Combustion characteristics of Al/PTFE materials with different microstructures

The microstructures play a crucial role in the combustion of aluminum/polytetrafluoroethylene (Al/PTFE) materials. Mechanically activated Al/PTFE typically demonstrates higher reactivity but a lower combustion rate compared to physically mixed Al/PTFE. Recently, the combustion performance of fuel-rich Al/PTFE has been well explained by the microexplosion mechanism. In this study, the combustion characteristics of stoichiometric Al/PTFE (26.5:73.5 wt%) materials with varying microstructures were investigated to further the understanding of their combustion mechanism and offer insights for their potential applications in metal cutting. The Al/PTFE materials with different microstructures were prepared using sonication and ball milling methods. The results of scanning electron microscope (SEM) analysis suggest that the sonicated Al/PTFE (s-Al/PTFE) containing spherical Al particles displayed a loosely dispersed structure, while the milled Al/PTFE (m-Al/PTFE) exhibited a densely layered structure with flake-like Al particles coated by the PTFE matrix. The milled Al/PTFE was found to be mechanically activated. Combustion in quartz tubes was recorded using a high-speed camera and a video. Combustion of s-Al/PTFE demonstrated a high-temperature flame (∼2346 K) and obvious microexplosions featuring hot particles ejection, while combustion of m-Al/PTFE showed a weak flame (∼2037 K) and slow-burning, featuring dense carbon smoke. Increasing the powder density was observed to slightly decrease (∼100 K) flame temperature. Microstructure and phase analysis of combustion products were systematically conducted to elucidate the combustion behaviors. The results suggest that the residue of s-Al/PTFE contained high AlF3 and low carbon content, whereas the residue of m-Al/PTFE exhibited the opposite condition. The results of the combustion tests suggest that microexplosions promoted the oxidation of hot Al particles and carbon products, consequently leading to a fast reaction, high flame temperature, and enhanced heat transfer capability.


Introduction
Aluminum (Al) particle is widely applicated in thermites due to its high reactivity, combustion enthalpy, and commercial availability [1][2][3][4].The oxidizers used in thermites vary in terms of particle sizes and morphologies to optimize combustion performance, including low ignition delay, high combustion rate, high heat release, and high combustion temperature [5][6][7][8][9].Extensive research has demonstrated that optimizing the interfacial properties of thermites can effectively enhance combustion performance, primarily by increasing the contact area between components [10] or employing surface modification on the metal particles [7,8,11].This knowledge applies broadly to thermite systems containing organic or inorganic oxidizers [12], and interface variations can sometimes lead to unique and complex reaction mechanisms [13].Therefore, it is crucial to analyze the reaction mechanism and combustion characteristics of thermites from a microstructural perspective.
The combustion performance of thermites is usually related to heat transfer.Thus, the study on the improvement of combustion characteristics of thermites could provide guidance for the application of metal cutting or degradation, where heat transfer occurs from the flame to metal targets.Chen et al [5] reported the combustion of Al/MnO 2 thermites under different tubular shell materials, and found that concentrating the heat and achieving high flame temperatures was possible by reducing heat transfer losses from the thermite reaction zone to the tube shell.
Aluminum/polytetrafluoroethylene (Al/PTFE) is a kind of energetic material that can release heat at 8678 J g −1 [14], which is higher than thermites with metal oxidants.It can produce carbon (C) and various gaseous products (e.g., AlF 3 , AlF 2 , F, AlF, and CF 2 ) [13,15,16].The gas generation property is found to help reduce particle agglomeration and two-phase flow loss in the application of solid propollants [17].The flame temperature of Al/PTFE could reach 3000 °C [10], which is well above the melting point of many metals such as aluminum (660 °C) and iron (1538 °C).Thus, Al/PTFE materials have the potential to be used as additives in thermites to improve the flame temperature and combustion rate in metal cutting applications.The study on the measurements of flame temperature and combustion rate of Al/PTFE can be found in [13,[17][18][19].Many studies have reported thermite compositions such as Al/PTFE/W [14], Al/PTFE [20][21][22], and Al/PTFE/Fe 2 O 3 [23], in which PTFE was utilized as the binder, demonstrating good mechanical properties [24].Therefore, PTFE can be used as both a binder and an oxidizer in thermites.Generally, the reactivities of Al/PTFE or other similar thermites are complex due to their diverse microstructures, which can be classified based on the factors such as the particle size [25][26][27], particle shape (e.g., flake-like [13,28]), and interface (coated [19,29] and layered [30,31]), among others.However, the reactivity and combustion properties of Al/PTFE could not be fully satisfied by a specific microstructure.Therefore, studying the effect of microstructure on the combustion performance of Al/PTFE can provide guidance for the application and design of energetic materials.
Generally, there are two methods to enhance the reactivity of Al/PTFE materials: nanoparticle mixing [30,32] and mechanical activation (MA) through ball milling [13,17,33,34].Nano Al powders are typically costly and prone to forming agglomerates.In contrast, the method of MA is cost-effective and easy to operate [13].The reactivity of mechanically activated Al/PTFE under slow heating rates (<10 3 K s −1 ) has been comprehensively studied by others.Mechanically activated Al/PTFE composites often have a lower onset temperature, lower activation energy, and higher energy release, similar to physically mixed nano-Al/PTFE materials.This is attributed to the close contact between Al and PTFE [35].Typically, the heat release, combustion rate, and ignition properties of mechanically activated Al/PTFE vary due to the irregular intraparticle structures.The milled Al/PTFE composites with PTFE content below 40 wt% exhibit a noticeable microexplosion phenomenon [36].Huston et al [13] reported that the combustion rates of mechanically activated Al/PTFE were faster when containing larger particles, samples with lower particle density, and absence of confinement, but were still four orders of magnitude slower than micro-and nano-sized physically mixed Al/ PTFE.However, the impact of microstructure on the flame temperature of mechanically activated Al/PTFE and the microexplosion intensity remains unclear.
The objective of this study is to investigate the effect of microstructure on the combustion properties of Al/ PTFE, specifically focusing on flame temperatures and combustion rates.The effect of powder density on the flame temperature was also studied.The combustion process of Al/PTFE materials was conducted in quartz tubes to determine their combustion behaviors and heat transfer capabilities.The flame was observed to comprehend the reaction mechanisms on a macroscopic scale.The combustion products were collected and analyzed to help elucidate the combustion characteristics from a microscopic perspective.The flame was recorded using a high-speed camera, and the temperatures were measured by an optical fiber spectrometer.

