassed interior of the vessel and the rapid melting of the inserted rod facilitated contamination-free transfer of the highly reactive liquid metal into the source oven. The seeded supersonic source previously used for sodium clusters (650°C) was incapable for producing lithium clusters which require a higher temperature (>1050°C). Therefore, we used a molybdenum alloy (TZM) which can sustain the requisite temperature to design a new source oven, however welding turn out to be a critical problem. Finally, we designed a stainless steel source in a compact version with much thicker walls and a set of new heaters. The new source has been tested to successfully work at 1050°C for around 48 hours, and deposition of lithium has been observed. 1 Chapter 1. Introduction 1.1 What are clusters? During the last several decades, study of atomic and molecular clusters has become an important branch in condensed matter physics, materials science and chemistry field. Clusters, also named nanoclusters, can also be considered intermediate materials between atoms or molecules and bulk materials. Normally, a finite number of atoms or molecules, from a few to thousands, can form nanoclusters which can exhibit fundamentally different properties compared with single atom and bulk solids. In principle, clusters are constituted by elements in the periodic table, thus clusters can be subdivided into different types: metal clusters, rare gas clusters, molecular clusters (for example, water clusters), etc. Why are clusters important? Currently, there are countless investigations including experimental and theoretical goals to study cluster types, which could provide a new region to extend the vision of classical and quantum physics. For instance, the milestone discovery of the shell structure, a striking order of 2 intense peaks of sodium clusters mass spectrum, was found by Knight in 1984 [1]. Compared with metal clusters, water cluster mass spectrum also shows a magic number around N = 21, which was first observed by Lin in 1973 [2]. Also more attractive examples include catalyst properties of gold cluster [3], a possible high temperature superconducting state of aluminum clusters T c > 100K observed by A. Halder [4]. These fascinating properties offer clusters could have considerable potential applications, such as new materials of catalyst and microelectronic apparatus and so on. 1.2 Why study water clusters? Water is one of the most common element in the space and is also most valuable and substance resource on earth. Furthermore, studying clusters of water creates an appropriate benchmark for a number of current problems, including the formation of acid rain, understanding of the nucleation of water droplets and biochemical processes [5-7]. Additionally, water cluster research delivers an appealing scene for significant scientific meaning in unravelling the molecular interaction properties, for instance, of hydrogen bonding cooperativity and structure fluctuation or rearrangement [8-12], which are prevailing in the liquid water. Specifically, two major issues have been studied extensively for understanding of the proton in the water: 1. It is proton bounded by one water molecule or two molecules bounded, which are referred as Eigen model (form H 3O + ) [13] and Zundel model (H 2O…H + …H 2O) [14]. In principle, these two types play an essential role in the large cluster and should behavior different IR absorption due to the proton located in the different position. 2. Determines the structure of water cluster, for instance, the hexamer owns a most stable cage-like structure, which can be representing the transition from cyclic to three-dimensional configuration, this is predicted and demonstrated by theory and experiments [8-12, 15-18]. Furthermore, 3 considerable amount of researches focused on the structure of “magic number” ((H 2O) 21H + ), initially Searcy and Fenn [19] suggested in 1974 that this can be owing to form a stable dodecahedral structure and this was verified by Miyazaki [20] and Shin [21] in 2004 with infrared spectra. However, it still exists controversies whether H 3O + occupies the center of the cage structure composed of the water dodecahedrons or sits on the surface by displacing a H 2O molecule to the center of the cage [22-29]. Apart from the IR technique, the beam deflection method can probe the dielectric properties of the water clusters, which can provide another effective way to investigate the water cluster structure and charge arrangement, thus the interaction of water molecules in cluster, and the action of dopants of the mixed water clusters can be analyzed [30-32]. For example, Kresin and coworker [33] found that the polarizabilities of the water clusters are size dependent and they found there exists a jump between n = 8 (1.3D, 1D is 3.3×10 −30 Cm) and n = 9 (1.6D) of the water clusters. More recently [34], the dipole moment of the mixed clusters of water with DCl were also measured and they discovered that a jump is between n = 5 to n = 6, which is possible related to the disassociation of the DCl molecule. Besides what we already introduced, studying water clusters also provides useful information on the size varying interaction between solvent and solute. While the theory that acid can be ionized in the bulk water was groundbreaking proposed by Arrhenius [35] more than one hundred year ago, who also received the Nobel prize in 1903, an interesting puzzle has been discussed with theory extensively: how many water molecules are necessary for dissociating the hydrogen halide molecule? Most theories [36- 40] show that the water cluster size n = 4 is essential for ionizing hydrogen chloride. In other words, the charge separation of HCl can be fulfilled by the larger cluster. However, recently a systematical theory [41] calculation showed that the HCl molecule can be at least partially dissociated for n > 4. This controversy still needs more precise experiments to resolve because the IR method cannot provide an effective or explicit means to confirm the structure for small water clusters produced by free jet expansion. The reason 4 is that the quantum interaction and thermal fluctuation of the molecules will become dominant for arranging the structure of the water cluster and the acid molecule. Therefore, when the cluster-acid is at low temperature, the separation distance between Cl - and H + should be larger than when the mixture is produced by the general molecular beam method (>100K). This is verified by Gutberlet et al. that a single HCl molecule can be fully ionized by n = 4 cluster in this ultracold environment (He droplet, T = 0.37mK), however, which has been disproven by a later experiment of Vilesov [42]. 1.3 Ionization properties of water clusters In most laboratories, water clusters are detected by mass spectrometric methods, so ionization properties of water clusters are fundamental and important in the molecule beam experiments and intensively studied by various techniques [43-50]. For example, electron bombardment ionization is often used for ionizing the water clusters and the typical electron energy is 70 eV which produces the maximum ion yield for most species [43]. Besides electron ionization [51], single-photon ionization [52-55], femtosecond multiphoton ionization [56], and special Na-doping photon-ionization are also common used [43, 57] in the researches of water clusters. A common viewpoint of the ionization of the water HOMO orbital in the bulk is that a proton prefers to transfer into a neighbor water molecule and valid as following equation: 𝐻 2 𝑂 + +𝐻 2 𝑂 → 𝐻 3 𝑂 + +𝑂 𝐻 ∙ The stable state is the hydronium cation 𝐻 3 𝑂 + and 𝑂 𝐻 ∙ radical, based on the recent paper, and the lifetime of the unstable 𝐻 2 𝑂 + cation is less than 40 fs [58, 59]. Consequently, the detected mass spectra of water clusters typically show the protonated (𝐻 2 𝑂 ) 𝑛 𝐻 + cluster peaks rather than the original water clusters. Subsequently one interesting question arises to the research: is it only one 𝑂 𝐻 ∙ radical lost during 5 the ionization? Or does this process accompany water molecules evaporated? Then how many molecules are lost during the ionization? A generalized process for water cluster based on the photoionization can be described as: (𝐻 2 𝑂 ) 𝑛 +ℎ𝑣 → [(𝐻 2 𝑂 ) 𝑛 + ]+𝑒 − → (𝐻 2 𝑂 ) 𝑛 −1 𝐻 + +𝑂 𝐻 ∙ +𝑒 − The fast formed protonated water cluster would possibly cause one water molecule evaporated in the mass spectrometer. The reason is because the protonated process could increase a small amounts of energy and which is enough to kick one water molecule out of the cluster body. Thus, the total energy can be reduced and we can write the stabilized process as: (𝐻 2 𝑂 ) 𝑛 −1 𝐻 + → (𝐻 2 𝑂 ) 𝑛 −2 𝐻 + +𝐻 2 𝑂 . Recent papers of Signorells’ group [60, 61] show the light fragmentation of the water clusters after near threshold VUV photoionization. Furthermore, the work from Belau [62] shows the metastable process might loss up to three water molecules rather than one molecule based on the measurement of the appearance energies of difference cluster sizes by vacuum ultraviolet ionization, then they analyzed the cluster decay near the VUV threshold. (𝐻 2 𝑂 ) 𝑛 −1 𝐻 + → (𝐻 2 𝑂 ) 𝑛 −𝑚 −1 𝐻 + +𝑚𝐻 2 𝑂 . m can be up to three. Also, small fragmentation are observed in the recent another paper which is also based on the VUV ionization [52]. There are some circumstances that the non-protonated (H 2O) n + ions were found upon ionization of water cluster solvated with Ar atom [63-64, 52]. However, these non-protonated clusters still have stable protonated structure like H 3O + with 𝑂 𝐻 ∙ radical established by the infrared spectroscopy [65-66]. The 6 possible reason is because Ar atom can be evaporated to cool down the clusters, thus OH cannot be sheared. While the photon ionization shows a small fragmentation near threshold, electron ionization can probably cause a more serious fragmentation of the water clusters from Jozef et al. [43] work. They verified the fragmentation exists by comparing the mass spectra of the water clusters from EI and NaPI methods. NaPI method is a special ionization method by doping water clusters with Na and the doped water clusters are ionized by photoionization. In principle, NaPI method provides the way detecting water clusters with fragmentation free since when water clusters pick up sodium atom, this picking process will form Na + and electron, which is named as solvated electron e - . This solvated electron is very loosely bound to the water cluster and can be kicked out of the cluster with relatively low energy 3-4 eV. Thus, the cluster (H 2O) nNa + ion can be collected as the mass spectrum, which should reflect the original mass spectrum of water clusters since the ionization energy only 3-4 eV cannot result in the fragmentation of the water clusters. Moreover, the authors also demonstrates that the fragmentation of water cluster does not reply on the electron ionization energy and they collected the mass spectra with the ionization energies between 15 eV and 90 eV, impressively 15 eV already caused a very serious fragmentation. This is similar to the results of Dong et al. [52], which shows that the water clusters fragmentation does not depend on the photoionization and only a small potion energy actually deposit into the water clusters, so they suggested the excess electron will take most energy out of the water clusters. However, the controversial point is that Dong’s work shows a small fragmentation with soft x-ray ionization while Jozef shows a serious fragmentation based on the EI method. It seems the fragmentation does not rely on the ionization energy and the ionization results are dependent on the ionization techniques and still puzzles researchers. 7 1.4 The work of this dissertation Although there are plenty of researches on the ionization properties of water clusters and the problem seems fundamental, the ionization process still not completely understood. In this dissertation, we developed a momentum pickup technique to assess the parent water clusters based on the traditional electron ionization. We confirm that the water clusters have severe fragmentation based on 70eV electron bombardment. The more detailed explanations are presented in the following chapters and the basic idea is: We doped the water clusters with foreign gas by using a capillary that is directed basically perpendicular to the cluster beam, ideally there’s no longitudinal momentum transfer from its effusing beam to the clusters, hence let’s see how much the mixed clusters are slowed down with respect to the neat water ones. If there’s no fragmentation, the amount of slowing down will be trivial, for example: 𝑚 (𝐻 2 𝑂 ) 𝑛 ∙𝑣 (𝐻 2 𝑂 ) 𝑛 =[𝑚 (𝐻 2 𝑂 ) 𝑛 +𝑚 (𝑓𝑜𝑟𝑒𝑖𝑔𝑛 𝑔𝑎𝑠 ) ]∙𝑣 (𝐻 2 𝑂 ) 𝑛 (𝑓𝑜𝑟𝑒𝑖𝑔𝑛 𝑔𝑎𝑠 ) If this held, it would confirm that the detected mass spectrum reflects the actual neutral size distribution, such as (𝐻 2 𝑂 ) 𝑛 𝑒 − → (𝐻 2 𝑂 ) 𝑛 𝐻 + +𝑂 𝐻 ∙ (1) In other words, there is no additional water molecules evaporated during the electron ionization. If this not held, there are probably some cluster fragmentation are involved during the ionization. In this dissertation, we focused on the speeds measurements of the water clusters size from 3 to 12, for even larger water clusters, the resolution of QMA is not ideal for discriminating the mixed and neat water clusters. We schedule the dissertation as following: 8 Chapter 2 demonstrates what foreign gases can be picked up by the water clusters in the capillary configuration and which foreign gas can give a detectable strong signal of mixed clusters. For instance, we tried a bunch of gases such as CO 2, CH 3OH, NH 3 and DCl, the mass spectra shows only DCl can be picked up. We also will explain why we use DCl doped with H 2O cluster rather than HCl in this chapter. In principle, (D 2O) n cluster should behavior similar properties with H 2O clusters, so we also performed the similar experiment for (D 2O) n cluster to check if the ionization properties are similar as we assumed, the more detailed discussion are shown in Chapter 5. In Chapter 3, we describe the details of our water cluster molecule beam machine. In Chapter 4, we show a Vollmer method to detect the water cluster velocities based on the time resolved flux recorded by MCS technique which enables detect neat and mixed H 2O clusters with DCl. In Chapter 5, the final results are discussed and we verify that water clusters have large fragmentation due to electron bombardment based on our measurements. More specifically, we can build a simple connection between daughter and parent clusters based on the momentum transfer pickup technique. In Chapter 6, we work on the lithium cluster project and introduce the background first. Then, designed of new loading system and lithium source are shown in this chapter, and we obtained some rudimentary progresses about this nasty materials. 9 Chapter 2. Mixed Water Clusters As we described in Chapter 1, doping the foreign gas can possibly enable us figure out the fragmentation information of water clusters by measuring the velocities of neat and mixed clusters. Thus, this chapter we describes the background of producing the mixed clusters, also demonstrates that only DCl can be easily picked up by H 2O clusters, although we tried a bunch of other gases CO 2, CH 3OH, NH 3. Finally we briefly explains why we chose (H 2O) nDCl and (D 2O) nHCl as the research objects rather than (H 2O) nHCl and (D 2O) nDCl. 2.1 Methods of doping clusters To prepare the mixed clusters, gas aggregation [67-69] has been extensively applied: the anticipated molecules are evaporated from the heated oven and then condensed with the cold aggregation gas. Additionally, there are two other popular methods to produce the mixed clusters, coexpansion [70-71, 52] and “pick-up” methods [72-73], see following sections. The mixed clusters produced by coexpansion 10 typically have much stronger intensity than the mixed cluster prepared by the “pick-up” methods. However, we use “pick-up” method (capillary) which enables us to investigate the collision between the neat water clusters and foreign gas for our purpose. 2.1.1 Producing mixed clusters by coexpansion We can easily get the mixed clusters by the coexpansion of two different chemical compound vapors though a small nozzle. For example, we attempted to get the mixed methanol-water clusters by mixing Figure 2. 1 Representative mass spectrum of water-methanol coexpansion, the red arrows show the peaks corresponding to the near H 2O clusters and the other peaks corresponding to the mixed clusters of CH 3OH and H 2O or neat CH 3OH clusters. 