A latent highly activity energetic fuel: thermal stability and interfacial reaction kinetics of selected fluoropolymer encapsulated sub-micron sized Al particles

Aluminum can enhance heat release of energetic composite in theory. However, the commonly used micron aluminum powder has several short comings like incomplete reaction and low reaction rate. Meanwhile, outer oxide shell of nano Al particle is thicker than micro Al, which leads to low active aluminum content and insufficient heat release. On the basis of previous research, reported fluoropolymers modified Al particles were compared and suitable F2311was chosen. Sub-micron scale Al (median particle size around 200 nm) was regarded as optimum coated object in consideration of activity content of aluminum powder changing with particle size. The super fine Al powder was prepared by electrical explosion method, and encapsulated in situ by selected fluorine rubber F2311. The experiments on thermal stability demonstrated F2311 coating thickness should be no less than 3.6 nm. These results were further confirmed by EXPLO5 thermo dynamic calculation. Calculated results showed that reaction characters of F2311 encapsulated Al exceeded conventional nano Al regardless of combustion and explosion. Scanning electron microscopy (SEM), transmission electron microscopy (TEM), laser particle size analyzer and X-ray photoelectron spectroscopy (XPS) were used to characterize coated products’ morphology, particle size distribution and interfacial bonding information. The results showed that the coated samples were generally spherical shape, with median particle size of 217.7 nm and coating thickness of 3.6 nm. The coating shell contained a small amount of alumina and aluminum fluoride besides fluoropolymer. The non-isothermal dynamic equations of Al/F2311 and Al/Al2O3 were deduced by TG/DSC simultaneous thermal analysis. Compared with conventional nano-Al, the apparent activation energy of Al/F2311 decreased by 45 kJ/mol and the first exothermic peak temperature was about 10 °C earlier. Moreover, heat release was nearly twice as conventional nano-Al. TG-DSC-MS coupled measurements certified that active Al was enveloped by ‘fluorine atmosphere’ while F2311 decomposed in range of 200–400 °C. Alumina was replaced with aluminum fluoride inside coating layer during 400–550 °C, which broadened the diffusion path and then accelerated the permeation of oxidizing gas. In addition, the exothermic of Al-F was obviously larger than Al-O. Consequently, the oxidation reaction was activated rapidly, especially in initial exothermic period. Fluoropolymer encapsulated sub-micron sized Al was a latent highly activity energetic fuel and a potential candidate for aluminum powder.

In addition, perfluoropolyether (PFPE), perfluorotetradecanoic acid (PFTD), and perfluorohexadecanoic acid (PFHD) were also involved in this literature. As well known to us, proportion of aluminum powder can be added up to 30% in energetic material formulation. As its coating agent, fluoropolymer should have certain flexibility, be beneficial to the reduction of mechanical sensitivity, have tensile strength to ensure the mechanical properties of the formula, have a moderate thermal decomposition temperature, not only to ensure the safety of storage, but also to avoid high temperature blocking the oxidation of aluminum. It should be specially pointed out that the coating experiment is carried out in the liquid phase environment, so it is necessary to investigate the solubility of coating agent. Some polymers with high fluorine content but poor solubility, such as PTFE, should be excluded. Table1 lists all the fluoropolymers mentioned in published literature to modify nano Al particles. The data in the table can clearly compare the strength, flexibility, oxidation element content, solubility and thermal decomposition temperature of various fluoropolymers. Considering the above factors, F2311 is suitable to be used as coating material for Al particles.
Al particle size to be coated. There are different thickness of oxidized shell on the aluminum surface.
