Highly insensitive and thermostable energetic coordination nanomaterials based on functionalized graphene oxides

School of Chemistry, Faculty of Exact Scie Israel. E-mail: cogozin@gmail.com (M. Goz +972-36405878 Wolfson Applied Materials Research Cent Israel School of Material Science & Engineering, 100081, China † Electronic supplementary information experimental details, characterization tec of oil extraction perforator charges, XPS s SEM images of GO and corresponding materials, DSC curves of physical mixture GO–TAG/Ni(II), and a Al–Cu phase diagra peak parameters, elemental analysis, F bonding information. See DOI: 10.1039/c6 Cite this: J. Mater. Chem. A, 2016, 4, 9941


Introduction
The reduced sensitivity and improved thermostability of high performance energetic materials (EMs) are essential requirements for their practical applications. 1 The EMs are still under fast development in terms of new energetic molecules, novel energetic compositions, advanced energetic composites, as well as unique energetic systems. 2 There has been a signicant amount of EMs being developed during the past decades, including new energetic polymers, 3 novel nitrocompounds, 4 energetic salts, 5 nitrogen rich compounds, based on azole and azine rings, 6 energetic ionic liquids, 7 nanothermites, 8 energetic supermolecules (host/guest complexes, cocrystals) 9 and polynitrogen compounds. 10 However, the newly developed energetic compounds, in most cases, are still not able to completely replace the currently used high EMs due to various problems including chemical incompatibility, instability, high sensitivity, and high cost. As one of the important research topics, the highly thermostable EMs are becoming more and more attractive, since the oil/gas industry is searching for more thermostable explosive charges that could be used in perforator guns for deeper well exploration (Fig. S1 †). 11 The currently used thermostable explosives including octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) with a melting point (m.p.) of 291 C (ref. 12) and 1,3,5-triamino-2,4,6-trinitrobenzene (TATB, Scheme 1) with a m.p. of 350 C are not sufficiently thermostable to full the upcoming requirements. 13 There are several strategies for the development of highly thermostable energetic compounds. The most common one is the formation of salts, where the potassium salt of 3,3 0 -diamino-2,2 0 ,4,4 0 ,6,6 0 -hexanitrodiphenylamine (HNDPA, m.p. ¼ 334 C) is a typical example. The second methodology is the introduction of amino groups into the structure of aromatic explosophore. For instance, the introduction of -NH 2 groups into a nitrobenzene ring results in more thermostable energetic compounds such as 1,3-diamino-2,4,6-trinitro benzene (DATB, m.p. ¼ 286 C) 14 and TATB. The third strategy is to form conjugations in the molecular structure, where the best example is hexanitrostilbene (HNS), 15 whose heat resistance can be further enhanced by the introduction of amino groups. 16 The fourth way is condensation with triazole ring/s, such as picryl-substituted and picrylaminosubstituted 1,2,4-triazoles by condensing 1,2,4-triazole or amino-1,2,4-triazole with picryl. 17 The insensitive energetic compounds 3-picrylamino-1,2,4-triazole (PATO) and 1,3bis(1 0 ,2 0 ,4 0 -triazol-3 0 -ylamino)-2,4,6-trinitrobenzene (BTATNB) with a m.p. of 310 and 320 C, respectively, are good representatives of this approach. By this strategy, very stable energetic compounds such as 4,4 0 -bis(1,2,4-triazol-3-ylamino)-2,2 0 ,3,3 0 ,5,-5 0 ,6,6 0 -octanitroazobenzene (BTDAONAB) and 5,5 0 -bis(2,4,6-trinitrophenyl)-2,2 0 -bi(1,3,4-oxadiazole) (TKX-55) were prepared (Scheme 1). 