11 the methanol and water solutions in ratio 1:5, the neat water cluster peaks are marked with the red arrows while the other peaks correspond to the mixed methanol-water or methanol clusters. With the co- expansion of mixed vapors of methanol and water, the formed mixed methanol-water clusters have the stronger intensities compared with the neat water clusters. The structure and proton properties of mixed water-methanol clusters produced by co-expansion method were investigated by Fujii group [71, 74]. For our purpose, we need to dope water clusters by “pick-up” method which typically has a relatively low intensity compared with neat water cluster. The reason why we produced a mixed water-methanol clusters by co-expansion is to find the possible mixed cluster peaks which can be as references for finding mixed water clusters with CH 3OH with “pick-up” method. 2.1.2 Pick up by capillary If the neat cluster beam passes through a cell with the low density gas, the clusters have the probability picking up the foreign atoms or molecules. This process provides another effective technique to dope the clusters and is extensively used in the helium droplet field which can offer the ideal low temperature environment for investigating chemical reaction between molecules and the structure of the other clusters in the droplet [75, 42], etc. Also, as we already mentioned considerable amount of the water cluster pick up experiments were also achieved in the last few decades. However, the gas in a pick- up cell can move in any direction and the speed directions of the foreign atoms can be randomly. Thus, after the water clusters pick up the foreign molecule, which can bring certain unknown longitudinal velocities to the mixed water clusters. In other words, this also can bring uncertainties to measure the mixed cluster speeds, the distribution of which could correspondingly be broaden. Therefore, the configuration of the perpendicular capillary is utilized and the effusive foreign gas can be picked up by the 12 water cluster beam. In the capillary configuration, the additional longitudinal velocities of foreign molecules can be neglected, see Figure 2.2. Now it is worth to make a small digression to explain how much distance we should put the capillary in front of the nozzle. We used a typical supersonic source and the details of experiment will be described in the next chapter. From the basic gas dynamic based on thermodynamics [76], the expansion of the vapor is adiabatic and isentropic. However, we should emphasize that the shape of the nozzle determines the expansion condition. In principle, the convergent shape has viscous influences on the beam and will break down “isentropic” condition, but if the nozzle length is not excessively long, it still can be considered approximately as “adiabatic and isent r op ic ” because viscosity of the boundary can be neglected. For example, the nozzle we used, see in Figure 2.2, is convergent-divergent shape and convergent part will cause the viscosity but we assume it can be neglected, therefore the supersonic expansion of water is Figure 2. 2 Schematic of convergent-divergent nozzle we used, DCl or HCl molecules can be picked up by the molecule beam. We used the distance between nozzle and capillary is about 5 mm which should be far enough to the “quitting” surface. 13 considered as adiabatic and isentropic. Consequently, some basic characters can be estimated from the isentropic (convergent nozzle) approximation and also will be true for our nozzle (convergent-divergent nozzle). Figure 2.3 shows the velocity, temperature, density and collision rate evolution with the distance ratio between X and d based on the theory of thermodynamics, where X is the beam distance to the nozzle exit and d is the diameter of the nozzle, for the nozzle we used, it is 75 micrometer. We can find that when X = 20d = 1.5 mm, the collision rate is already almost zero, which is typically known as “quitting surface”. In other words, the quitting surface represents a transition from continuous beam to discontinuous beam. Hence when we put the capillary 5 mm far to the nozzle, DCl or HCl should not be accelerated by the expansion of the water vapor and can be picked up by water clusters separately. Although we can put the capillary even farther to ensure the DCl or HCl molecule cannot be impacted by the expansion, for instance, X = 10 mm, the signal of mixed clusters can be lower and the background signal cannot be ignored, thus the mixed clusters signal could be strongly affected by the background noise. Finally we chose the Figure 2. 3 The collision rate and velocity of clusters are predicted by the isentropic approximation [76]. 14 distance of the capillary around 5 mm to the capillary as a compromising way and two representative mass spectra are shown in Figure 2.4 and 2.5. We can see that Figure 2.4 shows well separate peaks of neat and mixed H 2O clusters with DCl, while Figure 2.5 shows the similar separate peaks of the neat and mixed D 2O clusters with HCl. Definitely, the signal of mixed water clusters can be detected in the capillary configuration with distance around 5 mm. We also demonstrated that the mixed clusters produced from the co-expansion are much stronger than produced from the capillary configuration. Figure 2. 4 The schematic mass spectrum of H 2O clusters doped with DCl gas from a lecture bottle. The peaks at 73, 91 and 109 amu are corresponding to the clusters (H 2O) 4H + , (H 2O) 5H + and (H 2O) 6H + , while the peaks at 74, 92 and 110 amu are corresponding to the clusters, (H 2O) 4D + , (H 2O) 5D + and (H 2O) 6D + . 15 Figure 2. 5 The schematic mass spectrum of D 2O clusters doped with HCl gas from a lecture bottle. The peaks at 62, 82 and 102 amu are corresponding to the neat D 2O clusters (D 2O) 3D + , (D 2O) 4D + and (D 2O) 5D + , while the peaks at 61, 81 and 101 amu are corresponding to the mixed D 2O clusters with HCl: (D 2O) 3H + , (H 2O) 4D + and (H 2O) 5D + . 16 To increase the mixed water cluster intensity, we attempted to design a “wheel” shape capillary made of stainless steel, see Figure 2.6. There are eight small channels or capillaries near the center of the wheel and the right graph is an alignment procedure for the designed wheel, this alignment method also works for a single capillary tube. We put certain amount methanol or acetone solution in the source, then we send a carrier gas Ar or CO 2 at the pressure around 4 atmospheres, a very dedicated beam can be detected by eye, which is very helpful for making the alignment. Typically the beam can last 5 mins, we first roughly align the capillary or the wheel and subsequently we open the carrier gas valve, the methanol beam or acetone beam is ejected out for fine alignment. If we finish the fine adjustment, and fix the precise position of the capillary by a hose clamp. Original objective is to increase the pickup probability between acid gas and the water clusters. However, after we replaced the capillary tube with the wheel capillary, surprisingly the neat water clusters signals were totally eliminated and cannot be detected. The possible reason is since the initial wide water cluster beam hit on the wall of the wheel, the hot wall (supported by the rods which is connected to the nozzle nut) could cause the attached water clusters evaporate immediately. Thus, the center of the water Figure 2. 6 CAD drawing for the designed wheel, the center whole is for passing through the cluster beam while the smaller hole is for welding a 1/16” tube, there is one more 1/8” stainless cover plate. 17 beam can be exterminated by the evaporated water molecules, much less for the mixed cluster beam. For the capillary tube, the case is different since the capillary is more like a point and we can neglect the influence of the evaporated water molecules which might reduce a little intensity of the mixed water clusters. 2.1.3 Mixed cluster mass The previous research [77] found that a mixed water-HCl cluster always loses the Cl atom in electron- impact ionization and a protonated water cluster can be detected. If there is no fragmentation, we can describe as: (𝐻 2 𝑂 ) 𝑛 𝐻𝐶𝑙 𝑒 − → (𝐻 2 𝑂 ) 𝑛 𝐻 + +𝐶𝑙 − However, a pure water cluster also shows up as a protonated one, having lost its OH, as we described in the equation (1) in Chapter 1. Therefore, a protonated water cluster (H 2O) nH + can be from (H 2O) nHCl or (H 2O) n+1 and it would be impossible to separate a doped vs neat cluster in a mass spectrum. To circumvent this, we use deuterium: either (H 2O) nDCl or(D 2O) nHCl. See Figure 2.4 and 2.5. 2.2 Mixed clusters with other gases In the water molecule beam experiment, we tried a lot of different polar gases including CO 2, CH 3OH, NH 3 and DCl with the capillary configuration to dope with the water clusters. Since the polar molecular gases which are very easily dissolved in the bulk water and it possibly indicates these gases can be doped with the water clusters. Based on the experiment results, we found hydrogen chloride is easily doped with the water clusters and that is why we chose (H 2O) nDCl and (D 2O) nHCl as our research system which are 18 detectable, see Figure 2.4 and 2.5. Next we will show the mass spectra of the water clusters with dopants CH 3OH and NH 3, definitely we didn’t observe any additional peaks like dopant hydrogen chloride in Figure 2.4 and 2.5. 2.2.1 Methanol with water clusters The methanol liquid was heated in a “cross” shape glass bottle by a rope heater and kept at 35°C to get enough vapor pressure of the methanol. Then we sealed the glass bottle by a rubber plug with a 1/16” stainless steel tube which is connected to the existing capillary. The flow rate of the methanol vapor can be adjusted by the needle valve and we collected the mass spectrum in Figure 2.5, which shows the neat water cluster beam intensity was reduced about half, this means there are a lot of collision between methanol molecules and the water cluster beam. As we mentioned above, we expect the mixed methanol- water cluster peaks in the mass spectrum of the co-expansion method can also be observed in the capillary configuration. However, if we compare Figure 2.7 with Figure 2.4, we cannot observe the additional mixed methanol-water clusters in the capillary configuration. 19 2.2.2 Ammonia with water clusters Meanwhile, we also made an effort of doping other gases including NH 3 and CO 2, but the mass spectra also show those foreign gases cannot be picked up by the water clusters in the capillary configuration. For example, the mass spectrum of the water clusters with dopant NH 3 is shown in Figure 2.8 and definitely there are considerable collisions between NH 3 and the water cluster beam since the intensity of the pure water cluster is also reduced about half. The reason why we kept the dopant gases chopped the neat water cluster intensity as half after the foreign gas crossed the beam is that we found the intensities of Figure 2. 7 The mass spectrum of the water clusters when the CH 3OH was turned on and off. We can see the cluster intensity was reduced about half, but the additional mixed cluster peaks were not observed as in Figure 2. 4 and 2. 5. 20 the mixed water cluster with DCl were maximum. However, the possible additional peaks for the mixed water clusters with NH 3 are not observed in Figure 2. 8 and we got similar spectrum with CO 2. Impressively, we find only DCl is easily picked up by the water clusters and there exists possible interesting physics essence for future experiment or understanding. Figure 2. 8 The mass spectrum shows NH 3 may not be able to pick up by the water cluster beam and we also adjusted the needle valve to ensure the water cluster beam intensity chopped about half. 21 Chapter 3. Water Cluster Equipment In cluster science, it is important to note that progress has achieved which is much dependent on the development of the new experimental techniques. The predecessors obtained the clusters properties by means of depositing the clusters on the substrate or captured in matrices. However, the substrate or matrices could cause the structure of the cluster changed, include its exotic features. The molecule beam method provides an effective way to study isolated clusters in a free environment. In this chapter, we will describe the molecule beam machine of water cluster and explain how the mass spectra shown in Chapter 2 were scanned. First of all, a schematic representation of water molecule beam equipment in our group is summarized in Figure 3.1 and there are three sections are presented in this chapter to explain explicitly how a water cluster beam generated, size-selected by the quadrupole mass spectrometer and then detected by a synchronous detector based on this figure. 22 Figure 3. 1 The outline of the water molecule beam machine. 23 3.1 Water cluster source For different purposes, there are a number of important sources such as supersonic nozzle source, laser vaporization-flow condensation source and magnetron sputtering source and so on [78]. For instance, the laser vaporization-flow condensation source is a pulsed source which can be used for preparing small and medium size metal clusters. We use a typical seeded supersonic nozzle source which can produce relatively intense cluster beams of low temperature melting materials, for example, water! Figure 3. 2 shows the schematic representation of the source chamber of the water equipment. The purified water or deuterium water can be loaded into a stainless steel reservoir, which can be aligned by a X-Y-Z stage, through a 1/16” stainless steel tube. The source chamber pressure typically is around 2 × 10 -4 torr with liquid nitrogen, while the reservoir and the converging-diverging nozzle (75um orifice) Figure 3. 2 Outline of the water source chamber. 24 temperatures are kept constant at 408K and 448K 1 by PID controller to attain the enough stagnation pressure (around 2000torr) in the source. Thus, a supersonic water cluster beam can be created with the water vapor expansion which is adiabatic, through the small nozzle, so some basic characters can be estimated by the thermodynamic theory as we described in Chapter 2. A more detailed configuration of the source parts is shown in Figure 3.3. Two K type thermocouples are mounted to the tip of the nozzle and the back end of the reservoir, while Teflon is for heat insulation. The nozzle heater is home-built by wrapping the tantalum wires on a ceramic spool and the body heater is commercially manufactured and clamped by a hose clamp, the average powers are approximately 8W and 15W. The baffle plate is to avoid the water loaded into the nozzle directly and the bottom leg of the source is for locating the source in the right position. We use a Swagelok VCR gasket (SS-4-VCR-2-GR) and connector to mount the nozzle, so which can be disassembled conveniently for 1 The higher temperature of the nozzle is to avoid plugging of the orifice. Figure 3. 3 Schematic details of nozzle and capillary. 25 cleaning the nozzle and reservoir inside. A 1/16” stainless steel capillary tube is squeezed into an ellipsoidal cross shape(semi-major axis ID around 2 mm) and located 5 mm in front of the nozzle, the direction of the capillary tube vertically points downward to the beam and about 2 mm far to the beam as shown in the graph. Foreign gas (DCl/HCl) gets through the tube and the flow rate can be controlled by a two-stage pressure regulator(for corrosive use) and fine adjusted by a Swagelok dual-needle valve. The precise adjustment of the valve enables us obtain a maximum intensity of the mixed cluster, normally if the beam is chopper by half intensity by collision with DCl/HCl, as we shown in Chapter 2. 3.2 Quadrupole mass spectrometer DCl/HCl molecules are picked up by the water clusters after diffusing out of the capillary, then the mixed cluster beam can be collimated by a skimmer and a collimator, and then passes through a high voltage deflector (not used in this experiment, so not shown in the graph). The unparalleled clusters are eliminated by collimator which is designed as a rectangular shape with 0.25mm × 1.0mm to narrow the divergence of the beam. The following mixed beam passes through a chopper wheel which is spinning at 148Hz and driven by a mechanical motor with a three-phase power supply (the frequency is phase-locked to a quartz oscillator). Next the cluster particles are ionized with a ionizer by 70 eV and 20 mA electron bombardment, thus can be size selected from 0 to 300 amu by a quadrupole mass spectrometer, which is controlled with small DC analog(0-10 V, corresponding to 0-300 amu) voltages supplied by EG&G 5209(+/- 15.00V, 1mV resolution) lock-in amplifier. Meanwhile, the quadrupole should be set in the external program mode, subsequently a Labview program [79] “UTI-100c with serial V2-1 mx. vi” was written to drive the lock-in amplifier which plays an role for analog/digital conversation, thus the specific mass can be selected by the computer through an RS-232 serial port. An important note should be emphasized here, we found the scanning speed of the Labview program during experiment often becomes extremely 26 slow and the possible reason is because of switching off/on the relay of the PID controller(for keeping constant temperature, shown in the above right picture) can produce strong electromagnetic pluses resetting the amplifier. Accordingly we should keep the power supplies and PID controllers of the water source and nozzle as far as we can and a small copper box is made to insulate the possible radio E&M interruption with the voltage signal transferring in the naked bins of RS-232 connections, see Figure 3.4. 