Masters Project: Construction and Calibration of a Spark Discharge Generator to Create Mono-disperse Metallic Aerosol Nanoparticles By: Jonathan E. Pautler Project supervisors: Knut Deppert, Maria E. Messing, and Bengt O. Meuller Lund University 2012 1 Abstract: A Spark Discharge Generator is a versatile device which can be used to generate nanoparticles from any two conducting sources. The device constructed during this thesis project was to make metallic mono-disperse aerosol nanoparticles towards applications in catalysis and nanowire growth. The operating principle is two electrodes separated by a gap with an applied voltage above the discharge voltage. This high potential creates an ionization channel and a spark is discharged between the two electrodes causing a high local temperature on the ends of the electrodes which causes thermal evaporation of the material into the surrounding gas. These initial particles with sizes on the atomic scale collide and adhere by a process called coagulation creating larger nanoparticles which may then be filtered, reshaped, counted, and deposited. A theoretical introduction is presented, a description of the components is included, some techniques for calibration of the device are described, and initial data is shown. 2 Acknowledgments: I have been given the warmest welcome in Sweden by both the people and the department at Lund. I acknowledge those who have helped me most with my sincerest of gratitude and for those who I have forgot to mention here for brevity you are in my thoughts. First is of course my advisors; to Knut for giving me the opportunity to perform in the lab, for the time, concern, and effort he has continually given to help me be the best student, and to keep me on the right path in research. To Maria for all the laughs, wit, and encouragement enclosed within scientific discussions. You’re a source of never-ending energy and efficiency, and I find I am much more productive when I am having fun. To Bengt for being the person who was with me most days handling the real challenges of building the device; your wisdom and expertise has driven this project forward. Life is much easier when you have someone else to wonder, “what is going on with this thing” with! Bengt has been a founding designer of the system and completed the vast majority of the building of the device before and during my time here at Lund. To Sebastian who has been the best of partners, a close friend; as well as, the other fellow who has with me wondered “what is going on.” I am glad it seems we have figured most of the problems out. I wish you the best of futures and I hope we work together again. Sebastian also deserves credit for much of the work performed in this thesis, and is a contributor to several segments of the data, SEM images, and considerations on much of the analysis. Special thanks goes to Marcus from Prevas who was hired to create an updated structure to the Labview™ code, I have learned a lot from the new version and it certainly made the data acquisition at the end much faster. To Saiful, Mo, Henrick, Nick, Richard, Kristian, Dmitry and all the others who have given insight into many problems of this project and are great friends. Finally the personal side of things to my Mom and Dad for being a constant source of love and have guided me to be who I am and helped in making me into the person I am today. Lastly to Anna who has been at my side this time and is a reason for why I have chosen to do what I am doing. 3 Contents: 1. Introduction 2. Theory 2.1 Spark Discharge Generator Theory 2.2 Spark and Plasma Theory 2.3 Parameter Calculations 3. Methods 3.1 Machine Outline 3.2 Component Description 3.3 Lab-view/DAQ Summary 4. Results and Discussion 4.1 DMA Calibration and Accuracy 4.2 ESP and Deposition time calibration 4.3 Size distribution for various parameters such as current and gap size 4.4 Reshaping Temperature 4.5 Varying Electrode Size 4.6 Other Elements 5. Conclusions 6. Future work 7. References 8. Appendix 8.1 Major Component List 8.2 SDG Operating Procedure 4 1. Introduction/aim: Nanoparticles are a major source of interest in modern research due to the large surface to volume ratio and the capacity to demonstrate reduced dimensionality giving rise to quantized states of matter. What this means is that nanoparticles are different in many ways to their bulk counterparts; for example some gold nanoparticles are red or purple in color, many nanoparticles are sticky, and many new phenomena happen in this territory that baffles the mind and sends the imagination into overdrive. Many next generation technologies could be centered on this already multibillion dollar industry. Electronics have a high potential to be formed from nanoparticle and nanowire technology especially those based on semiconductors due to the incredibly small size. These prototype nano-devices have demonstrated that they can change their state of matter faster and consume far less power than existing devices. This means better video games, faster simulation speeds, and less time until we can see what our friends are doing on Facebook. Additionally many Solar Technologies have nanoparticle centered cell types [1]. There are many methods to create nanoparticles: evaporation, sputtering, annealing of thin films, and so forth. The thesis focuses on the Spark Discharge Method. The aim of this thesis is to give a brief review of the theory behind nanoparticle generation in the Spark Discharge Generator, the specific components and technology used within the device constructed during the thesis, additionally the initial calibration data is presented and described; as well as, a short summary of future work. 5 2. Theory: 2.1 Spark Discharge Generator: The Spark Discharge Generator (SDG) comes in many forms. The basic form consists of a carrier gas through a chamber, with two electrodes separated by a small gap, and a voltage source applied to the electrodes to create a breakdown voltage. After the breakdown voltage is reached, a high energy spark is created across the gap. The electrical energy is converted to heat causing a high local temperature that rapidly evaporates part of the electrodes resulting in nanoparticles in the carrier gas. Figure 2.1.1: A SDG while in operation Special thanks to Saiful Islam for photography In a SDG many parameters can be adjusted to affect particle generation. Some examples such as the type of carrier gas, flow of the carrier gas, pressure which is often near atmospheric, and spark energy are critical to the type of particles produced. Additionally after the particles leave the surface of the electrode the small initial particles collide and stick together through a coagulation phase which leaves them as a stretched out chain of connected primary particles which are called agglomerates as seen in Image 2.2. After 6 generation the particles are sent through a charger to give a known electrical distribution, transported to a Differential Mobility Analyzer (DMA) for size selection, a compaction furnace to reshape the agglomerated particles into spherical particles, a second DMA, an Electrometer to give a particle count, and finally an Electrostatic Precipitator (ESP) to focus charged particles onto a collector electrode where a substrate is placed for deposition [2]. Finally the gas exits the chamber and is filtered before returning to the atmosphere. Image 2.1.2: Large Coagulation of gold nanoparticles: The photo exemplifies how both large and small clusters can form together. These particular particles have undergone a surface coagulation technique so some of the formations are larger than the aerosol particle clusters. 