And the phase transition temperature of oxidized shell is higher than that of aluminum. The structure analysis of aluminum powder shows that the oxide layer is mainly composed of octahedral Al(O 1/6 ) 6 and tetrahedral Al(O 1/4 ) 4 . The average density of the oxide layer is 2.98 g/cm 3 , about 3/4 of crystal Al 2 O 3 38 . The active aluminum content can be calculated combined with the pure aluminum density (2.70 g/cm 3 ). After processing by Origin, the relationship among diameter d, the oxide thickness δ and the active aluminum content can be described according to the results of TEM, as shown in Fig. 1. Piecewise function relations between aluminum oxide layer thickness and particle size can be seen from Fig. 1. After fitting (Appendix A), the correlation of shell thickness and particle size of alumina can be obtained. www.nature.com/scientificreports/ In Fig. 1, the thickness of the oxide layer is about 3 nm when d ≤ 70 nm. The oxide layer's thickness is approximately exponential with the particle size when d is between 70 nm and 1 μm. The thickness and active aluminum content both increases rapidly at first and then slows down after d exceeding 200 nm. Thus, it is speculated that submicron aluminum powder of 200 nm has both high active aluminum content and high reactivity like nano powder. Therefore, in this paper, aluminum powder with a median particle size of 200 nm was selected for encapsulation in situ.
Instruments and materials. Instrument: microphotographs were taken using a scanning electron microscope (SEM, SU8020, Hitachi, Japan) to observe the morphology of the samples. High resolution transmission electron microscopy (HTEM, Tecnai G2 F20, FEI, USA) was used to characterize the core-shell structure. To measure the particle size of samples, the dynamic light scattering nanometer particle size analyzer (Nanotrac Flex, Microtrac, USA) were used. The X-ray photoelectron spectrometer (XPS, Thermoescalab 250Xi) was used to analyze the chemical state of elements on the sample's surface. The thermal reaction kinetics and the thermal fragments during the reaction were characterized by the simultaneous thermal analyzer-quadrupole mass spectrometry linkage device (TG-DSC-MS, STA449F3TG-DSC, QMS403D, NETZSCH, Germany).
In-situ encapsulation. The preparation process of F2311 encapsulated submicron aluminum powder is shown in Fig. 2: firstly, the submicron aluminum powder was produced using electrical exploding wires method. The pulse power generator provided an instant high voltage (1.5 kV), and then the aluminum wire (diameter 0.2 mm) in the explosion chamber vaporized due to the electric explosion. The explosion chamber was filled  www.nature.com/scientificreports/ with argon gas, in which aluminum vapor cooled down and gathered to form submicron particles. The product was collected by separating-buffer tank-powder collection device, then directly dispersed into organic solvent. A certain amount ethyl acetate solution of F2311 was added while stirring. Changing the weight ratio of F2311 and aluminum powder would gain products with different thickness of shell. The solvent was dried at 60 °C to gain the F2311 encapsulated submicron aluminum powder. The final product is black powder. Ordinary aluminum powder was prepared by electric explosion as control under the same conditions.

Results and discussion
Morphology and particle size. SEM images were captured and coupled with EDS mapping results to characterize the micro-structural differences between F2311@Al and ordinary Al. As shown in Fig. 3a, the F2311 @Al is spherical with different sizes ranging from 1 to 400 nm. 70% of particles are within 200 nm, but there were also 2% of particles are above 400 nm, which was caused by the unstable voltage during the electrical exploding.
In terms of morphology, there is agglomeration and adhesion on the encapsulated samples compared with raw aluminum powders, which is due to the presence of polymer on the surface. Figure 3c is the SEM image of raw aluminum powder we can see that the F2311 coated aluminum powder remains spherical after the encapsulation just like raw aluminum powder. Fluorine (F) distributes evenly on the F2311@Al's surface according to Fig. 3b. Oxygen (O) existed on the surface of both samples, indicating that there is oxide on both samples. However, the O content of the encapsulated samples is lower comparing Fig. 3b with d. HTEM was used to observe the core-shell structure of F2311@Al. The samples were ultrasonic dispersed to weaken the agglomeration. Then, the encapsulated samples are dispersed in ethyl alcohol and dropped on micro-grid after drying. HTEM images shown in Fig. 4a demonstrates the core-shell structure of the samples. Aluminum was clearly separated from the outer organic material. The thickness of the outer layer is 3.5 nm, which may be the amorphous oxide of Al or F2311. The thickness of the coating layer was measured by HTEM, and the result showed that the average thickness is 3.6 nm.