3.3 Detection system After passing through the quadrupole spectrometer, the ionized clusters are detected by a pulse- counting electron multiplier-DeTech Inc. Model 311, which produces the quasi-Gaussian distribution of pulse amplitudes and fed through a single-threshold discriminator (ARI FT-100D). Then the generated TTL- level pulses from the discriminator pass into a isolator(NVE, IL710GMR) to eliminate possible ground loops(extra noises), which is wired to a line driver module. The final ion pulses from the module can be recorded by the synchronous detector and Ortec MCS-pci multi-channel scaler (MCS) with software of Ortec’s MCS-32. Figure 3. 4 A copper box is built to insulate the possible E&M interruption from the PID controller that is put in the bottom of the rack, while the EG&G amplifier is in the middle of the rack (not shown). Furthermore, a piece of metal plate is used to eliminate the possible E&M interruption. 27 3.3.1 Synchronous detector The synchronous detector is built based on the PCI 6602 that are arranged as the two counters (counter A and counter B, see Figure 3.5), which are synchronized to the chopper by an LED-photodiode. For now, we assume that the channeltron outputs a square wave with the same frequency as the chopper which modulates the beam. Counter A is supposed to count channeltron pulses when the wave is maximum and the other when the wave is minimum, corresponding to the times when the chopper is 'open' and 'closed'. Accordingly the whole system runs for many chopper cycles, for example, 12 chopper cycles in the figure, actual cycles may be thousands, which can be set up by the Labview program and gated by counter 5 in the PCI 6602 to make sure the counter A and counter B have collected same cycles of the chopper. Thus, the counter A measures all 12 channeltron pulses when the chopper is open, while the B counter measures all the 12 pulses when the chopper is closed. In principle, the noise in the system is uncorrelated with the chopper, and assume that it appears equally in the A and B counters: this allows us to obtain the synchronous signal is A-B. 28 Figure 3. 5 Schematic representation of synchronous detector system. 29 3.3.2 Chopper phase However, a small wrinkle exists: the molecule beam travels at a finite speed, so which means that it reaches the channeltron certain time (depending on the clusters speeds) after the chopper opens or closes. There is a photodiode, closely aligned to the chopper actually cuts through the beam, which measures the time that the chopper opens and closes, red square wave of chopper synchronization signal in the Figure 3.5. Hence we can't use the real chopper signal from the photogate to switch between the A and B counters--it has to be delayed by the amount of time it take for the beam to transit from the chopper to the channeltron. This is known as chopper phase = 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑐 ℎ𝑜𝑝𝑝𝑒𝑟 𝑎𝑛𝑑 𝑐 ℎ𝑎𝑛𝑛𝑒𝑙𝑡𝑟𝑜𝑛 𝐶𝑙𝑢𝑠𝑡𝑒𝑟 𝑠𝑝𝑒𝑒𝑑 , which is an important experimental parameter in the LabVIEW program. The chopper phase and the number of the chopper cycles are realized by an internal 20MHz oscillator and gated with the gate generator, which affords an accurate delay with resolution 50ns for triggering the counter A, B and counter 5. 3.3.3 Synchronous mass spectrum The mass spectrum can be obtained by scanning the analog voltage of UTI-100c based on Labview program ”counter spectrum”, which collected two important parameters of the Dip and Total rate: Dip= 𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐴 −𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐵 (𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐴 +𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐵 )/2 ; Total Rate= 𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐴 +𝐶𝑜𝑢𝑛𝑡𝑒𝑟 𝐵 𝐺𝑎𝑡𝑒 𝑑 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 𝑡𝑖𝑚𝑒 𝑙𝑒𝑛𝑔𝑡 ℎ ; Dip represents the ratio of the actual water cluster counts to the total counts including the background, so it has certain probabilities the dip is negative if the gate valve between the source and detector 30 chamber is closed. In other words, the total rate contains both the background and the real beam ion intensities. Therefore, the final synchronous signal named Lock-in Rate = Dip * Total Rate. There are some important parameters need to be explained in the “counter spectrum” program. For example: Window 1: 35-50 amu, 150 ms dwell, 150 steps Window 2: 50-200 amu, 300-500 ms dwell, 1500 steps For 35-40 amu or 50-200 amu, it is the range of the mass spectrum can be defined as we need. Also, we should keep in mind that the spectrum below about 50 amu is very congested and intense--therefore we probably need to use different settings for the two spectrum windows (i.e., reduce the dwell time and points so the channeltron doesn't burn out. For 50-200 amu, there should be less junk and dwell time can be longer to eliminate the background fluctuation. Furthermore, don't start or end on an intense mass because the program sometimes spends extra time there). The parameter of the total step is also required (150steps or 1000steps, 15amu/150steps = 150amu/1500steps = 0.1amu/step), dwell time is the approximate amount of time spent measuring each step. A mass spectrum with 150 steps and 150 ms dwell time takes roughly 22.5 seconds to measure, however, the actual time may be longer than 22.5 seconds because there are a bunch of delays to make sure that the spectrometer is on the right mass, etc. If dwell time is 150 ms and the chopper runs at 140 Hz (this would be a typical chopper speed--more in a minute), then each measurement would be 140 Hz * 150 ms = 21 cycles of the chopper. For a light mass range, it requires short dwell time because the background has very strong intensity. If the dwell time is set to long scale, considerable amounts of ions feeding into the channeltron might burn the channeltron, which is very expensive and sensitive. Last but not the least, the parameter “pass” number controls how many scans will be processed. In principle, if we obtain the mass spectrum with the more scanning times, the spectrum could be more decent because the backgrounds can be eliminated by statistical average. However, if we set the pass number too large, the water in the reservoir might be run out of. If we just 31 want to check that the beam is on for a deflection measurement, one pass should be fine (if the beam is too weak to see with one pass, then we won't be able to do anything with it anyway). That would let we check the spectrometer calibration so that we know where the peak is for a deflection experiment (and in this case, we want probably 10 points per amu). There are probably not very many cases where it would ever make sense to have more than 10 passes. If we collect a spectrum over a wide range, we probably want fewer points, so that it will finish before we run out of water--maybe 3 to 5 points per amu. The resolution of the spectrometer is only about 1 amu, so more points are only really useful if we are trying to fine-tune the spectrometer setting for a deflection experiment. In any case, there is always a trade-off between how much time you want to spend on the spectrum and how much resolution and noise you want. The purpose of having multiple passes is to throw away noise--10 passes of 1 second are better than one 10 second pass because some disturbance that corrupts the measurement will probably only affect one pass. Generally, it's a good idea to not have small numbers of points and short dwell times- -the UTI does not respond instantly to changes in the mass setting. There are two synchronous mass spectra of pure H 2O and D 2O clusters, obviously the background intensity is much lower in the large mass range compared with the small clusters, see Figure 3.6 and 3.7. 32 Figure 3. 6 Representative mass spectrum of pure H 2O clusters. Pass number = 5; 35-50 amu, 150 steps, dwell time = 100 ms; 50-200 amu, 1500 steps, dwell time 300 ms. Scanning time is approximately (150 * 100 ms + 1500 * 300 ms) * 5 ≈𝟒𝟎 𝒎𝒊𝒏𝒔 . However, the actual scanning time is about 2-3 hours because the time delay of amplifier should convincing UTI scan the right mass, etc. So if we scan 8 passes, the water probably could be used up and the mass spectrum could be contaminated by background because a typical one time loading could run about 4-5 hours. 33 Figure 3. 7 Representative deuterium water cluster mass spectrum. The nozzle and source temperature is 5 Celsius higher than H 2O clusters to obtain enough evaporation pressure in the reservoir. Pass number = 5; 21-45 amu, 240 steps, dwell time 150 ms; 45-143 amu, 980 steps, dwell time 300 ms; The mass range we scanned for D 2O is a bit shorter than H 2O to save scanning time(2 hours), because we consider the D 2O run out, we have to load D 2O again which is very expensive. 34 Chapter 4. Measurement of Cluster Velocity The produced cluster speeds can be predicted by the thermodynamic theory and are associated with the growth mechanism. Thus determining the velocities not only can enable us study the formation mechanism, but also it is fundamental for measuring the water deflection experiment. Especially, we developed a pickup method to assess the possible parent clusters by measuring the cluster velocities. To measure the velocity of cluster, different methods have been developed in the molecule beam experiments. Undoubtedly, time-of-flight (TOF) which measures the time of the neutral clusters flying through a fixed distance by employing the velocity definition v = d/t, is one of these techniques used for velocity measurement. Therefore, a slotted disk or mechanical chopper is often used to measure the time which can be derived from the period of chopper spinning, while the distance is already known or can be measured between the chopper and the detector. As we described in Chapter 3, a mechanical chopper is used in the experiment to module the water cluster beam and we can use Vollmer method, which measures the beam speeds distribution by using the time-resolved detection, to derive the beam velocity we will describe more details in the following sections. 35 4.1 MCS We already mentioned the discriminator pulses finally are also fed into a Multichannel Scaler (MCS), see Figure 3.1. A time-resolved cluster flux signal is recorded by the MCS in discrete time intervals which is typically around 50 to 200 microseconds per channel. As we already described in the previous chapter, the photodiode trigs the MCS first channel and the total channels can be set by the MCS software. For example, Figure 4.1 shows a typical MCS spectrum for the neat and mixed deuterium hexamer ((D 2O) 5D + and (D 2O) 6H + ion and we used a typical setting of MCS, channel width is 50 microseconds and the total channel number is 100. We take 2-20 mins to take these MCS spectra of neat and mixed clusters and the scanning time is dependent on the noise ratio and the signal intensity. For example, the mixed cluster typically has low signal and we need to take about 20 𝑚𝑖 𝑛 𝑠 ~ 50 𝑢𝑠 ∗100∗200000 𝑠𝑡𝑒𝑝𝑠 to get the decent the MCS spectrum, while the neat cluster MCS spectrum can be scanned in 2 𝑚𝑖𝑛𝑠 ~ 50 𝑢𝑠 ∗ 100∗20000 𝑠𝑡𝑒𝑝𝑠 . According to Figure 4.1, we also can find that the mixed hexamer shows higher noise ratio (17500/25000) than the neat one (2000/10000). This means the mixed cluster has low signal and the background will be more dominant than the neat cluster case. We also can observe the mixed cluster intensity by looking at the mass spectrum as in Figure 2.5. 36 Figure 4. 1 The two graphs show MCS spectrum of neat deuterium hexamer (top) and mixed with HCl (bottom). 37 4.2 Vollmer method In 1988, Vollmer [79] developed an equation to describe the time resolved cluster flux signal which is recorded as MCS spectrum, see Figure 4.1. The method implements that a typical supersonic beam velocity distribution is modeled as 𝑓 (𝑢 )=𝐶 𝑢 3 𝑒 −( 𝑢 −𝑣 𝑤 ) 2 , (4.1) where C is the normalized constant to satisfy ∫𝑓 (𝑢 )𝑑𝑢 =1, while w is the width of the distribution and v is the limited value when w=0. Based on Vollmer’s calculation, the MCS spectrum profile can be expressed as function of v and w 𝑆 (𝐿 ,𝑡 )=𝑆 0 −𝑆 0 { 1 2 [erf( 𝑣 𝑤 )±erf(𝑧 )− 𝑤 𝑣 𝑒 −𝑧 2 𝑔 ( 𝑤 𝑣 ,𝑧 )]}. (4.2) 𝑆 0 is the unchoppered the beam intensity and the formula valids for fully depleting the beam when the chopper is close, L is the distance between the chopper and the detector. erf(𝑧 )= 2 √𝜋 ∫ 𝑒 −𝛼 2 𝑑𝛼 𝑧 0 is a tabulated error function, where 𝑧 =− 𝑣 𝑤 + 𝐿 𝑤𝑡 , while ± corresponding to the positive and negative z seperately. 𝑔 ( 𝑤 𝑣 ,𝑧 ) is defined as 𝑔 ( 𝑤 𝑣 ,𝑧 )= 2 √𝜋 1+ 3 2 ( 𝑤 𝑣 )𝑧 +( 𝑤 𝑣 ) 2 (1+𝑧 2 )+ 3 8 ( 𝑤 𝑣 ) 3 (𝑧 + 2 3 𝑧 3 ) 1+3( 𝑤 𝑣 ) 2 + 3 4 ( 𝑤 𝑣 ) 4 . (4.3) Obviously, there are two important parameters w and v can be obtained by fitting the measured MCS spectrum to the equation. Thus, the mean velocity of cluster <u> can be calculated as <𝑢 > =∫𝑢𝑓 (𝑢 )𝑑𝑢 =𝑣 1+3( 𝑤 𝑣 ) 2 + 3 4 ( 𝑤 𝑣 ) 4 1+ 3 2 ( 𝑤 𝑣 ) 2 . (4.4) 38 4.3 Fitting procedure To extract v and w, we need to fit the measured MCS spectrum to the equation 4.2. To explain the fitting procedures [79], we will take neat deuterium hexamer MCS spectrum in Figure 4.1 as an example. First, we can simplify the equation by fixing S 0 as unity and thus S(x, t) is located in the range 1 to 0, which behaviors as falling edge of the beam flux. In addition, the equation 4.2 is derived from the assumption of depleting the beam completely when the chopper is close. However, the measured MCS spectrum shows the depleted signal which is from the background noise always exists and breaks the assumption, so first what we need to do is to subtract the depleted signal and to simplify the fitting procedure. Therefore, the measured MCS spectrum should also be normalized in the range 1 to 0 and defined as: 𝑀𝐶𝑆 𝑛𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 = 𝑀𝐶𝑆 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 −<𝑀𝐶𝑆 𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 > <𝑀𝐶𝑆 𝑢𝑛 −𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 >−<𝑀𝐶𝑆 𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 > (4.5) , where <MCS undepleted> and <MCS depleted> are the average value of the water cluster intensity in the un- depleted and depleted channels. However, MCS spectrum profile is dependent on the cluster speed and noise ratio, also the recorded channel number is not fixed, see Figure 4.1 and Figure 4.4. In other words, the depleted or un-depleted range might occupy different channel numbers, and we have to correct each MCS spectrum individually. There we use following procedures to circumvent the embraced range happens, which even work for very noisy MCS spectrum, also for longer channel MCS spectrum. To realize the normalization of MCS to the range 0 to 1, we need to prenormalize the MCS spectrum. Subsequently the measured MCS first was prenormalized by subtracting the average intensity of the MCS spectrum based on every channel and the range was set to 2. For instance, the middle graph of Figure 4.2 shows the neat hexamer was prenormalized to the range around [-0.6, 1.4]. Then we use a least squares method to fit a cosine function to the prenormalized MCS, see the pink line. Three parameters (amplitude, frequency and phase) can be varied to best fit with the prenormalized MCS. Then we set two ranges as 39 intro (0, Max(10, 1/8f +α/f)/channel width) and outtro (Min(E-10, (7/16f+ α/f)/channel width), Min(N, (5/8f+ α/f)/channel width)) to calculate the <𝑀 𝐶 𝑆 𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 > and <𝑀𝐶𝑆 𝑢𝑛 −𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 > and these two ranges are maked with blue and green solid points, where f is the frequency by fit, α is the phase. Based on the calculated <𝑀𝐶𝑆 𝑑𝑒𝑝𝑙𝑒𝑡𝑒𝑑 > and <𝑀𝐶𝑆 𝑢𝑛 −𝑑𝑒𝑝𝑙𝑒𝑡 𝑒𝑑 >, we can normalize the MCS by the equation (4.5) and the channels are normalized in the first 5/8 of the cosine period. Then the final step apply the Matlab function lsqcurvefit to extract the v and w in the equation 4.2. For the neat hexamer example, the parameters of v and w are equal to 1011.397070 and 134.640239 m/s, so the mean cluster speed of the hexamer can be calculated as 1037.818522 m/s. To compare the fitting results, we also shows the fitting results of the MCS spectrum of the mixed deuterium hexamer with DCl in Figure 4.3. For instance, the middle graph of Figure 4.3 shows the mixed deuterium hexamer was also prenormalized to the range around [-0.6, 1.4]. Similarly, we got the parameters of v and w equal to 909.085889 and 163.978448, so the mean cluster of the mixed hexamer can be calculated as 952.076519 m/s. Finally, we also shows the fitting results of the MCS spectrum of the neat deuterium hexamer with the longer channel number in Figure 4.4. For instance, the middle graph of Figure 4.3 shows the mixed deuterium hexamer was also prenormalized to the range around [-0.9, 1.1], this is because the un- depleted channel number are more dominant than Figure 4.3 and Figure 4.4. However, we still similarly obtained the parameters of v and w equal to 1012.293074 and 131.175408 m/s, therefore the mean cluster of the mixed hexamer can be calculated as 1037.372501 m/s. 40 Figure 4. 