7 Gas in Spark Charger DMA 1 Circuit Furnace DMA 2 ESP Electrometer Pump out Figure 2.1.3 Schematic of the Spark Discharge Generator System In Summary: 1. The Spark Generator creates the particles. 2. A charger gives a Boltzmann distribution of the particles charges [3]. 3. The first DMA is used to filter out particles based on their mobility diameter in an electric field [3]. 4. A compaction furnace is used to control the shape of the particles by reshaping them into spheres [2]. 5. The second DMA is used to filter out particles based on their mobility diameter but now that they are more uniformly spherical we have a more precise distribution. This step is critical to getting mono-disperse particles because the furnace reshapes the particles and when the furnace is over 500°C it might alter the number of charges on some of the agglomerates [3]. 6. A) An Electrometer is used to give a particle count. 8 B) Or an Electrostatic Precipitator (ESP) is used to focus charged particles onto a collector electrode where a substrate is placed for deposition [2]. Each component will be explained in detail later on. 9 2.2 Spark and Plasma Theory: During the spark process a constant flow of a carrier gas such as N2 goes between two electrodes that are sparking. A plasma channel seen by the spark is caused by the wellknown phenomenon of having a high potential difference between two electrodes. As the current passes from metal to carrier gas back to metal a large resistance causes energy from the circuit to be transferred to the metal as heat causing electrode loss, which generates the particles. Figure 2.2.1: Circuit Diagram of the SDG The circuit consists of a power source with a capacitor in parallel to cause regular discharges based on the total capacitance (Figure 2.2.1). The spark itself can be thought of by its equivalent circuit as a resistor that blocks the circuit until the breakdown voltage is reached as a kind of switch resistor. The inductance in the wire determines the length of spark duration. A diode is needed to protect the circuit from a current backflow, and there is a resistor used to create a load which causes the capacitors to charge properly. 10 The Breakdown voltage, Vb, is governed by Pashcen’s Law: ππ = π΄∗π∗π ln(π ∗ π) + π΅ where p = Pressure d = Gap Distance A & B are constants with values of: A=4.36*10^7 V/(atm*m) B=12.8 [4] This breakdown voltage is reached when the electric field of the spark exceeds the dielectric field strength of the surrounding gas forming an ionized conducting channel [5]. The conducting channel is formed by the excessive field, which first causes large numbers of electrons to be released. Then the free elections collide with the neutral gas. The collisions can be split into two categories Elastic and Inelastic collisions. Elastic collisions do not change the internal energy of the neutral species but raise their kinetic energy. Inelastic collisions modify the electronic structure of neutral species when energy is high enough turning an innate gas into either an excited species or an ion at the cost of kinetic energy [6]. These ions if given sufficient energy will collide and cause more excited species to form. These excited species create a conductive channel, which then discharges current forming a spark. The spark itself is the current flowing through this conductive channel. In practice the spark has a delay from when breakdown voltage is reached and the discharge occurs, and thus has an overvoltage. It sparks at a real voltage called the discharge voltage ππ = ππ + ππ which is generally greater than the breakdown voltage. 1 The energy transferred via the spark is calculated using πΈ = 2 ∗ πΆ ∗ ππ ^2 The spark frequency is given via π = πΌπ πΆ∗ππ where Ic is the current, C is the capacitance, and Vd is the discharge voltage. The total energy released, as a rate of time, is a function of the frequency times the energy, which is the power [7], [8]. 11 2.3 Parameter Calculations The breakdown voltage is completely governed by the electrode gap and the pressure, so we vary the gap size in Table 2.3.3 the pressure in Table 2.3.4. The current or frequency can be varied to give a larger or smaller amount of energy transfer. These calculated values are shown in Table 2.3.5 and Table 2.3.6. This can be compared to the energy required to evaporate the metal to give a rough estimate of the evaporation rate of the electrodes via the specific heat Q=m*c*ΔT which is in J/Kg*K. The specific heat tells us the required energy to melt bulk metal. The specific heat can be found in table 2.3.1 and is useful because it can estimate the amount of time it takes under different conditions to generate a mass of nanoparticles, also it indicates that there will be more or less mass loss for certain elements meaning smaller or larger particle counts for identical conditions. For example with gold and copper we see copper is over 3 times as high so we would expect less mass to be lost for the same conditions. To determine how much of a finished product can be produced is a different question. Additionally losses should be factored such as the filtering process in the DMAs and losses in the tubes, but this works as a rough estimate as to how quickly electrodes disappear. c (J/kg*K) Melting Point (K) Q (J) Copper 386 1,083 408,002 Gold 126 1,064 130,788 Silver 233 962 218,088 Table 2.3.1: Specific Heat of Metals: Room temperature was assumed to be 26 °C [9] Another consideration we can see from Figure 2.3.2 is a physical limitation to the frequency of around 1 MHz where the plasma follows thermal laws instead of the electric field; however, most frequencies tested in the lab are on the order of 300 Hz because higher frequencies than 1kHz requires the capacitors in the circuit to be able to bear the strain of a large power throughput. The value in Table 2.3.6, which is at 800 kHz, frequency makes it clear why larger frequencies are not performed in the lab since the power consumption is quite extreme. 12 Figure 2.3.2: Plasma frequency and behavior: fpi and fpe are the names of the regimes where the plasma is ion or electron dominated [6] Gap Size Breakdown (mm) Voltage (kV) Current (mA) 0.5 2 1.0 3 1.5 4 2.0 6 2.5 7 3.0 9 3.5 10 4.0 11 4.5 13 5.0 14 10.0 28 20.0 54 30.0 80 50.0 132 Table 2.3.3: Varying Gap Size: Energy per spark (mJ) Frequency (Hz) 30 27 30 104 30 229 30 400 30 617 30 879 30 1,186 30 1,537 30 1,933 30 2,372 30 9,131 30 35,172 30 77,435 30 209,353 Calculated from Paschen’s Power (W) Power (kW/Hr) 998 2.7 0.2 509 5.3 0.3 344 7.9 0.5 260 10.4 0.6 209 12.9 0.8 175 15.4 0.9 151 17.9 1.1 133 20.4 1.2 118 22.8 1.4 107 25.3 1.5 54 49.7 3.0 28 97.4 5.8 19 144.6 8.7 11 237.8 14.3 Law at Atmospheric Pressure and subsequent calculations. The capacitance was 24 nF. The voltage shown is assumed to be the breakdown voltage Vb but the real voltage or discharge voltage Vd is different due to the overvoltage, the energy per spark will be shown to be much lower because of this. Larger gap sizes produce larger particles due to the higher energy per spark. 