Materials
Al powder was purchased from Shanghai Xiaohuang Co., Ltd with an average particle diameter of 2 m.The PTFE powder used was Solvay F5 with an average particle diameter of 15 μm.Ethanol and hexane were purchased from Tianjin Yongda Chemical Reagent Co., Ltd Ethanol was used as dispersant in the sonication process.Hexane served as a process control agent (PCA) in the milling process.

Samples preparation
In this study, the sonicated Al/PTFE (s-Al/PTFE) and milled Al/PTFE (m-Al/PTFE) mixtures were prepared with the stoichiometric ratio of 26.5/73.5 wt%.The diagram illustrating the preparation procedures is shown in figure 1.For the s-Al/PTFE, the sample was ultrasonicated and simultaneously physically mixed using an electric stirring rod at a speed of 300 revolutions per minute (rpm).Firstly, 13.25 g Al and 36.75 g PTFE powders were sonicated and stirred for 30 min using 200 ml of ethanol as the dispersant.Then, the slurry was dried using a vacuum drying oven at 80 °C for 24 h to remove ethanol.For the m-Al/PTFE, the sample was ball-milled using a planetary mill (MITR YXQM-1L, China).About 15 ml of hexane was added to the container as PCA.The rotating speed was 500 rpm.Briefly, 10 g of s-Al/PTFE were milled using 10 mm stainless steel balls at a ball-topowder mass ratio (BPR) of 20:1 inside a 250 ml hard alloy (19NiCr1Ti) container filled with argon.Another identical container was used to maintain symmetrical loading on the rotary table.The total milling time was 6 h which was close to the critical time leading to reaction.After the milling process, the Al/PTFE slurry was collected inside an argon-filled glove box and slowly passivated with diluted air.To remove the hexane, the mixture was dried using a vacuum drying oven at 60 °C for 24 h.
To investigate the combustion behaviors of the two samples, 5 g of each powder was cold-pressed into a quartz tube with an inner diameter of Φ20 mm and a wall thickness of 5 mm.The pressed samples in quartz tubes are shown in figure 2. Pure PTFE powder (white area) was filled and pressed at the bottom of the tube to form a plug.The applied pressure for the Al/PTFE samples was controlled at 20 MPa.Finally, the cylindrical Al/ PTFE blocks were formed with the dimensions of Φ20 mm × 6.5 mm inside the tubes, achieving 64% of the theoretical maximum density (TMD).