2 The extracted parameters of cosine function: frequency = 182.7842Hz, amplitude = 1.0636, phase = -0.0566, and the bottom graph is the final normalized neat deuterium hexamer MCS spectrum. 0 10 20 30 40 50 60 70 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 Final Normalized Neat Hexamer MCS and Fitted by Vollmer Function Channel Number Intensity(a.u.) 0 10 20 30 40 50 60 70 80 90 100 -1.5 -1 -0.5 0 0.5 1 1.5 Channel Number Ion Intensity(a.u.) Prenomalized Neat Hexamer MCS Spectrum 41 0 10 20 30 40 50 60 70 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 Final Normalized Mixed Hexamer MCS Spectrum and Fitted by Vollmer Function Channel Number Intensity(a.u.) Figure 4. 3 The extracted parameters of cosine function: frequency = 183.1381Hz, amplitude = 0.9848, phase = -0.0381, and the bottom graph is the final normalized mixed deuterium hexamer with HCl MCS spectrum. 0 10 20 30 40 50 60 70 80 90 100 -1 -0.5 0 0.5 1 1.5 Channel Number Intensity(a.u.) Prenormalized Mixed Hexamer MCS Spectrum 42 Figure 4. 4 The extracted parameters of cosine function: frequency = 157.7374Hz, amplitude = 1.2047, phase = -0.1169, and the bottom graph is the final normalized neat deuterium hexamer MCS spectrum scanned with longer channel number. 0 10 20 30 40 50 60 70 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 Channel Number Intensity(a.u) Final Normalized Neat Hexamer MCS with Longer Channel and Fitted by Vollmer Function 0 20 40 60 80 100 120 140 -1.5 -1 -0.5 0 0.5 1 1.5 Channel Number Intensity(a.u.) Prenormalized Neat Hexamer MCS Spectrum with Longer Channel 43 Chapter 5. Measurements and Discussion In Chapter 4, we described a time resolved flux technique based on Vollmer method for measuring water cluster velocity. We measured a complete speeds of neat water clusters (H 2O) n-1H + vs mixed water clusters with DCl ((H 2O) n-1D + , the measured ions) and also the other complete speeds of neat D 2O clusters (D 2O) m-1D + and mixed D 2O clusters with HCl ((D 2O) m-1H + , the ion measured), the measured cluster speeds are shown in the table 1 and 2 with size dependent. 5.1 Results Table 1: The column 2 and 4 show the mean speeds of neat (H 2O) n-1H + clusters and mixed (H 2O) n-1D + clusters, while the column 3 and 5 show the width of the corresponding size cluster speed, the speed distribution is described in formula 4.1. Since the capillary is directed basically perpendicular to the cluster beam, there’s no longitudinal momentum transfer from its effusing beam to the clusters, so mixed clusters 44 are slowed down with respect to the neat water ones. If there’s no fragmentation, the amount of slowing down will be trivial, for example: 𝑚 (𝐻 2 𝑂 ) 𝑛 ∙𝑣 (𝐻 2 𝑂 ) 𝑛 =[𝑚 (𝐻 2 𝑂 ) 𝑛 +𝑚 𝐷𝐶𝑙 ]∙𝑣 (𝐻 2 𝑂 ) 𝑛 𝐷𝐶𝑙 𝑣 (𝐻 2 𝑂 ) 𝑛 𝐷𝐶𝑙 = 𝑚 (𝐻 2 𝑂 ) 𝑛 𝑚 (𝐻 2 𝑂 ) 𝑛 +𝑚 𝐷𝐶𝑙 ∙𝑣 (𝐻 2 𝑂 ) 𝑛 = 18∙𝑛 18∙𝑛 +37.5 𝑣 (𝐻 2 𝑂 ) 𝑛 The coefficient 18∙𝑛 18∙𝑛 +37.5 is listed in the column 6 based on the different n. If the momentum conservation held, it would confirm that the mass spectrum reflects the actual neutral water size distribution, such as (H 2O) n+e - -> (H 2O) n-1H + and (H 2O) nDCl+e - -> (H 2O) nD + as we described in Chapter 1. In other words, if there is no fragmentation during the ionization, the measured mixed water cluster speeds should simplify follow the expected speeds. However, as always happens, what we ended up seeing doesn’t fit the above: the mixed clusters are moving much faster than predicted ones, see the table 1. Cluster size n (H2O)n-1H + <V> (H2O)n-1H + <V> width (H2O)nD + <V> (H2O)nD + <V> width Coefficient Expected (H2O)nD + <V> 3 1152.8564 ±5.9191 169.0723 0.5905 680.7239 4 1120.8144 ±13.6535 162.1501 1020.1082 ±25.1400 171.1734 0.6578 737.2903 5 1107.0044 ±10.2036 152.8002 1024.8670 ±13.1874 166.9691 0.7061 781.7031 6 1104.3858 ±15.4059 156.4980 1028.0352 ±11.1732 170.1491 0.7425 820.0152 7 1106.2990 ±24.4845 157.1076 1035.4983 ±17.5049 175.1585 0.7709 852.8059 8 1096.4377 ±13.9898 149.4830 1024.5944 ±17.3840 151.5503 0.7936 870.1263 9 1094.1420 ±12.7316 145.6898 1030.7812 ±25.4180 153.3649 0.8122 888.6855 45 10 1096.0809 ±14.1830 156.5738 1040.3727 ±12.3661 165.7581 0.8278 907.2975 11 1088.5498 ±8.4916 150.1358 1053.0664 ±24.6919 176.9695 0.8409 915.3966 12 1081.0189 ±5.6959 148.5345 1053.0024 ±20.5080 195.3671 0.8522 921.2756 Table 1: the measured speeds of the neat, mixed H 2O cluster ions, the expected speeds are shown in the different columns. Table 2: Similarly, the column 2 and 4 show the mean speeds of neat (D 2O) m-1D + clusters and mixed (D 2O) m-1H + clusters, while the column 3 and 5 show the width of the each specific size cluster speeds. Meanwhile, we should get the expected speeds for D 2O cluster as 𝑣 (𝐷 2 𝑂 ) 𝑚 𝐻𝐶𝑙 = 𝑚 (𝐷 2 𝑂 ) 𝑚 𝑚 (𝐷 2 𝑂 ) 𝑚 +𝑚 𝐻𝐶𝑙 ∙𝑣 (𝐷 2 𝑂 ) 𝑚 = 20∙𝑚 20∙𝑚 +36.5 𝑣 (𝐷 2 𝑂 ) 𝑚 However, the measured mixed (D 2O) mH + speeds are also much higher than expected speeds. See the table 2. Cluster size m (D2O)m-1D + <V> (D2O)m-1D + <V> width (D2O)mH + <V> (D2O)mH + <V> width Coefficient Expected (D2O)mH + <V> 3 1090.5019 ±6.3395 149.2699 969.9748 ±27.1531 174.9741 0.6218 678.0741 4 1061.9671 ±2.2344 138.7477 948.8819 ±12.3065 180.9244 0.6867 729.2528 5 1052.9105 ±1.8087 137.7287 934.1753 ±27.2945 158.9254 0.7326 771.3622 6 1043.4005 ±4.8674 135.5863 944.1259 ±14.5825 164.4060 0.7668 800.0795 7 1041.6942 ±6.9443 138.6601 952.4930 ±5.4935 162.6585 0.7932 826.2718 8 1036.3669 ±5.8471 136.9790 944.8466 ±14.9558 152.2790 0.8142 843.8100 9 1033.2854 ±8.2723 136.3760 945.9478 ±12.0714 158.8884 0.8314 859.0735 10 1030.5519 ±5.0771 135.6611 973.5265 ±6.6942 175.1541 0.8457 871.5377 46 11 1024.0912 ±7.6993 136.0713 970.5989 ±11.0956 172.7533 0.8577 878.3631 Table 2: the measured speeds of the neat, mixed D 2O cluster ions, the expected speeds are shown in the different columns. 5. 2 Discussions We also draw the velocities vs cluster size based on the data in the table 1 and 2, which are more straightforward for viewing, see Figure 5.1 and 5.2. Figure 5. 1 The summarized speeds of H 2O clusters are shown in the graph, the measured mixed H 2O cluster speeds (red curve) do not follow the predicted mixed H 2O cluster speeds (black curve). 47 Unfortunately, the both measured mixed cluster speeds from H 2O and D 2O clusters do not follow the expected speeds. How do we explain this phenomena? One convincing explanation is that the actual measured neat and mixed water clusters (which can be detected after the ionization) can be from the larger clusters which are neutral and original from the beam expansion or collision with DCl/HCl. Meanwhile, the previous research [43] already shows there exists fragmentation of water cluster during the electron bombardment, while the water cluster ionization threshold is around 12 eV from the reference. However, the parent clusters should have a number of different ways to decay into the various smaller fragments. To derive those parent clusters, we may still use momentum conservation law with admitting fragmentations: Figure 5. 2 The summarized speeds of D 2O clusters are shown in the graph, the measured mixed D 2O cluster speeds (red curve) do not follow the predicted mixed D 2O cluster speeds (black curve). 48 [(𝑚 +𝑚 ′)∗18]∗𝑉 (𝑚 )=[37.5+(𝑛 +𝑛 ′)∗18]∗𝑉 (𝑛 ) [𝑄 ∗18]∗𝑉 (𝑚 )=[37.5+𝑄 ∗18]∗𝑉 (𝑛 ) , where m, m’, n and n’ are integers, which can refer to (H 2O) m, (H 2O) m+m’, (H 2O) n and(H 2O) n+n’ clusters. From the above equation, the actual measured pure and mixed cluster size are m and n instead of the original cluster size m + m’ and mixed n + n’. If the momentum conservation is valid for cluster Q, we must have m + m’ = n + n’ = Q because we should apply the momentum law for the same cluster. This means that the pure cluster Q with velocity V(m) picked up a DCl with zero velocity, thus the measured mixed cluster speed should be retarded to velocity V(n). Therefore, velocities V(m) and V(n) are actually experimental measured from the fragmentation cluster neat m and mixed n which can be originally from parent cluster Q. Therefore, we can get the parent cluster Q for H 2O cluster: 𝑄 = 37.5 18 𝑉 (𝑛 ) 𝑉 (𝑚 )−𝑉 (𝑛 ) For D 2O, the parent cluster Q should be little bit different: 𝑄 = 36.5 20 𝑉 (𝑛 ) 𝑉 (𝑚 )−𝑉 (𝑛 ) In this way, we get a relationship between parent clusters and detected clusters. See following Q value table is calculated based on different m and n: 49 Neat Mixed m=3 m=4 m=5 m=6 m=7 m=8 m=9 m=10 m=11 m=12 n=4 16 21 24 25 25 28 29 28 31 35 n=5 17 22 26 27 26 30 31 30 34 38 n=6 17 23 27 28 27 31 32 31 35 40 n=7 18 25 30 31 30 35 37 36 41 47 n=8 17 22 26 27 26 30 31 30 33 38 n=9 18 24 28 29 28 33 34 33 37 43 n=10 19 27 33 34 33 39 40 39 45 53 n=11 22 32 41 43 41 51 53 51 62 78 n=12 22 32 41 43 41 51 53 51 62 78 Table 3: the original parent H 2O clusters are calculated and shown in the table. How do we understand this table? If we take an example in the table 3 of m = 12 and n = 11, the Q value is equal to 78. This means (H 2O) 78 can possibly decay into (H 2O) 12, while (H 2O) 78DCl can possibly decay into (H 2O) 11DCl. Of course it might decay into other type smaller cluster. For row n = 11 and n = 12, we can see that the Q table is identical, which may be due to the close speeds of mixed clusters. We should emphasize that for larger cluster n = 10 to 12, the speeds are less accurate and can be contaminated by the neat water clusters. From the mass spectrum in Figure 2.4, the mixed cluster and neat water cluster peaks can be distinguished, however, for larger clusters n=10 to 12, the resolution of quadrupole spectrometer is not ideal for discriminating the neat and mixed clusters by 1 amu. If we take one more example, Q = 30 appears five times. This means (H 2O) 30 can possibly decay into (H 2O) 5, (H 2O) 7, (H 2O) 8, (H 2O) 10, while (H 2O) 30DCl can be ionized into (H 2O) 5DCl, (H 2O) 7DCl and (H 2O) 8DCl. Correspondingly, the final detected clusters are ions (H 2O) 4H + , (H 2O) 6H + ,(H 2O) 7H + and (H 2O) 9H + , while for mixed detected ions are (H 2O) 5D + , (H 2O) 7D + and (H 2O) 8D + because OH or Cl is sheared during the ionization. Furthermore, (H 2O) 30 or (H 2O) 30DCl can decay into different fragments, the probability of decaying into 50 them are probably different. It would be interesting to know the probabilities, then we might know how much actual energy deposited into the water cluster during the ionization. One thing is clear that the electron carries most energy away and this energy is varied dependent on the cluster size. In other word, only partial energy of 70eV was used to ionize the water cluster. Also, column m = 3 shows Q is equal to 16, 17, 18, 19, 22 and so on. The value of Q is almost continuous which means most parent clusters can decay into the trimmer. However, column m = 12 shows Q equals 35, 38, 47, 78, etc. This means only some very larger cluster can decay into cluster m = 12, that is why we can obtain very strong intensity of trimmer and relatively low intensity of m = 12 in the mass spectrum, see Figure 3. 6. We also calculated Q table of D 2O: Neat Mixed m=3 m=4 m=5 m=6 m=7 m=8 m=9 m=10 m=11 m=12 n=3 15 19 21 24 25 27 28 29 33 34 n=4 12 15 17 18 19 20 21 21 23 24 n=5 11 13 14 16 16 17 17 18 19 20 n=6 12 15 16 17 18 19 19 20 22 22 n=7 13 16 17 19 19 21 22 22 24 25 n=8 12 15 16 17 18 19 19 20 22 23 n=9 12 15 16 18 18 19 20 20 22 23 n=10 14 19 21 24 24 26 27 28 32 33 n=11 14 18 20 22 23 24 26 27 29 31 Table 4: the original parent D 2O clusters are shown in the table. One more interesting aspect is that we measured the size of mixed H 2O clusters is from 4 to 12, while we obtained the size of mixed D 2O clusters is from 3 to 11. In principle, we can also measure the mixed D 2O cluster to 12 or even larger, but the resolution of the quadrupole cannot discriminate well for the larger mass as we described above. Thus, it is insignificant to measure even larger mixed cluster above 12. 51 Also, we find the mixed trimer of H 2O cluster, the possible reason is because the mass of (H 2O) 3D + is 56 and (D 2O) 3H + is 61. The background for H 2O cluster near mass 55 amu is always high (near 200K), so the noise may cover the mixed cluster signal. To compare the Q table of H 2O and D 2O clusters, we calculated the fractional difference (Q(H 2O)- Q(D 2O))/Q(H 2O): Neat Mixed m=3 m=4 m=5 m=6 m=7 m=8 m=9 m=10 m=11 m=12 n=4 0.25 0.375 0.4375 0.4375 0.375 0.5 0.5 0.5 0.5 0.6875 n=5 0.375 0.5625 0.75 0.6875 0.625 0.8125 0.875 0.9375 0.9375 1.125 n=6 0.3125 0.5 0.6875 0.6875 0.5625 0.75 0.8125 0.8125 0.8125 1.125 n=7 0.3125 0.5625 0.8125 0.75 0.6875 0.875 0.9375 1.0625 1.0625 1.375 n=8 0.3125 0.4375 0.625 0.625 0.5 0.6875 0.75 0.6875 0.6875 0.9375 n=9 0.375 0.5625 0.75 0.6875 0.625 0.875 0.875 0.9375 0.9375 1.25 n=10 0.3125 0.5 0.75 0.625 0.5625 0.8125 0.8125 0.8125 0.8125 1.25 n=11 0.5 0.875 1.3125 1.3125 1.125 1.6875 1.6875 2.0625 2.0625 2.9375 Table 5: the fractional difference of H 2O and D 2O original clusters The overlapped size for H 2O and D 2O clusters are from m = 4 to 11 and we find the overall of the original H 2O clusters are larger than D 2O clusters. The source temperatures of producing hydrogen and deuterium water clusters was different only by 5°C. In principle, the cluster size distributions of H 2O and D 2O clusters should be relatively same. We estimated the average cluster size of the H 2O cluster should be around 36 while D 2O should be even larger and around 39, see the following session. The estimation of the H 2O cluster size are matched to our Q tables attractively, but the calculated size of D 2O do not match to experimental Q table. We can conclude that the D 2O clusters Q table are different with H 2O Q table and estimated size, this is probably from the experiment imperfect. 52 5.3 Correction of speeds. Figure 5.1 and Figure 5.2 shows the speeds of water clusters decrease with the cluster size escalation. The speeds are extracted from fitting the MCS profile to the Vollmer method, we used the fixed length of our machine 0.7m. However, we didn’t consider the length of QMA part and we neglected the flight time of water ion spending in the quadrupole since this time is small compared with the total TOF. We can estimate the TOF that the ion spends in the QMA based on Figure 5.3, the ionized water cluster ions can be accelerated by the focus plate (-20V) to a large speed: 𝑣 =( 2𝑒𝑉 𝑚 ) 0.5 Therefore, we can estimate the time spent in the QMA: 𝑡 𝑞𝑚𝑎 = 𝑑 ( 2𝑒𝑉 𝑚 ) 0.5 =2.5µ𝑠 ∗𝑚 0.5 (5.1) , d is the length of QMA and V is the focus plate voltage. We found the t qma is dependent on the mass of cluster ion and the speeds of water clusters are corrected, see Figure 5.4. Figure 5. 3 The ionized water cluster ion will be accelerated by the voltage of focus plate, we neglected this time for measuring the speeds of water cluster. 53 Figure 5. 4 We used the equation 5.1 to correct all the water speeds and the summarized corrected speeds of H 2O and D 2O clusters are shown in the graph. 