The particle size distribution of coating samples was characterized by dynamic light scattering (DLS) particle size analyzer. 1.0 mg samples were dispersed for 0.5 h by ultrasonic in 150 mL ethyl alcohol before the measurement. The particle size distribution shown in Fig. 4b indicates that the particle size distribution of F2311@Al basically met the normal distribution and the median particle size D 50 was 217.0 nm.
Thickness of coating layer. The thickness of coating could significantly affect their application. Thus, the coating should be as thin as possible to assure the high energy release. However, the storage safety of the samples would be better with a thicker coating. Aluminum powders with different coating thickness could be prepared   Fig. 5, when the coating thickness is about 3.6 nm, it barely increased after being heated. When the coating thickness is less than 3 nm, the shell (coating and alumina) increased significantly after being heated, indicating that the coating is not dense when less than 3 nm, which means oxygen in the air can diffuse through the shell easily to trigger the slow oxidation on core-shell interface. And the growth of oxide will result in the shell increasing after heated. Therefore, the critical thickness of F2311 is 3.6 nm. Based on the above analysis, the parameters of F2311@Al were confirmed, namely, the median particle size of aluminum powder is about 200 nm, and the thickness of F2311 is 3.60 nm. In order to figure out the amount of ordinary aluminum powder and F2311@Al that could affect energetic composites, EXPLO5 6.03 was applied to calculate the combustion and explosion parameters (theoretical specific impulse and ideal detonation heat) of aluminum-containing energetic composites under the same condition based on BKW equation 39,40 and the principle of minimum free energy 41 . When calculating the detonation heat, we assumed that the system insists of 80 wt% CL-20 and 20 wt% Al 42 . When calculating the adiabatic constant pressure combustion, the system is composed of 15 wt% HTPB, 65 wt% AP, and 20%wt Al 43 . Al represents F2311@Al and ordinary Al, respectively. The F2311 in F2311@Al and alumina in ordinary Al share the same mass ratio. Because the density of fluororubber (1.85 g/cm 3 ) is less than that of alumina (2.98 g/cm 3 ), the thickness of F2311 is larger and denser. The results are shown in Fig. 6.  www.nature.com/scientificreports/ Figure 6a is the calculation result of detonation heat. It can be seen that within the range of 20-1000 nm, the detonation heat of F231@Al is higher than that of raw Al. The detonation heat of the system increases with the increase of particle size, which is due to the more active aluminum with the growth of particle size. Since fluorine will participate in exothermic reaction, as shown in Eqs. (1) and (2), when the mass of outer shell is settled, the detonation heat of F2311@Al is higher than that of ordinary Al. Based on the calculation results, we could see that when the particle size of aluminum powder is at the nanometer scale (≤ 100 nm), the content of active aluminum is relatively low, so the detonation heat is not low. When the size of aluminum powder is between 100 and 400 nm, the detonation heat of the two aluminum powders is significantly different, especially in the vicinity of 200 nm. When the particle size is above 400 nm, the detonation heat of the system does not increase significantly, and the detonation heat tends to be stable, which is gradually close to the calculation results of micron aluminum powder. Thus, it is confirmed that the aluminum powder with a median particle size of 200 nm used for encapsulated in-situ has excellent potential of releasing energy in the energetic composites. The calculation results of theoretical specific impulse show similar trends. When the particle size of aluminum powder is within the range of 100-400 nm, the theoretical specific impulse's difference is the largest. The formula of F2311@Al is significantly higher than that of ordinary Al, and the specific impulse remains unchanged after 400 nm. In conclusion, when the average particle size of submicron aluminum powder is 200 nm, and the fluororubber is used for encapsulated in-situ, the combustion and detonation efficiency of the energetic composites is better, and the reactivity of products is higher accompanied with better thermal stability. X-ray photoelectron spectroscopy analysis. The results of SEM show that there is a certain amount of O on the F2311@Al coating sample. XPS were used to analyze the surface valence bond states of F2311@Al. The measured depth of XPS is 3-5 nm, which is similar to the thickness of the coating layer. Therefore, the coating layer information can be inferred. Firstly, the presence of C, O, Al and F on the surface is determined by widespectrum scanning. Then C, Al and F were narrow scanned, and the XPS spectra is shown in Fig. 7. Figure 7a shows that there are four chemical states in C, which are C1s (284.8 eV), C-Cl (286.2 eV), CF (288.8 eV) and CF 2 (291.0 eV), all of which are from F2311. Figure 7b shows that there are four chemical states in Al, which are Al 2 O 3 (74.2 eV), aluminum (72.1 eV) 44 , Al-F (75.1 eV) 45 and Al 2p1 (73.1 eV), indicating that there are a small amount of alumina on the surface. The results of Al and F elements showed that there are Al-F bonds in the coating layer, (75.1 eV, 684.9 eV). It is speculated that there are two main ways to generate Al-F compounds. One is that Al atoms reacted with fluororubber at the interface. Another is that the alumina on the surface reacted with fluoride. Since F is more electronegative than O, F replaced O to form AlF 3 . In addition, Fig. 7c demonstrates that the peak strength of Al-F bond is much lower than CF 2 , indicating that the F2311 exists mostly as polymer, and will form chemical bonds only when it contacts with Al directly. According to the XPS results, there are F2311, AlF 3 , and a small amount of alumina on the surface of F2311@Al.