13 Gap Size Pressure Breakdown (mm) (Atm) Voltage (kV) 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 0.01 0.1 0.5 1 2 3 4 10 20 Current (mA) 0.1 0.6 2.9 5.8 11.3 16.8 22.2 54.1 106.3 30 30 30 30 30 30 30 30 30 Energy per spark (mJ) Frequency (Hz) 0.05 4.59 104.06 399.83 1,537.43 3,381.48 5,916.27 35,172.12 135,573.48 Power (W) 22,546 2,427 509 260 133 89 68 28 14 Energy (kW/Hr) 0.1 1.1 5.3 10.4 20.4 30.2 40.0 97.4 191.3 0.07 0.67 3.18 6.23 12.22 18.13 23.98 58.47 114.79 Table 2.3.4: Varying Pressure: Calculated from Paschen’s Law. As we can see from the table a change in pressure allows us to increase or decrease the breakdown voltage required for a spark, also there are various advantages and disadvantages to different pressures. Gap Size Breakdown (mm) Voltage (kV) 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 Current (mA) 5 10 15 20 25 30 35 40 45 50 100 1,000 5,000 10,000 Energy per spark (mJ) 400 400 400 400 400 400 400 400 400 400 400 400 400 400 Frequency (Hz) 43 87 130 173 217 260 303 346 390 433 866 8,662 43,311 86,621 Power (W) 1.7 3.5 5.2 6.9 8.7 10.4 12.1 13.9 15.6 17.3 34.6 346.5 1,732.4 3,464.9 Energy (kW/Hr) 1.0 2.1 3.1 4.2 5.2 6.2 7.3 8.3 9.4 10.4 20.8 207.9 1,039.5 2,078.9 Cu Mass Loss Au Mass Loss (Hr/Kg) (Hr/Kg) 3,925.2 1,962.6 1,308.4 981.3 785.0 654.2 560.7 490.6 436.1 392.5 196.3 19.6 3.9 2.0 1,258.2 629.1 419.4 314.6 251.6 209.7 179.7 157.3 139.8 125.8 62.9 6.3 1.3 0.6 Table 2.3.5: Varying Current: A higher current means a much higher amount of mass particle production; however, as larger amounts of energy are inputted the size distribution is altered. Specific heat was used to calculate mass loss over time, but there are many losses in the system so it simply acts as an estimate of electrode material lost over time. 14 Gap Size Breakdown (mm) Voltage (kV) 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 Current (mA) 69 139 277 693 1,385 6,927 13,853 69,267 110,827 Energy per spark (mJ) 400 400 400 400 400 400 400 400 400 Frequency (Hz) 500 1,000 2,000 5,000 10,000 50,000 100,000 500,000 800,000 Power (W) 20 40 80 200 400 2,000 4,000 20,000 32,000 Energy (kW/Hr) 12 24 48 120 240 1,200 2,400 12,000 19,200 Cu Mass Loss Au Mass Loss (Hr/Kg) (Hr/Kg) 340.0 170.0 85.0 34.0 17.0 3.4 1.7 0.3 0.2 108.99 54.50 27.25 10.90 5.45 1.09 0.54 0.11 0.07 Table 2.3.6: Varying Frequency: The table clearly shows how an increase in frequency will cause a larger theoretical gain in particle mass. However achieving the upper limits will be a severe test for the circuits used. Specific heat was used to calculate mass loss over time, but there are many losses in the system so it simply acts as an estimate of electrode material lost over time. To reach industrial scale production optimizing the mass transferred from bulk material into nanoparticles is critical. Several things become apparent from the formulas and the tables. First is the larger the gap size, the larger the voltage, and thus the larger the energy transferred. Second is the higher the pressure, the higher the voltage, the higher the energy transfer. Third is the higher the current, the higher the frequency, the higher the energy transfer. Since the total energy release is a function of the frequency measuring the current versus production levels was tested. A fourth way out of the scope of the tables to increase the energy would be to alter the inductance to have longer sparks. Another would also be to add an external electric field to allow the plasma form easier. Of all of these parameters that can be changed testing each one individually is needed to see how it affects size distribution as parameter changes to size distributions is not always as direct and simple and many applications of nanoparticles requires specific sizes for their effects to work. Also there are stability problems near the extreme ends of the spectrum; even in the early phases of testing several processes were unstable due to being at the lower edge of where a spark was created. By unstable we mean that the spark would change voltages, the frequency would vary dramatically during the process, or the device would simply not spark at all. 15 2.4 Growth processes The initial particles formed by the localized thermal process on the electrodes and are on the atomic size range. At particle diameters of smaller than 100 nm it is assumed all particles adhere with every collision. Provided there is enough particle density for collisions they will form clusters of particles via coagulation, which is sometimes called agglomeration. That is when particles collide in a gas and form larger particles. For a monodisperse flow this process is well modeled but for a non-homogeneous flow (which the spark process is) an exact description is complicated and geometry dependent. To the authors knowledge simplified numerical solutions are available but were not used in the course of this thesis work [10]. A full theoretical model would be incredibly useful towards understanding the effect of parameters on size distribution. After formation the particles are flushed by the chamber flow. The rate at which they are flushed is based on the flow of the process, which is typically a few liters per minute, and the collision rate is affected by the flow. Under higher levels of particle generation a higher flow may be necessary to keep the particles from coagulating too much and exceeding the nanoparticle regime. Conversely when larger particles are desired a lower flow, or a coagulation chamber may be added to increase particle diameter. In this system a coagulation chamber may be added after the charger before sorting and sintering to create larger particles by giving them a larger volume where they can collide and form larger particles. 16 3. Methods 3.1 Machine Outline Figure 3.1.1: Highlighted Picture of Spark Discharge Generator at Lund University 17 Equipment list: 1) Spark Chamber 2) MFC 1 3) MFC 2 4) Charger 5) DMA1 6) Furnace 7) DMA 2 8) MFC 3 9) MFC 4 10) ESP- Electrostatic Precipitator 11) EM-Electrometer 12) Capacitor Bank 13) DAQ 14) HV Source 15) Pump Most of the connections used were VFC© or Swagelok© Our range of capacitances can be varied, but for all experiments used in this thesis the value is fixed to 24 nF. Component description: ο· 1) Spark Chamber: A pressurized chamber that holds the electrodes of the device and protects the surrounding environment from exposure to nanoparticles; as well as, containing the sheath flow of an inert carrier gas such as pure nitrogen or a noble gas. Nitrogen was used for all parts of the work performed during this document, and if the resultant nanoparticles are reactive to nitrogen should be considered in future work. ο· 2,3,8,9) MFCs: Mass Flow Controller a device used to control gas flow inside a pressurized system. 18 The operating principle for a mass flow controller is a small bypass tube with two resistance thermometers or thermocouples. The gas flowing through the tube and the cooling rate creates a temperature differential that is measured electronically 3.1.2 Operational Principle of a Mass Flow Controller [11]. The temperature differential is based upon mass flow, which is a function of density, specific heat, and flow rate. It may be important to consider that during rapid changes in values the measurement is inaccurate; however, for this SDG it is kept constant at about 1.7 l/min because this is approximately 1/6 of the default 10 l/min sheath flow of the DMAs and is a requirement for particle selection [3], [11]. ο· 4) Charger: Radioactive source which bombards particles passing through the device giving a known Boltzmann charge distribution [3]. ο· 5,7) DMAs: A Differential Mobility Analyzer is a metal tube with a sheath flow and an inlet and outlet connected to the particle flow. Inside the metal tube is a rod that is 19 charged to create an electric field between the wall of the tube and the inner rod. The electric field sorts the incoming particles via their electrical mobility diameter by only allowing specific values to continue and is thus a