54 From the corrected speeds, we can find the red curve of the neat water cluster speeds are more flat than uncorrected one in Figure 5.1 and 5.2. For instance, deuterium water cluster speeds are around 1085 m/s and while the H 2O cluster speeds are around 1143 m/s. From the thermodynamic theory, we can check our neat cluster speeds accuracy. In principle, the speeds can be predicted as 𝑣 =√ 2𝑘 𝑚 ( 𝛾 𝛾 −1 )(𝑇 0 −𝑇 ) , k is the Boltzmann constant, 𝛾 is the specific heat ratio, m is the molecular mass, 𝑇 0 is the original temperature of water cluster and T is the final temperature of water clusters after the expansion. We kept the reservoir at 175°C and 180°C for the H 2O and D 2O clusters separately, and we can assume (𝑇 0 −𝑇 ) is same for deuterium cluster and H 2O cluster because 5°C can be neglected. Thus, 𝑣 𝐻 2 𝑂 𝑣 𝐷 2 𝑂 = √ 𝑚 𝐷 2 𝑂 𝑚 𝐻 2𝑂 Therefore, we can substitute H 2O and D 2O neat cluster speeds into this equation to check the accuracy of them: 1143 1085 = √ 20 18 1143 1085 ≈1.053, while √ 20 18 ≈1.054. We can conclude that the neat water cluster speeds should be accurate. We also calculated the new Q table based on the corrected speeds, we can find the results are not changed dramatically, the original parent clusters are a tiny bigger than the table 3 and 4. 55 H 2O Neat Mixed m=3 m=4 m=5 m=6 m=7 m=8 m=9 m=10 m=11 m=12 n=4 16 21 23 24 24 25 25 22 24 23 n=5 18 23 26 27 27 28 28 25 27 26 n=6 19 25 29 30 30 31 31 28 29 28 n=7 21 29 35 36 36 37 37 33 35 34 n=8 20 26 30 32 32 32 32 29 31 30 n=9 22 30 35 37 37 38 38 34 36 35 n=10 25 37 46 49 49 50 50 43 47 44 n=11 32 51 70 78 78 81 82 63 73 67 n=12 33 55 77 86 86 90 91 69 81 73 D 2O Neat Mixed m=3 m=4 m=5 m=6 m=7 m=8 m=9 m=10 m=11 m=12 n=3 14 16 17 17 16 16 16 15 15 15 n=4 12 14 15 15 14 14 14 13 14 13 n=5 12 13 13 14 13 13 13 13 13 12 n=6 13 16 16 16 15 15 15 15 15 14 n=7 15 18 19 19 18 18 17 17 17 17 n=8 15 18 18 18 17 17 17 16 17 16 n=9 16 19 19 20 19 19 18 18 18 17 n=10 24 33 33 35 31 31 30 28 29 28 n=11 25 34 34 36 32 32 30 29 30 29 Table 6: we also corrected the original parent cluster Q based on the corrected speeds. The top is the Q table for H 2O clusters and the bottom is the Q for D 2O clusters. If we compare with the Table 3 and 4, the results are not changed much. 5.4 Cluster mean size As we already noticed the H 2O parent cluster sizes are overall larger than D 2O cluster sizes, we can estimate the cluster size [80-81] based on the formulas: <𝑛 >=(𝐷 𝛤 ∗ 1000 ) 𝑎 𝛤 ∗ represents 56 𝛤 ∗ = 𝛤 𝐾 𝑐 ℎ 𝛤 is the critical parameter for the cluster size 𝛤 =𝑛 0 𝑑 𝑞 𝑇 0 𝑠𝑞 − 𝑓 2 (0<𝑞 <1) 𝑛 0 , 𝑇 0 and 𝑑 are the source water vapor density, the temperature of the nozzle and d is the nozzle diameter: 𝑛 0 = 𝑝 𝑘 𝑇 𝑟𝑒𝑠𝑒𝑟𝑣𝑜𝑖𝑟 𝑑 𝑒𝑞 =0.933 𝑑 𝑡𝑎𝑛𝛼 𝐾 𝑐 ℎ is the reduced scaling parameter 𝐾 𝑐 ℎ =𝑟 𝑞 −3 𝑇 𝑐 ℎ 𝑞 −3 𝑟 and 𝑇 𝑐 ℎ are molecular parameters. We summarized the parameters from the paper: Fitted parameters Molecule parameters Nozzle Calculated parameters D = 11.6 r = 3.19A d = 75um d eq = 75.03909585 a = 1.886 T = 5684K Α = 43° K = 3.8417E + 13 q = 0.634 The nozzle we used is not exactly the conical shape nozzle, but for estimation we can take the angle as half and that is about 43deg. Therefore, we can get the mean cluster size under the source conditions for H 2O clusters (T nozzle = 135°C and T source = 175°C and above assumptions for the nozzle) corresponds to 36. This size is in nice agreement with the sizes around 30 to 40 as shown in our table. For D 2O clusters (T nozzle = 140°C and T source = 180°C), we probably just need to change the vapor pressure by the ratio 1.001 57 and the heat vaporization by 1.021. Thus, we also can estimate the D 2O mean cluster size which should be around 39, this is not matched to our table. 5.5 Future work Basically, we developed a technique for assessing the water cluster original size by pickup momentum method. From the results described above, we find that the errors in determination of the speed of neat water clusters are less than for mixed water clusters. If the errors of the speeds of mixed cluster are large, the results of the water cluster original sizes may not be as accurate. Additionally, we already described that the resolution of QMA for discriminating the larger cluster is lower, especially for size 10 to 12. We can take n = 12 as an example, the neat (H 2O) 11H + cluster whose mass is equal to 199amu while the mixed (H 2O) 11D + cluster whose mass is equal to 200 amu. The two mass spectrum peaks of 199 and 200 amu can be overlapped, so the mixed cluster (H 2O) 11D + speed may be contaminated by the neat water cluster speed. Especially, (D 2O) 10H + and (D 2O) 11H + have very small error bars, this hints that the data might be already contaminated by the neat D 2O cluster speeds. The mixed cluster speeds error bars are large, there are several possible reasons: 1, The capillary configuration can bring uncertainty of mixed cluster speeds. More specifically, the capillary is not so strong and easily bended during assembling the source, so the capillary may not be absolutely perpendicular to the beam. In the future, if we can design a stronger “L” shape capillary which, additionally, can be mounted in a fixed position, we can possibly get more accurate data. 2, As we described in Chapter 2, we put the capillary about 5mm in front of the nozzle, which is a compromise way to avoid mixed cluster acceleration by the gas expansion without sacrificing the mixed water cluster beam intensity. By altering the experiment so that the capillary can be put in a detector chamber, thus the distance between the 58 capillary and nozzle may be around 50-60 mm, we can completely get rid of possible acceleration of mixed water clusters from the gas expansion. However, then we should consider new ways to increase the mixed cluster intensity. The QMA that we used is a UTI 100C with a mass range of 0-300amu which is not a high resolution mass quadrupole. If we need to extend to higher masses of water cluster, we should replace the current quadrupole with a high resolution one, which may also improve our data accuracy of the size between cluster size ranges from 3 to 9, and reduce the errors of the mixed water cluster speeds. Last but not the least, the evaporated water molecules can be from the ionization process and also can be from the attachment of the acid molecules. In other words, evaporation of water molecules can be induced by acid attachment and electron ionization, both giving the same effect. We can expect the ionization to be strongest effect, so changing the dopant or the electron energy may give evidence to discriminate which effect is dominate. The second possible method is to expand some similar gas where the magic numbers of the neutral and the ions are known to be different. This will tell us if the ionization is likely to induce the effect we see. Although we tried other gases which are supposed to be picked up easily (CO 2, ammonia and CH 3OH) and none of them produced mixed cluster when sprayed at the beam from the capillary. In the future, it is still worth trying different other possible gases including HBr, CH 2O, NO 2, NO and SO 2 so on. If we can dope a light gas molecule, the mixed water cluster may have stronger intensity because there is less rebounded force which may reduce the beam intensity. 59 Chapter 6. Design of Source with Loading System for Lithium Clusters In this chapter, we introduce background why we studies lithium clusters and then we show an updated loading system that enables us obtain the neat pure lithium sample in the high temperature. Finally, we describe the different sources were designed and used, a final version were used and we got the rudimentary results. 6.1 Recent theory The recent theories [82] based on the observation that for certain alkali cluster sizes (n = 14, n = 23, etc.), electronic term has two equilibrium positions (prolate and oblate) that have comparable in depth, producing an almost symmetric double-well potential, see Figure 6.1. Under such potential, a quantum oscillation could be arising between oblate and prolate configuration, this is oppositely to previous viewpoint that for this specific size clusters are composed by mixture of two static isomers. 60 This exotic property can be an analogy to the well-known Kekule structure: As we know from above, the deformed clusters usually have oblate or prolate configurations that often have different the effective oscillator strength. For instance, the prolate configuration should persist the ratio 1:2 of the amplitudes of the photon absorption peaks. However, the experimentally observed picture is quite different. This could be explained by the superposition of prolate and oblate configurations. At small shell filling, the prolate structure is dominant and the ratio is close to 1:2. On the other hand, close to half-filling the amplitudes of the prolate and oblate configurations are almost equal, then both of prolate and oblate configurations will yield the same peaks. As a result, near a half-filled shell, one should expect both peaks to display similar intensities, as observed experimentally. Consequently, this quantum mechanical characteristic could also bring up a peculiar isotope effect: the strength of the collective electronic resonances is dependent on the atomic mass. This interesting Figure 6. 1 Symmetric double-well potential of certain cluster sizes (n = 14, 23, etc.) Figure 6. 2 Kekule structure. 61 phenomenon also could be pursued experimentally. To search such effect, lithium which has isotope atomic mass 6 and 7 in alkali metals is probable the best candidate because Li 6 and Li 7 for cluster Li 23 will cause a noticeable shift of value B = I 1/I 2-1/2 about 14%, while the shift for potassium is only 5%, where I 1 and I 2 corresponds to the collective electronic resonances strengths of different axis, more details can refer the paper. 6.2 Loading system for lithium clusters Lithium clusters, which has very possible fascinating properties as we described above, are few investigated due to experiment difficulties. Studying them is important and challenging for both experimental and theoretical interest. The first difficulty is to prepare an extraordinarily fresh environment and delicate sample to avoid the contamination of lithium. To build such clean surrounding, it is significant to know the basic properties of lithium in different condition [83]: 1. Although the lithium is least reactive in the alkali metal, molten lithium in high temperature can react vigorously with concrete and other ceramic materials rapidly in moister environment. 2. Lithium is most sensitively with water evaporation while dry nitrogen can keep inert up to 160°C. Moreover, dry air, CO 2 and oxygen can even keep inert up to 250°C. 3. Lithium can react with glass and the reason is because the lithium can steal the oxygen atom from the glass. 62 The product of lithium oxide and lithium nitride also possess high corrosive properties, so the less contamination can enable us have more cleaning circumstance. Previously, L. Wöste group [84] designed a special oven built with titanium-zirconium-molybdenum materials and lithium was preloaded in a boat in atmosphere. During assembling the source, lithium is possibly already contaminated by the air. Furthermore, K. Bowen group [85] loaded lithium in an Ar atmosphere with a glove box and the loaded source can be capped and easily transferred to the vacuum chamber. Although Bowen’s method can obtain a less contamination than Wöste, there are still some possible dirty gases (especially H 2O), which can be released from the oven wall during the heating up. Thus, lithium still can be polluted in the high temperature. For example, we can see the mass spectrum from previous paper of Wöste, the rising background is possible due to the residue which can hinder investigating the properties of large clusters, because the background intensity can cover the cluster intensity. In order to verify the theory prediction stated from above, we need to produce the lithium clusters that could be generated from the seeded supersonic nozzle source already described by de Heer [87] in his thesis. In our group previous work, most focused on sodium clusters prepared. Although previous Figure 6. 3 The rising background should be due to the contamination of residues, Li 23 peak shows almost covered by the background residues [86]. 63 members also tried to prepare lithium clusters, they do not provide sufficient experience for us. During the process of preparing lithium clusters, we also find some difficulties due to the high reactive property of lithium. First, I will illustrate our experimental setup that is constituted with several components. A loading system was already designed to transfer the melted metal to the cluster beam source, see Fig. 6; the detail of this loading system is described in Ref. [88] Figure 6. 4 Loading System for supersonic source. In this note, we updated our previous loading system which enables us loading sodium conveniently, the detail of this loading system is described in reference. The original idea of this loading system is that the alkali metal in an ampule is baked about 2 hours until the metal totally become liquid, then the blade of plunger can break the ampule and the molten alkali can be transferred to the source with the ultra- pure Ar gas (99.999%, Gilmore). However, to our best effort, we failed to find the company produces the glass ampule for lithium and the behind reason, as we described above, might be due to strong corrosion properties of lithium can react with glass. (Thus, even we assume the ampule of lithium is available, it still may contaminate lithium during the heating up the loading system.) 64 Although no lithium ampules seems not playing an important role during the whole experiment, it actually brings us difficulties for loading system. For instance, we bought the lithium bars sealed in the mineral oil and they were pre-cleaned in the hexane. Then we put the cleaned lithium in the vessel and pumped the vessel immediately to 10^-3 torr. At this moment, the lithium bar was still in good situation with shiny color, we found that the lithium can keep un-oxidized in 1-2 mins [89]. However, even we filled the vessel with protection Ar gas, the lithium still were gradually becoming dark during the baking and the oxidized lithium can form a hard core that prevents the lithium melted. Although we tried many other methods, we got the same annoying results. To solve the problem, we updated our loading system from Figure 1 (a) to (b). Figure 6. 5 (a) Heated alkali loading vessel employing a plunger to break molten metal in ampule. Fig. (b) Modification for use with lithium, with a long drop tube used to insert clean lithium bar into a hot prebaked overpressured vessel. The body of the vessel is heated by a mineral-insulated strip heater, and the liquid feed tube and the bellows valve (Swagelok) by a tightly wood rope heater. The vessel cover, the rectangular pyrex window, and the lid of the tall tube are sealed with viton o-rings. 65 From the Fig. 1 (b), we can see the plunger is replaced with a stainless steel tube whose length is about 15” and the diameter is 1”. At the top of the tube, it has stainless steel cover which is held by 4 screws. In this case, the vessel behaviors like an Argon Trap. Now we also need to change our experimental procedures slightly: primarily we bake the vessel out about 2 hours at 250°C rather than putting lithium bar first, the bottom vessel is around 300°C. During the baking, we always repeat the procedure for 8 times-pumping the vessel and then refilling with argon to remove the dirty gas including water evaporation, which is the fatal for the lithium as we already described. After 2 hours baking, then the vessel is filled with refresh pure Ar gas and the pressure is kept above the atmosphere to keep the air enter the vessel. This moment the whole system has already been cleaned and baked in Ar environment for 2 hours, the most water molecules should be removed. Next, it is securely that we can open the top cover, put the cleaned lithium bar into the vessel and close the cover from the top of the long tube promptly. Immediately the next step is that we need to close the argon gas valve and start to pump the vessel immediately to remove some air possibly coming into the vessel when we opened the cover, although the Ar pressure is kept above 1 atm. This step can be repeated several times to remove the air completely and we would emphasize that even there are some air sneak into the vessel, they can be diluted by the argon and have no severe influence on the lithium bar which was just threw in to the vessel. Surprisingly, we found that the lithium bar is melting with decent color after 30seocnds or 2mins which is dependent on the bottom vessel temperature. We would emphasize that the bottom vessel temperature is critical for melting the lithium bar: the room temperature lithium bar dropped on 300°C vessel bottom, which is similar to an ice bar fallen on a hot plate. In other words, if the bottom vessel’s temperature is not high enough, the lithium still can be possibly contaminated because it needs longer time to absorb enough heat to melt, while there is no vacuum system perfected seal. When we see the lithium bar totally melting into liquid, then we could start to load the lithium with the pressure difference of the vessel and source (not shown)! Last but not the least, even the lithium was accidently oxidized a bit and form very 66 thin layer on the bar, the pure lithium still can flow out and the dirty layer will be filtered by the filter mesh. It is worth to deliberate that why this design can work. Actually, the function of this design is more or less same with the ampule method. Considering that the glass ampule could prevent the alkali metal invaded by the dirty gas released from the stainless steel or the atmosphere air (no system can have perfect seal). After 2 hours bake out of the vessel and with fresh argon gas, the dirty gas density is almost zero. So we can feel safely to break the ampule. The updated system first is baked out about 2hours to remove all the dirty gas, then we can feel safely throw the lithium in it. Because the vessel is already in the high temperature, so the lithium can be melted immediately. The long tube has two functions: 1, It can prevent the air contact to the lithium effectively. 