Experimental Vaporization and Computational Modeling of Ionization of Hydroxylammonium Nitrate in Vacuum Conditions IEPC /ISTS-2015-b-237

Transcription

1 Experimental Vaporization and Computational Modeling of Ionization of Hydroxylammonium Nitrate in Vacuum Conditions IEPC /ISTS-2015-b-237 Presented at Joint Conference of 30th International Symposium on Space Technology and Science 34th International Electric Propulsion Conference and 6th Nano-satellite Symposium, Hyogo-Kobe, Japan July 4 10, 2015 Forrest G. Kidd, III 1, Matthew J. Baird 2, Greg Neff 3, and Kristina Lemmer 4 Western Michigan University, Kalamazoo, MI, 49008, USA Abstract: The Aerospace Laboratory for Plasma Experiments (ALPE) at Western Michigan University is currently investigating methods to efficiently vaporize ionic liquids. The ionic liquid hydroxylammonium nitrate (HAN) is investigated in this paper. A preliminary study of the vaporization of HAN has been completed. Variations from decomposition reactions observed at elevated pressures were seen when HAN was vaporized in vacuum. Possible products of HAN vapor are hypothesized but not conclusively proven. Simultaneously, a rigorous modeling effort to understand the ionization processes of vaporized ionic mixtures has been begun using COMSOL Multiphysics software. The initial model based on composition estimates from residual gas analysis (RGA) is also presented. a A B CO 2 C p E f() F k H 2 HNO 3 H 2O j J e n e n N 2 Nomenclature = switching function which is 1 when electron flux is directed towards wall, 0 otherwise = magnetic potential = magnetic flux density = carbon dioxide = heat capacity at constant pressure = electric field = electron distribution function = body force = thermal conductivity = forward rate coefficient = hydrogen = nitric acid = water = imaginary quantity = externally generated current density = mass flow rate = electron density = electron energy density = nitrogen 1 Doctoral Research Assistant, Department of Mechanical and Aerospace Engineering, forrest.g.kidd@wmich.edu 2 Undergraduate Research Assistant, Department of Mechanical and Aerospace Engineering, matthew.j.baird@wmich.edu 3 Graduate Research Assistant, Department Mechanical and Aerospace Engineering, greg.neff@wmich.edu 4 Assistant Professor, Department of Mechanical and Aerospace Engineering, kristina.lemmer@wmich.edu

2 NH 2OH = hydroxylamine NH 3 = ammonia NH 3OH-NO 3 = hydroxylammonium nitrate NO = nitric oxide N 2O = nitrous oxide O = oxygen OH = hydroxyl p = local pressure = outlet pressure q = heat transferred per unit time Q = general heat source supplied from inductively coupled plasma module Q gen = generated heat source term Q vh = viscous heating source R e = electron source term S en = electron energy source term T = local temperature T v = temperature of incoming vapor T w = chamber wall temperature u = velocity vector v = velocity vector of plasma v e,th = electron thermal velocity W p = pressure work = ratio of specific heats p = secondary emission coefficient for the pth positive species e = electron flux = electron energy flux p = flux of pth species to the wall = electron energy 0 = permittivity in vacuum r = relative permittivity = dynamic viscosity 0 = permeability in vacuum r = relative permeability = local density st = standard density = electrical conductivity k() = electron impact cross section = frequency I. Introduction NTEREST in dual mode propulsion has been rising over the past several years. The ability to expand mission Icapability through high thrust and high specific impulse modes is increasingly important as the cost of space missions grows. Ionic liquids are a promising propellant for dual mode propulsion systems. They are energy dense and their large molecular weight makes them suitable for electric propulsion. Further most ionic liquids are liquid at room temperature with negligible vapor pressure. This makes them easier to handle and store. However, most electric propulsion systems require the propellant to be vaporized and then ionized. This makes the low vapor pressure of ionic liquids a weakness as well as an asset. As such a detailed analysis of the vaporization and ionization processes and products of ionic liquids is required. The Aerospace Laboratory for Plasma Experiments (ALPE) at Western Michigan University (WMU) has developed and built an experimental setup that combines heat addition with vacuum application to increase the vapor pressure of ionic liquids and successfully vaporize hydroxylammonium nitrate (HAN). Combined with this experimental work, a comprehensive modeling effort to understand the properties of the plasma formed in future ionization experiments is underway. The initial results from this study are also presented. The multi-physics modeling software package COMSOL was utilized to build an inductively coupled plasma simulation. A combination of heat

3 transfer, laminar flow, and inductively coupled plasma modules was used in this model. This allows for approximations of expected electron temperature, plasma potential, electron density, ion density, and particle flux. II. Background HAN is the ionic liquid of interest in the presented research. The molecular formula of HAN is NH 3OH-NO 3. At atmospheric pressure it vaporizes at roughly o C depending on its molarity. 1 The decomposition reactions of HAN have been extensively studied due to its use by the nuclear industry. 2-6 The decomposition reactants of HAN vapor are considered to be HNO 2, H 2O, N 2O and N 2. HAN can self-catalyze because the intermediate product of HNO 3 and NH 2OH can react to accelerate decomposition. 7,8 Overall reaction processes for HAN were proposed as shown below: 6 4NH 3OH-NO 3 N 2 + 7H 2O + 2HNO 2 (1) 3NH 3OH-NO 3 N 2O + N 2 + HNO 3 + 5H 2O (2) However, these studies have primarily focused on HAN decomposition under high pressure or at atmospheric conditions. 3,9. If HAN is used as a propellant for gas phase electric propulsion it will be exposed to vacuum conditions. This may allow for reactants to vaporize before forming the traditional decomposition products or to form entirely new species. Therefore, this study focuses on the vaporization of HAN in vacuum. III. Experimental Setup The experimental setup at ALPE has been developed to successfully vaporize HAN in vacuum. Experiments are performed in a cylindrical stainless steel vacuum chamber that is inches-long by inches-diameter. The chamber is fitted with a Varian 600DS roughing pump with a pumping capacity of 21.2 cfm and a Varian Turbo V 550 turbo pump with a pumping capacity of cfm. The pumps are capable of maintaining pressures of 7e-3 and 1e-6 Torr respectively. The vacuum chamber is shown in Fig. 1. Figure 1. Vacuum Chamber for Vaporization Experiments. Facilities at ALPE at WMU for vaporization of HAN experiments. Residual gas analyzer is marked by a yellow arrow. Figure 2. Vaporization experiment schematic. Ceramic cup is surrounded by a tungsten heater. A thermocouple is imbedded in the ceramic to give the approximate temperature of the sample cup. Residual gas analysis (RGA) is currently enabled in the vacuum chamber. RGA is performed using a Stanford Research Systems RGA-100 residual gas analyzer. The RGA-100 is connected to a small side chamber fit with a TRIVAC D16B roughing pump with a 9.4 cfm pumping capacity and a Varian Turbo V 70 LP turbo pump with a cfm pumping capacity. These pumps provide the pressure gradient necessary for the RGA-100 to receive a

4 sample of gas from the vacuum chamber. RGA provides a measurement of the mass of species in a gaseous sample. The residual gas analyzer consists of three parts. A high energy electron source ionizes the sample gas. A quadrupole mass filter which consists of four electrodes induces stable oscillations in the species with correct mass/unit charge and causes all other species to ground themselves. A faraday cup collects the ions that have been filtered and detects them as an electrical current. For vaporization experiments a ceramic and tungsten heater was developed to heat the HAN. The heater consists of a inch-diameter ceramic cylinder with a 0.25-inch-diameter and 0.6-inch-deep cylindrical cup cut from the center. The ceramic is wrapped with 0.02-inch-diameter tungsten wire. A current limited EMS power supply provides the power to heat the tungsten wire. At vacuum the ceramic heater can reach 600 o C. Temperature is measured by a K-type thermocouple imbedded below the base of the cylindrical cup in a recess cut from the ceramic. Samples are contained in an aluminum foil cup and placed in the heater. A diagram of the heater system is shown in Fig. 2. IV. RGA was performed on the ionic liquid HAN. Relative species concentrations were measured continuously up to 250 C and at 50 C intervals above that up to 500 C. RGA was used to determine whether HAN vaporizes, and if so at what temperature. Further, RGA was used to determine an initial estimate of the products of HAN vaporization as a function of temperature. If HAN decomposes as previous studies suggest the spectra will show peaks at 18, 28, 44, and 63 AMU/unit charge marking H 2O, N 2, N 2O, and HNO 3 respectively. Fig. 3 shows the difference between RGA traces measured with water for a baseline and with 91% HAN by mas in a water solution in the chamber. Each spectra was measured at a heater temperature of 200 C. The data from the baseline scan show peaks at 2, 16, 17, 18, 28, and 44 AMU/unit charge. These most likely are due to H 2, H 2O, N 2, and CO 2. The peaks at 16 and 17 AMU/unit charge may be due to species such as O and OH. However, they are more likely due to fragmentation of H 2O by the Residual Gas Analysis Results Figure 3. Comparison of HAN and H2O traces. Data for vaporization of each fluid were measured at 200 C residual gas analyzer. The high energy electrons created by the residual gas analyzer are more than enough to ionize most species and in some cases to doubly ionize them. However, these high energy electrons can also cause molecules to fragment. For example, H 2O can fragment into H 2 and O or H and OH. The fragmentation spectra for many molecules are fairly well known. 10 The fragmentation profile for H 2O taken from the RGA-100 analysis software is shown in Fig. 4. This fragmentation of molecules is why RGA traces often show more species present than would be expected. However, by taking into account the fragment profile of H 2O it becomes clear that the lines at 16, and 17 AMU/unit charge are due to H 2O fragmentation. Further, keeping in mind the large concentration of H 2O, the lines at 28 and 44 AMU/unit charge could be due to N 2 and CO 2 respectively which may be dissolved in H 2O in the HAN solution. The peak at 2 AMU/unit charge is most likely due to the small mass of H 2 as it is fairly difficult to keep it out of even the most carefully sealed vacuum system. Taking into account species that are connected to H 2O the primary components of HAN appear to occur at 16, 17, 28, 30, 33, 44 and 46 Amu/ unit charge. The peak at 44 AMU/ unit charge is quite small, and can be seen in the HAN spectra. However, it is difficult to determine whether this peak is due to CO 2 or N 2O, and experiments to determine the acidity of HAN and to test for the presence of nitrates will be performed to help determine this. Strong peaks at 17 and 28 AMU/unit charge were seen in scans with HAN and in scans with only H 2O, however, the intensity of the 17 AMU/unit charge peak is higher than the fragment profile of H 2O would predict. The fragment profile estimates that the 17 AMU/ unit charge peak would have 25% of the current of the 18 AMU/ unit charge peak. This is the case for the experimental water trace. However, for the trace with HAN,

5 the 17 AMU/ unit charge peak is roughly 31% of the current of the 18 AMU/ unit charge peak. This increase is most likely due to the presence of a species in the HAN vapor. Therefore, the 17 AMU/ unit charge line is considered to be a product of HAN. Further, if the 28 AMU/ unit charge peak was due to the residual N 2 it should have decreased in a similar fashion as the H 2, and if it is due to N 2 dissolved in H 2O it should have decreased along with H 2O. Therefore the peak at 28 AMU/ unit charge is considered to be a possible product of HAN. Possible species that have singly charged ions with AMU/ unit charge matching the peaks seen in the HAN trace and not the water trace or for reasons given above are shown in Table 1. The RGA- 100 was used to obtain qualitative estimates of the concentration of the species shown in Table 1. Their relative concentrations as a function of current collected by the residual gas analyzer are shown in Fig. 5a. The peak at 18 AMU/unit charge is shown with a separate axis because the large current made comparison difficult. Fig. 5b shows the current of each species normalized by the composition fraction of the species with the greatest current which is in this case H 2O. Examination of the profiles of the different species from Fig. 5a shows that the species at 17, 18, 44, and 46 AMU/ unit charge all follow the same qualitative shape. H 2O is shown to increase with temperature before decreasing and leveling. It is difficult to determine from Fig. 5a if this is due to a larger amount of H 2O being vaporized from the HAN solution or due to the formation of H 2O. For H 2O vaporization would increase with temperature but would level off as the supply of H 2O is expended. Studies with larger amounts of HAN Figure 4. Fragmentation profile of H2O. This profile shows the way gaseous species can dissociate when impacted by a high energy electron. Table 1. Possible Chemical Components of HAN Vapor AMU Products 16 O, NH 2 17 OH, NH 3 28 N 2 30 NO 33 NH 2OH 44 CO 2, N 2O 46 NO 2 could be used to determine the cause of this shape. Fig. 5b shows two distinct paths that the species present when HAN is vaporized follow as the temperature increases. The first is a saddle shape seen most strongly in 16 and 33 AMU/ unit charge species, and somewhat at 30 AMU/unit charge. The second path is a more Gaussian-type profile with an extended wing as temperature increases. This is seen in species indicated by 28 and 44 AMU/unit charge. These two shapes could correspond to two competing reactions making the species at 17 and 46 AMU/unit charge components in both reactions due to their flat profiles. The peak at 30 AMU/unit charge appears to be a product of a third reaction as well due to its large increase with temperature. Equation 2 showed a decomposition reaction of HAN that created N 2, HNO 3, H 2O and N 2O. Current was measured at 18, 28, and 44 AMU/unit charge which could mark H 2O, N 2, and N 2O respectively with the Gaussian-like paths, but no current marking HNO 3 was observed at 63 AMU/ unit charge. As such it is probable that the reactions governing HAN decomposition in vacuum are different from those previously observed in HAN decomposition at atmospheric pressures and above, resulting in HNO 3 no longer being a product. 3,6

6 a) b) Figure 5. Relative Composition from RGA: a) Raw b) Normalized by H2O. Two competing reactions appear to be taking place during HAN decomposition. One has a more Gaussian profile as seen at 28 AMU/ unit charge, while the other takes on a saddle shape as seen at 33 AMU/ unit charge. V. COMSOL Modeling A further goal of this project is to study the ionization processes of ionic liquid mixtures after vaporization. Therefore, an initial model of the plasma created when the ionic vapor mixture is ionized is undergoing development. At the time

7 of this paper, the code is capable of simulating an inductively coupled plasma based on mole fractions of molecular species input into the simulation. Currently, these are chosen as H 2O, NO, and NH 3 due to the likelihood of their presence in HAN vapor as determined from RGA previously discussed. The model consists of a coupled set of physics modules within the multi-physics simulation software COMSOL. The modules that are utilized are the laminar flow, heat transfer, and inductively coupled plasma modules. The physics is defined in 2D axisymmetric cylindrical coordinates and solved simultaneously in time. However, due to the low pressure of the model, interactions between neutral species and unit charged species are considered to be minimal. Therefore, steady state equations are solved for the laminar flow module, and time dependent equations are solved for the heat transfer and the inductively coupled plasma modules. COMSOL uses the PARDISO nonlinear solver to solve the system of equations before progressing to the next time step through its time-dependent solver 11. A. Model Description Fig. 6 shows the complex mesh geometry used in the COMSOL simulation. The letters correspond to boundary condition definitions described below. A free triangular mesh is used for the majority of the model with a quadrilateral mesh being used in the coils and boundary layer to accurately define the plasma-rf power interactions. The majority of mesh elements are focused on the coils and boundary layers. The model is divided into roughly two sections: (1) the ionization chamber (ABCD) and (2) the exterior (DCFE). Laminar flow, heat transfer and most inductively coupled plasma equations are solved only in the ionization chamber. Ampere s law, as part of the inductively coupled plasma module, is solved in both sections. Laminar flow is in the positive z-direction with the gas composition mole fraction initially being 0.2 NH 3, 0.3 NO, and 0.5 H 2O. These initial mole fractions are used to demonstrate the effectiveness of the model and simulation and are based on the initial values found from the RGA vaporization study. Figure 6. Mesh geometry used in the COMSOL simulation of the ionization process. The letters correspond to locations on the boundary where specific boundary conditions are described. Currently, elastic and ionization electron impact reactions for each species are being considered, and reverse reactions are being ignored as the boundary conditions are a more efficient means of implementing them. The reaction coefficients are calculated from electron impact cross section data acquired from the LXCAT database 12 using the following equation: (3)

8 B. Governing Equations and Boundary Conditions The model is symmetric around the centerline (AB as shown in Fig. 6) with no gradients in the radial direction. The governing equations for the laminar flow module consist of the 2D Navier-Stokes and total mass continuity equations: (4) (5) Boundary conditions for the laminar flow module are shown in Table 2. The pressure at the outlet (BC) is fixed at the vacuum set pressure; however, if the local pressure were to drop below the vacuum set pressure the outlet pressure is set to drop to avoid inflow. A no slip wall condition is set on the surface of the interior walls (CD). The inlet (AD) flow is controlled by mass flow rate based on standard density. The mean molar mass required to calculate standard density is supplied by the inductively coupled plasma module. Table 2. Boundary Conditions of Neutral Flow Module Boundary AB BC CD AD The heat transfer module is governed by the time dependent thermal energy equation (6) The boundary conditions for the heat transfer analysis are shown in Table 3. At the outlet (BC), heat is allowed to leave the system through convection, but radiation is ignored due to the low gas temperature. The interior boundary of the ionization chamber (CD) is set to the vacuum chamber wall temperature. The interior of the ionization chamber (ABCD) is set to the temperature of the incoming vapor. No heat flux is allowed through the inlet (AD). Thermal conductivity, density and specific heat needed for calculations are provided by the inductively coupled plasma module. The pressure and velocity distribution of the gas needed for calculations are provided by the laminar flow module. Table 3. Boundary Conditions for Heat Transfer Module Boundary AB AD CD ABCD BC The governing equations for the inductively coupled plasma module are described by

9 (7) (8) (9) (10) The boundary conditions for the plasma model are shown in Table 4. The electrons are allowed to leave through outlet (BC) to preserve quasi-neutrality, and no secondary emission or thermionic emission is allowed from any interior walls (CD). Table 4. Boundary Conditions for the Plasma Module Boundary CD BC ABCD Eq. (10), Eq. (11), AD (11) B. Parametric Study A parametric study of outlet pressure and area normalized flow rate (nmf) was performed using the COMSOL model. The electron density at two normalized settings is shown in Fig. 7. Above a critical nmf the electron density increases by over 8 orders of magnitude. This is most likely due to the inability of energetic electrons to achieve a sufficient number of collisions before this point. This along with the ovular shape of the plasma density expected from inductively coupled plasmas seems to show that the plasma model accurately models inductively coupled plasma physics. The second goal for the COMSOL simulation is to model the chemistry of the plasma. COMSOL uses the electron impact collision reactions along with any other reactions defined by the user to model plasma chemistry. As such the modeled chemistry will change, possibly quite drastically, if the inupt reactions are changed. Therefore, modeling the chemistry is an iterative process. As more information is provided from future ionization experiments the modeled reactions will be updated to more closely match experimental data. The ionization fraction for the various species currently being modeled is shown in Fig. 8. With the existing chemistry, NO ionizes the most readily, indicating that NO is a species on which future ion ionization experiments should focus.

10 a) b) Figure 7. Electron density of ionization chamber: a) 150 nmf b) 200 nmf A large increasea in electron density is seen above a critical geometry specific flow rate. Figure 8. Ionization fraction of plasma species. Computational modeling of the plasma allows for estimates of plasma behavior including the ionization of species. VI. Future Work An issue that was seen in this study is that there are multiple possible species which have the same AMU/ unit charge. This made it difficult to differentiate with only RGA data. Therefore, other diagnostic methods are needed. Experiments are planned with optical emission spectroscopy (OES) and Fourier transform infrared absorption spectroscopy (FTIR). OES works by measuring the emission of fluorescence from a vapor and then correlating that to

11 both its composition and other physical traits. By determining which peaks in fluorescence are seen, the composition of the HAN vapor can be determined more accurately. FTIR works by passing an infrared beam through the sample. This beam is then measured by a sensor. By determining at what wavelengths radiation is absorbed by the sample, information on the bonds making up the various species can be determined. This allows for better resolution of the species as it can complement OES when fluorescence from more than one species overlaps. Future modeling efforts will focus on changing the reactions implemented into COMSOL until the plasma chemistry is accurately modeled. This will be an iterative process as results from OES and FTIR give more information on the vapor composition both during and after vaporization and during ionization. VII. Conclusion An initial study of the vaporization processes of the ionic liquid HAN in vacuum has been performed. RGA indicates that the reactions governing HAN have changed from those seen at atmospheric and higher pressures. Two primary reactions appear to take place with distinct shapes as functions of temperature. The difference in these reactions from those seen previously is apparent from the large amount of NO present in the vapor and the lack of HNO 3. As RGA takes place downstream of the vaporization these differences could be caused by reactions that take place after decomposition. Additionally, an initial model for the plasma physics and chemistry of HAN vapor has been developed using COMSOL Multiphysics based on initial composition estimates from RGA. The plasma physics showed expected trends for inductively coupled plasma, and the chemistry shows that NO is a species upon which future experiments should focus. Acknowledgments The authors would like to thank the Air Force Office of Scientific Research for support of this research. References 1 Sasse, R., Thermal Characteristics of Concentrated Hydroxylammonium Nitrate Solutions, BLR-MR- 3651, Zhang, C., Thermal Decomposition Study of Hydroxylamine Nitrate During Storage and Handling, Masters Thesis, Chemical Engineering Dept., Texas A&M Univ., College Statiion, TX, Wei, C., Thermal Runaway Reaction Hazard and Decomposition Mechanism of the Hydroxylamine System, Ph. D. Dissertation, Chemical Engineering Dept., Texas A&M Univ., College Station, TX, Schoppelrei, J. W., and Brill, T. B., Spectroscopy of Hydrothermal Reactions. 7. Kinetics of Aqueous [NH 3 OH]NO 3 at K and 27.5 MPa by Infrared Spectroscopy, J.Phys. Chem. A 101, 1997, pp Rafeev, V. D., and Rubtsov, Y. L., Kinetics and mechanism of thermal decomposition of hydroxylammonium nitrate, Russian Chemical Bulletin, vol. 42, 1993, pp Oxley, J., and Brower, K., Thermal Decomposition of Hydroxylamine Nitrate, Proc. SPIE, Propulsion, vol. 63, Los Angeles, CA, Gowland, R., and Stedman, G., Kinetic and product studies on the decomposition of hydroxulamine in nitric-acid, Journal of Inorganic and Nuclear Chemistry, vol. 43, 1981, pp Chervin, S., and Bodman, G., Phenomenon of autocatalysis in decomposition of energetic chemicals, Thermochimica Acta, vol , 2002, pp

12 9 Djik, C., and Priest, R., Thermal Decomposition of Hydroxylammonium Nitrate at Kilobar Pressures, Combustion and Flame, vol. 57, 1984, pp RGA-100, Programming Reference Models RGA100, Residual Gas Analyzer, Stanford Reseach Systems, Sunnyvale, CA, Luisier, M., and Schenk, O., Fast Methods for Computing Selectred Elements of the Green s Function in Massively Parallel Nanoelectronic Device SImulations, Springer-Verlag, 2013, pp Hayashi, M., Hayashi database, Electron Impact Cross Section Database [online database]. URL: [cited 17 February 2015].