size selection device [3]. An important consideration when operating a DMA is to remember that the particles are sorted by electrical mobility and not their geometrical diameter. For spherical particles the mobility diameter is the geometrical diameter but because the particles come out as agglomerates or fluffy elongated particles a furnace is used to reshape the particles so the mobility diameter is uniform with the geometrical diameter [10], [2]. The transfer function governs the voltage required to get specific electrical mobility diameters. As this is a problem solved by my predecessor I will simply ask you to see Martin Karlsson’s thesis for a detailed explanation of the transfer function and its derivation. The DMA’s used in the device measure between 0-100 nm in range, because of the way the signal is analyzed to achieve particles below 10 nm or above 90 nm a special low voltage control was needed. The voltage divider functions by splitting the signal and then using a second 0-10 volt control to represent 0%-10% of the input. This should give the DMA’s accuracy down to 1% range or 1 nm particles; although at that range the particles have other sorting problems such as charging [12]. 20 Figure 3.1.3 Schematic of DMA 1 (red circle number 5 in Fig. 3.1.1) TSI model 3080 Long DMA [13] ο· 6) Sintering Furnace: A furnace is used to exceed 2/3 the melting point of the bulk property of the nanomaterial to allow reshaping of the particles to form spherical objects. There are two primary reshaping processes a compaction process and a reshaping process. 600 °C seem to be a sufficient temperature to reshape gold particles from agglomerates into spherical particles [2]. ο· 10) ESP: Electro-Static Precipitator uses an electric field to direct the charged particles towards a center ring where it forms a spray for deposition either on a substrate such as thin film silicon, a liquid, or whatever the final form would be [14]. ο· 11) An Electrometer, abbreviated as AE or EM, is a tool for measuring the total number of particles in the gas flow. As particles come into the chamber they are 21 collected by a Faraday cup and the current is used to calculate the number concentration via πΌ = π ∗ ππ ∗ π ∗ ππ where πΌ is the electrometer current, π is the aerosol concentration, ππ is the average number of charges per particle, π is the elementary unit charge, and ππ is the volumetric aerosol flow [15]. ο· 12) Capacitor Bank: A set of parallel capacitors used in the circuit. ο· 13) DAQ: Data Acquisition device from National Instruments. Connects the voltage as a digital or analogue signal to the computer via USB port where Labview™ controls the hardware. Several power supplies are required for voltage to control the signals to the various devices, as well as a ground. An external ground is critical to an accurate signal as the SDG creates a huge electric field interfering with the interpretation of the wire signals. The output signal is compared with the external set point signal, which directs the connected device. The output signal is measured as a percentage of the total volume normalized to the maximum capacity i.e. for MFC using a control voltage of 0-10 V on a 20 l/min maximum flow 3 volts would correspond to 3/10 of the total flow or 30% of 20 l/min which is a flow of 6 l/min. Signals in and out of the DAQ are accurately measured based on the gain but for this system between 10-90% is accurate and the control circuit becomes less predictable between the extreme regions of 0-10% and 90100%. ο· 14) High Voltage Source: Provides the power required to cause the potential difference for the spark. ο· 15) Vacuum pump: Drives flow through the system and removes excess particles from the chamber. Particles are filtered before they are returned to the atmosphere to prevent inhalation of nanoparticles. 3.2 Labview™/DAQ Labview™ was used to control many of the components via a computer interface. Digital and Analogue signal input and output was performed through the NI-DAQ. 22 4. Results and Discussion: The operational method is shown in the appendix. To reach full functionality a new SDG’s following parameters must be tested or calibrated: 4.0 First Data Sets 4.1 DMA Calibration and Accuracy 4.2 ESP and Deposition Time Calibration 4.3 Reproducibility of Size Distributions 4.4 Reshaping Temperature 4.5 Varying Electrode Size and Gap Size 4.6 The Other Elements 4.0 First Data Sets Particle Count in millions [#/cm3] Size Distribution for a 2 mm gap 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 10 20 30 40 50 60 70 80 90 100 Particle Diameter [nm] Figure 4.0.1: Particle density for a gold electrode with 3 mm diameter with bypassed DMA 1 and furnace. Settings of 2 mm gap, 5.5 kV, and 30 mA. The first sets of data were performed on 3 mm gold electrodes. The primary research interest for this device is predicted to be with gold nanoparticles so making system settings for gold electrodes seemed a proper place to start. Figure 4.0.1 shows the size distribution notice the y-axis is in units of millions (10^6). Here the particles are gold which has a much lower specific heat than for copper which might explain why there are more particles in the gold data than for copper or silver. 23 4.1 DMA Calibration and Accuracy The new SDG system was tested in order to calibrate all the components to ensure they will work properly. The scans were performed by fixing one DMA at a defined particle diameter value while the other is scanned across the range of particle diameters the EM was used to collect a particle count. Furthermore particles were reshaped to spherical ones, in a furnace at 600 °C, scanned with DMA 2, and deposited on a silicon surface. The particles on the surface were investigated by SEM and analyzed with nanoDim: a Matlab program developed by Kristian Storm that measures particle size and shape [16]. The average particle diameters were compared with the values of the mobility diameters gained from DMA 2. DMA Calibration Graph 1 Mobility diameter of scanning DMA 1 [nm] 100 90 80 70 60 y = 0.8857x + 0.0267 R² = 0.9996 50 40 30 20 10 0 0 20 40 60 80 100 120 Mobility diameter of fixed DMA 2 [nm] Figure 4.1.1: DMA Calibration Graph 1 showing the maximum mobility diameter values of DMA 1 while DMA 2 was fixed at several set points DMA 2 was set at fixed mobility diameters (10, 20, 30, 40, 50, 60, 70, 80, 90 and 100 nm) while DMA 1 was scanned for each of these set points. Gold electrodes and a gap of 2 mm were used. A voltage of 5.5-6 kV and a current of 30 mA were applied between the electrodes which led to a breakdown voltage of 2.3-2.5 kV and a spark frequency of 300 Hz. The experiments were accomplished under room temperature (ca. 20 °C) and standard pressure (ca. 1013 kPa or 1 atm). The nitrogen carrier gas flow was set to 1.68 24 l/min. Figure 4.1.1 above shows a linear increase of maximum mobility diameter values with increasing fixed diameters of DMA 2. The increase is lower than 1, which indicates that the DMA’s have a small divergence between each other. This means DMA 2 might show mobility higher than in reality or DMA 1 displays lower mobility diameter than reality. Because DMA 2 is the final sorting device its accuracy is critical to process control. Additional information on the accuracy of the DMA’s is provided by several scans with fixed DMA 1 diameter and varying DMA 2 the positions are shown in Figure 4.1.2 below. Because these results are not a perfect correlation we know at least one of the DMAs is offset. DMA Calibration Graph 2 Mobility diameter of scanning DMA 2 [nm] 120 100 80 y = 1.2233x - 1.7222 R² = 0.9861 60 40 20 0 0 10 20 30 40 50 60 70 80 90 100 Mobility diameter of fixed DMA 1 [nm] Figure 4.1.2: DMA Calibration Graph 2 showing the maximum mobility diameter values of DMA 2 while DMA 1 was fixed at several points SEM was performed to determine actual size of the DMA’s to the produced particles. Spherical particles were deposited on silicon surfaces, for the deposited particles the original gold electrodes (3 mm diameter) were employed and 600 °C furnace temperature was chosen. Also, due to the high erosion rate and the significant cost of gold larger copper electrodes were utilized (6 mm diameter) for later scans. 