2, Keep the top cover temperature not very high to operate easily with hand. More specifically, the vessel bottom is about 250-300°C while the top cover is only about 40°C. 6.3 The different sources designed for lithium clusters 6.3.1 Typical seeded supersonic nozzle source Figure 6. 6 The typical seeded supersonic source with wrappings and protected tank for sodium clusters. 67 In Figure 6.6, the molten alkali are loaded into a stainless steel supersonic source which is described in details by de Heer’s thesis [87]. The reservoir part is typically kept at 650°C for sodium while we need higher temperature around 950°C (or even higher) to get enough evaporation pressure for lithium clusters. However, after tried with a bunch of sources for several experiments, we found that this type source can easily get burned during the source heating up, see Figure 6.7. The second problem is that the heaters for the source body and nozzle are also not stable, and these heaters can be easily broken down in the high temperature. We summarized the possible reasons: 1. The heating mantles for the source body and nozzle are alumina cylinder (ceramic) threaded on outside, a double stranded 0.015” tantalum wires are wrapped on the ceramic groove. The ceramic cylinder are hard machined to perfectly match to the source, there exists a large gap between ceramic cylinder and the source body. Consequently, this gap can cause ineffectiveness thermal conduction from the heater to the source, “hot spot” can easily form and destroy the stainless steel source. Figure 6. 7 We heated the typical seeded supersonic source to 1000°C with the wrapping materials graphite and tantalum films. 68 2. To fix this problem, we wrapped different materials to fill up the gap between the heater and source. Thus, the thermal energy from the heater can be distributed more uniformly, which can correspondingly enhance the thermal conduction. We tried three kinds of materials including stainless steel, graphite and tantalum films. We found a caution that wrapping materials still can be a serious problem to the source. If we use stainless steel and graphite are thin and thermally isolated, the temperature in the wrapping or shielding material itself can get much higher than intended. These can be disastrous. For example, if we use stainless steel, this wrapping material can melt and drop onto our oven, which is already thermally stressed, and lead to even more terrible things. In general, if we use the wrapping approach, try to keep the melting point of the wrapping/shielding material as high as possible. Tantalum may be appropriate because the foil is easily workable, can be spot welded, and its thermal conductivity is reasonable (about twice that of SS316). Also, if very tight fits are held, we should consider the thermal expansion properties of various materials. We found when we use graphite for wrapping film for filling the gap, the source is burned more severally and see Figure 6.7. The carbon atoms might participate with stainless steel in the high temperature. In the other aspect, the baffle plate which prevents the loaded alkali to block the nozzle directly was welded in the middle of source wall, and this welding part is also possibly the weak part of the source in the high environment. 3. For the heaters are also easily broken down, the possible reason is that during the source heating up, the tantalum wires are gradually oxidized based on that we found the wires are very fragile after heating up, although they are in the vacuum environment around 10^-4 torr. If the wires are partially oxidized in certain part, the remaining un-oxidized wire is obviously thinner and easily even hotter than the normal wire. The “hot spot” can happen and break the wire. 69 6.3.2 Upgrade to a TZM source Based on the possible reasons we described in above, a new TZM source was designed to deal with these drawbacks. We changed several belongings as following, the old stainless steel source drawing and new TZM source drawing are shown for comparison, see Figure 6.8 and 6.9: 1. The stainless steel source was altered to TZM materials which is the alloy composed of titanium- zirconium-molybdenum, and the melting point is around 2600°C and much higher than stainless steel melting point around 1350°C. Therefore, we expected the TZM source can sustain longer time than stainless steel source in the high temperature. 2. We changed the position of the baffle plate which is supported by a TZM rod welded on the bottom plate that is much thicker than the fragile side wall. Changing the baffle plate to the new position which was welded on the bottom plate, the source should be stronger than stainless steel configuration. Figure 6. 8 To compare with the new designed TZM source, we also show the original seeded supersonic source, the heaters and wrappings are not shown here. 70 3. The most remarkable change was that the new TZM source has outside diameter 2” which is half inch larger than the old stainless steel source, and the larger OD of TZM source can provide enough space for drilling 20 holes with diameter 0.190”. Therefore, we can insert the commercial ceramic rods with four bores with parameters as 0.825"(𝑙𝑒𝑛𝑔𝑡 ℎ)∗0.109"(𝑂𝐷 )∗0.020"(𝐼𝐷 ) can be bought from Coorstek Company into the drilled holes, see Figure 6.10. The advantage of this design is that we put the heaters in the source, thus the thermal conduction should be more efficiently and the heat should be distributed more uniformly. Meanwhile, this design is also for removal of the wrapping, which might be disaster of our source during the heating up as we described in above. Figure 6. 9 The schematic drawing of the new TZM source, we changed the position of the baffle plate which may weaken the wall in the high temperature. Also, the heaters are in the source body and the wrappings are removed. 71 Figure 6. 10 The new TZM source was drilled with 20 holes which enable us insert 20 ceramic rods with four bores for the tantalum wires. The new heater is around 8ohm and can be covered by the ceramic paste. 4. We string the ceramic rods with the tantalum wires following the drawing in Figure 6.11, the left graph is shown the top cross view of the TZM source while the right one is shown the bottom cross view. This configuration also enables us seal the top tantalum wires strikingly by ceramic paste, since there is no crossed wires connected between two ceramic rods. The ceramic paste cannot seal the crossed wire connections in the bottom as well as the top tantalum wires. Figure 6. 11 We made the heater with the method as shown in the illustration, rod 1 and rod 2 are the lead wires. In this way, we keep some distance between 1 and 2 to avoid the hot spot. Also, the wire comes from the rod 1 and rod we never bend them, we enhance the lead wire by bending the tantalum wires from rod 3 and 4. 72 Therefore, we keep the tantalum wires crossed over the ceramic rods at the bottom of the source to protect the tantalum wires, since the lowermost of the source has lower temperature than the top. The ceramic paste will be stabilized and solidified during the source heating up. One more important thing is that we introduce lead wires out from ceramic rod 1 and 2, and we load the separate independent tantalum wires into the ceramic rod 3 and 4 to enhance the thickness of lead wires. After we tested the heaters for a number of times, we found that the lead wires are always the weak parts and easily get burned. Consequently, we preserve appropriate distance between two lead wires to prevent the hot spot. Furthermore, we make sure the single wires from rod 1 and 2 are not bended or folded to avoid the stress in the wires since the stress can also cause the heater wires broken in the high T. The five stranded wires grouped of two lead wires are then insulated by ceramic beads, which can be easily contaminated or metalized during the heating up of the source. The metalized ceramic beads could cause the short circuit and break the heaters, even the source. To circumvent such problem, we wrap a thin 0.015” tantalum film out of the ceramic beads to avoid them being invaded from the contaminations effectively due to the oil from DP, the oxygen molecules and unknown residues in the source chamber. 5. We also designed a similar heater for the source nozzle and see Figure 6.12, the difference is that the nozzle heater designed as a hollow cylinder with smaller size: length is 0.5”, the inside diameter is ¼’’ and the outside diameter is 0.625”. There are 16 holes in the nozzle heater and each hole with diameter 0.09”. Ceramic rods with parameters 0.6”(𝑙𝑒𝑛𝑔𝑡 ℎ)∗0.085”(𝑂𝐷 )∗ 0.024”(𝐼𝐷 ) from Coorstek Company are used and we still string ceramic rods by the 0.015’’ tantalum wires with the similar method as source body heater. 73 However, one major issue of this source with TZM materials is the welding problem, we found there exists few welding company working on TZM materials. In this design, finally we delivered our source to the Advanced Technology Company (Pasadena, CA) that used electron beam method welded our source, the welded parts are marked with red color in Figure 6.9. However, the frustrated aspect is that the welded strength is not strong enough since the penetration depth of welding part is extraordinary shallow. When we just received the new welded source, we found there exists leaking from the bottom tube. After the leak of the bottom tube was fixed, we discovered that although TZM materials can sustain in the high temperature, all the welding parts are very sensitive to the temperature variation. More specifically, if the temperature of the TZM source is reducing very fast or increasing rapidly, these welding parts became fragile and also can cause leaking. The discouraged work is that this expensive TZM source took us plenty of efforts (one and half year for designing, purchasing, machining and welding), but there are several parts of the TZM source still leak. Furthermore, not only the welding parts are fragile, but also the actual physics properties of TZM materials are awfully fragile in the high temperature (like glass) and also very sensitive to the oxygen above the 200°C. Figure 6. 12 We designed the similar TZM heater for the nozzle, and only difference is that this heater is dependent of the nozzle. 74 6.3.3 Modified French Source We also tried to modify a French source to our machine, see the Figure 6.13. This French source was delivered from previous group [90] and it was made of TZM source with nozzle orifice 100um and can work with a special cooling water system. To adapt our source chamber, the bottom of the source can be sealed by a commercial stainless steel Swagelok nut with ferrules, see the drawing. A ¼” stainless steel tube (not shown in the drawing) is welded on the center of the nut, accordingly we still can connect the stainless steel tube to the previous feedpipe which is for the carrier Ar gas and loading lithium. Also, we designed a special baffle plate (not shown) in the middle of the source to avoid the lithium loaded to the nozzle directly. However, the big problem of this source is also the leaking issue that if we heated the source around 250°C, the leak will happen around the ferrule parts due to the difference of the thermal expansion coefficient between TZM(5.3 × 10^-6 m/K -1 ) and stainless steel materials(16 × 10^-6 m/K -1 ). Meanwhile, we also designed the new source and nozzle heaters for the modified French source and they will be used for the final version of the designed source, we will describe more details of them in the following session. In the meantime, we also modified water cooling water jacket cover with thinner wall and protected tank as a whole piece for the modified source, which are not dramatically different with before. We will keep those designs for the following new stainless steel source. 75 Figure 6. 13 The old French source was adapted to our machine, however, the Swagelok nut and ferrules part had the leak due to the thermal expansion difference between TZM and stainless steel materials. 76 6.3.4 A newly designed stainless steel source Based on the unsuccessful experiences of TZM materials, we finally get back to stainless steel source again, but we design a new source, see Figure 6. 14. There are two main improvements for this new source, the first one is that we design this source more compact with less welding part. In other words, we can see there is a single piece of source body but the welding area are much reduced compared with TZM source. Also, we need to emphasize that the welding of stainless steel penetration depth is much deeper and stronger than TZM welding and our machine shop’s technique of welding stainless steel materials was also very mature. The second change is that we alter the wall thickness of this source which is much thicker than previous designed sources, especially the nozzle part which needs to be kept at 100°C higher than source body. The nozzle wall is about 0.25” thick which is much stronger than previous stainless source. The tip of nozzle with the hole diameter 100 um, which is calibrated as SS Gland type (SS-4-VCR-3-CAL), is bought commercially from the Lenox Laser Company. Figure 6. 14 The final version of stainless steel source is designed and shown in the drawing. 77 Figure 6. 15 The final version of stainless steel source which is adapted to our machine, and the new heaters for nozzle and source were designed which also can work more stable than previous ceramic groove heaters. 78 This new source needs to be adapted to our machine and show the schematic drawing of the new source with the water cooling jacket. Basically, we need to make an alignment for the new SS source by changing the supporting legs length and we also need to consider the length of source to make sure the shielded tank can cover the source. Two new TZM heaters, the ceramic rods embedded in the drilled holes and tantalum wires covered by the paste, are designed to the new source. Since after the heating up, those ceramic rods in the drilled were awfully rigid to take it out due to the thermal expansion and solidified ceramic paste. If we need to change the wires or clean the source inside, this is not very convenient and also the cleaning solution may contaminate the tantalum wires. To deal with this problem, we change the previous designs for TZM heaters and the heater for the source body is shown in the Figure 6.16, the cross views can be seen in the drawing while the side view is shown in Figure 6.15. Since this new heater is now separate and independent with the source body (the source OD is 1.25” and the ID of the heater is 1.3”), therefore we keep a gap about 0.025” between the heater and source to avoid the crash due to the thermal expansion. This design enables us remove the wrapping materials and is also convenient for cleaning the source separately without possible contamination from the ceramic paste. Moreover, a shoulder for the body heater is designed to avoid the heater sliding off during assembling Figure 6. 