1. Introduction
Nanothermites (NTs), which originally were referred to as metastable interstitial composites [
1], are one of the newest classes of pyrotechnic systems. NTs have found a broad array of prospective applications, both civilian and military in nature. Among potential applications of NT systems, their use in the construction of igniters, microthrusters, primers, gas generating systems and safety switches [
2,
3,
4] is expected. Moreover, NTs may find use in biocidal applications, welding and in the construction of detectors [
5,
6,
7]. This significant practical potential has brought about many research efforts into NTs, encompassing their combustion mechanisms, optimising their performance and designing novel NT systems [
8,
9,
10,
11].
Among the various NT formulations reported in the literature, most reports are dedicated to Al/CuO [
12], which is a nigh-model system. Similarly, many works have been dedicated to Al-based NTs, such as Al/NiFe
2O
4 [
13], Al/WO
3 [
14] and Al/Mn
2O
3 [
15]. The variety of oxidising agents used for NTs is not matched by the variety of fuels, with Al-based NTs being the most commonly reported in the literature. Recently, however, Ti was reported as a potential alternative to Al [
16].
Although the Ti/CuO system has been studied both theoretically and practically [
17], virtually no information is available on the mechanism of its combustion and on the impact of gas-generating additives on this mechanism. Tests have been performed to investigate the effect of Ti nanopowder addition to Al-based thermites [
18,
19] and limited studies have been conducted on the Ti/CuO system [
20].
The lack of an established mechanistic description of the combustion of titanium-based thermites may lead to the flawed assumption that their combustion takes place along the same route as the much more comprehensively described aluminothermites. To exemplify, for aluminothermites, the formation of an aluminium oxide shell on the fuel grains rapidly becomes irrelevant during the evolution of combustion due ot the low melting point of Al. Conversely, for Ti-based (melting point of 1650 K) nanothermites, even during the main combustion stage, temperatures may not be sufficient to liquefy the entirety of the fuel, greatly increasing the influence of the Ti oxide shell on shaping the reactivity of the fuel.
Although macroscopic processes, such as the above-mentioned formation of and changes in the titanium oxide shell, cannot govern the occurring chemical reactions, they shape the kinetics of these reactions to a significant extent. Consequently, even an initial combustion mechanism needs to corroborate these processes with the occurrence of individual chemical reactions, so as to provide an accurate picture of the ongoing processes. This task is made even more complex by the inclusion of additives (e.g., cellulose nitrate (NC)) that significantly affect the macroscopic mass and heat transfer conditions in the combusting system. In this work, we have endeavoured to achieve such a mechanism, encompassing both the combustion of a “pure” Ti/CuO nanothermite, as well as the combustion of a NC-supplemented Ti/CuO nanthothermite.
3. Discussion
Based on the performed tests, the following combustion mechanism appears plausible. The initial system is composed of CuO, Ti and TiO2 in the form of anatase. TiO2 is present as an oxide layer on the Ti grains, which limits access to the reactive fuel. During heating, the preliminary combustion reaction between Ti and CuO takes place up to temperatures of 350 °C. At the same time, the agglomeration of Ti particles and CuO decomposition take place, as confirmed by the occurrence of mass loss (DSC/TG) and a change in the sample morphology (SEM-EDS).
In the case of NC-doped compositions, the sintering and agglomeration process is significantly limited by gases generated during NC decomposition, whereas a high heat of reaction (in comparison with pure NT for temperatures below 250 °C) results from NC decomposition.
The main combustion reaction starts around 420–450 °C and leads to products composed of multi-walled sintered TiO
2 particles covered with smaller granular and spherical CuO and Cu particles. The low ignition temperature indicates two important conclusions. Firstly, the reaction starts in the solid–solid phase and runs in both solid–solid and solid–liquid phase systems due to the high melting temperatures of the components (1668 °C for titanium and 1085 °C for copper [
24,
25]). Secondly, the low ignition temperature may stem from not only the high reactivity of nanometric titanium but also from the catalytic effect of Ti oxides on oxidising agent decomposition [
16,
26], whereas the decomposition temperature of nanometric CuO is equal to ∼530 °C [
27].
The main combustion reaction manifests as a double peak in the DSC curves, except for the case of NC-doped NTs. This indicates that diffusion of the oxidising agent into the fuel core acts as a rate-limiting stage. During fuel oxidation, the increasing thickness of the Ti oxide layer hinders further diffusion of oxygen into the reactive core. Eventually, however, thermal expansion and changes in the chemical composition of the Ti oxides result in fracturing the oxide shell, leading into rapid movement of the oxidising agent deep into the reactive fuel core. The mentioned sealing/unsealing process of the reactive fuel core is related to the multi-step oxidation of Ti through lower oxides up to the stable TiO
2. It is important to note that the intermediate oxides are characterised by different melting points and densities (
Table 3).
The lack of this effect for the NC-doped compositions stems from lower agglomeration and sintering of the system during the previous reaction, which leads to higher intermediate contact between the components and a higher surface area of the components and, thus, shorter diffusion lengths and better availability of the fuel.
The direction of mass transfer into the fuel core is also confirmed by changes in the mass loss for different heating rates during the tests. A higher heating rate led to significantly lower mass losses, as the mass loss process is related to oxygen escaping from the system during CuO decomposition. In other words, the most violent reaction has a shorter time scale that does not facilitate the escape of oxygen. Additionally, the presence of areas with elevated content of Ti confirms diffusion deep into fuel grains; otherwise, the distribution of Ti atoms would be even throughout the whole system.
The results of the XRD tests indicate a combustion timescale that is sufficient for full oxidation of the Ti grains. Titanium spontaneously covers itself with a durable and tight layer of titanium oxides. Both titanium and anatase when annealed at a temperature of 550 °C indicated a transition to the most thermodynamically stable phase, which is rutile. Anatase is a metastable form and the process of its transformation into rutile is an irreversible reaction that takes place in the temperature range of 400–800 °C, which was confirmed by the XRD patterns. In the case of annealing at temperatures of 350 °C and 550 °C, the CuO phase was revealed, which proves the complete reaction of the copper particles present in the analysed compositions. The components of samples annealed at 750 °C of TiO2, Cu2O and CuO point to limited participation of the oxidising agent in the reaction. This was confirmed in the XRD patterns by the presence of Cu2O and Cu phases, which resulted from a deficiency in the oxygen needed for the complete oxidation of the copper. Two types of processes may occur during crystallisation, i.e., the incorporation of admixtures into the structures of the molecules included in the composition or the formation of structural defects that accelerate the transformation process of individual phases. Moreover, the above results are influenced by macroscopic factors related to the contact between components and the morphology, which undergoes rapid changes during the reaction of sintering processes at higher temperatures, which is also clearly visible in the SEM and SEM-EDS photos.
The decrease of the activation energy of this reaction stems from higher fuel content and macroscopic factors like contact between the fuel and the oxidising agent, which may be optimal at higher values of . The rapid increase in the activation energy for NC-doped compositions may result from two main reasons: firstly the oxidising agent and fuel particles may be covered with solid residues from NC decomposition; secondly, the limited sintering of components due to the gas-generating effect of the NC may deteriorate the initial contact between components.
The last combustion reaction results in the covering of the whole surface with remelted copper with a small amount of CuO and Cu2O granular particles. The reaction is related to a massive unsealing of reactive Ti cores covered with oxide layers as an effect of thermal expansion leading to the presence of cracks that facilitate violent penetration of the reactive fuel bed by oxygen and by the oxidising agent. The observed effects on the heating rate and this reaction’s contribution to the total heat confirm this hypothesis. With an increase in the heating rate, the contribution of this reaction dipped down, which stems from the lower content of reactive Ti remaining in system after the more rapid and violent main combustion reaction.
The residual CuO undergoes decomposition and, along with the Cu from the combustion reaction, covers the whole surface of the sample during solidification, which points to the presence of gaseous or liquid copper in the reaction system. The Ea values for the compositions without the NC additive are quite constant, whereas for the C3-1NC composition, it decreases almost by half, which may result from a lower reaction rate than those of the other reactions—therefore, there is higher reactive fuel availability and lower sintering of the whole system.
The morphologies of products annealed at different temperatures are quite similar between samples for all conditions. Nevertheless, we may notice an enhanced level of sintering with increased fuel content in compositions and a strong influence of gases from NC decomposition on the morphology. Therefore, we may point out that the combustion mechanism of the Ti/CuO and Ti/CuO/NC systems is in line with a reactive sintering (RS) mechanism [
30]. For the NC-doped composition, the reaction mechanism most likely is slightly different due to effect of the presence of the gas-generating NC. NC decomposition results in less sintering and the grouping of the reaction components into smaller clusters due to the presence of a strong pressure wave. For Ti/CuO systems, the heat transfer is based on conduction, while for Ti/CuO/NC, the large amount of gases enables a significant contribution of convective heat transfer. We can expect a reaction mechanism similar to the one described in [
12], which proved the formation of reacting micro-clusters composed of micrometric nanothermite agglomerates that are “held” by hot gases for a time on the order of the timescale of full combustion.
In terms of the initial chemical reaction mechanism, the first reactions take place at or below 350 °C. At this stage, an initial solid phase reaction between
Ti and
CuO (Equations (
1), (
2) and (
4)) and the evolution of oxygen being dissolved in
Ti upon the oxidation of
Ti (Equation (
3)) are observed [
31,
32] For samples containing nitrocellulose, a third reaction, i.e., the decomposition of nitrocellulose, is also observed.
TiO2 was observed by both Raman spectroscopy and XRD, with detection of both rutile and anatase indicating that partial oxidation of titanium took place in addition to the occurrence of the oxide shell on the surface of the Ti grains prior to the reaction. The TiO and TiOx, whose presence during the combustion reaction is hypothesised, cannot be detected by XRD if it is amorphous. Additionally, between annealing and measurement, the samples characterised by nanometric sizes could have been oxidised by air to the most stable product. Similarly, Cu2O was detected only by Raman spectroscopy, suggesting only a minor degree of conversion being achieved in the timescale of the experiment. The amount of dissolved oxygen that is released due to Ti oxidation corresponds to the degree of conversion and accounts for the observed mass loss of approximately 1 wt.%.
In the case of NC-containing samples, a much greater mass loss (approximately 2.5 wt.% for the C3-1NC sample) is observed. While part of this mass loss is accounted for by the 1 wt.% NC content undergoing a reaction to gaseous combustion products, the remaining mass loss is likely caused by the resulting convection promoting the release of oxygen dissolved in the
Ti.
The next stage, i.e., the main stage of combustion, takes place at or below 550 °C and follows Equation (
5); it is related to further oxidation of intermediate titanium oxides. Contrary to the case of 350 °C, the degree of conversion in this reaction is much higher, being nigh-100%, as for samples conditioned at this temperature, Ti is no longer observed by XRD. A similar mechanism was reported in the literature [
17], where gradual oxidation to stable
TiO2 was suggested. The reduction of the
TiO2 layer to
TiOx and
TiO during heating, which facilitates oxygen migration into the fuel core, was, in turn, postulated in another report on the subject [
16]. Moreover, the double peak present in the DSC thermograms may be attributed to rapid oxidation of
TiOx to
TiO2, which acts as a significantly stronger barrier to oxygen diffusion [
16,
17]. The subsequent cracking and unsealing of the
TiO2 layer at higher temperatures facilitates a further combustion process.
Interestingly, the main combustion stage is followed by afterburning (most likely combustion of residual
TiOx) and follow-up decomposition of the oxidising agent (Equations (
7) and (
8)). This is seen by the manifestation of metallic copper in the combustion products and the decrease in mass.
The effect of the NC additive on the combustion process is mostly macroscopic and was discussed in detail in previous parts. However, the presence of gases from NC decomposition may lead to partial reduction of Ti oxides at the combustion front. On the contrary, the reaction of NC decomposition products (e.g., carbon black or nitrogen) with the reactive
Ti core [
33] can be neglected (Equations (
9) and (
10)). Nevertheless, due to the low content of NC in the tested composition (1 wt.%), the effect may be insignificant in terms of quantitative and qualitative analyses.
4. Materials and Methods
4.1. Materials
The components (
Table 4) were used to produce compositions C2–C4 and C3-1NC
Table 5. All components were used as received, without any purification.
4.2. Preparation of Compositions
The compositions were prepared for three different values of : 1.2, 1.4 and 1.6 for C2, C3 and C4, respectively. Additionally, the composition characterized by equal to 1.4 was prepared with a 1 wt.% NC additive to form the C3-1NC composition.
All compositions were prepared using the same procedure. Firstly, all components were weighed and dried for 12 h at a temperature of 60 °C. The solid NC was obtained via drying of collodium solution at a temperature of 90 °C for 12 h. After drying, the powders were transferred into glass bottles and filled with propan-2-ol in such a manner as to achieve solid loading equal to 125 mg/cm3. In the next step, each composition was mixed for 30 min with a magnetic stirrer, was sonicated for 30 min (UH50 sonicator with a 0.7/0.3 work/break cycle) and was stirred for another 48 h. The prepared suspensions were electrosprayed in small portions (1 cm3) with the following conditions: voltage difference equal to 19 kV, distance between the collector and nozzle equal to 10 cm, nozzle internal diameter equal to 0.50 mm and flow rate of 3 cm3/h. After the deposition process, the compositions were gently removed from the collectors with a spatula, were dried for 12 h at a temperature of 90 °C and were poured into sealed containers.
4.3. Preparation of Samples for SEM-EDS and XRD Tests
The compositions were thermally conditioned in an argon 5.0 atmosphere in an oven for 6 h. A heating rate of 20 K/min and a cooling rate of 0.5 K/min were used for all samples. Based on the results of the DSC/TG measurements (
Figure A36,
Figure A37 and
Figure A38), the conditioning temperatures were selected:
350 °C—Below this temperature all samples showed significant mass loss, which we attributed to a preliminary reaction between Ti and CuO along with NC decomposition in the case of the C3-1NC composition. Hence, this temperature was chosen to verify that hypothesis.
550 °C—Below this temperature, the main combustion reaction took place, which is connected to significant mass loss and a high heat of reaction. Therefore, this temperature was chosen to assess the main combustion phase.
750 °C—Below this temperature, the last combustion reaction was observed, while mass remained fairly constant during this event. This temperature was chosen to verify the composition of the residues after all of the reactions.
All samples are described with labels appended with the conditioning temperature, i.e., C2-350 refers to composition C2 that was conditioned at 350 °C.
4.4. SEM-EDS Tests
The chemical compositions and morphologies of the thermally conditioned compositions were investigated using an FEI Inspect S50 (FEI, Hillsboro, OR, USA) scanning electron microscope (SEM) and an X-ray energy dispersive spectrometer with an EDS Octane Elect Plus detector and an EDAX Z2-i7 analyser (Bruker, Billerica, MS, USA), enabling the simultaneous acquisition of micrographs and energy dispersive X-ray (EDS) maps of the investigated samples. The working distance was equal to 10 mm, the acceleration voltages of the incident electrons were in range of 5–30 kV, the current intensity of the incident electronic beam was around 95 µA and the electronic beam spot size was equal to 3.
4.5. Raman Spectroscopy
Raman spectra were recorded with a Renishaw inVia Raman spectrometer and a 532 nm laser (Renishaw, diode laser, 380 mW at the source). The acquisition time for each scan and the number of accumulations were varied to improve the signal-to-noise ratio. Additionally, a large percentage (typically greater than 95 %) of the laser power was attenuated in order to prevent damage to the samples. The spectral resolution was ±1.2 cm−1. Spectra were typically collected using ×20 long-focus objectives (LEICA). Calibration of the spectrometer was carried out using a 520 cm−1 line of silicon. Data acquisition was carried out using the Renishaw WiRE 5.6 software package.
4.6. XRD Tests
X-ray diffraction measurements of the selected samples were performed on a Rigaku Miniflex 600 diffractometer using a D/teX Ultra silicon strip detector. The X-ray diffraction (XRD) patterns were recorded over a range of 20–90° 2 using Cu-K radiation ( = 1.5406 ) operated at 40 mA and 15 kV using Bragg–Brentano geometry with a scanning step size of 0.01°. The exposure time at each point was 1.67 s without sample rotation. For all scans, the IHS slit = 5 mm, Soller slits = 2.5° and DS slit = 1.25° were used.
4.7. DSC/TG Tests
DSC/TG was performed with a TA Instruments SDT Q600. The argon 5.0 flow was equal to 200 cm3/min, the sample masses ranged from 9 to 10 mg, and the heating rates were 5, 10, 15 and 20 K/min. Ceramic crucibles closed with gold disc (with a 0.3 mm hole in the centre) were used in all tests.
4.8. Kinetic Analysis
The Kissinger and Ozawa methods were applied to the DSC curves to calculate the energy of activation [
34]. The rate of reaction can be described by
k(
T) and
α(
T) functions and can be written as:
where
dα/
dT is the conversion rate,
is the conversion and
k(
T) is a temperature-dependent rate constant. The function of the conversion can be described as:
where:
Comparison of the equations leads to:
4.8.1. Kissinger Method
After appropriate transformations, which are described in detail in [
35], it is possible to present Equation (
15) as:
where:
The Kissinger method assumes that the
value does not affect (1 −
)
n−1. So after logarithmic transformation of both sides, we have:
The activation energy can be calculated from the slope of the line in the plot of lnβ/T2m vs. 1/Tm by a k-multiple of the R constant.
4.8.2. Ozawa Method
Equation (
17) can be written for a linear
as:
The above equation can be written as:
The activation energy can be calculated from the slope of the line (y = kx + b) in the plot log vs. 1/Tm by k−/(−0.4567) multiplied by the R constant.
5. Conclusions
The combustion mechanism of the Ti/CuO system is in line with the RS mechanism and exhibits an unusual direction of mass transfer, i.e., into the fuel core, which is opposite of the trend observed for Al-based nanothermites. The NC-doped composition is characterized by a slightly different combustion mechanism originating from the fact that the evolution of hot gases interferes with the sintering process. The sintering process itself is clearly evidenced by changes in the sample morphology (SEM-EDS); moreover, it is more advanced for compositions with high Ti content (C4 composition).
The multi-step combustion process is limited mostly by macroscopic (e.g., homogeneity of the composition, contact between the fuel and oxidising agent and the equivalence ratio), mass and heat diffusion factors. A higher leads to a lower Ea, higher reaction heats and a more violent combustion process. An interesting feature of the Ti/CuO system is the repeated sealing/unsealing of the reactive core of the Ti grains, which is caused by the formation of a titanium oxide shell on their surface, followed by cracking of this shell due to changes in the unit volume of the oxides (caused by changes in the chemical composition of the oxide shell).
The above effects is in line with the proposed reaction cycle, where the subsequent oxidation of Ti through TiOx oxides explains the limited access of the oxidising agent to the reactive Ti core.
The tested systems, owing to their favourable properties, may find application both in civil and military industry. The tested systems can be used in primers and igniters, for the construction of NPED-type detonators, for microthrusters, for delay systems, in welding applications or as a base NT for EM-doped NSTEX systems or gas-generating systems, e.g., for ejection seats or airbags.