Structure characterization
Particle morphologies of s-Al/PTFE and m-Al/PTFE, as well as their combustion products, were characterized using a field emission scanning electron microscope (FE-SEM, Hitachi S-4800) with an accelerating voltage of 5  kV.The element distributions of the combustion products were analyzed by energy dispersive spectrometer (EDS) line scanning.Brunauer-Emmet-Teller (BET) specific surface area measurements for loose powders were conducted using a Micromeritics APSP 2460 system.The samples of ∼100 mg were degassed at 393 K in nitrogen (N 2 ) for 6 h before specific surface analysis.The x-ray photoelectron spectroscopy (XPS) analysis was conducted using Thermo ESCALAB 250XI with Al kα (hv = 1486.6eV) radiation to analyze the element valence of both samples and combustion products.The working voltage was 12.5 kV, and the filament current was 16 mA.The peak of C1s at a binding energy of 284.8 eV was used for peak calibration.The phases of the samples and their combustion products were analyzed by x-ray diffraction (XRD) using a Rigaku Ultima IV system with Cu Kα (λ = 1.5438Å) radiation.Data of 2θ was collected by continuous scanning from 10°−80°at 10°/min with a step of 0.02°.

Thermal analysis
Thermogravimetric and differential scanning calorimetry (TG-DSC) analysis was performed using a NETZSCH STA-449F3 system to measure the reactivity of the samples.Each sample was heated under argon flow (purity = 99.99%,flow rate = 50 ml min −1 , atmospheric pressure) from 100 °C to 780 °C with varying heating rates of 20, 30, and 40 K min −1 to investigate the activation energy.The sample mass tested was about 5 mg.Samples were slightly flattened at the bottom of an alumina crucible to enhance heat conduction.Baseline subtraction of the DSC curve at the heating rate of 20 K min −1 was also considered for the analysis of heat release.

Combustion experiments
The setup of the combustion experiments is shown in figure 3. To visualize the flame and distribution of burning particles, an AI-030G300M high-speed camera equipped with an LC0820-5M 8 mm lens was used.The highspeed flame monochrome images recorded were captured at 200 fps (frames per second).Meanwhile, a Hikvision color video with a frame rate of 25 fps was used to record the combustion process.The motion characteristics of burning particles are found to be related to the exposure time.When the exposure time is decreased to an aproporiate threshold, the contours and distributions of hot burning particles could be identified [37].Thus, to better capture the motion features of the burning particles, the image exposure time was set to 10 μs.Samples were ignited using electric igniters triggered by a 9 V DC battery.The flame temperatures of pressed powders in quartz tubes were characterized using an OHSP-350UV optical fiber spectrometer.The spectrometer was calibrated using a standard blackbody furnace at a temperature of 3000 K.Meanwhile, the flame temperatures of the loose powders were also measured to evaluate the impact of porosity on the flame temperature.
At the end of the combustion tests, the combustion products of the pressed samples were collected and characterized to evaluate the combustion efficiency and analyze the reaction mechanism.Worth mentioning, the thermal conductivity of Al (238 [18].Thus, increasing the porosity of Al/PTFE may help enhance heat transfer, leading to improvements in ignition properties and heat release [18].The increase in interfaces between Al and PTFE in m-Al/PTFE shortened the reactive length compared with the bare particles in s-Al/PTFE.However, the Al particle with the flake-like structure may lead to a low content of active Al due to its damaged passivation layers when not sufficiently coated by the PTFE matrix.Therefore, it is necessary to investigate and compare the reactivity and combustion performance of s-Al/PTFE and m-Al/PTFE. The N 2 adsorption-desorption curves of the loose powder samples are depicted in figure 5.The two curves both represent a typical IVa isotherm and H3 hysteresis loop [38].The pore size distribution curves showed that the pores of s-Al/PTFE were mainly mesopores, while those of m-Al/PTFE were mainly macropores.The macropore diameter of ∼80 nm of m-Al/PTFE represents the scales of layered gaps inside the flake particles, and the surface of the flake was smooth.According to the Brunauer-Emmett-Teller (BET) method, the specific surface area of the samples was calculated, and the mesopore size distribution was analyzed using the Barret-Joyner-Halenda (BJH) method.The results presented in table 1 suggest that m-Al/PTFE exhibited a larger specific surface area and pore volume compared to s-Al/PTFE.The high pore volume of m-Al/PTFE is attributed to their layered structure inside the flake particles.Considering that m-Al/PTFE had a layered intraparticle structure, flake-like Al debris may be coated by the PTFE matrix, resulting in a low volume of mesopores.Therefore, m-Al/PTFE powder exhibited a higher pore volume and specific surface area than s-Al/PTFE.
Figure 6 shows the XRD patterns of the two prepared samples.The diffraction peaks correspond well to the crystal planes of Al and PTFE.The pattern showed that the AlF 3 and Al 2 O 3 phase were not detected in m-Al/ PTFE, indicating that no reaction occurred between the deformed Al particles and PTFE.
Figure 7 presents the elemental analysis of s-Al/PTFE and m-Al/PTFE.Al 2p spectra were selected to compare the differences in the valence of Al in the samples.The peak at ∼74.PTFE matrix.Meanwhile, the direct interface contact of Al(0) and PTFE atoms tends to react at a lower energy threshold.

Thermal analysis
Figure 8 shows the TG-DSC curves of these two samples heated under an argon atmosphere.The m-Al/PTFE exhibited a lower onset temperature of 524 °C compared to the 666 °C of s-Al/PTFE.In figure 8(b), the weak peak at 501.5 °C represents the pre-ignition reaction (PIR) [39], while the main exothermic peak at 579.9 °C confirms the successful activation of m-Al/PTFE.However, the s-Al/PTFE displayed a primary endothermic peak at 594 °C and a main exothermic peak at 676 °C (figure 8(a)).The results suggest that the reaction of m-Al/ PTFE under slow heating rates was more complete than that of s-Al/PTFE.The thermal analysis of mechanically activated Al/PTFE has been widely reported, showing similar results of a pre-ignition reaction and a lower onset temperature for the exothermic peak [40,41].The TG curve of s-Al/PTFE (figure 8  Figure 9 presents the DSC curves of the two samples under varying heating rates (20, 30, 40 K min −1 ) to investigate their activation energy (E), which is related to the ignition properties of energetic materials.A low activation energy indicates that the reaction of energetic materials can be initiated by a low-energy stimulus.In this research, the activation energy was studied using the isoconvertional Flynn-Wall-Ozawa's method [42] shown in equation (1).
where, β i is the heating rate, A is the pre-exponential factor, G(α) is the integral of the reaction mechanism function which is a constant at a particular conversion rate (α), R is the gas constant, and T(α) is the temperature.In this study, the specific conversion rate was determined at the temperature corresponding to the exothermic peak.The results in figure 9 show that the m-Al/PTFE exhibited an activation energy of 231.45 kJ mol −1 , which is 95.57kJ/mol lower than that of s-Al/PTFE.It is suggested that m-Al/PTFE could be ignited with a lower threshold of energy stimulation.The reactivities of the samples under an air environment are crucial for practical applications.Hence, the heat release of the loose sample powders was measured using microcalorimetry under N 2 /O 2 flow.A N 2 /O 2 flow at 80/20 vol.% was used to simulate air conditions.Table 2 presents the results of microcalorimetry analysis.The m-Al/PTFE generated heat release nearly 1.7 times higher than that of s-Al/PTFE, whereas the peak temperature of 509.6 °C corresponding to the peak heat release rate was reduced by ∼40 °C.The peak temperature of m-Al/PTFE was very close to the onset temperature of 524 °C in the DSC curve (figure 8(b)), while that of s-Al/PTFE was significantly lower than the onset temperature of 666 °C shown in figure 8(a).The  reaction mechanism of the samples in microcalorimetry was the same as in TG-DSC analysis because the samples were slowly heated in both tests.The peak heat release rate of m-Al/PTFE is lower than that of s-Al/ PTFE.It is hypothesized that m-Al/PTFE has a slower heat release rate and burning rate during the combustion process.

Combustion test 3.3.1. Combustion process
In order to capture the combustion process of the sample more effectively, the distance between the high-speed camera and the sample was maintained at 2 m.The first image showing ignition was defined as the initial frame.Both pressed samples were successfully ignited, and a sequence of high-speed camera images is shown in figure 10.The combustion of m-Al/PTFE exhibited a longer burning time of approximately 7 s compared to s-Al/PTFE which burned for ∼3.5 s.Initially, the quartz tubes remained dark and the flame appeared irregular.Subsequently, the upper surface of the samples ignited.The bright and fierce flame of s-Al/PTFE occurred at 550 ms, while that of m-Al/PTFE appeared at 900 ms.The height of the flame initially increased and then decreased over time.The flame height corresponds to the energy loss of the reaction caused by heat convection from the flame to the tube wall.The increase in the wall temperature led to a decrease in heat convection, which consequently resulted in a reduction in the energy loss of the flame.The flame grew higher and more intense.The flame of s-Al/PTFE ejected hot particles at 550 ms and continued ejecting until the end (figure 10(a)).The ejection of hot particles resulted from the microexplosion of Al particles.The decrease in the height of the flame at the end was due to the insufficient heat release rate of the reaction.However, the ejection of hot particles was not observed in the flame of m-Al/PTFE throughout the entire burning process (figure 10(b)).It might resulted from the near fuel-lean formulation of Al/PTFE used in this study that affects the microstructure of m-Al/ PTFE.Generally, other researchers have reported two main conclusions: (1) the microexplosion tendency was widely observed in the combustion of fuel-rich m-Al/PTFE (e.g., 40:60 wt% and 30:70 wt% [13]) and physically  (2) the combustion rate of both fuel-rich and fuel-lean m-Al/PTFE was found to be slower than physically mixed Al/PTFE [13].The combustion performance in this study was consistent with these opinions, but the combustion characteristics of near fuel-lean m-Al/PTFE were first analyzed.The reaction of s-Al/PTFE was incomplete, but it still resulted in a violent flame in air.It is inferred that the reaction of Al hot particles in air generated significantly more heat, and rate of the heat release was faster.For the m-Al/PTFE, the uniformly distributed PTFE and Al particles may result in stable combustion.However, the thick coating of PTFE matrix may make it easier to preclude the heating and ejection of Al particles.Thus, the m-Al/PTFE exhibited a longer burning time and weaker flame, despite its reaction being more complete than that of s-Al/PTFE.Figure 11 displays the color images of flames to further investigate the burning characteristics of Al/PTFE.The flame of m-Al/PTFE released a lot of black dense smoke, while the flame of s-Al/PTFE produced less black smoke.It is deduced that the smoke was mainly composed of carbon products.The results indicate that the combustion of m-Al/PTFE generated a higher quantity of carbon products than s-Al/PTFE.The carboncontaining products of s-Al/PTFE may exist in the forms of gas-phase carbon oxides and carbon fluoride.

Flame temperature measurements
The flame temperature is an important characteristic of Al/PTFE.Conventional thermometers like thermocouples have a limited measurement range for detecting high temperatures (above 1500 K).Therefore, an optical non-contact thermometer is required to measure the flame temperature of both loose and pressed Al/ PTFE samples.In previous research, the temperature of pyrotechnic flames could be accurately measured using optical fiber spectrometers [43,44].According to Planck's law, the radiant existence is a function of wavelength and temperature, as is shown in equation (2).
where, M is the radiant existence of the grey body, W⋅m −2 ⋅μm −1 , λ is the wavelength, μm, T is the temperature of the hot body, K, ε is the emissivity that relates to λ and T, c 1 and c 2 are the first and second radiation constant respectively, c 1 = 3.742 × 10 8 W⋅μm 4 ⋅m −2 , c 2 = 1.4388 × 10 4 μm⋅K.Therefore, the flame temperature could be determined when the parameters of M λ , T and λ are specified.However, the detected spectra were the irradiance (E λ,T ) curves of the flame.The irradiance is linearly correlated with the radiant existence of the grey body.Then, a coefficient of A was defined and their relation was shown in equation (3).
According to Xu's study [43], the coefficient is related to the flame area and the distance between the optical fiber probe and the flame.Therefore, different irradiances of the flame can be compared under the same test condition.By selecting a radiation curve of the gray body to fit the irradiance spectrum, the value of T could be calculated, thereby obtaining the flame temperature of the samples.The fitting curve was calculated using the Levenberg-Marquardt algorithm, which is a widely used nonlinear least square method.
The typical radiation spectra curves of Al/PTFE flame in this study are shown in figure 12.The collection frequency of the spectrometer used in the experiment was 1 fps.The integral time was set at 5 ms to prevent signal overflow.The radiation spectrum of s-Al/PTFE shows two main peaks at 275 nm and 425 nm, indicating that the spectrum contains not only continuous spectra but also atomic and molecular characteristic spectra [44].According to the NIST spectra database [45], the peak near 275 nm corresponds to the line of Al ions and C atoms.The wide band near the peak of 475 nm consists of the atomic lines of Al and C atoms.The wide band near 500 nm represents the line of Al-O molecules.The results indicate that the reaction in air was Al O AlO O.
2 +  + The continuous spectra ranging from 650 nm to 700 nm result from the thermal irradiance.The emissivities of the flame spectra within this continuous range are stable and they can be assumed to be the same (ελ 1,T = ελ 2,T ), treating the flame as a gray body [43,46].To record the peak flame temperature, the irradiance spectrum during the stable burning stage was chosen.The measured temperatures are presented in table 3. The results show that the flame of loose s-Al/PTFE reached the highest temperature, followed by the pressed s-Al/PTFE.When the sample was pressed, the flame temperatures of s-Al/PTFE and m-Al/PTFE decreased slightly by 3.7% and 5.3%, respectively.It is due to the reduced porosity that hinders heat and mass transfer.Compared to s-Al/PTFE, the flame temperatures of m-Al/PTFE decreased by 13.2% and 14.5% for the loose and pressed powder, respectively.This indicates that the microstructure is the crucial factor affecting the flame temperature.

Structure and phase analysis of combustion products
The combustion products were collected inside the quartz tubes.The SEM images of the combustion residues of Al/PTFE are presented in figure 13.The cubic products in s-Al/PTFE (figure 13(c)) were identified as AlF 3 , which was not prominently find in the products of m-Al/PTFE (figures 13(d)-(f)).These results suggest that more intense combustion and the presence of microexplosions led to an increased production of AlF 3 in the residue.The agglomerates of nanoparticles were carbon produced by the PTFE.The C agglomerates in the products of m-Al/PTFE had a floc-like structure, while in the products of s-Al/PTFE, they had a granular structure at the micron scale.In figure 13(f), the C agglomerations of m-Al/PTFE exhibited several hollows, which is due to the production of gas-phase aluminum fluoride.The presence of these hollows can provide strong evidence that the dense PTFE matrix on the Al surface precluded the microexplosion of Al particles, subsequently inhibiting the reaction between Al fragments and O 2 in air.Therefore, the primary heat release in m-Al/PTFE was produced by the intra-particle reaction.Similarly, the reduced microexplosion tendency (hot particle ejection) was also observed in the combustion tests conducted in Xu's study as the PTFE content increased from 35 wt% to 45 wt% in PTFE-coated Al particles, where the PTFE coating was noticeably thickened and agglomerated [19].
To gain further insight into the microstructure and reaction mechanism, elemental line scanning analysis of the combustion products was conducted, and the results are displayed in figures 14 Figure 14(c) presents the XRD analysis results of the combustion residues.The major peak at 25.7°c orresponded to the phase of AlF 3 , while crystalline carbon was not detected.The presence of residual Al and Al 4 C 3 in the combustion products of s-Al/PTFE indicates incomplete reaction.This could be due to the agglomeration of Al particles and the limited surface contact between Al and PTFE particles.
XPS analysis was conducted to determine the elemental valence to further confirm the phase analysis mentioned above.Figures 14(d)-(e) presents the spectra of C1s, O1s, F1s, and Al2p of the combustion products.The results of C1s, O1s, and F1s combined, as shown in figures 14(d)-(e), confirmed the presence of carbon oxide and trace amounts of carbon fluoride.This indicates that the reduction of C in s-Al/PTFE resulted from the reaction between C and O 2 in air.The microexplosion of Al particles in s-Al/PTFE not only enhanced the combustion rate but also promoted the oxidation of C, leading to a higher energy release and more complete combustion efficiency in air conditions compared with m-Al/PTFE.In figure 14(d), the weak peak of F1s at the

Combustion mechanism
The analysis of microstructure, flame temperature, and combustion products could help understand the combustion mechanism of Al/PTFE [19,21,47].The pressing process had a slight effect on flame temperature, and the microstructure of Al/PTFE dominates the reaction and combustion behaviors.The flake-like structure in m-Al/PTFE particles results in a lower activation energy and a more complete reaction under slow heating rates, while the s-Al/PTFE containing spherical Al leads to a more intense combustion and microexplosion tendency, consequently resulting in a higher flame temperature under fast heating rates.
During the combustion of m-Al/PTFE, the dense structure and coating of PTFE matrix in m-Al/PTFE hindered the microexplosion of Al.The reaction of broken Al particles was limited inside the PTFE matrix, preventing their ejection and further reactions with O 2 .The passivation layer on the surface of flake-like Al particles may be incomplete and flexible, thereby reducing the internal stress of Al, which also hinderd the microexplosion.Meanwhile, the carbon product of m-Al/PTFE formed large agglomerates which may further impede heat and mass transfer.During the combustion of s-Al/PTFE, the burning of Al particles in air promoted the heat release because the combustion heat of Al (1675 kJ mol −1 ) [48] is higher than the heat release of Al/PTFE.The oxidation of carbon products could further increase the heat.Therefore, the microexplosion in s-Al/PTFE improved the flame temperature and combustion rate by enhancing the reactions of Al/O 2 and C/O 2 .

Conclusions
This study investigated the combustion characteristics of sonicated and milled Al/PTFE (26.5:73.5 wt%) materials.The combustion process was conducted in quartz tubes with an inner diameter of 25.4 mm to better observe the combustion behaviors and heat transfer phenomenon.The microstructures, phases, and elements of the prepared samples and combustion residues were analyzed to elucidate the combustion behaviors and mechanisms.The peak flame temperature and combustion rate are considered as the indices to evaluate the combustion characteristics of Al/PTFE with different microstructures.The m-Al/PTFE particle had a flake structure with Al debris coated by the PTFE matrix and a large scale of inter-particle interval.The s-Al/PTFE powder was evenly distributed with spherical Al particles and irregular PTFE particles.The m-Al/PTFE exhibited more complete reactions and a lower activation energy of 231.45 ± 24.58 kJ/mol under slow heating rates compared with s-Al/PTFE of 327.02 ± 50.35 kJ mol −1 .It is found that the combustion of s-Al/PTFE in air ejected hot Al particles and exhibited a shorter combustion time of 3.2 s and a flame temperature of ∼2346 K than m-Al/PTFE of 6.8 s and ∼2037 K, respectively.However, the ejection of hot particles throughout the combustion process of m-Al/PTFE was not observed, but the dense carbon smoke was evident.The results suggest that the combustion of s-Al/PTFE led to more intense heat transfer.By analyzing the combustion products, it is inferred that the difference in combustion performance lies in the intra-particle reaction of m-Al/ PTFE and the inter-particle reaction of s-Al/PTFE combined with microexplosions.The combustion of s-Al/ PTFE could introduce O 2 gas from air to further oxidize the hot Al particles and carbon to enhance the combustion and increase the heat release, leading to a higher combustion rate and flame temperature compared to m-Al/PTFE.The reaction of m-Al/PTFE was confined within the composite flake particles, leading to restricted heat release and a low rate of product generation.By comparing the flame temperature of loose and pressed powders, it is also found that the microstructures of Al/PTFE are the dominant factors influencing the flame temperature, rather than the porosity.
This study expanded the understanding of the effect of microstructure on the combustion characteristics of Al/PTFE energetic materials.Especially, the combustion mechanism of Al/PTFE was explained by analyzing the microstructures of combustion products.The results are helpful for the design and application of Al/PTFEcontained energetic materials.

Figure 1 .
Figure 1.Diagram of the sonication and milling procedures.

Figure 2 .
Figure 2. Pressed samples (in red frame) in quartz tubes.

Figure 3 .
Figure 3. Diagram of the setup of combustion experiments.

3 .
Results and discussion 3.1.Structure characterization The SEM images of the prepared samples are shown in figure 4.There was an obvious difference in morphology between these two samples.For the loose s-Al/PTFE powder (figure 2(a)), spherical micro-Al and irregular PTFE particles were uniformly mixed, while slight agglomerations of sub-micron Al particles were observed.The pressed s-Al/PTFE (figure 2(c)) exhibited a close contact between Al and PTFE particles, with the Al particles retaining their spherical shape.For the loose m-Al/PTFE powder (figure 2(b)), Al and PTFE particles formed flakeswith a width of ∼10 μm and a thickness of ∼2 μm.The flakes were uniformly distributed agglomerates of deformed Al fragments within a PTFE matrix.The agglomerates have nano-scale intra-particle interfaces and a micrometer inter-particle scale.The structure of m-Al/PTFE was similar to that reported elsewhere.For the pressed m-Al/PTFE powder (figure 2(d)), the flake-like particles were in close contact with each other.

Figure 4 .
Figure 4. SEM images of Al/PTFE of (a) and (b) loose powder; (c) and (d) pressed powder.
5 eV was related to the chemical bond of Al-O, which could be attributed to the Al 2 O 3 layer coated on the Al surface.The Al 2 O 3 passivation layer is necessary for commercial Al particles to prevent further oxidation.The peak at ∼72.0 eV is attributed to Al (0) atoms in the inner cores of Al particles.The Al 2 O 3 content on the Al surface is associated with the ratio of the peak intensity of Al-O to that of Al(0) because the probing depth of XPS analysis is approximately 3 nm.Therefore, the high relative intensity of the Al(0) peak in figure 7(b) suggests that the flake Al particles in m-Al/ PTFE had a thinner Al 2 O 3 passivation layer and were not oxidized by O 2 in air due to the compact coating of the

Figure 6 .
Figure 6.XRD patterns of two prepared samples.
(a)) shows a one-stage weight loss of 67.1 wt% ranging from 400 °C to 600 °C, while the m-Al/PTFE (figure 8(b)) exhibited a two-stage TG curve ranging from 350 °C to 500 °C with a weight decrease of 9.3 wt% and from 500 °C to 580 °C with a weight decrease of 48.5 wt%.The weight loss resulted from the pyrolysis of PTFE and vaporization of reaction products.The main exothermic peak in figure 8(a) resulted from the reaction between molten Al and the gaseous products of PTFE.Combined with the two-stage weight loss of m-Al/PTFE (figure 8(b)), the weak exothermic peak resulted from the decomposition of PTFE, and followed by the PIR between the Al 2 O 3 layer and the gaseous products of PTFE.Subsequently, the main exothermic peak was attributed to the reaction between the solidphase Al core and the gaseous products of PTFE.The evident exothermic reaction of m-Al/PTFE suggests that

Figure 10 .
Figure 10.High-speed images of the flame of (a) s-Al/PTFE and (b) m-Al/PTFE with an exposure time of 10 μs.The samples were ignited at 5 ms.The flame of s-Al/PTFE generated many hot particles (in red circle).

Figure 11 .
Figure 11.Color images of the flame of pressed samples of (a) s-Al/PTFE and (b) m-Al/PTFE.

Figure 12 .
Figure 12.Typical irradiance spectra of the flame of Al/PTFE.
(a)-(b).The elements C, O, F, and Al were detected.The results revealed that the products of s-Al/PTFE consisted of a high atomic content of AlF 3 and low-content of C, while those of m-Al/PTFE consisted of low content of AlF 3 and Al 2 O 3 , and high content of C. Additionally, the high F/Al ratio of ∼5.2 and the O/Al ratio of ∼3.0 shown in figure 14(b) suggest that trace amounts of carbon fluoride and carbon oxide were present in the residue of m-Al/PTFE.

Figure 14 .
Figure 14.SEM images and elemental line scanning of the combustion products of (a) s-Al/PTFE and (b) m-Al/PTFE; (c) XRD patterns of the combustion products of two samples; XPS spectra of the combustion products of (d) s-Al/PTFE and (e) m-Al/PTFE.

Table 1 .
Calculation results of BET-specific surface area and pore size distribution of loose powder.

Table 2 .
Microcalorimetry results of the loose sample powder.