16 The heater for the source is designed with a “shoulder”, which can prevent the heater sliding on the source. The twenty holes are opened a bit to avoid the ceramic stuck into the hole after the heating up. 79 the source to circumvent the potential short circuit from the bottom lead wires. The last change is that we open the holes a bit to avoid the ceramic rods stuck in the holes and then we stringed the tantalum wires with similar method as we described, see Figure 6.16. For nozzle heater, it is very similar to the source body heater but there is no special shoulder for it. Equally, we put a ring on the bottom of the nozzle heater to avoid the heater sliding off. There is not enough room for designing a shoulder for nozzle heater. Of course, we also changed the OD (1.2”) and ID (0.785”) for adapting the heater to the nozzle diameter, so we keep a similar gap between the nozzle and the heater to circumvent the crash due to the thermal expansion. The ceramic rods are also reformed into different geometry as 0.825"(𝑙𝑒𝑛𝑔𝑡 ℎ)∗ 0.109"(𝑂𝐷 )∗0.020"(𝐼𝐷 ). See Figure 6.17. Figure 6. 17 The drawing of the nozzle heater which is similar to the source heater but without the “shoulder”. 80 6.4 Results The new stainless steel source and TZM heaters were tested and we acquired the rudimentary promising results. After we loaded the lithium into the source, then we spent about 10 hours heating the new source up to 1050°C (nozzle) and 950°C (source body) in the first day. There are two reasons why we heated the source very gradually: 1, if we heated it the source rapidly, the heat may be accumulated and cannot distribute in the source uniformly and immediately, and thus the hot spot may be formed which is very dangerous to our source. 2, the lithium in the source if they are heated rapidly, the liquid of lithium may also jump around and could block the nozzle. Subsequently we kept the source at the temperature 1050°C (nozzle) and 950°C (source body) overnight to the next day. Since the experiment lasted about 36hours without burning the heater in the first trying with the new stainless steel source, the source was in this actual high temperature (1050°C) for 24 hours. Finally we stopped the experiment because we still cannot detect the lithium cluster beam by our detecting system which is composed of QMA, UV-lamp and PMT, see Figure 6. 18, which shows our machine is built with 2.5 meter differentially pumped vacuum chambers. We kept the carrier gas Ar 99.999%(ultrapure from Gilmore) at 6 atmospheres and in principle, the neutral lithium clusters are produced from supersonic expansion of lithium vapor with argon flow through the 100 um nozzle. Meanwhile, the temperature of the nozzle should get 100 C more than source temperature to avoid the clog of the nozzle hole. Then, the ideal cluster beam pass a molecular beam skimmer (0.4”) heated to approximately 450 C. After this, the beam will fly through collimated aperture in series of chambers. In principle, the produced clusters should get through the detector aperture, and then are ionized by UV light. The UV light could be produced by an Hg-Xe arc lamp (Oriel Instruments) and passed through a UV filter transparent for the light wavelength between 250 nm and 400nm. In this photon energy regime, the clusters can be ionized almost without having fragmentation. The resulting ions enter a quadrupole mass analyzer (QMA), and the selected size of the clusters will be detected by a 81 photo multiplier tube. The signal pulse can be converted to digital signal by the NI box and then can be collected by the Labview program “qma 0. 2.8vi”. Although we cannot detect the lithium clusters directly by the PMT system yet, we indeed observed a considerable amount of lithium coming out of the new source after we dissembled the chamber, see Figure 6.19. The top picture was taken after we just opened the chamber, there were a lot of black lithium powders on the cooling water jacket. After two days, those powders gradually reacted with oxygen and probably formed a layer of Li 2O. We also saw a lot of oxidized lithium on the skimmer which was not blocked since we kept the skimmer temperature at 450°C, see the middle picture. One thing we need to emphasize is that during the high temperature of the nozzle (1050°C), the nozzle has certain probability been blocked and opened back and forth. This was still more dependent on the fortune, and one possible way to fix the problem is to increase the carrier gas pressure (maximum 700 kPa) to circumvent the block of the nozzle. Figure 6. 18 The schematic drawing of our metal cluster machine. 82 After this promising experiment, we run the second experiment to test the new source and heaters. The new SS source was kept in the same high temperature about another 24 hours. We found the heaters and source still sustained, this is the best attempt we got. Previous designed sources and heaters were easily broken during the source heating up. Although we still failed to observe the signal of lithium clusters from the detector, the heaters and source already run about 48hours in 1050°C and considerable lithium powders are observed in the chamber including the cooling water jacket, skimmer and chamber wall. If we consider the time of heating up, the heaters and source have worked about 68hours without breaking down. After we run the second heating experiment, we found the tantalum wires are a little bit fragile and heaters were slightly dark but not worse as previous ceramic groove heaters. This is probably the critical point for changing the wires and ceramic insulating materials. The fragile wires means the tantalum wires probably already partially oxidized from the oxygen because the ceramic paste cannot seal the tantalum wire perfectly. To summarize, we designed a new stainless source with TZM heaters which can work in high temperature (around 1000°C) for 48 hours without breaking up. Meanwhile, we observed substantial amount of lithium powders were ejected out from the source, although we cannot detect the lithium beam. 83 Figure 6. 19 After the heating up the source, the lithium particles are observed on the water cooling jacket and skimmer. After two days, the lithium was oxidized and formed a white layer. 84 6.5 Future work Lithium clusters, which have the potential to display fascinating properties, are not extensively investigated due to experimental difficulties. Investigating them is significant and challenging for both experimental and theoretical interest. Indeed, as we made substantial efforts on producing the lithium cluster beam, we successfully obtained the lithium particles expanding out from the new source but failed to observe them by our detection system. There are several possible ways for future work on lithium clusters detection. The next step is to perfect the source alignment. Specifically, we can use a quartz film thickness monitor (INFICON XTC) to detect the beam flow rate in the precollimator-chamber, then we can continue with alignment for the nozzle with the aperture in the precollimator chamber. After this part is finished, we can apply the same alignment procedures for the remaining chambers. Meanwhile, we undoubtedly need to design a dozen of sources and heaters with stainless steel and this is very time consuming and expensive! The reason is that the heaters and sources typically can sustain 48hours in the high temperature as we found, then they gradually oxidize and contaminated by oil from the diffusion pump and other origins. Therefore I would change the source, heaters, ceramic tubes and tantalum wires for two runs. This is tedious, but in principle it should stabilize the source and heaters. Otherwise, we can consider a way to seal the cover tank more nicely, but we think this might be also extremely difficult. We also can try to design a stainless steel source with even larger volume, which should have even thicker wall to make the source stronger. To match such a large volume source, we also can design new heaters with a larger ceramic diameter of the bores, so we can use double stranded wires to make the heaters. Meanwhile, we probably require a high-power supply to provide enough power to the new 85 source. In this way, we may even obtain a large volume source and sturdier heaters in the higher temperature (>1050°C). The reason why we need the larger volume source is that the surface plasma experiment typically requires around 15 hours. 86 References: [1] W. D. Knight, K. Clemenger, W. A. de Heer, W. A. Saunders, M. Y. Chou and M. L. Cohen, Phys. Rev. Lett. 52, 2141 (1984) [2] S. S. Lin, Rev. Sci. Instrum. 44, 516 (1973) [3] G. C. Bond, C. Louis and D. T. Thompson, Catalysis by Gold (Imperial College Press, London, 2006) [4] A. Halder, A. Liang and V. V. Kresin, Nano Lett. 15 (2), 1410 (2015) [5] C. E. Kolb, J. T. Jayne, D. R. Worsnop, M. J. Molina, R. F. Meads and A. V. Viggiano, J. Am. Chem. Soc. 116, 10314 (1994) [6] H. R. Carlon, Infrared Phys. 19, 549 (1979) [7] H. R. Carlon, J. Phys. D 17, 1221 (1984) [8] N. Pugliano and R. J. Saykally, Science 257, 1937 (1992) [9] R. J. Saykally and G. A. Blake, Science 259, 1570 (1993) [10] K. Liu, J. D. Cruzan and R. J. Saykally, Science 271, 929 (1996) [11] J. D. Cruzan, L. B. Braly, K. Liu, M. G. Brown, J. G. Loeser and R. J. Saykally, Science 271, 59 (1996) [12] K. Liu, M. G. Brown, J. D. Cruzan and R. J. Saykally, Science 271, 62 (1996) [13] M. Eigen, Angew. Chem. Int. Ed. 3, 1 (1964) [14] G. Zundel, The Hydrogen Bond – Recent Developments in Theory and Experiments. II. Structure and Spectroscopy, Vol. 2, P. 683, (North-Holland Pub. Co., 1976) [15] K. Liu, M. G. Brown, C. Carter, R. J. Saykally, J. K. Gregory and D. C. Clary. Nature 381, 6 (1996) [16] C. Pérez, M. T. Muckle, D. P. Zaleski, N. A. Seifert, B. Temelso, G. C. Shields, Z. Keisiel and B. H. Pate, Science 336, 897 (2012) [17] R. N. Barnett and U. Landman, J. Phys. Chem. A 101, 164 (1997) [18] K. Kim, K. D. Jordan, and T. S. Zwier, J. Am. Chem. Soc. 116, 11568 (1994) 87 [19] J. Q. Searcy, and J. B. Fenn, J. Chem. Phys. 61, 5282 (1974) [20] M. Miyazaki, A. Fujii, T. Ebata and N. Mikami, Science 304, 1134 (2004) [21] J. W. Shin, N. I. Hammer, E. G. Diken, M. A. Johnson, R. S. Walters, T. D. Jaeger, M. A. Duncan, R. A. Christie and K. D. Jordan, Science 304, 1137 (2004) [22] R. W. Bolander, J. L. Kassner and J. T. Zung, J. Chem. Phys. 50, 4402 (1969) [23] J. L. Kassner and D. E. Hagen, J. Chem. Phys. 64, 1860 (1976) [24] M. P. Hodges and D. J. Wales, Chem. Phys. Lett. 324, 279 (2000) [25 ] C. C. Wu, C. K. Lin, H. C. Chang, J. C. Jiang, J. L. Kuo and M. L. Klein, J. Chem. Phys. 122, 074315 (2005) [26] A. Khan, Chem. Phys. Lett. 319, 440 (2000) [27] T. Kus, V. F. Lotrich, A. Perera and R. J. Bartlett, J. Chem. Phys. 131, 104313 (2009) [28] S. S. Xantheas, Can. J. Chem. Eng. 90, 843 (2012) [29] M. J. Ryding, R. Izsák, P. Merlot, S. Reine, T. Helqaker and E. Uqqerud, Phys. Chem. Chem. Phys. 17, 5466 (2015) [30] S. Heiles and R. Schäfer, Dielectric Properties of Isolated Clusters (Springer Netherlands, 2014) [31] W. A. de Heer and V. V Kresin, in Handbook of Nanophysics, edited by K. D. Sattler, P. 1 (CRC Press, Boca Raton, 2010) [32] M. Broyer, R. Antoine, I. Compagnon, D. Rayane, and Ph. Dugourd, Physica Scripta 76, C135 (2007) [33] R. Moro, R. Rabinovitch, C. Xia, and V. Kresin, Phys. Rev. Lett. 97, 1 (2006) [34] N. Guggemos, P. Slavíček and V. V. Kresin, Phys. Rev. Lett. 114, 043401 (2015) [35] S. A. Arrhenius, Z. Phys. Chem. 1, 631 (1887) [36] C. T. Lee, C. Sosa, M. Planas and J. J. Novoa, J. Chem. Phys. 104, 7081 (1996) [37] D. E. Bacelo, R. C. Binning and Y. Ishikawa, J. Phys. Chem. A 103, 4631 (1999) [38] A. Milet, C. Struniewicz, R. Moszynski and P. E. S. Wormer, J. Chem. Phys. 115, 349 (2001) 88 [39] S. Odde, B. J. Mhin, S. Lee, H. M. Lee and K. S. Kim, J. Chem. Phys. 120, 9524 (2004) [40] M. Masia, H. Forbert, D. Marx, J. Phys. Chem. A 111, 12181 (2007) [41] W. Lin and F. Paesani, J. Chem. Phys. A 117, 7131 (2013) [42] A. Gutberlet, G. Schwaab, O. Birer, M. Masia, A. Kaczmarek, H. Forbert, M. Havenith, and D. Marx, Science 324, 1545 (2009); S. D. Flynn, D. Skvortsov, A. M. Morrison, T. Liang, M. Y. Choi, G. E. Douberly and A. F. Vilesov, J. Phys. Chem. Lett, 1 (15), 2233 (2010) [43] J. Lengyel, A. Pysanenko, V. Poterya, J. Kočišek and M. Fárník, Chem. Phys. Lett. 612, 256 (2014) [44] Z. Shi, J. V. Ford, S. Wei and A. W. Castleman, Jr., J. Chem. Phys. 99, 8009 (1993) [45] H. Shinohara, N. Nishi and N. Washida, J. Chem. Phys. 84, 5561 (1985) [46] L. Belau, K. R. Wilson, S. R. Leone and M. Ahmed, J. Phys. Chem. A 111, 10075 (2007) [47] S. Q. Wei and A. W. Castleman, Jr., Int. J. Mass Spectrom. Ion Processes 131, 233 (1994) [48] S. Barth, M. Ončák, V. Ulrich, M. Mucke, T. Lischke, P. Slavíček and U. Hergenhahn, J. Phys. Chem. A 113, 13519 (2009) [49] J. Kočišek, J. Lengyel, M. Fárník and P. Slavíček, J. Chem. Phys. 139, 214308 (2013) [50] A. Golan and M. Ahmed, J. Phys. Chem. Lett. 3, 458 (2012) [51] U. Nagashima, H. Shinohara, N. Nishi and H. Tanaka, J. Chem. Phys. 84, 209 (1986) [52] F. Dong, S. Heinbuch, J. J. Rocca and E. R. Bernstein, J. Chem. Phys. 124, 224319 (2006) [53] C. Y. Ng, D. J. Trevor, P. W. Tiedemann, S. T. Ceyer, P. L. Kronebusch, B. H. Mahan and Y. T. Lee, J. Chem. Phys. 67, 4235 (1977) [54] H. Shinohara, N. Nishi and N. Washida, J. Chem. Phys. 84, 5561 (1986) [55] H. Shiromaru, H. Shinohara, N. Washida, H. S. Yoo and K. Kimura, Chem. Phys. Lett. 141, 7 (1987) [56] P. P. Radi, P. Beaud, D. Franzke, H. M. Frey, T. Gerber, B. Mischler and A. P. Tzannis, J. Chem. Phys. 111, 512 (1999) [57] C. Bobbert, S. Schütte, C. Steinbach and U. Buck, Eur. Phys. J. D 19, 183 (2002) 89 [58] O. Marsalek, C. G. Elles, P. A. Pieniazek, E. Pluhařová, J. VandeVondele, S. E. Bradforth and P. Jungwirth, J. Chem. Phys. 135, 224510 (2011) [59] O. Svoboda, D. Hollas, M. Ončák and P. Slavíček, Phys. Chem. Chem. Phys. 15, 11531 (2013) [60] B. L. Yoder, J. H. Litman, P. W. Forysinski, J. L. Corbett and R. Signorell, J. Phys. Chem. Lett. 2, 2623 (2011) [61] J. H. Litman, B. L. Yoder, B. Schläppi and R. Signorell, Phys. Chem. Chem. Phys. 15, 940 (2013) [62] L. Belau, K. R. Wilson, S. R. Leone and M. Ahmed, J. Phys. Chem. A 111, 10075 (2007) [63] H. Shinohara and N. Nishi. J. Chem. Phys. 84, 10 (1986) [64] S. Denifl, F. Zappa, I. Mhr, F. F. da Silva, A. Aleem, A. Mauracher, M. Probst, J. Urban, P. Mach, A. Bacher, O. Echt, T. D. Mrk and P. Scheier, Angew. Chem. Int. Ed. 121, 9102 (2009) [65] G. H. Gardeinier, M. A. Johnson and A. B. McCoy, J. Phys. Chem. A 113, 4772 (2009) [66] K. Mizuse and A. Fujii, J. Phys. Chem. A 117, 929 (2013) [67] K. Sattler, Surf. Sci. 156, 292 (1985) [68] T. P. Martin, J. Chem. Phys. 80, 170 (1984) [69] M. Kinne, T. M. Bernhardt, B. Kaiser and K. Rademann, Z. Phys. D 40, 105 (1997) [70] T. Habteyes, L. Velarde and A. Sanov, J. Chem. Phys. 126, 154301 (2007) [71] K. Mizuse and A. Fujii, Phys. Chem. Chem. Phys. 13, 7129 (2011) [72] R. Moro, R. Rabinovitch and V. V. Kresin, Rev. Sci. Instrum. 76, 056104 (2005) [73] L. Poth, Z. Shi, Q. Zhong and A. W. Castleman, Jr., Int. J. Mass Spectrom. Ion Processes 154, 35 (1996).* [74] D. Bing, T. Hamashima, A. Fujii and J. Kuo* J. Phys. Chem. A 114, 8170 (2010) [75] S. Vongehr and V. V. Kresin, J. Chem. Phys. 119, 11124 (2003) [76] P. Milanni and S. lannotta, Cluster Beam Synthesis of Nano-structured Materials (Springer, 1999) [77] R. Moro, R. Rabinovitch, and V. V. Kresin, J. Chem. Phys. 124, 146102 (2006) 90 [78] W. A. de Heer, Rev. Mod. Phys. 65, 611 (1993) [79] M. Vollmer, K. Selby, V. V Kresin, J. Masui, M. Kruger, and W. D. Knight, Rev. Sci. Instrum. 59, 1965 (1988); N. G. Guggemos, Ph. D. Dissertation, University of Southern California, Los Angeles (2014) [80] Personal communication with Prof. Michal Fárník [81] C. Bobbert, S. Schutte, C. Steinbach and U. Buck, Eur. Phys. J. D 19, 183192 (2002) [82] V. Z. Kresin, J. Chem. Phys. 128, 094706 (2008) [83] D. W. Jeppson, J. L. Ballif, W. W. Yuan and B. E. Chou, (1978). “Lithium Literature Review: Lithium’s Properties and Interaction”, this review gave an impressive summarization for lithium propreties. [84] J. Blanc, V. Bonačić‐Koutecký, M. Broyer, J. Chevaleyre, Ph. Dugourd, J. Koutecký, C. Scheuch, J. P. Wolf and L. Wöste. J Chem Phys. 96, 1793 (1992) [85] H. W. Sarkas, S. T. Arnold, J. H. Hendricks, V. L. Slager and K. H. Bowen, in Proc. SPIE 1858, 240 (1993) [86] C. Brechignac, Ph. Cahuzac, M. de Frutos, J. Ph. Roux and K. Bowen, in Physics and Chemistry of Finite Systems: From Clusters to Crystals (Springer, 1992) [87] W. A. de Heer, Ph. D. Dissertation, University of California, Berkeley (1984) [88] A. Scheidemann, V. V. Kresin. Rev. Sci. Instrum. 62, 2046 (1991) [89] It is possible because the weather in southern California is dry. [90] J. Blanc, V. Bonačić‐Koutecký, M. Broyer, J. Chevaleyre, Ph. Dugourd, J. Koutecký, C. Scheuch, J. P. Wolf and L. Wöste. J. Chem. Phys. 96, 1793 (1992)

