A. Plummera,
V. A. Kuznetsovab,
J. R. Gascooke
a,
J. Shapter
ac and
N. H. Voelcker
*d
aCentre of Expertise in Energetic Materials, School of Chemical and Physical Sciences, Flinders University, 5042, Bedford Park, Australia
bWeapons and Combat Systems Division, Defence Science and Technology Group, Edinburgh, SA 5111, Australia
cCentre for Nanoscale Science and Technology, School of Chemical and Physical Sciences, Flinders University, 5042, Bedford Park, Australia
dFuture Industries Institute, University of South Australia, 5095, Mawson Lakes, Australia. E-mail: nico.voelcker@unisa.edu.au; Tel: +61 8 8302 5508
First published on 23rd January 2017
Nanoporous silicon (pSi) films on a silicon wafer were loaded with sodium perchlorate (SP) and perfluoropolyether (PFPE) oxidising agents to generate a pyrotechnic energetic material. The potentially violent reaction between the silicon and the loaded oxidising agent was studied using correlated differential scanning calorimetry (DSC) and FTIR spectroscopy for samples heated continuously between ambient and 500 °C. We observed that the energetic reaction between pSi and SP depended on the presence of various hydride species on the surface of freshly etched pSi, and on formation of volatile free radical species released during either oxidation of the surface in the presence of air at about 200 °C or during desorption of the hydride above 270 °C in the absence of oxygen. However, energetic reactions between pSi and PFPE were delayed until pyrolysis of the PFPE above 390 °C in the absence of oxygen, suggesting PFPE's suitability for pyrotechnics applications. Correlated thermal and spectroscopic methods of analysis gave new insights into the earliest stages of the reaction of these energetic materials.
A variety of oxidising agents (typically nitrate or perchlorate salts) have been shown to react energetically with pSi. Sodium perchlorate (SP) has emerged as the oxidiser of choice since it delivers the most powerful and consistent energetic reaction.2 Conveniently, controlling the material properties of the pSi substrate (by tuning the etching conditions) permits control over the stoichiometry of the system by limiting the volume of the pores available to be filled. The burning rate of pSi energetic films has been reported from as high as 3600 m s−1 (ref. 11) to less than 1 m s−1.12 The choice of oxidising agent, porosity, pore depth and pore diameter are all interdependent factors which govern the burning rate of the system. Fluorinated polymers (such as perfluoropolyether (PFPE)) have recently been investigated as an alternative to ionic oxidising agents, with reported burning rates typically less than 1 m s−1.12
It is recognised that pSi EM's can be excessively sensitive, rendering those materials ill-suited for real world applications such as MEMS actuators or air-bag initiators envisaged in earlier works.2,10,13 Knowledge of the reaction mechanism between pSi and an oxidising agent is an important factor to help understand the excessive sensitivity, and may assist in the selection of oxidising agent or material treatment regimes in order for these materials to be translated into real-world applications.
Simple bomb calorimetry can determine the total energy released during reaction of a material.14 However, the reaction of an energetic material involves sometimes complex decomposition stages and formation of intermediate species. Investigation of these processes by controlled thermal analysis methods can yield important information, including reaction endo- and exotherms, or intermediate transition or decomposition temperatures all of which gives an insight to the mechanism of the reaction.15
Investigating the reaction between the pSi surface with liquid oxygen, Kovalev et al.16 first proposed that hydride termination of the freshly etched pSi surface was an essential passivating barrier to prevent spontaneous oxidation. Rupturing the Si–H surface bonds or sub-surface Si–Si bonds by various means created free radicals susceptible to reaction with oxygen. The exothermic reaction between a single Si atom and oxygen molecule releases sufficient energy to disrupt neighbouring bonds. A single point of initiation can then rapidly cascade up to an explosive reaction.
Churaman et al.17 investigated the mechanism of the reaction between a pSi film (approx. 4 nm pore diameter, 40 μm layer thickness) and SP. Using differential scanning calorimetry (DSC) dehydration endotherms of the monohydrated SP within the pSi pores were observed at approximately 51 °C and 149 °C, a phase transition of the SP at 307 °C, and finally an explosive reaction at 320 °C. No significant exotherms were observed below the temperature of explosion. They concluded that the mixture commenced reaction initially by formation of reactive ions through dissociation of the SP or H2O during these dehydration steps, which are then free to react with the pSi surface. Again it was observed that the presence of the hydride termination on the pSi surface was essential to promote a strong explosive reaction.
Becker et al.18 investigated the influence of atmospheric conditions on the reactivity of pSi films impregnated with SP. Using bomb calorimetry, freshly etched pSi prepared by galvanic corrosion (up to 150 μm thick layers, 62 to 69% porosity, pore sizes 2.7 to 3.1 nm) loaded with SP yielded gross reaction enthalpies of 9.9 kJ g−1 of pSi in a N2 atmosphere and 27.3 kJ g−1 in O2. They further observed that oxygen back bonding beneath the hydride-terminated pSi surface resulted in a weak exotherm at about 250 °C. This was confirmed by FTIR spectroscopy. In the absence of the hydride termination, no energetic reaction was observed. Hence, the presence of the hydride termination on the pSi surface was essential to enabling the material to react explosively according to the authors. They further concluded that the substantial difference in the total energy of the reaction recorded between N2 and O2 atmospheres was due to the mixture being substantially fuel rich (a limitation imposed by the available pore volume within the pSi layer), thereby allowing remaining Si to react with the oxygen after the main reaction with the SP.
Parimi et al.19 studied the reactivity of various pSi EM's using DSC and TGA, investigating the influence of doping type and doping concentration of the original Si wafer. Wafer dopant concentration and type influenced the reactivity of the material, not directly by contributing to the reaction between the pSi and the oxidising agent, but indirectly by impacting on the morphology and specific surface area of the pSi generated by means of anodisation. Studying the propagation of the flame through pSi EMs of different morphology,20 the authors observed that the peak temperature of the first and strongest exotherm (which was attributed to the solid phase reaction between pSi and SP) shifted to lower temperatures as the specific surface area of the pSi increased, from approximately 400 °C at a surface area of 284 m2 g−1 to approx. 280 °C at a surface area of 730 m2 g−1. A calculation of the activation energies for these two materials gave values of 124.5 kJ mol−1 for the low surface area pSi and 72.1 kJ mol−1 for the high surface area pSi. From those calculations and others, it was determined that the propagation of the flame through pSi was due to a combination of both conductive and convective burning, not simply a conductive mechanism as observed in other solid-phase energetic materials.
Infrared spectroscopy is a versatile technique often applied to pSi research due to the wealth of information available about the chemical structure of this high surface area material. Oxidation of a pSi surface heated continuously in an O2 atmosphere was observed by Mawhinney et al.21 with the results showing that the reaction proceeded initially by insertion of oxygen into the Si–Si bonds behind the Si–H surface species when the temperature exceeded 300 °C. As the oxidation proceeded, the hydride species were gradually replaced by hydroxyl groups and eventually, the surface was completely oxidised. Likewise, Salonen et al.22 used FTIR in combination with DSC, showing a strong exothermic reaction of pSi heated in air commencing at about 200 °C, peaking at approximately 300 °C and attributed this to back bond oxidation of the pSi surface. Furthermore, a broad exothermic tail observed in the DSC rising from about 400 °C and terminating at approx. 650 °C was attributed to the formation of surface hydroxyl groups. However, in their study Salonen et al.22 cooled the samples after heating, prior to FTIR analysis.
Both DSC and FTIR studies independently yield valuable information about the structure of pSi and the possible reaction pathways. Significantly, Mawhinney et al.21 recognised the need to continuously record spectra during the heating process to minimise kinetic effects caused by cooling of a heated sample prior to measurement. Such problems would be exacerbated for the fast reacting pSi EM's studied here. It is therefore the aim of this work to combine these two techniques to obtain valuable real-time spectroscopic information regarding the reaction of pSi, and to correlate that to observed reactions in the thermal analysis.
Secondly, a stock solution of the PFPE (Fomblin Y HVAC 25/9) was prepared at a concentration of 200 g L−1 in the fluorinated solvent perfluoro(butyltetrahydrofuran) (Fluorinert FC-75). Fomblin Y is a high viscosity liquid at room temperature. Impregnation of this material into the pores required the stock solution to be pipetted onto the surface, followed by evaporation under vacuum in a desiccator with subsequent back-filling of the vessel using N2 – three successive steps were required in order to load the wafer.
The mass of the EM (i.e. the net explosive quantity) was determined gravimetrically. Samples were weighed prior and subsequent to loading with the oxidising agent, and the unreactive supporting wafer was weighed subsequent to analysis once the pSi layer had been stripped away using dilute aqueous potassium hydroxide.
All spectra collected for an individual sample were background corrected to compensate for the decreasing transparency of Si at the temperatures of these experiments – details of the correction method are provided in the ESI.† Table 1 lists the characteristic spectral peaks for various species involved in this investigation.
Wavenumber (cm−1) | Transition | Ref. |
---|---|---|
616, 621, 643 | ClO4− vibration | 26 |
615, 622, 666 | Si–Hx deformation, overlapped with Si crystal modes | 21 |
671, 730, 776 | C–C/C–O/C–F/C–F2/C–F3 stretch | 27 |
840 | Si–N asymmetric stretch (>800 °C) | 28 |
841, 877 | O–SiHx (unknown mode) | 21 |
906 | ClO4− vibration | 26 |
915 | Si–H2 scissoring | 21 |
940 | Si–F stretch | 29 |
950–1250 (broad) | Amorphous Si–O2 absorbance | 21 |
980 | Overlapping C–C/C–F2/C–F3 stretch | 27 |
1011, 1127, 1192 | ClO4− vibration | 26 |
1015, 1121 | Si–O–Si vibration | 21 |
1029, 1190, 1826 | Si–F4 (unknown mode) | 30 |
1108 | Si–O–Si crystal stretch | 21 |
1110 | ClO4− vibration | 26 |
1131–1345 | Multiple overlapping C–C/CF2/CF3 stretch | 27 |
1157 | C–F3 (unknown mode) | 30 |
1200 | C–F stretch | 31 |
1283, 1539, 2186, 2561 | C–F4 (unknown mode) | 30 |
1300–1150 | C–O–C; C–F stretch | 31 |
1372 | Overlapping C–C/CF2/CF3 stretch | 27 |
1929 | COF2 (unknown mode) | 30 |
2090, 2115, 2141 | Si–Hx stretch | 21 |
2214, 2273 | OSi–Hx stretch | 21 |
3736 | SiO–H stretch | 21 |
3745 | SiO–H stretch | 21 |
3750–4250 | H–F stretch | 30 |
![]() | ||
Fig. 1 DSC scans of (a) blank pSi in air, (b) blank pSi in N2, (c) blank SP in air, (d) pSi/SP EM in air, (e) pSi/SP EM in N2, (f) blank PFPE in air, (g) pSi/PFPE EM in air, and (h) pSi/PFPE in N2. |
However, for pSi heated in N2 (Fig. 1b and 2b), the DSC showed a weak exothermic peak with an onset from approximately 270 °C and peaking at approximately 380 °C. FTIR spectra displayed the typical triplet of peaks at 2109 cm−1 associated with multiple Si–Hx stretching reducing gradually to a single peak centred at 2082 cm−1 between 350–400 °C. The onset temperature of this transition observed on the FTIR spectra (≈350 °C) was higher than observed on the DSC (≈270 °C), however the FTIR resolution may prevent subtle changes from being observed at lower temperatures. Similarly, the peak at 910 cm−1 (Si–H2 scissoring) decreased over the temperature range tested, disappearing completely by ≈400 °C. It seems that pSi heated in N2 simply desorbs the surface hydrogen slowly commencing at ≈270 °C. By the end of the experimental run at 503 °C the hydride peaks still had not completely disappeared, and the absence of gaseous oxygen prevents further oxidation of the surface. A very weak peak at 850 cm−1 appears at 389 °C and increases intensity with temperatures up to 468 °C. This peak may be attributed to incorporation of nitride species on the Si surface based on the study by Scardera et al.28 However, their study was conducted on samples of thin silicon nitride films deposited at 400 °C and annealed to over 800 °C, and revealed multiple fine peaks centred around 840 cm−1 that were assigned to modes of Si–Nx asymmetric stretching depending on the annealing temperature. Thus, we tentatively assign our observed peak at 850 cm−1 to surface nitride species.
When SP was incorporated into the pSi layer to create an EM, significant changes in the FTIR spectra were observed compared to blank pSi and pure SP (Fig. 2a and c). Initially, the strong absorption band centred at 1108 cm−1 due to SP was clearly distinguishable from relevant pSi peaks at 910 cm−1 and the triplet band at ≈2109 cm−1 for samples heated in either air or N2 atmosphere (Fig. 2d, e and 3c).
For pSi/SP heated in air (Fig. 1d and 2d), the FTIR spectra displayed a clear transition commencing at approximately 200 °C – this correlates with the onset temperature recorded in the DSC (203 °C). This change was rapid and complete by approximately 250 °C (with DSC recording a maximum at 233 °C). These changes are associated with strong oxidation of the pSi surface as evident from the disappearance of the Si–Hx triplet band and associated appearance of the OSi–Hx back bonding (approximately at 2254 cm−1). The broad absorption band between ≈800–1200 cm−1 was also attributed to Si–Ox vibrations. Changes in the perchlorate absorption peaks at ≈1100 cm−1 were masked at temperatures above 200 °C due to broad Si–Ox absorption (800–1200 cm−1). These changes occurred at a temperature far lower than the melting point of SP (477 °C) and its associated thermal decomposition. As suggested by Churaman et al.17 this may include release of volatile O and OH radical species from decomposition of the perchlorate which may then interact with either of the two solid phase materials present, leading to a sustained reaction.
In a N2 atmosphere (Fig. 1e and 2e), energetic pSi/SP displayed a similar sudden change in the FTIR spectra with an onset of approximately 270 °C, 70 °C higher than in the presence of air. This correlated with an exothermic reaction observed by means of DSC commencing at around 270 °C with a sharp peak at 300 °C. DSC and FTIR results for blank pSi in N2 (Fig. 1b and 2b) indicate that the surface hydride desorption commences at approximately 270 °C, and this desorption appears to be an important first step leading to commencement of the energetic reaction, in the absence of oxygen. Becker et al.18 reached the same conclusion. Again, these reactions occur at a temperature lower than the melting point or decomposition point of the SP.
An additional observation recorded on the DSC scans of EM's run in air (Fig. 1d and 2d) is a shoulder in the main exothermic peak at 272 °C, a feature not observed in the samples in N2 (Fig. 1e and 2e). This shoulder may be due to afterburning reactions of unreacted pSi with the atmospheric O2 within the DSC. Afterburning reactions are common in fuel-rich samples which is a possibility with the smaller samples prepared in this study. As noted in the experimental section, loading of small samples of pSi was difficult due to precipitation of the SP on the exterior face of the wafer, leading to variability of the quantity and distribution of the SP within the pSi pores. Additionally, the porosity of the pSi in this investigation (65%) was slightly lower than the calculated ideal porosity required to achieve a stoichiometrically balanced mixture (70–72%),2 indicating that the system investigated here was fuel rich, thereby facilitating afterburning reactions.
The FTIR scans of pSi/PFPE EM heated in air (Fig. 1g, 2g and 3e) were almost indistinguishable from the appearance of blank pSi in air except for the presence of the PFPE peak at 985 cm−1 and the band at 1200 cm−1. Oxidation of the pSi surface due to the oxygen in the air was evident at the same transition temperature of approximately 250 °C but appeared less intense compared to the spectra of blank pSi heated in air. In accordance with this observation, the DSC results of pSi/PFPE in air showed the same exothermic process as blank pSi in air but with a broader transition. No clear exotherms were observed to indicate an energetic reaction. One possible explanation for this is that the presence of the PFPE within the pores limits the rate of diffusion of oxygen into the surface such that the pSi reacts slowly with O2 prior to pyrolysis of the PFPE.
In contrast, pSi/PFPE EM in N2 displayed a different profile (Fig. 1h, 2h, and 3e). The typical pSi peaks (910, 2213 cm−1) and PFPE peaks and bands (985 & 1200 cm−1) were present and persisted until above 350 °C. The strong exothermic transition observed in the DSC occurring at 397 °C correlated with the loss of most of the PFPE peaks in the FTIR spectra from 390 °C onwards and also strongly attenuated pSi hydride peaks centred at 910 and 2213 cm−1. Simultaneously, a peak at 1029 cm−1 (very weak prior to the transition) increased, potentially indicating production of SiF4. Some remnants of the pSi surface were seen in the FTIR spectra above this temperature and may be due to the presence of unreacted pSi left after all of the PFPE has been consumed. The peak at 1029 cm−1 was the only indication of a reaction product (SiF4). Alternative species (such as HF at 3750–4250 cm−1) were not observed.
When the pSi/PFPE results are compared to the pSi/SP heated in N2 it is apparent that the energetic reaction occurred at a far higher temperature (≈400 °C for PFPE compared to 300 °C for SP). For pSi/SP, it was concluded that desorption of the surface hydride from the pSi was the initial step in the energetic reaction. For pSi/PFPE, it appears that the reaction did not occur even above the temperature required to commence hydride desorption, but was delayed until pyrolysis of the PFPE occurred at about 400 °C. Release of oxidising species from the PFPE during pyrolysis appears to be necessary to induce the energetic reaction. This mechanism is very similar to the reaction of other perfluorocarbon-based energetic materials.33
It was found that breaking of the Si–Si back bonds behind the hydride surface upon reaction with atmospheric oxygen was sufficient to permit the SP to attack the surface and commence the energetic reaction. However, in the absence of oxygen, desorption of the hydride species from the pSi surface was a necessary step to initiate reaction of the pSi with SP. Turning our attention to pSi loaded with PFPE, the pSi surface was oxidised in an air atmosphere before having the opportunity to react with PFPE. Yet in a N2 atmosphere, the reaction between pSi and PFPE was delayed until the onset of pyrolysis of the PFPE. Heating a sample slowly under these controlled conditions is a departure from the normal mode of the explosive reaction of these materials, but the information gained in this way provides important insights into the earliest stages of the reaction.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra27028j |
This journal is © The Royal Society of Chemistry 2017 |