Non-isothermal reaction kinetics.
Previous studies about Al/F reaction were conducted in argon or other inert gas, while Al/F composites are usually used in open systems or in aerobic environments. Therefore, we investigated the adiabatic and non-isothermal reactions between F2311@Al and ordinary aluminum in air systematically in this study. Figure 8 shows the TG/DSC curves of two aluminum powders at different heating rates. The TG curves show that there are two stages of weight gain (550-650 °C and 750-850 °C), corresponding to the two obvious exothermic peaks in the DSC curves.
(1) 2Al + 3CF 2 → 2AlF 3 + 3C + 1481 kJ/mol (2) 2Al + 1.5O 2 → Al 2 O 3 + 455 kJ/mol www.nature.com/scientificreports/ At the heating rate of 10 °C/min, the peak temperature of F2311@Al in the first stage is 11.4 °C lower than that of ordinary Al, and the heat release is 2.80 kJ/g higher. At the heating rate of 15 °C/min, the peak temperature of F2311@Al in the first stage is 10.6 °C lower, and the heat release increases by 3.44 kJ/g. The reaction temperature of F2311@Al is lower, and the heat release is around twice as high as that of ordinary aluminum. However, there is a slight delay in the reaction of F2311@Al in the second stage of weight gain. The reaction processes of two samples are quite different. After 1000 °C, the weight gain of both samples is close to 150%.
To further investigate the response mechanism of the F2311@Al and ordinary aluminum, the thermal reaction integral model functions were fitted using non-isothermal chemical reaction dynamics. Activation energy Ea and pre-exponential factor A were calculated by Ozawa method, and 41 types of kinetic model functions and α-T (α-T data is the conversion degree (α) varies with temperature (T)) data were calculated 46 . Eventually, the two stages' most probable mechanism functions and kinetics parameters were selected and shown in Table 2. More detailed analysis could be seen in Appendix B.