25 The sparking conditions were standard pressure, room temperature, 2 mm gap distance, 300 Hz frequency, 4-8 kV applied voltage, 30 mA current and a nitrogen flow of 1.68 l/min. Due to long spark times which caused significant ablation of the thin electrodes there were deviations of up to 20% of those base values. Particles with diameters of 10, 20, 40, 60 and 80 nm were deposited to investigate the accuracy of the DMAs. The image below shows a deposition of copper on silicon at a set DMA 2 mobility diameter of 40 nm employing the already mentioned conditions but with 1000 °C reshaping temperature. For this paper anytime a specific value for particle size is mentioned instead of a range it should be taken to mean that the particles have been reshaped through the furnace and through both DMA’s set to that value which dramatically lowers the overall particle concentrations. For all samples in this thesis the ESP deposition voltage was set to the maximum voltage of 10 kV in order to deposit the highest possible amount of particles. In practice the device reads 9.66 kV when set to 10 kV so all values seen as 10 kV should be assumed to be 9.66 kV. The particle diameters were investigated by utilizing SEM and nanoDim. The image 4.1.3 below shows 3 of the thousands of SEM images that were captured during the project. 26 Image 4.1.3: SEM Images A) 40 nm Copper particles which were deposited on silicon for 5 minutes B) NanoDim analysis program calculating radius of 74 particles and giving a distribution of diameters varying between 5 and 10 nm C) 20 nm Copper particles on silicon deposited for 5 minutes D) Distribution of Gold nanoparticles on silicon made during calibration to a density of 1 particle/ π m 2 The particle sizes were plotted against the mobility diameters seen in Figure 4.1.4 below. The graph demonstrates a slope of almost 1 and the values show a maximum of 12% deviation on the extreme value. It should be noted that as the particle size increases the number of particles captured per image decreases so the last two data points had less particles for statistics while the early points had several thousand points. There were only 200 points for 60 nm and 10 points for 80 nm and thus these values are less accurate. The author assumed the 12% deviation is from this and thus from the statistics the output of DMA 2 is well known and is extremely accurate, and thus the optimization values for DMA 1 are now known from the same data. 27 Average Size Distribution Deposited diameter [nm] 80 70 60 50 40 30 Average Size Distribution 20 10 0 0 20 40 60 80 100 DMA 2 diameter [nm] Figure 4.1.4: Statistically analyzed spherical particles with SEM are demonstrated towards the diameter given by DMA 2. Each data point is a statistical average over thousands of particles except for the 60 nm and the 80 nm. The author ascribes the nonlinearity of data point 80 nm to the lack of statistics. 4.2 ESP and Deposition Time Calibration Another parameter to investigate is the time required to achieve specific levels of surface density. To determine the time required samples of each of the conditions above were deposited at different lengths of time (30 s, 1 min, 2 min, 5 min) and a linear regression was performed. Figure 4.2.1 below shows not the surface density but a quotient of density and electrometer current observed before and after the deposition, due to the fact that when depositing many of the particles are diverted to the sample. As expected, the graph shows higher increases in the slope of smaller particles. 28 Surface Density/Electrometer Current [#/pA*µm2] ESP-Calibration 30 y = 2.4448x R² = 0.9988 25 20 nm Density/Current 40 nm Density/Current 20 60 nm Density/Current 15 Linear (20 nm Density/Current) y = 0.6974x R² = 0.9884 10 Linear (40 nm Density/Current) Linear (60 nm Density/Current) 5 y = 0.2297x R² = 0.8448 0 0 2 4 6 8 10 12 Time [min] Figure 4.2.1: Surface density, measured with a SEM and analyzed with nanoDim, divided by the electrometer current before and after the deposition is plotted against the deposition time The increase of the slopes of the 3 different particle sizes over time is used as a calibration factor and introduced in the following figure. With those three points an exponential function describing the behavior of the ESP system to different particle sizes is approximated. This function shown in Figure 4.2.2 is inserted to the LabView™ program controlling the SDG and DMA-scanning process and now the calculated values for surface density are inside the program so it can determine the proper time required for a desired surface density with this specific system. 29 Calibration Factor [#/pA*s*µm2] Density-Size Regression 9 8 7 y = 7.78823e-0.05913x R² = 0.99878 6 5 Density-Size Regression 4 3 Expon. (Density-Size Regression) 2 1 0 0 20 40 60 80 100 120 Particle Size [nm] Figure 4.2.2: A calibration factor gained from this figure is plotted against the particle size to get a density-size regression for the deposition time calibration 4.3 Reproducibility of Size Distributions For the parameters of current and gap distance, size distribution data was taken consecutively to see if they would be identical. The gold electrodes were used up in the previous data set so larger 6 mm copper electrodes were used for the reproducibility tests. All conditions were standard pressure, room temperature, and a nitrogen flow of 1.68 l/min. Other settings were the minimum voltage required for a spark which varied from 4, 5, and 6 kV for 1, 1.5, and 2 mm gap size respectively and the standard 24 nF for capacitance. The current was varied and thus the frequency varied with it as well as the gap size as seen in the data. 30 Particles in millions [#/cm3] Size Distribution for a 1 mm gap 1st run 0.25 0.2 5 mA 0.15 10 mA 0.1 20 mA 0.05 30 mA 0 40 mA 0 10 20 30 40 50 60 70 80 Particle diameter [nm] Figure 4.3.1: Size distributions of particles generated at different currents at a gap distance of 1 mm with a copper electrode with 6 mm diameter Particles in millions [#/cm3] Size Distribution for a 1 mm gap 2nd run 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 5 mA 10 mA 20 mA 30 mA 40 mA 0 10 20 30 40 50 60 70 80 90 Particle diameter [nm] Figure 4.3.2: Size distributions of particles generated at different currents at a gap distance of 1 mm with a copper electrode with 6 mm diameter 31 Particles in millions [#/cm3] Size Distribution for a 1.5 mm gap 1st run 1.2 1 5 mA 0.8 10 mA 0.6 20 mA 0.4 30 mA 0.2 40 mA 0 0 20 40 60 80 100 120 50 mA Particle diameter [nm] Figure 4.3.3: Size distributions of particles generated at different currents at a gap distance of 1.5 mm with a copper electrode with 6 mm diameter Particles in millions [#/cm3] Size Distribution for a 1.5 mm gap 2nd run 1 0.8 5 mA 0.6 10 mA 0.4 20 mA 30 mA 0.2 40 mA 0 0 20 40 60 80 100 120 50 mA Particle diameter [nm] Figure 4.3.4: Size distributions of particles generated at different currents at a gap distance of 1.5 mm with a copper electrode with 6 mm diameter 32 Particles in millions [#/cm3] Size Distribution for a 2 mm gap 1st run 1.4 1.2 5 mA 1 0.8 10 mA 0.6 20 mA 0.4 30 mA 0.2 40 mA 0 0 20 40 60 80 100 120 50 mA Particle diameter [nm] Figure 4.3.5: Size distributions of particles generated at different currents at a gap distance of 2 mm with a copper electrode with 6 mm diameter Particles in millions [#/cm3] Size Distribution for a 2 mm gap 2nd run 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 5 mA 10 mA 20 mA 30 mA 40 mA 50 mA 0 20 40 60 80 100 120 60 mA Particle diameter [nm] Figure 4.3.6: Size distributions of particles generated at different currents at a gap distance of 1 mm with a copper electrode with 6 mm diameter As we can see from the size distribution a larger current means a higher overall number of particles. The frequency varies with the current and with the circuit used here is limited to less than 1 kHz. The size distribution shifts upward to the right as higher current is reached meaning more particles are created and larger nanoparticles are generated. The peak production for smaller nanoparticles is given by lower currents as seen in the distribution data in Figures 4.3.1-4.3.6. One important aspect is that under identical conditions a significant difference is seen in the number of particles produced sometimes in excess of 30% of the peak particle densities. Figures 4.3.7 and 4.3.8 are the peak curve graphs: created from this distribution data taking the highest data point for 33 each of the current curves. This makes it quite clear that as current is increased the number of particles increases as does the average diameter of the particles. However from the reproducibility tests we can clearly see that the total number of particles at a given size varied heavily from scan to scan as well as that the peak particle diameter is not completely reproducible and some of them are exceptionally wide peaks, but is in general in the same range. Particle diameter [nm] Average Peak Particle Diameter 80 70 60 50 40 30 20 10 0 1 mm 1.5 mm 2 mm 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.7: Maximum particle diameters of 3 investigated gap sizes gained from the size distributions at different currents for a copper electrode with 6 mm diameter Particles in millions [#/cm3] Average Peak Particle Density 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 1 mm 1.5 mm 2 mm 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.8: Maximum particle densities of 3 investigated gap sizes gained from the size distributions at different currents for a copper electrode with 6 mm diameter The breakdown voltage and frequency data was measured during the distributions. The breakdown voltage is clearly decreasing as the current increases. Additionally the frequency chart shows a very linear regime early on, but later begins to increase more 34 rapidly for the smaller gap sizes. More data is needed but these two trends could be indicative that the ability to form a plasma channel becomes easier at a given point since the larger gap sizes remains linear for a higher current. Breakdown Voltage 3 Voltage [kV] 2.5 2 1 mm 1.5 1.5 mm 1 2 mm 0.5 0 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.9: Breakdown voltages of 3 investigated gap sizes gained for a copper electrode with 6 mm diameter at different currents Frequency Frequency [Hz] 1000 800 600 1 mm 400 1.5 mm 200 2 mm 0 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.10: Frequencies of 3 investigated gap sizes gained at different currents for a copper electrode with 6 mm diameter 35 Energy per Spark Energy/Spark [mJ] 120 100 80 60 1 mm 40 1.5 mm 20 2 mm 0 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.11: Energies per spark of 3 investigated gap sizes gained at different currents for a copper electrode with 6 mm diameter Energy per Second Energy per Second [W] 25 20 15 1 mm 10 1.5 mm 5 2 mm 0 0 10 20 30 40 50 60 70 Current [mA] Figure 4.3.12: Energies per second of 3 investigated gap sizes gained at different currents for a copper electrode with 6 mm diameter Figure 4.3.11 and figure 4.3.12 show the energy per spark and the energy per second respectively. These graphs were calculated from the frequency data including the 24 nF capacitance using the formulas presented in the theory section and the voltage as measured across the gap which was typically between .5-3 kV. The energy per spark is a decreasing function due to the increased frequency, but the energy per second is an increasing function also due to the increased frequency as the total energy input has been increased. 36 4.4 Reshaping Temperature Particle diameter [nm] Copper Reshaping 90 80 70 60 50 40 30 20 10 0 80 nm 50 nm 30 nm 0 200 400 600 800 1000 1200 Furnace temperature [°C] Figure 4.4.1: Particle diameter scanned by DMA 2 is shown after varying temperatures in the reshaping furnace for 30, 50, and 80 nm particles determined with DMA 1 Values for 30, 50, and 80 nm particles under various temperatures were taken in order to determine the amount of reshaping that occurs over time. From the graph we can see that the values level off around 600 °C where they become spherical which is roughly 2/3 of the bulk material property as expected from the group’s previous work [2]. This also provides an indication that the particle material is pure copper whose melting point is 1085 °C, because this does not correspond to the values expected for copper oxide which is 1200 °C and oxides reshapes at a much higher temperature unlike pure metals [9],[17]. 4.5 Varying Electrode Size and Gap Size Similar distribution data was acquired for copper with a smaller 3 mm diameter electrode. The length of time the electrodes lasted was as we expect. Over dozens of hours of run time one pair of the thinner 3 mm gold electrodes were eroded away completely, and a second pair of 3 mm copper electrodes was eroded visibly while the thicker 6 mm copper electrodes showed little wear over a much larger time frame. A future experiment would be to precisely measure the electrode weight over time to see how 37 closely it corresponds to specific heat calculations, but in general this trend seems to be true. Size Distribution for a 1 mm gap Particles in millions [#/cm3] 0.7 0.6 0.5 5 mA 0.4 10 mA 0.3 20 mA 0.2 30 mA 0.1 40 mA 0 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.5.1: Size distributions of particles generated at different currents at a gap distance of 1 mm with a copper electrode with 3 mm diameter Particles in millions [#/cm3] Size Distribution for a 1.5 mm gap 0.7 0.6 0.5 0.4 5 mA 0.3 10 mA 0.2 20 mA 0.1 30 mA 0 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.5.2: Size distributions of particles generated at different currents at a gap distance of 1.5 mm with a copper electrode with 3 mm diameter 38 Particles in millions [#/cm3] Size Distribution for a 2 mm gap 0.7 0.6 0.5 5 mA 0.4 10 mA 0.3 20 mA 0.2 30 mA 0.1 40 mA 0 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.5.3: Size distributions of particles generated at different currents at a gap distance of 2 mm with a copper electrode with 3 mm diameter Peak Particle Diameter Particle diameter [nm] 80 70 60 50 40 1 mm 30 1.5 mm 20 2 mm 10 0 0 10 20 30 40 50 60 Current [mA] Figure 4.5.4: Maximum particle diameters of 3 investigated gap sizes gained from the size distributions at different currents for a copper electrode with 3 mm diameter 39 Millions Particle Count[#/µm2] Peak Particle Density 0.7 0.6 0.5 0.4 1 mm 0.3 1.5 mm 0.2 2 mm 0.1 0 0 10 20 30 40 50 60 Current [mA] Figure 4.5.6: Maximum particle densities