Introduction

One, relatively new approach to the preparation of fast-burning thermites is preliminary mechanical activation (MA) of micron-sized metal + solid oxidizer powder mixtures in high-energy ball mills [1,2,3,4,5,6]. The process ensures the fragmentation and intermixing of mixture components at the submicron level and the formation of structural defects, which allows one to raise the rate of the chemical reaction on the surface of the reactants and, accordingly, to control the energy release. The materials thus produced are referred to as mechanically activated energetic composites (MAECs). A recent review [6] summarized the first data on the formation and properties of defects resulting from the MA of a number of thermite mixtures (MoO3, (–CF2–CF2–)n, Al, Mg). In particular, MA was shown to favor the formation of “weakly bound” oxygen in MoO3 and considerably reduce the temperature of oxygen release to the gas phase. The process ensures the self-ignition of the thermite at lower temperatures.

This paper focuses on the MA of Al + CuO thermite. This composition offers one of the largest heat effects per unit volume of the mixture (over 20 kJ/cm3). In addition to the large heat effect, extremely high burning velocities, up to 2500 m/s, were obtained using low-density compositions based on Al + CuO nanopowders (density of charges, ~ 5% of the maximum density) [7]. The ignition and combustion behavior of Al + CuO nanopowder mixtures, including those prepared using MA, has been the subject of rather extensive studies (see, e.g., Refs. [3, 5, 8,9,10,11,12]).

The MA process may involve not only the fragmentation of components, defect generation, and the formation of a composite but also metal–oxidizer chemical interaction processes. In this work, considerable attention is paid to these “parasitic” processes, and optimal mechanical activation conditions are found that maximize the burning velocity and brightness temperature. To quantitatively determine the degree of conversion, the amount of “unutilized” oxygen in CuOn was evaluated from the weight loss due to the hydrogen reduction of the system. The hydrogen reduction of CuO and Cu2O has been the subject of systematic studies [13,14,15,16,17].

Materials and methods

As starting materials, we used micron- and nanometer-sized powders: PP-2l Al pyrotechnic powder (flat particles 50–100 × 2–5 μm in dimensions), CuO (analytical grade) ranging in particle size from 20 to 50 μm, V-Alex nanoparticles (100–200 nm), and nCuO (50–80 nm) (Advanced Powder Technologies LLC, Tomsk). X-ray diffraction characterization showed that the starting CuO powder consisted of not only CuO but also Cu2O (25 wt%) and Cu (0.5 wt%).

We prepared powder mixtures with Al/CuO weight ratios of 18.5/81.5 and 19.6/80.4. The theoretical maximum density (TMD) of the former composition is 5.13 g/cm3.

The mixtures were prepared and activated using two distinct types of ball mills: Aronov vibro mill and Aktivator-2SL planetary mill. The volume of the grinding vials was 108 and 250 mL, the total ball weight was 200 and 300 g, and the powder mixture weight was 15 g in both cases. To prevent overheating of the mixtures, they were milled in 20- to 60-s cycles, with 5- to 10-min breaks. During milling in the Aktivator-2SL, the vials were water-cooled. The total activation time ta was varied from 1 to 60 min. To eliminate the risk of explosion, the mixtures were milled in an inert atmosphere in the presence of a liquid additive (20–60 mL of hexane).

To compare the degree of MA in different types of activators, use is made of the specific milling dose D = J × ta (kJ/g), where J (W/g) is the power intensity of the activator and ta (s) is the activation time. J can be determined by the method of test objects [18] from the rate at which the specific surface area of a material with known properties increases during milling. In this study, the mixtures were milled in liquid, so the absolute values of D are only approximate, but the relative values of D allow for an objective estimate of the degree of activation in different mills. The estimated power intensity of the Aronov mill is J = 3.7 W/g, and that of the Aktivator-2SL at its full power is J = 9.7 W/g. In what follows, we specify both the activation time and milling dose. The following designations of compositions are used: Al/CuO(v) and Al/CuO(p) are micron-sized components activated in the vibro and planetary mills, respectively, and nAl/nCuO is nanopowder activated in the planetary mill.

The structure of the MAECs was determined by X-ray diffraction (XRD) on a DRON-3 diffractometer (Cu target X-ray tube). Their reactivity during slow heating (10 °C/min) was assessed by differential scanning calorimetry (DSC) and thermogravimetry (TG) using a simultaneous thermal analysis system (Netzsch STA 449 C Jupiter) in combination with a quadrupole mass spectrometer (QMS 403 C Aelos).

The burning velocity was measured using several experimental implementations. In the case of low-density charges, with porosities from 50 to 75%, experiments were carried out in metallic, plastic, and glass tubes 5–10 mm in diameter. The charges were ignited by a heated Nichrome wire, spark, or standard electrical pyrotechnic igniter MB-2N (EVF-1).

The burning velocity in the metallic and plastic tubes was measured by detecting the emission from the combustion front using photodiodes with optical fiber. The experimental data were used to determine the average burning velocity 20–200 mm from the ignition site. In the case of the glass tubes, the emission was detected by a Cordin 226-16G high-speed electron-optical camera.

The brightness temperature Tb of the combustion products was measured in the visible range using a four-channel optical pyrometer (effective wavelengths λ = 500, 600, 700, and 800 nm). About 5 g of MAEC powder was poured in portions into a metallic tube 15–20 mm in diameter and densified to a preset density (0.3–0.4 TMD). For ignition, we used either a heated Nichrome wire or a discharge between two electrodes. At the end of the charge, a transparent window and 5-mm-diameter orifice were made. The light from emitting products was directed to the pyrometer by a system of mirrors and optical filters.

The MAEC ignition temperature and delay were determined using a PM-700 muffle furnace. A weighed amount of the powder (30 ± 1 mg) was placed in a container, introduced into a furnace, and dropped on a heated copper plate. The temperature was monitored with an accuracy of 1 °C. The highest temperature was 430 °C. The ignition delay was determined with an accuracy of 1 s using a stopwatch. The operation of the setup was checked using the known ignition temperature of hexogen.

Results and discussion

Optimization of mechanical activation conditions

Figure 1 shows XRD patterns of two samples of a 2Al + CuO (19.6/80.4 wt%) mixture milled for 2 (1) and 20 (2) min.

Figure 1
figure 1

XRD of the Al/CuO(v) mixtures milled for (1) 2 and (2) 20 min. Milling dose D = (1) 0.44 and (2) 4.4 kJ/g

Full size image

Both XRD patterns have lines of four phases: Al, CuO, Cu2O, and Cu. Note that a considerable amount of Cu2O and trace levels of Cu were already present in the starting CuO. Mechanical activation does not change the phase composition of the system, but the relative intensity of the lines of Cu and Cu2O increases.

It is reasonable to assume that milling is accompanied by partial reaction between the mixture components, CuO → Cu2O → Cu reduction, and the formation of Al2O3 as the reaction product. Unfortunately, the forming alumina is X-ray amorphous, so it is inappropriate to use XRD data for assessing the degree of conversion.

To quantify conversion, we determined the concentration of the oxygen persisting in the system in the form of CuO or Cu2O (remaining unreacted with aluminum). To this end, the milled samples were heated in DSC crucibles under a hydrogen atmosphere. As a result, hydrogen detached oxygen from the copper oxides, resulting in water release to the gas phase, and the sample weight decreased. The constant sample weight at higher temperatures indicated that the copper reduction process reached completion. Curve 1 in Fig. 2 corresponds to the reduction of the unmilled CuO during heating in H2. It is seen that the weight of the CuO stops decreasing near 550 °C, with ΔM/M = 18.26%. After reduction, only one phase, Cu, was identified in the sample by XRD, and the weight loss coincided well with the total oxygen content of the initial CuO (~ 17.9 wt%), as can be evaluated from XRD data (see “Materials and methods” section). In further analysis, we assumed that H2 did not detach oxygen from Al2O3 and that the amount of oxygen combined with copper could be estimated in this way.

Figure 2
figure 2

Weight loss of unmilled (1) CuO and (2) Al/CuO and milled (3–6) Al/CuO(v) mixtures during heating in H2

Full size image

Curve 2 in Fig. 2 corresponds to the hydrogen reduction of the unmilled Al/CuO mixture. It coincides with the curve for the reduction of CuO (curve 1). Curves 3–6 in Fig. 2 correspond to heating of the Al/CuO mixtures in hydrogen after MA to various doses. It is seen that the weight loss curves of the activated mixtures are shifted to lower temperatures by about 70 °C. Further research is needed to find out the nature of the shift of the reduction curves to lower temperatures. It cannot be ruled out that the shift can be understood in terms of the position of Al and Cu in the electrochemical activity series. The electrochemical potential of Al is − 1.7 V, and that of Cu is + 0.522 V [19]. The MA-induced close contact between the mixture components and the resulting potential difference may facilitate hydrogen reduction of the copper oxides.

The XRD data obtained after heating MA mixtures Al/CuO in H2 (Fig. 3) confirm that the reduction of copper reaches completion. The XRD patterns in Fig. 3 are dominated by lines of Cu and Al. At high milling doses, there are also trace amounts of CuAl2 and Cu9Al14, without even trace levels of the copper oxides. Thus, gravimetric data on hydrogen reduction can be used to assess the amount of oxygen persisting in CuOn after MA.

Figure 3
figure 3

XRD after reduction in H2 of Al/CuO(v) mixtures. Milling dose D = (1) 0.88, (2) 2.2, and (3) 4.4 kJ/g

Full size image

Row 1 in Table 1 summarizes the weight loss data for the samples milled to various doses and then reduced in H2. The oxygen content of the starting mixture [O]ini was 18.26 × 0.804 = 14.68 wt%. It is seen from row 1 in Table 1 that, for sample with dose D = 0.44 kJ/g, the weight loss due to oxygen detachment equal to 14.58 wt%. With increasing milling dose, the amount of residual oxygen decreases, indicating that the aluminum oxidizes during milling. The data given in row 1 in Table 1 are presented in row 2 as the percentage of oxygen consumed for the oxidation of the aluminum. The fraction of reacted oxygen, [ΔO]R = [O]ini – [O], divided by the total amount of oxygen, [O]ini, increases from a fraction of a percent at low milling doses to about 10% at doses near 2 kJ/g, reaching ~ 28% at the highest dose. Thus, mechanical activation is accompanied by reaction between the mixture components and, at milling doses above 4 kJ/g, the degree of conversion can reach tens of percent.

Table 1 The degree of conversion of 2Al + 3CuO reaction ([ΔO]R/[O]ini) under milling
Full size table

Figures 4 and 5 present examples of TG and DSC curves of the unmilled (Fig. 4) and milled (Fig. 5) (Al/CuO)v mixtures in He. The weight of the starting mixture (Fig. 4, curve 1) decreases by 0.2% in the range ~ 300–400 °C, which is due to CO2 release. At temperatures above 700–750 °C, we observe further weight loss, due to oxygen release to the gas phase (see curve 5) as a result of the thermal decomposition of CuO. The DSC curve (3) shows endothermic peaks due to aluminum melting and CuO decomposition, accompanied by oxygen release, but there are no well-defined exothermic peaks attributable to aluminum oxidation. During a second heating of the same sample, its weight increases slightly (curve 2), presumably due to the oxidation of the Al by O2 impurities in He, and the DSC curve shows only the Al melting peak (curve 4), without high-temperature O2 release (curve 6).

Figure 4
figure 4

Weight change (1, 2), heat effect (3, 4), and O2 (m/z = 32) ion currents (5, 6) during first (1, 3, 5) and second (2, 4, 6) heating of the unmilled Al/CuO mixture

Full size image
Figure 5
figure 5

Heat effects (1,2) and weight change (3,4) during the first (1,3) and second (dash) heating of the (Al/CuO)v MAEC (8 min, D = 1.8 kJ/g)

Full size image

It follows from XRD data that the high-temperature oxygen release is due to the 2CuO → Cu2O + O decomposition and that the only phase present after first heating is Cu2O. The weight loss due to O2 release from this sample is 4.77%.

Table 1 (row 3) compares the amounts of oxygen released from different samples to the gas phase above 750 °C (ΔM (> 700 °C)). With increasing milling dose, ΔM (> 700 °C) decreases. Starting at a milling dose of 1.8 kJ/g, we observe no O2 release to the gas phase. It is reasonable to assume that, at milling doses above 1.8 kJ/g, the mixture is well homogenized and that all of the oxygen released participates in aluminum oxidation reactions.

Figure 5 presents data for the sample milled for 8 min (D = 1.8 kJ/g). The lines 1 and 3 represent the first heating in He, and the lines 2 and 4 represent the second heating. The TG curve 3 shows a weight loss in the range 200–400 °C, due to CO2 release. At higher temperatures, the sample weight remains essentially unchanged. The formation of CO2 seems to be due to the thermal decomposition of hexane, which is present in the mixture as a result of the milling in the presence of liquid hexane and is oxidized by oxygen impurities in helium in the calorimeter. The DSC curve 1 shows a small, broad exotherm below 500 °C; an endothermic peak at 540 °C; an exotherm in the range 550–700 °C, overlapping with the aluminum melting endotherm (660 °C); and two exotherms at 750 and 850 °C. The temperature of the endothermic peak at 540 °C coincides with the eutectic temperature in the Al–Cu system. All of the exothermic peaks correspond to aluminum oxidation. During reheating of the sample, no exotherms were detected, and it DSC curve (2) showed small endotherms at 540, 590, 620, and 650 °C. In this process, the sample weight (curve 4) increases slightly, which seems to be caused by the oxidation of the residual aluminum by trace amounts of oxygen in helium.

The MA of Al/CuO thermite mixtures is known to result in the fragmentation of their components, defect generation, mixture homogenization, and the formation of a composite with a large contact area [3, 5, 6]. It is reasonable to assume that the amount of free high-temperature oxygen (Fig. 4; Table 1, row 3) qualitatively characterizes the degree of homogeneity of the mixtures and the increase in contact area. At temperatures above 750–800 °C and a low oxygen pressure, CuO decomposes to give Cu2O + O and the oxygen reaches the gas phase. In the case of an unmilled mixture, this oxygen is released in the gas phase (Fig. 4) and does not contribute to the oxidation of aluminum. (O2 release is an endothermic process.) With increasing milling dose and contact area, the amount of free high-temperature oxygen decreases (Table 1, row 3), and there is no free oxygen at a milling dose of 1.8 kJ/g or above. All of the released high-temperature oxygen reacts with aluminum, and the reaction is exothermic (see the maximum at ~ 850 °C in curve 1 in Fig. 5).

Thus, at a milling dose of 1.8 kJ/g, there is already a rather large contact area between the components, which is capable of “captured” all of the oxygen released.

A negative consequence of the formation of a large contact area is the acceleration of the reaction between the components during milling. At low milling doses, the contact area is small and the degree of milling-induced conversion is insignificant. At milling doses above 1.8 kJ/g, the rate of interaction between the components during milling sharply rises and the degree of conversion caused by mechanical activation approaches almost 30% even at a milling dose of 4.4 kJ/g. Thus, to optimize the mechanical activation of the thermite system, one should stop milling as soon as there is a sufficiently large contact area. The data in Table 1 strongly suggest that optimal milling doses are 1.8–2 kJ/g.

Effect of milling dose on the exothermic heat effect, burning velocity, and brightness temperature

Figure 6 compares the areas of the “main” exothermic peak in the range 550–700 °C at different milling doses. It is seen that the curve has a maximum: the highest heat release is observed after milling for 8–10 min (milling dose of about 2 kJ/g).

Figure 6
figure 6

Exothermic heat effect in the range 550–700 °C as a function of milling time (dose D) for (Al/CuO)v

Full size image

The maximum in the heat effect as a function of milling dose is due to the competition between the fragmentation and intermixing of the components on the one hand and partial reaction between them on the other. At low milling doses, the contact area between the components is insufficient, so a considerable part of the oxygen released from CuO during heating escapes to the gas phase without reacting with aluminum. Starting at D = 1.8 kJ/g, all of the oxygen participates in the reaction. At the highest milling dose, a significant degree of conversion is reached in the course of activation in the mill, so the heat effect decreases (Fig. 6).

Table 2 and Fig. 7 present measured burning velocities in porous samples for a number of mixtures.

Table 2 Burning velocity (u) of the mechanically activated mixtures in tubes
Full size table
Figure 7
figure 7

Burning velocity as a function of milling time (dose) for (Al/CuO)p

Full size image

On the whole, the present burning velocity data demonstrate that mechanical activation to optimal milling doses ensures a considerable increase in burning velocity. In particular, whereas the burning velocity of conventional mixtures does not exceed a few tens of centimeters per second, the low-density MAECs burn with a transition to explosion, leading to the disintegration of thin-walled plastic and metallic shells. Burning velocity depends on the particle size and shape of the starting components, the density of the mixture, and the milling dose.

The highest burning velocities were obtained for the Al/CuO(v) and Al/CuO(p) mixtures at a relative density near 0.3. The burning velocity of the MAECs depends to a significant degree on the shell material and diameter, the ignition technique, and the distance between the ignition and velocity measurement sites. In most experiments, combustion was initiated by pulsed heating (30–50 ms) of a Nichrome wire (2.8–3 Ω). The process accelerates from slow burning with velocities at a level of 10 m/s to fast burning at 400–700 m/s.

Figure 8 shows the measured brightness temperature Tb of the combustion products of different MAECs. Tb varies from 2500 to 3800°K, depending on the MAEC composition and milling conditions. The highest Tb (up to 3800°K) was obtained for Al/CuO(p) at D near 2 kJ/g. Further increasing D leads to a decrease in Tb. The brightness temperatures of all the mixtures studied here vary strongly with wavelength (by up to 400°K), which can be interpreted as evidence that the state of the reaction products is significantly displaced from equilibrium.

Figure 8
figure 8

Brightness temperature as a function of milling dose D for (1) Al/CuO(v), (2) Al/CuO(p), and (3) nAl/nCuO

Full size image

Thus, despite the slight distinctions, on the whole the burning velocity and brightness temperature data are consistent with the DSC + TG data, which indicate that the highest reactivity is offered by the pyrotechnic powder-based MAEC at a milling dose of about 1.8–2.2 kJ/g.

Effect of milling dose on the ignition delay and the role of “weakly bound” oxygen

Figure 9 shows the self-ignition delay of the activated composites in contact with a hot surface maintained at temperature T. Line 1 refers to the Al/CuO(v) MAEC, with an initial particle size in the micron range, after milling to D = 2.2 kJ/g. The temperature 340 °C corresponds to the lowest temperature at which self-ignition occurred in ~ 1 s. The arrows indicate temperatures at which no self-ignition occurred. The lowest temperature at which self-ignition was detected in run 1 in Fig. 9 is 263 °C.

Figure 9
figure 9

Ignition delay t as a function of substrate temperature for the MAECs. Mechanical activation of (1) micron-sized particles and (2–4) nanoparticulate components. Milling doses of (1) 2.2, (2) 1.5, (3) 2.9, and (4) 4.3 kJ/g

Full size image

Table 3 illustrates the effect of milling dose on the lowest thermite self-ignition temperature. It is seen that, at a low milling dose, no self-ignition was detected, because the self-ignition temperature exceeded the highest substrate temperature (430 °C). At the optimal milling dose (2.2 kJ/g), the lowest self-ignition temperature is 263 °C (as seen from line 1 in Fig. 9). At higher milling doses (3.1 and 4.4 kJ/g), however, self-ignition occurs at noticeably lower temperatures (down to 175 °C), but the self-ignition knock intensity decreases.

Table 3 Minimum ignition temperatures Tign of the Al/CuO(v) MAEC in contact with a hot surface
Full size table

To shed light on the last series of experiments, Table 4 shows the measured amount of residual oxygen bonded to copper after heating in an inert atmosphere to different temperatures. Before this, the sample was reheated in a hydrogen atmosphere after the first heating in He. The last column in Table 4 indicates the amount of “weakly bound” oxygen, M1, which enters into the reaction below 500 °C, for two samples (with the optimal and maximal milling doses). It is seen that, with increasing milling dose, the amount of “weakly bound” oxygen persisting in the material after milling rises even though the degree of conversion also increases with milling dose.

Table 4 Weight loss during reduction in H2 after preheating in He to different temperatures
Full size table

Another piece of evidence that, with increasing milling dose, the amount of the most “weakly bound” oxygen in the sample increases, in spite of the reaction occurring in the system, is the systematic shift of the onset of reduction in hydrogen to lower temperatures, as seen from curves 3–6 in Fig. 2.

Thus, the optimal milling dose maximizes the heat of the reaction, burning velocity, and brightness temperature. At the same time, a small amount of “weakly bound” oxygen is responsible for self-ignition. Raising the milling dose increases the amount of oxygen in the “weakly bound” state.

Experiments with nanoparticulate starting components

We carried out several experiments aimed at activating a mixture of nanopowders [nCuO + nAl (Alex)]. Comparison of Figs. 10 and 5 demonstrates that, after mechanical activation of the nanopowders, interaction is observed at higher temperatures in comparison with the mixtures of the micron-sized particles. In the former case, there are two sharp exothermic peaks at temperatures higher than 590 and 660 °C, which correspond to the melting points of the intermetallic phase CuAl2 and Al. Thus, in the case of the nanoparticulate starting components, transformations accompanied by heat release begin only upon liquid phase formation. The brightness temperature (Fig. 8, data point 3) and burning velocity (Table 2) of the nAl/nCuO composites are similar to or lower than those of the composites prepared using micron-sized particles. Self-ignition experiments with activated nAl/nCuO mixtures in contact with a hot substrate (Fig. 9, curves 2–4) also showed that these compositions did not have any advantages. In those experiments, a large scatter in the data was observed, there was no clear correlation with the milling dose, and the self-ignition temperature always exceeded that of the mixtures prepared using micron-sized particles. Thus, the use of nanopowders as starting components offers no significant advantages.

Figure 10
figure 10

DSC curves of an activated nAl + nCuO nanopowder mixture

Full size image

Conclusions

The MA of Al/CuO thermite mixtures at room temperature is accompanied by not only the fragmentation of their components, defect generation, mixture homogenization, and formation of dense composites but also chemical reaction between the mixture components. In the range of milling doses studied, the degree of milling-induced conversion can reach almost 30%. The formation of dense composites can be qualitatively characterized by the amount of oxygen released during high-temperature (750–850 °C) CuO decomposition to Cu2O. Optimal MA conditions are determined on the one hand by the condition that the contact between the mixture components be as close as possible, so that all of the oxygen released from the CuO reacts with the aluminum, and on the other hand by the condition that the consumption of the components in the “parasitic” reaction during milling be minimized. Under the conditions of our experiments, the optimal milling dose is 1.8–2 kJ/g. These milling doses ensure the maximum heat release, burning velocity, and brightness temperature, even though these parameters have slightly different dependences on the milling dose.

As shown earlier [6], the mechanical activation of oxides, including MoO3, is accompanied by the formation of “weakly bound” oxygen, leading to a low-temperature oxygen release. The process initiates self-ignition of thermite mixtures at low temperatures. In this study, we have obtained the first indirect evidence for the formation of “weakly bound” oxygen in CuO during the activation of an Al/CuO thermite mixture. We have detected a low-temperature exothermic peak below 500 °C. That reaction with aluminum occurs at these temperatures is also evidenced by a chemical method: hydrogen reduction. The amount of “weakly bound” oxygen, capable of reacting during heating below 500 °C, increases with milling dose, even though there is active chemical interaction during milling. Thus, as a result of the reaction, more and more oxygen becomes “weakly bound.” Finally, the fact that, when a sample is brought into contact with a hot substrate, the self-ignition temperature decreases with increasing milling dose without extrema also seems to be due to the accumulation of “weakly bound” oxygen. It is worth noting that another possible cause of the increased reactivity of the oxygen in CuO when there is a close contact between the Al and CuO is the difference in electrochemical potential between Al and Cu. Further research is needed to separately assess the contributions of the two mechanisms in question.

The first experiments on the MA of composites with the use of nanoparticulate starting components show that this approach offers no additional possibilities of accelerating the process and does not appear reasonable.