It can be seen from Table 2 that the mechanism of F2311@Al first-stage oxidation process is the Jander equation of n = 1/2, and the process is controlled by the three-dimensional diffusion rate. The integral formula is G(α) = [1 − (1 − α) 1/3 ] 1/2 , and the corresponding differential form is f(α) = 6(1 − α) 2/3 [1 − (1 − α) 1/3 ] 1/2 . At this time, the fluoride in the shell has completely decomposed, and because of the lack of a tight layer of alumina binding, aluminum particles are surrounded by "fluorine atmosphere" and hot air. The rate of Al oxidation depends on the diffusion rate of these oxidizing gases. The most probable mechanism function is differential equation f(α). The average Ea and A were substituted into the equation dα/dt = Af(α)e −E/RT , and the kinetic equation was obtained as dα/dt = 10 22.74 The Mapel Power law is applicable to the first-stage reaction process of ordinary Al. The control step is onedimensional phase boundary diffusion, the integral formula is G(α) = α, and the differential formula is f(α) = 1. This is due to the existence of alumina on the ordinary Al surface, which is relatively steady and will not participate in the reaction. Therefore, the oxidation reaction process depends on the diffusion rate of oxygen in the shell. The data in Table 2 showed that the apparent activation energy of F2311@Al in the first phase is 45 kJ/ mol lower than that of ordinary Al (Ozawa method). It is reported 27 that PVDF delayed the aluminum's oxidation temperature. However, F2311 promoted the oxidation reaction of aluminum. Because fluorine polymer effectively reduces the content of alumina, and F is more electronegative than O. Thus, the pre-ignition reaction (PIR) will occur in the first stage, and the activation energy will decrease. Meanwhile, due to the higher reaction heat release between Al and F, the total heat release in the first stage of F2311@Al increases. In general, the encapsulated sample not only has lower activation energy, but also has higher output of heat (3.44 kJ/g higher).  www.nature.com/scientificreports/ The most probable mechanism function of the second-stage oxidation process of F2311@Al powder was the three-dimensional shrinkage spherical equation of n = 3, and the control step is spherically symmetric phase boundary reaction, following the deceleration type α-t curve. The integral formula is G(α) = 3[1 − (1 − α) 1/3 ], and the corresponding differential form is f(α) = (1 − α) 2/3 . The kinetic equation is: dα/dt = 10 13.49 × (1 − α) 2/3 e −30939.4/T . The most probable mechanism function of the second-stage oxidation process of ordinary Al is G-B equation, Reaction speed control step is spherically symmetric 3d diffusion equation, following the deceleration type α-t curve. In the second stage, the reaction activation energy and pre-exponential factor of the two aluminum powders is similar, and the kinetic mechanism function is similar, indicating that the reaction in this stage is mainly the process of oxidizing atmosphere (O and CO 2 ) diffusing to Al core to form alumina.

Al-F interface reaction mechanism.
In recent years, the conjecture and description of reaction mechanism of the Al-F interface have been reported. It is generally accepted as the "pre-ignition" theory, that is, before the beginning of aluminum's oxidation, a "shucking" like solid-phase chemical reaction between alumina and fluoride will occur. In order to further reveal the details of the interface reaction of F2311@Al, the sample was tested from room temperature to 1000 °C in air at a heating rate of 10 °C/min combined with mass spectrometry  www.nature.com/scientificreports/ (MS). Figures 9 and 10 respectively shows the mass spectrum of oxygen-containing fragments and fluorinecontaining fragments. Figure 9 shows the MS spectrum with O containing fragments. The intensity of O containing ions slowly decreases from 200 °C. This is because with the increase of temperature, oxygen is more likely adsorbed on the surface of the particles. At about 550 °C, which is the first stage of reaction, oxygen intensity falls sharply and then rebounds after 600 °C. It falls again in the second stage at about 800 °C. The two decreases in signal intensity of O containing ions correspond to the aluminum oxidation in the first and second stage.
As the MS spectrum with F containing fragments shown in Fig. 10, almost all fragments are generated between 200 and 400 °C. The signal intensity of m/z = 20 is very intense. It is speculated that this fragment is HF + . HF + is the main product of decomposition of polyvinylidene fluoride 32 , which means that F2311 begins to decompose at 200 °C. Unlike the oxygen ions of which signal intensity decreases at first, HF + remains high concentration even at 500-800 °C, when the aluminum powder began to oxidize. HF + concentration does not decrease with the fluoride decomposition it is speculated that it is the product of subsequent thermal reaction.
In addition to HF fragments, there are also some fragments with large molecular weights, such as m/z of 50, 64 and 66, corresponding to CF 2 =CH 2 , F 2 C and COF 2 . Before 400 °C, these fragments occur with the decomposition of F2311 and gradually enriches, which corresponds to the slight weight loss before 400 °C on the TG curve (Fig. 8). After 400 °C, CF 2 =CH 2 , F 2 C and COF 2 ions flow intensity decreases dramatically, while the intensity of HF ions flow is still rising. TG curve does not increase significantly in 400-550 °C. Therefore, the fragments above do not react with the Al core at this time. They just react with alumina on the surface, and alumina transforms into AlF 3 . This reaction leads to the weight loss mass in 400-550 °C. This process only occurs on the surface, and we call it pre-ignition reaction (PIR).   www.nature.com/scientificreports/ When system is further heated up to 580-600 °C, the reaction of aluminum initiates, and releases a large amount of heat. The signal intensity of CF 2 =CH 2 , COF 2 and F 2 C quickly drops to the bottom. Fragments of fluoride no longer stay on the surface to participate in PIR, they go through the interface, and reacts with the Al inside. However, there is no sign of AlF 3 , which should be seen at this time in mass spectrometry. This is because AlF 3 is still in solid phase due to its high melting point (1040 °C), and MS device only collects fragments of the gas phase. It can be seen that in the first stage (550-650 °C) of oxidation, after that they all rise up. The first weight gain reaction is more intense. We assumed that F mainly reacts with Al in the first stage. In addition, HF and other products are released due to the reaction between fluoride and aluminum, so the concentration of these fragments increased after 650 °C.
In conclusion, according to TG-DSC-MS analysis, it is inferred that the interface reaction of F2311@Al can be divided into the following four stages: 1. Between 200 and 400 °C, oxygen slowly diffuses and is adsorbed in the coating layer, and amorphous alumina gradually thickens. At the same time, F2311 began to decompose slowly with the temperature rising, releasing small molecular monomers of vinylidene fluoride and trifluorochlor oethylene, such as CF 2 =CH 2 , COF 2 , F 2 C. The sample slightly lightens at this stage, as shown in Fig. 11a. 2. When the temperature ranging from 400 to 550 °C, the F containing fragments produced by F2311 decomposition diffuses into the alumina lattice, replacing amorphous Al 2 O 3 with AlF 3 . Pre-ignition reaction occurs between F2311 decomposition products and Al 2 O 3 in the coating layer to generate AlF 3 and HF, as shown in Fig. 11b. Since the ion current strength of F-containing monomer decreases at this stage, and the weight of samples decreases, these fluorine-containing fragments do not react with the Al core, but only reacted with the alumina in the coating layer. The reaction only happens in the interface. 3. Since AlF 3 density is less than Al 2 O 3 , the replacement of Al 2 O 3 by AlF 3 results in an increase of voidage on the shell, which provides a channel for oxygen diffusion. Oxygen go through the shell and diffuses to the active aluminum, triggering the first stage of oxidation. Severe oxidation of aluminum powder occurred at 560-650 °C, as shown in Fig. 11c. The first step of exothermic reaction of ordinary Al is one-dimensional phase boundary reaction, and oxygen diffusing is slow. However, the reaction speed control step of F2311@ Al is three-dimensional diffusion, the activation energy is lower, the reaction rate is accelerated, so oxygen can contact the active aluminum. At the same time, some fluorine-containing fragments also contact with Al core, generating a large amount of AlF 3 . The heat release of Al-F reaction is greater than that of Al-O, so the heat release of F2311@Al in first stage is much higher than ordinary Al. 4. As the reaction proceeding, a new alumina shell regrows and again hinders oxygen diffusion and also passively absorbs heat from the system, thus ending the reaction. Aluminum melts at about 660 °C, and a very small endothermic peak can be observed on the DSC curve. Aluminum changes from solid phase to liquid phase resulting in the growth of volume, which can enlarge the shell 47 , as shown in Fig. 11d. After 800 °C, the liquid Al breaks the shell and erupts, and they quickly react with oxygen. Thus, the second exothermic peak occurs. The Ea and A of the second exothermic stage is very similar to ordinary Al, and the there's no sign of F reacting with Al in second stage according to MS, we assumed that second exothermic reaction is mainly the reaction of Al and O 2 .

Conclusion
In this work, we systematically studied the microstructure, surface bonding structure, thermal stability and the interface reaction mechanism of F2311 in-situ encapsulated submicron aluminum powder. The following conclusions were obtained: 1. By studying the correlation between aluminum particle size and activity, it is concluded that submicron aluminum powder (median particle size 200 nm) has the advantages of rapid reaction kinetics and high active metal content, and is a promising raw material for encapsulated in situ process. F2311 with better comprehensive performance is selected as the coating material.