of 3 investigated gap sizes gained from the size distributions at different currents for a copper electrode with 3 mm diameter The peak of the size distributions seems unaffected by the change in electrode diameter, but for two sets of data this is inconclusive. One point of note is the gap size changes visibly over time when smaller electrodes are used and this is very clear from acquiring the reshaping data as the frequency changed by several hundred hertz over hours of usage on the smaller electrodes; however, for the data here the gap is adjusted to be kept constant and to keep the frequency steady at increasing values relatively consistent with each other. The most unexpected and perhaps interesting part of the data is that for the larger gap size there is an inconsistency in the peak particle diameter. This seems due to the smaller gap distance having a very broad peak while at the largest gap size the peaks are sharper. Additionally for the 1.5 mm gap data in figure 4.5.2 the size distribution decreases for an increased current from 20 mA to 30 mA. This is not terribly surprising as the consistency tests in section 4.3 showed large variance in particle densities from run to run, and the larger sets of 6 mm copper electrode data have several instances of crossover between increasing current scans. 40 4.6 The Other Elements The initial silver data although far from conclusive displays a few general trends, primarily the peak points and size distribution data for silver is roughly similar to that of copper and a larger difference would be expected if the particle density followed a linear correlation to the specific heat. A higher voltage was required to cause a spark to form at values of 5, 6, and 6 kV for 1, 1.5, and 2 mm gap. More data or identical data is needed to make a strong statement on any comparisons between elements, but current data shows that as the current is increased the particle density increases and the peak diameter is increased with increasing current. This should be true for increasing gap size, but for the 2 mm gap the measured voltage had a sharper drop than expected and despite the increased current and decreased voltage the frequency became lower. Too low to agree with the previous data and went to 70 Hz on the 40 mA setting when the expected value should have been 400-500 Hz to be consistent with the other 2 mm silver scans and should be between 600-800 Hz to be consistent with the rest of the data sets. Size Distribution for a 1 mm gap Particles in millions [#/cm3] 0.7 0.6 0.5 5 mA 0.4 10 mA 0.3 20 mA 0.2 30 mA 0.1 40 mA 0 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.6.1: Size distributions of particles generated at different currents at a gap distance of 1 mm with a silver electrode with 3 mm diameter 41 Particles in millions [#/cm3] Size Distribution for a 1.5 mm gap 1 0.8 5 mA 0.6 10 mA 0.4 20 mA 0.2 30 mA 0 40 mA 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.6.2: Size distributions of particles generated at different currents at a gap distance of 1.5 mm with a silver electrode with 3 mm diameter Particles in millions [#/cm3] Size Distribution for a 2 mm gap 0.6 0.5 0.4 5 mA 0.3 10 mA 0.2 20 mA 0.1 30 mA 40 mA 0 0 20 40 60 80 100 120 Particle diameter [nm] Figure 4.6.3: Size distributions of particles generated at different currents at a gap distance of 2 mm with a silver electrode with 3 mm diameter. 42 Peak Particle Diameter Particle diameter [nm] 80 70 60 50 40 1 mm 30 1.5 mm 20 2 mm 10 0 0 10 20 30 40 50 60 Current [mA] Figure 4.6.4: Maximum particle diameters of 3 investigated gap sizes gained from the size distributions at different currents for a silver electrode with 3 mm diameter Particle Count[#/µm2] Peak Particle Density 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 mm 1.5 mm 2 mm 0 10 20 30 40 50 60 Current [mA] Figure 4.6.5: Maximum particle densities of 3 investigated gap sizes gained from the size distributions at different currents for a silver electrode with 3 mm diameter 43 Energy per Spark Particle Count[#/µm2] 120 100 80 60 1 mm 40 1.5 mm 2 mm 20 0 0 5 10 15 20 25 30 35 40 45 Current [mA] Figure 4.6.6: Energies per spark of 3 investigated gap sizes gained at different currents for a silver electrode with 3 mm diameter The measured values for the frequency did not increase according to the theory, and the measured voltage dropped significantly. This can be seen in Figure 4.6.6 since the energy here is calculated from the measured voltage across the gap. Perhaps the plasma formation takes more energy because the settings were at the lower edge of what creates a spark. The frequency was very low so perhaps ion channel formation had a higher overall resistance because of recombination between sparks. Another possibility is because of the very small electrodes there could be a formation of localized turbulent flow. Also due to the geometry of the smaller electrodes there is a difference in the electric field; perhaps this difference is enough to cause changes in ion channel formation and particle formation. 44 5 Conclusions The methods and tools described in the methods section can be applied to all new Spark Generator Systems. ο· For the 6 mm copper data an increase in the current causes an increase in the frequency and an increase in the number of particles formed. The increased current means there is a larger amount of energy in the process and causes a shift in the curve towards larger particle formation. For an increase in the gap size we find an increase in the peak particle size, and the total particle density was increased. ο· For the 6 mm electrodes an increase in the current causes the number of particles formed to increase and a shift towards larger particles as expected. However, for some the data acquired during the thesis work the gap size has the opposite effect than the 3 mm electrodes! Both sets of particle peak data show a dual peak in the 1 mm gap and a decrease in the 2 mm gap. Additionally the max particle data is inconclusive as well. There is much less data for the 3 mm electrodes, and perhaps this anomaly is just due to lack of data and having an unstable process. Still this happening repeatedly could be indicative of a real physical process and further investigation is needed. ο· One of the more interesting trends is that for the 3 mm electrodes the 1 mm gap on both silver and copper has a dip in the maximum values. A more detailed investigation into smaller gaps would be interesting to see if there is a dual peak in the production which might indicate there are two competing stable particle formations. ο· Also the measured frequency is not linear to the current increase, this could be due to a lack of data, or these could be indicative that plasma channels form easier under higher frequencies since the ion channel may not have time to dissipate fully. ο· Electrode material loss seems to follow specific heat calculations. A comparison of the weight of the electrodes before and after use under a stable process to be conclusive about how correlated to the energy required to evaporate metal is to particle formation would also be of future interest. 45 6 Future Work Future work includes varying the types of material and making various samples for use in work such as catalysis and nanowire synthesis. Creating known catalyst particles such as Gold-55, silver, and nickel particles will be necessary. Core shell particles are a possibility by adding a particle formation process in the furnace to create a shell, also creating bimetal particles by having two different electrodes. Janus particles or particles with their own p-n junction, or other shaped particles could also be a possibility. Control over the synthesis process would be needed but nanoparticles with desired electrical properties such as GaAs, InAs, or GaInAs and varying the doping in these particles to engineer band gaps could be interesting. Developing a method to create solutions out of aerosol nanoparticles, and measuring the refractive index by matching dielectric constant of the solution could also be of interest. Additionally a solar cell made by nanoparticles that are hydrophobic or hydrophilic is a possibility. 46 7. References: [1] Tsakalakos, (2010). Nanotechnology for Photovoltaics. 1st ed. Boca Raton Fl. USA.: CRC Press. [2] M. Messing: Engineered Nanoparticles Generation, Characterization, and Applications. Lund University, Division of Solid State Physics, Department of Physics 2011 [3] Martin Karlsson, (2004). Methods to Generate Size- and Composition Controlled Aerosol Nanoparticles. 1st ed. Sweden: Media Tryck. ISBN 91-628-6034-8 [ONLINE] Available at: http://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=467000&fileOId=1779 071 [Last Accessed 6/19/2012]. [4] Friedrich Paschen,(1889). Ueber die zum Funkenübergang in Luft, Wasserstoff und Kohlensäure bei verschiedenen Drucken erforderliche Potentialdifferenz. Annalen der Physik. 273 (5), pp.69-96 [ONLINE] Available at: http://onlinelibrary.wiley.com/doi/10.1002/andp.18892730505/abstract [Last Accessed 6/19/2012]. [5] J. M. Meek, (1940). A Theory of Spark Discharge. Physical Review. 57 (57), pp.722 [ONLINE] Available at: http://prola.aps.org/pdf/PR/v57/i8/p722_1 [Last Accessed 6/19/2012]. [6] Claire Tendero, Christelle Tixier, Pascal Tristant, Jean Desmaison, Philippe Leprince, (2006). Atmospheric pressure plasmas: A review.Spectrochimica Acta. 61 (B), pp.2-30 [7] N. S. Tabrizi, M. Ullmann, V. A. Vons, U. Lafont, A. Schmidt-Ott: Generation of nanoparticles by spark discharge. J Nanopart Res., 2009, DOI 10.1007/s11051-008-9407-y [8] B. Mueller: Generation of Nanoparticle Aerosols by Spark Discharge. Aerosol Science & Technology. AST-REV-2011-18, 2011 [9] Carl Yaws, (October 1, 1998). Chemical Properties Handbook: Physical, Thermodynamics, Engironmental Transport, Safety & Health Related Properties for Organic & Inorganic Chemical. 1st ed.: McGraw-Hill Professional. [10] Hinds, ( January 19, 1999). Aerosol Technology. 2nd ed. New York: Wiley-Interscience. 47 [11] (2012). Fundamentals of Mass Flow Control. [ONLINE] Available at: http://www.advanced-energy.com/upload/File/White_Papers/SL-MFCFUND-270-01.pdf. [Last Accessed 6/19/2012]. [12] R. Nave (2012). Voltage Divider. [ONLINE] Available at: http://hyperphysics.phy- astr.gsu.edu/hbase/electric/voldiv.html. [Last Accessed 8/22/2012]. [13] TSI (2009). Series 3080 Electrostatic Classifiers . [ONLINE] Available at: http://cires.colorado.edu/jimenez-group/Manuals/SMPS_3080_manual.pdf. [Last Accessed 9/12/2012]. [14] TSI, June 2002, Model 3089 Nanometer Aerosol Sampler Instruction Manual Revision A, TSI Incorporated, Saint Paul, Mn. USA. [15] TSI, Oct. 2009, Model 3068B Aerosol Electrometer User’s Manual Revision B, TSI Incorporated, Shoreview, Mn. USA. [16] K. Storm (2012). About NanoDim. [ONLINE] Available at: http://www.nanodim.net/. [Last Accessed September 5th 2012]. [17] M. N. A. Karlsson, K. Deppert, L. S. Karlsson, M. H. Magnusson, J. -O. Malm and N. S. Srinivasan, (2005). Compaction of agglomerates of aerosol nanoparticles: A compilation of experimental data. Journal Of Nanoparticle Research. 7 (1), pp.43-49 48 8 Appendix: 8.1 Major Component list: ο· MFC: o Aero FC – 7700 CU o Aero FC - 7710 CU o 2x Brokhorse NL 7261 ο· DMA: o 1) TSI 308010 ο· High Voltage Source: o Technix CCR15-P-750 ο· Spark Chamber: ο· DAQ: o NI 9403 o NI 9205 o NI 9264 ο· Power Supply: o FUG HCN 7E – 3500 o FUG HCN 7E - 6500 ο· Sintering Furnace: o Linn Eurotherm ο· ESP o TSI 3089 ο· EM o TSI 3068 B 49 8.2 SDG Operating Procedure By Jonathan E. Pautler Startup 1. Turn on Computer a. ID: Gotta b. Pass: Ask 2. Set Electrode Distance 3. Connect Oscilloscope Probe a. Turn on Oscilloscope be sure to set channel 1 to frequency and channel 3 to current 4. Turn on Lab view Program (Version 2011) a. close all windows (3 opens close 2) except main 5. Check all things are plugged in and gas flow from wall is on. 6. Simultaneously (or as close as possible) a. Turn all pneumatic valves DOWN i. Note: Bypass valve is labeled off as on. Turn it off (down) now. ii. EXCEPT gas flow 2 (last lever) leave that off unless you are using it. b. Turn on Vacuum Pump 7. Check Pressure and flow a. Stabilize pressure at 1015±5 mbar by adjusting needle valve (See program for value) b. Stabilize flow by adjusting the outlet valves if needed (Read from EM screen) i. Values for MFC’s should be ( 1.68 , 0 , 10 , 10 ) 8. Turn key to on position at HV source 9. Test Spark by (using the program) i. Disable “Disable” button by clicking 2x and then turn “On” button on. ii. The value from the source to create a breakdown voltage depends on the gap size and other things but setting to say 6kV and 30mA is a good starting point. This should generate a signal around 2.5kV in the program as read on the screen. iii. SPARK IS LIMITED TO UNDER 1KHz DO NOT GO OVER THIS LIMIT OR CAPACITOR BANK COULD BREAK! 10. Furnace to desired temperature i. Note: Takes ~20+ minutes to get desired temperature 50 Operating Procedure 1. Check values for flow and pressure 2. Connect system to proper gas flow TO CHANGE ANY VALVES: BE SURE TO OPEN FIRST AND CLOSE SECOND SO THE GAS ALWAYS HAS A PATH!!! a. To Bypass DMA 1 i. Open Valves 1 ii. Close Valve 2, 3 b. Bypass Furnace, Use DMA 2 i. Open Valves 4 outward, 5, 6, 9 ii. Close Valve 7, 8 c. Use Furnace, Bypass DMA 2 i. Open Valves 4 inward, 7, 8 ii. Close Valve 5, 6, 9 d. Use Furnace, Use DMA 2 i. Open 4 inward, 6, 7, 9 ii. Close 5, 8 e. Bypass Furnace, Bypass DMA 2 i. Open 4 outward, 5, 8 ii. Close 6, 7, 9 f. To Open ESP i. Open Valves 10 ii. Close Valve 11, 12 g. To reverse any of these simply switch the open and close numbers. I.E. to close the ESP you would OPEN 11,12 and Close 10. 3. Turn Spark to Desired Settings 4. Check that EM is working properly 5. To Run Scans a. Goto Scan Menu and adjust Settings 6. To Deposit a. Place Sample in ESP b. While ESP is offline turn on and set to desired deposition voltage c. Open gas line to ESP (See 2f) d. Turn on ESP e. Turn off ESP when Deposition time has been reached f. Close gas line to ESP 51 Shutdown 1. Turn off Spark by using Program 2. Turn off and disable HV Source (via program) 3. Turn off key to HV Source 4. Turn off Furnace 5. Simultaneously (or as close as possible) Close down gas flow a. Turn all pneumatic valves UP i. Note: Bypass valve is labeled off as on. Turn it on now. b. Turn off Vacuum Pump 6. Close Lab view Program (hit Stop then Exit) 7. Turn off Oscilloscope 8. Turn off power to All plugs (on side) Valve Labels 1 DMA 1 Bypass 2 DMA 1 Inlet 3 DMA 1 Outlet 4 Three way valve for Furnace Inlet, DMA 2 Inlet, or DMA 2 Bypass 5 DMA 2 Bypass 6 DMA 2 Inlet when Furnace is bypassed 7 DMA 2 Inlet 8 DMA 2 Bypass 9 DMA 2 Outlet 10 Electrometer 11 ESP 12 ESP 13 Outlet Bypass IN CASE OF EMERGENCY: CONTACT BENGT MUELLER @ --- --- ---OR JONATHAN PAUTLER @ --- --- ---- Known Bugs 1. If the Program FREEZES a. Immediately turn off spark and gas line manually. i. Turn off Spark by the key and button on HV supply ii. Turn off Gasline by closing pneumatic valves and pump 52 b. Restart the program 2. If the power goes out a. Turn off system so when the power is restored the machine will not start 53