Abstract (if available)

Abstract We have developed a technique for assessing water cluster sizes by pickup momentum transfer, revealing abundant fragmentation of water clusters is due to electron impact-ionization (70eV). ❧ Neat water clusters (H₂O)n were produced by a supersonic nozzle source, while neutral acid-water clusters were produced with the same method by attaching a deuterium chloride molecule to neat water clusters. Other gases including CO₂, CH₃OH and NH₃ were also explored, but the mass spectra did not show efficient attachments. ❧ The speeds of the water cations (H₂O)n-1H⁺ covering the size from n = 3 to 12 were investigated. By applying the momentum conservation law, we show that most of the detected water cluster ions originate from an original size of around n = 30-40. To check the validity of this conclusion, we also carried out the similar experiment for neat deuterium water clusters and corresponding mixed clusters with hydrogen chloride. Here, we found that the original deuterium water cluster sizes are slightly smaller than the original hydrogen water cluster sizes. ❧ In another part of the project, we designed a system for transferring a load of high purity lithium metal into a molecular or cluster beam source. A hot loading vessel was thoroughly baked out while empty and overpressured with argon. A clean Li rod was then dropped in through a long narrow tube. The thoroughly degassed interior of the vessel and the rapid melting of the inserted rod facilitated contamination-free transfer of the highly reactive liquid metal into the source oven. ❧ The seeded supersonic source previously used for sodium clusters (650℃) was incapable for producing lithium clusters which require a higher temperature (>1050℃). Therefore, we used a molybdenum alloy (TZM) which can sustain the requisite temperature to design a new source oven, however welding turn out to be a critical problem. Finally, we designed a stainless steel source in a compact version with much thicker walls and a set of new heaters. The new source has been tested to successfully work at 1050℃ for around 48 hours, and deposition of lithium has been observed.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A measurement of the dielectric properties of water nanoclusters

PDF

Dissociation energy and dynamics of HCl and aromatic-water clusters

PDF

Fabrication, deposition, and characterization of size-selected metal nanoclusters with a magnetron sputtering gas aggregation source

Asset Metadata

Creator Huang, Chuanfu (author)

Core Title A technique for assessing water cluster sizes by pickup momentum transfer and a new source with loading system for lithium clusters

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Physics

Publication Date 07/24/2016

Defense Date 05/11/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag equipment design,ionization properties,lithium cluster,OAI-PMH Harvest,water cluster

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Kresin, Vitaly V. (committee chair), Dawlaty, Jahan M. (committee member), Haselwandter, Christoph (committee member), Shakeshaft, Robin (committee member), Takahashi, Susumu (committee member)

Creator Email chuanfuh@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-273809

Unique identifier UC11281139

Identifier etd-HuangChuan-4591.pdf (filename),usctheses-c40-273809 (legacy record id)

Legacy Identifier etd-HuangChuan-4591.pdf

Dmrecord 273809

Document Type Dissertation

Format application/pdf (imt)

Rights Huang, Chuanfu

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA