Development and Optimization of a Catalytic Thruster for Hydrogen Peroxide Decomposition

ABSTRACT: Hydrogen peroxide is a non toxic, low cost, green monopropellant with a significant potential for applications in satellite microthrusters. It presents a density-specific impulse similar to hydrazine. A new bulk mixed oxide catalyst for decomposition of hydrogen peroxide has been developed and tested in a 2 N thruster with modular parts, allowing the use of catalytic beds with different diameters and lengths. A method of experiment design with a star configuration was implemented to optimize the catalytic bed geometry, in order to yield complete peroxide decomposition, reduce thruster size and volume and provide maximum thrust.


INTRODUCTION
Nowadays, most satellites in Earth orbit utilize hydrazine catalytic decomposition within their propulsion systems.Hydrazine (N 2 H 4 ) is an expensive and highly toxic (carcinogenic) monopropellant.In the last decade, due to growing concerns about the environment, there has been a significant interest for storable and nontoxic liquid propellants.Hydrogen peroxide (H 2 O 2 ) is one of the most important candidates for application as a non-toxic, low cost, green-propellant in satellite propulsion systems (Wernimont, 2009).
The traditional catalyst used for H 2 O 2 decomposition is made of superposed silver gauzes which form the catalytic bed.The use of solutions with high concentration of H 2 O 2 leads to the increase of the adiabatic decomposition temperature of the monopropellant (632°C for 85%; 755°C for 90%; 953°C for 98% in mass), making not viable the application of pure silver or silver coated catalysts in long duration operations.The melting point of silver (962°C) is very close to the decomposition temperature of H 2 O 2 and, besides, the formation of silver oxide does not allow the extended use of this type of catalyst (Kappenstein et al., 2002).However, it is notorious that the H 2 O 2 purity level, meaning the amount of stabilizers present in the monopropellant, has high influence in the efficiency of the catalytic system.
The reaction of H 2 O 2 decomposition for propulsion applications is very fast, where the induction velocity is very dependent on the active sites exposed to the reaction.Since this reaction is extremely exothermic, a few miliseconds after the contact of the monopropellant with the catalyst, the catalytic bed reaches a temperature high enough to make the reaction almost entirely of thermal nature instead of catalytic.
A large number of transition metal oxides are catalytically active for H 2 O 2 decomposition (Goldstein, 1974).Nevertheless, a catalyst frequently considered for H 2 O 2 decomposition, is MnO 2 .However if there is oxidation of this oxide, it will change to Mn 2 O 3, that presents a low catalytic activity.On the other hand, higher temperatures lead to the formation of Mn 3 O 4 , which is as active as MnO 2 (Hasan et al., 1999).Rusek (1996) proposed manganese oxides, supported on alumina and silica, to replace the traditional silver catalysts.The results obtained suggest that the catalytic properties of manganese oxides are influenced by the type of support, hence the interest in new, supported catalyst, containing high concentration of surface active species, which can be stable under reaction conditions, typical of space propulsive applications.
Mixed oxides, containing cations of transition metals, are frequently used as bulk catalysts in different reactions.These oxides are generally prepared by thermal decomposition of different precursors, such as hydroxides, carbonates, nitrates and oxalates (Vaccari, 1998).
Taking into account these considerations and the results of preliminary tests, bulk catalysts were developed.Extrudates of cobalt/manganese based mixed oxides were prepared for H 2 O 2 decomposition, employed in satellite microthrusters.The materials were tested in a modular 2 N microthruster, specially designed and built with catalytic beds of variable sizes (diameter and length), for catalytic decomposition of stabilized H 2 O 2, concentrated to 90% in mass.
A method of experiment design, using a star configuration, was employed to optimize the catalytic bed dimensions, aiming to obtain complete decomposition of propellant, the reduction of thruster volume and mass, and to yield maximum thrust.
The modeling of the process was made by the response surface method, adjusting quadratic models to the experimental results, obtained from a factorial planning.Through this modeling, the response sensitivity to the variables can be estimated, besides determining the levels where the response is optimal.The interest factors studied were diameter and length of the catalytic bed, and the response of interest was the thrust generated by the system.

HYDROgEn pEROxiDE COnCEnTRATiOn
Stabilized 50% H 2 O 2 , supplied by Solvay Chemicals, was concentrated to 90%, in mass, by a counter flow of hot dry air.A deionization treatment may be done before concentration, in order to remove the stabilizers present in the peroxide.Stabilizers can reduce catalytic stability, however, in this work, stabilized peroxide has been used to evaluate the new catalyst.

CATAlYST pREpARATiOn
Extruded material was prepared from a bulk catalyst, based on manganese and cobalt oxides, capable to spontaneously decompose cold H 2 O 2 .The catalyst was prepared by co-precipitation of solutions of cobalt, manganese and aluminum nitrates (Sigma-Aldrich) in a Na 2 CO 3 solution at room temperature, keeping a constant pH of 10, by addition of a NaOH solution.A mole ratio of 4:1:1 (Co:Mn:Al) was used.The product of this reaction was filtered, washed, dried and macerated.The powder was then peptized with HNO 3 solution and finally extruded, followed by stove drying and calcinations at 900°C temperature.The final product presented a cylindrical form with approximate dimensions of 2 mm diameter and 3 mm length.

MiCROTHRuSTER
Performance parameters of the new mixed oxide bulk catalyst for H 2 O 2 decomposition were obtained by tests with a 2 N thruster, manufactured with modules, including different chambers, nozzles and ring adapters built with 316 stainless steel (Fig. 1).To vary the dimensions of the catalytic bed, three chamber modules were designed and manufactured, besides two nozzles and several rings of different lengths and thicknesses, to hold the catalytic beds in place.
The NASA CEA rocket performance code (McBride & Gordon, 1996) was used to calculate the thruster theoretical performance.Chamber pressure was chosen as 5 bar, and the nozzle expansion ratio was calculated as 1.45, to adapt the nozzle at 600 m altitude, where the tests were performed.Assuming frozen flow, with 90% H 2 O 2 , the calculated specific impulse was 1040 m/s, theoretical characteristic velocity was 940 m/s and ideal thrust coefficient was 1.106.Figure 2 shows a computer view of thruster and thrust balance.Pressure transducers and thermocouples were placed before and after injection and after the catalytic bed, as can be seen in Fig. 3. Thrust was measured by a 5 N load cell.A graphical user interface, written with LabView software, was used to monitor and control the tests.

TESTS
Before tests, the mass fl ow rate of propellant, as a function of diff erential pressure in the feeding line, was determined to estimate the peroxide tank pressure required to provide the desired mass fl ow rate during each test.Aft er this procedure, the microthruster was assembled and mounted on the thrust balance, the load cell HBM 1-PW4C3/500G-1 (error±0.1g)was calibrated using known weights and, fi nally, continuous and pulsed test fi rings were performed.

RESULTS AND DISCUSSION
The mixed oxides bulk catalysts prepared in this work had high mechanical strength and low specifi c area (~6 m 2 /g) due to the high temperatures in which they were calcined.
Nevertheless, they showed no signifi cant mass loss during tests in the microthruster.
Th e plan of experiments followed the methodology described by Barros-Neto et al. (1995).Table 1 presents   From this model, the infl uences of the variables in the test conditions were represented by level lines of the response surface, as depicted in Fig. 4. Th e responses are indicated on z-axis (thrust), with a color bar, as function of the independent variables indicated in the x (diameter) and y (length) axes.Table 2 shows the regression analysis to fi t the response function to experimental data.
As seen on Table 2, L and LD present lower coeffi cients than D, indicating that bed diameter presents a more signifi cant infl uence than bed length in this experiment.Th e analysis of variance, depicted on Table 3, demonstrates that the model is signifi cant,  as evidenced from the p-value (0.001) obtained.Th e model showed a high determination coeffi cient, which explains 96% of the variability in the response.
According to the contour graphic (Fig. 4), monopropellant microthrusters using the new bulk mixed oxide catalyst for rocket grade stabilized H 2 O 2 decomposition yield optimum thrust when the catalytic bed has length from 29 to 33 mm and diameter from 15 to 17 mm.Th rust, specifi c impulse and characteristic velocity curves obtained within these optimized dimensions (D = 15 mm and L = 30 mm), with chamber pressure P c ~ 5.3 bar, are shown, respectively, in Figs. 5, 6 and 7.
Th e propellant mass fl ow rate ( ) was determined from injection pressure (∆P) considering the following polynomial equation, obtained from a regression of 16 experimental data (up to 9 bar) using water injection, prior to fi re tests in the thrust balance: (1) Th e thruster specifi c impulse (Isp) was then determined by: (2)     where F is the measured thrust and g o = 9.8065 m/s 2 is the standard gravity acceleration (Sutton and Biblarz, 2000).
An average experimental specific impulse of 97 s was obtained (Fig. 6).This value is close to 106 s, the value calculated with the NASA CEA code (1996).Therefore, the experimental result is satisfactory, since various losses occur in the process.
Th e experimental characteristic velocity was obtained by: (3) where P c is the measured chamber pressure and A t = 3.6 mm 2 is the nozzle throat area.
Since the characteristic velocity is a function only of temperature and gas properties in the combustion chamber, one
the plan of experiments with a star confi guration built from the factorial planning 2 3 (2 factors and 3 study levels), added to three replicates at the central point, totalling 11 tests.A quadratic model was obtained using a factorial design, describing the infl uences of catalytic bed length (L) and catalytic bed diameter (D) on thrust (F): F = 2.22 + 0.084L + 0.247D -0.344L 2 + 0.0925LD -0.573D 2 Pressure transducers ermocouples F igure 3. Cut view of thruster with thermocouples and pressure transducers.

Figure 4 .
Figure 4. Contour graphic for the variable response studied: thrust F (N).

Figure 2 .
Figure 2. Computer view of thruster and thrust balance.
can use this parameter to examine the chamber independently of the nozzle.The theoretical value obtained was 940 m/s and the experimental average value was 822.3 m/s (Fig.7), corresponding to a characteristic velocity effi ciency of 87%, which indicates incomplete catalytic decomposition.However, this can be considered a high c* effi ciency for hydrogen peroxide decomposition, when compared to results ofPasini et al. (2007) as depicted on Table4.Finally, a continuous firing of 30 seconds (Fig.8) and a sequence of 12 pulsed firings of 5 seconds, at intervals of 5 seconds (Fig.9), were performed to check the stability of the CoMnAl catalyst, both tests adjusting P c ~ 5 bar.There was no evidence of catalyst deactivation with respect to time, despite the use of stabilized hydrogen peroxide.No fragmentation of the catalyst grains was observed after tests.Pressure drops in the catalytic bed were about 0.5 bar and approximately constant (Fig.10).However, long duration and shorter pulse tests, at various mass flow rates and pressures, will still be performed to evaluate thruster performance on a broader operational range.

Table 1 .
Plan of experiments.

Table 2 .
Regression analysis using a 2 3 full factorial central composite design.

Table 3 .
Analysis of variance.
SS = sum of squares; MS = mean square; F and P = statistical parameters which indicate how far away a result is from the hypothesis