18 These explosive molecules are symmetric and contain conjugation. BTDAONAB has the unique distinction of being the most thermostable explosive reported so far (with a broad exothermic peak at 550 C), 18a while TKX-55 normally starts to decompose at 335 C. 18b More recently, another thermostable energetic guanidine salt of 3-nitro-1-(2H-tetrazol-5-yl)-1H-1,2,4-triazol-5-amine (HANTT) was prepared, which has a decomposition peak temperature of 321 C, with density (1.65 g cm À3 ) even lower than that of RDX. 19 However, most of the abovementioned thermostable EMs have not been applied in the perforating guns due to difficult multi-step preparation and low synthetic yields. Recently, alternative superior ways to stabilize the energetic molecules have been developed, including the construction of energetic super structures such as co-crystals and coordination polymers. 20 It has been also reported that graphene oxide (GO) could stabilize HMX and some other energetic materials, resulting in reduced sensitivity and improved mechanical strength. 21 In addition to mechanical mixtures, GO was modied into several novel energetic supermolecules, because GO itself is energetic and can readily undergo fast exothermic decomposition. 22 Typically, GO can be covalently functionalized with selected small molecules or polymers through activation and amidation/ esterication reactions of either the carboxyl or hydroxyl functional groups on the GO surface. 23 The GO carboxylic acid groups could be rst activated by thionyl chloride (SOCl 2 ) or by a mixture of 1-ethyl-3-(3-dimethyl-aminopropyl)carbodiimide (EDC) and N-hydroxysuccinimide (NHS), followed by the coupling of amines, and the resulting functionalized products were found to be soluble in common organic solvents. 24 The epoxy groups in GO nanomaterials can also be covalently functionalized with amines. 25 Alternatively, the noncovalent functionalization of GO can also be accomplished via p-p stacking, cation-p, van der Waals interactions, or hydrogen bonding. 26 In addition, GO has also shown strong oxidizing properties and could be used as an effective oxidant in a broad range of reactions. 27 Many N-containing compounds are very good ligands in metal coordination chemistry. Among them, diaminoguanidine (DAG) and triaminoguanidine (TAG) are promising ligands, in protonated, deprotonated and neutral forms depending on their pH value, binding Cu(I) or Cu(II) as the typical metal centers. 28 The complexes Cu(DAG)(NO 3 ) 2 , Cu(DAG) 2 (NO 3 ) 2 , and Cu(DAG) 2 (NO 3 ) 2 $HNO 3 have been reported, where the DAG ligand is coordinated in the rst two compounds. For the latter complex, both the DAG molecule and the DAGH + cation are coordinated, 29 and it was also the case in the [Cu(DAGH)Cl 3 ] complex. 30 The reaction between Cu(NO 3 ) 2 and TAG in aqueous solution results in the formation of the Cu(TAG)(NO 3 ) 2 complex. 29 The coordinated compounds containing protonated TAGH + were prepared by the reaction of CuCl 2 and TAG in concentrated HCl. 31 However, the energetic properties of all the reported metal complexes of DAG and TAG were not explored.
In this work, we report the preparation of novel energetic coordination nanomaterials (CNMs) based on DAG and TAG functionalized GO, using the abovementioned strategies for GO functionalization. In particular, GO-TAG-Cu(I) and GO-TAG-Cu(II) CNMs were synthesized by the amidation process, followed by coordination reactions with CuCl or Cu(NO 3 ) 2 . This method is facile and avoids the issues of multi-step preparation processes, typical to many other highly thermostable energetic molecules listed in Scheme 1. The rst key process is to chemically gra DAG or TAG moieties onto GO akes by EDC/ NHS coupling. The second process is reacting the functionalized GO akes (GO-TAG and GO-DAG) with metal ions (Fig. 1), where the resulting structures are supported by characterization of the mononuclear 4-coordinated TAG-Cu(II) complexes.

Preparation of GO-TAG
40 mg GO was dispersed in 60 mL H 2 O by high power ultrasonication in a glass reactor. 10 mL aqueous mixed solution of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, 8 mg) and N-hydroxysuccinimide (NHS, 6 mg) was added to the dispersed GO, and stirred for 30 min. A dark grey agglomeration was formed. Then 400 mg of triaminoguanidine hydrochloride (TAG$HCl) was added to the obtained mixture in the glass reactor, which should be kept in a water bath at a controlled temperature of 75-80 C for four hours with magnetic stirring. Black occulent precipitates were gradually formed, which were ltered and washed with distilled water three times. The ltered liquid has a pH value of 1.5-2.0 due to the presence of HCl molecules. The as-prepared product (231 mg) is the TAG functionalized GO via a bridge of -CO-NH-(GO-TAG).

Chemical structure of the ECPs
Successful TAG functionalization was rst conrmed by elemental analysis (EA) and SEM/EDS analysis (Fig. 2), with the compositions and formula listed in Table S1 in the ESI. † As fabricated, GO has a C GA /O ratio of 2.0, with a water content of less than 3.0 wt% aer drying under vacuum for 48 h. In contrast, the water content for the GO-TAG was slightly increased to 5.3 wt%, while the C GA /O ratio was signicantly increased to 4.6 due to the loss of H 2 O during the amidation of carboxyl groups by the TAG moiety. The newly functionalized GO-TAG has a N TAG /C GA ratio of 0.293. GO-TAG was further coordinated with copper ions.
It has been demonstrated that TAG$HCl readily reacts with copper dichloride in concentrated HCl, forming an energetic complex, whose crystal structure is composed of [Cu 3 (GO-TAGH)Cl 3 ] + cations, NO 3 À anions and water molecules. 31 Our materials exhibit a similar coordination chemistry. By examining the morphologies of our CNMs by SEM and EDS (Fig. 2), it is clear that GO-TAG, GO-TAG-Cu(II) and GO-TAG-Cu(I) CNMs are homogeneous materials containing N and Cu atoms, indicating successful functionalization and coordination. The molar ratio of C/O/Cu is found to be very different for GO-TAG-Cu(II) and GO-TAG-Cu(I)/Cu(II) CNMs, while the ratio of Cu/Cl was maintained to be the same (1 : 1), which was also conrmed by XPS analysis of these materials. During functionalization, the GO akes were linked together by the TAG moiety bridge, resulting in a much higher thickness (150-200 nm) as compared to 2-5 nm of stacked GO in 1-5 layers. 24, 25 The coordination of Cu(II) did not change the thickness of GO-TAG due to simple doping in the voids between the layers. The crystal clusters of the TAG-Cu(I) and TAG-Cu(II) complexes were formed on the GO-TAG surfaces when extra TAG$HCl was added during the complexation reaction. The formed TAG-Cu(I) complex has smaller crystal sizes (<500 nm) as compared to TAG-Cu(II) complex crystals, with 1-4 mm in size. Interestingly, when the Cu(I) and Cu(II) ions reacted with GO-TAG in aqueous solution, several crystals were formed on the surface or between the layers of GO-TAG even without using extra TAG$HCl during complexation. It could be assumed that the reduction of GO-TAG occurred in the presence of Cu(I) ions as a reducing agent, which was then converted to Cu(II), as conrmed by XPS analysis (Fig. S3 †). During the reduction, the oxygen functionalities on the surface of GO were almost completely removed and some of the TAG ligands were detached and coordinated with Cu(II), forming the crystals on the surface. To study the functionalization and the resulting disorder on the GO akes for energetic GO-TAG-M(II) CNMs, we employed Raman spectroscopy (Fig. 3), and the corresponding peak parameters are summarized in Table S1. † In general, a higher disorder in graphite leads to a broader G band, as well as to a broad D band, with relatively higher intensity as compared to the G band. 32 The functionalization of GO with TAG and the corresponding metal ion coordination process slightly shi the peak position and largely change the intensity of both D and G bands (Table S1 †). All the CNMs based on GO-TAG exhibit a similar couple of Raman-active bands, with the D band inbetween 1343-1350 cm À1 , corresponding to a breathing mode of k-point photons of A 1g symmetry and the D band at 1575-1589 cm À1 , related to the rst order scattering of the E 2g phonon of sp 2 in C atoms. 32 The G band of natural graphite could be attributed to the in-phase vibration of the graphite lattice. 33 Aer oxidation, the intensity of the D band is increased, and the G band of GO becomes much broader, owing to the presence of isolated double bonds that resonate at higher frequencies than that of graphite. 32,33 Two bands could be observed in the samples, in a comparable D/G intensity ratio (I D /I G ) to that of GO, suggesting that the skeleton structure of GO remains in GO-TAG aer functionalization. With the evaporation of the solvent, abundant TAG groups bound to the edges of GO akes could make GO-TAG nanosheets aggregate together as multi-layered platelets with signicantly increased thickness (Fig. 2). The peaks at 273, 279 cm À1 and shoulder peaks correspond to A g and B g modes of Cu-O bonds, while the small bands at around 620 cm À1 are probably due to coordination bonds between N atoms in the TAG ligand and copper. However, there are no strong peaks observed for nickel(II), cobalt(II) and iron(II) CNMs, which is also the case for glycine functionalized GO during water purication. 34 One could also notice that the I D /I G values for Ni-and Co-based nanomaterials (Table S1 †) are very close to that of GO-TAG, due to partial van der Waals interactions between the metal ions and the TAG-graed GO surfaces.
To further address the binding chemistry of GO-TAG with metal ions, in order to generate energetic CNMs, XPS was used. Fig. 4 presents C 1s and N 1s curve tted spectra of GO before and aer its functionalization. It shows that the functionalization of GO with the TAG moiety was successful, showing additional peaks for N-bonding (Fig. S3 †). The C 1s and N 1s curve tting also shows the presence of HO-C]O and -HN-C]O bonding components (288.9 and 400.9 eV, respectively). It enables direct quantication of the degree of GO functionalization with TAG. 35 The GO surface oxygen content was reduced during the TAG functionalization and the following coordination processes, causing the disappearance/decrease of the C-O (or/and C-O-(C) peak, with simultaneous appearance of C-N, C]N, -HN-C]O and C-O-Cu peaks ( Fig. 4e and f). Although C-O (or C-O-C) and C]N peaks could be in a close proximity to each other, it is difficult to resolve the observed shi of peak binding energy in the C 1s structure from 286.6 eV (C-O or/and C-O-(C)) to 286.4 (C]N). This nding is supported by the observation of the appearance of C]N in N 1s spectra. All these data allow us to deduce that the covalent bonding of TAG-NH 2 to GO-COOH, and the coordination reaction occurred for both GO-TAG-Cu(II) and GO-TAG-Cu(II)/Cu(I). Concurrently, four new peaks corresponding to C-N, C]N, NHC]O and C-O-Cu bonds appear in the C 1s XPS spectrum for GO-TAG-Cu(II) (Fig. 4b). These chemical bonding types are further conrmed by FTIR analysis (Fig. S4 and Table S3 †). Fig. 5 shows XRD patterns of GO, GO-TAG based ECP materials, and the TAG-Cu(I) complex. The Miller indexing and corresponding plane distance data of our ECP materials are summarized in Table S5 † (determined from PXRD spectra by using Jade soware). As shown in Fig. 5a, the peak at 10.63 displays the characteristic peak of GO, corresponding to the Miller plane of 100 with a distance of 8.21Å. The corresponding Fig. 3 Raman spectra (using 3.0 mW laser) with D, G, and 2D peaks. The decreased D peak intensity of GO compared to TAG functionalized GO (GO-TAG) and its coordinated nanomaterials, indicates more structural defect sites on GO after functionalization and complexation. plane distances for the ECPs are smaller than GO, resulting in higher density ( Table 1). The peaks at about 43.5 represent the peak of the graphite phase, with the atom Miller planes indexed as 202 for GO, 131 and 061 for GO-TAG-Cu(II) and GO-TAG-Cu(II)/Cu(I), respectively. If we compare the XRD patterns of GObased ECP materials with pristine GO, the difference in the GO/ graphite peak ratio is signicant, indicating loss of oxygencontaining groups in GO planes during TAG functionalization, in agreement with the nding that there are more graphitic fractions of GO aer the emulsion process. 36 In the meantime, the peaks representing the GO planes in GO-TAG ECP materials shied to higher angles. The second and third peaks in the curves (b) and (c) correspond to the Cu-Cl complexation structure in these ECPs, which are much different from those of the pure TAG-Cu(I) complex crystal. The latter complex has a space group of P2 1 /c with cell parameters of a ¼ 6.8581(4)Å; b ¼ 3.6173(1)Å; c ¼ 8.3926(2)Å; b ¼ 98.765(5) ; a ¼ g ¼ 90 ; V ¼ 189.42(3)Å 3 with Z ¼ 4. The XRD results further conrmed the successful synthesis of a new structure of GO-TAG ECPs, instead of a mixture of GO with the TAG-Cu(I) complex.

Thermal reactivity and stability
Aer coordination with metal ions, the hygroscopicity was substantially decreased, especially in the case of GO-TAG-Cu(II)/Cu(I) CNM due to the loss of oxygen functional groups of GO. There is a negligible mass loss for these two CNMs below 140 C (Fig. 6c), which was likely due to the increase of their hydrophobicity caused by a decrease in the polar functionality on GO akes. 37 The observed rapid mass losses between 180-200 C for GO and GO-TAG are due to the pyrolysis of the oxygen-containing functional groups. This feature accounts for 25% of the initial mass of GO, whereas it is about 21% for GO-TAG, resulting in a slightly higher C GO /O ratio, which is in good agreement with XPS and EDS elemental analysis. The TAG curve of GO-TAG also shows a slow mass-loss process in the temperature range of 310-430 C in contrast to pristine GO, which is attributed to the detachment and following thermolysis of the TAG moiety, accompanied by a broad exothermic process (Fig. 6b).
TAG$HCl decomposes at about 240 C with a heat release of 2310 J g À1 . For the GO-TAG-Cu(I) and GO-TAG-Cu(II)/Cu(I) CNMs, the endothermic peaks are observed at 255.6 and 262.3 C, corresponding to the loss of water molecules upon reduction of GO. These H 2 O molecules contribute a mass loss of 4.1%, as shown in the TGA curve (Fig. 6c). The large exothermic peaks were covered with a heat release of 1792, 1192, and 847 J g À1 for GO-TAG-Cu(II), GO-TAG-Cu(I) and GO-TAG-Cu(II)/ Cu(I), respectively. The lower heat releases of the latter two CNMs again correspond to the slow reduction of the oxygen  functional groups by Cu(I). In comparison, the decomposition of nickel(II)-, cobalt(II)-and iron(II)-based materials show almost the same decomposition peak as GO-TAG in the temperature range of 170-200 C (Fig. S5 †), which further supports the partial physisorption of these metal ions on the GO-TAG surface. However, there is a weak exothermic peak for iron(II)based materials, probably due to coordination with the TAG ligand. There is a sharp endothermic peak in all cases at about 545-547 C, due to the phase transition of the CuAl 2 alloy (Fig. S6 †), which is also the case for the solid-state reaction of CuCl with metallic Al upon heating (Fig. 6a). We assume that CuCl and CuO are the main decomposition products of the described CNMs. These products could be further reduced in the presence of metallic Al to metallic copper upon heating. In the case of functionalized GO with DAG, further coordination was not so successful for all metals, although DAG could readily react with Cu(II) ions, forming an energetic complex with two exothermic peaks at above 250 C (Fig. S5 †).

Detonation performances and mechanical sensitivity
In order to predict the detonation performance of our new energetic CNMs, the EXPLO-5 soware was used based on the experimental heat of formation (DH f ) and density (r) data ( Table  1). As shown in Table 1, pristine GO is not detonable due to low energy output and low achievable detonation temperature. When 20 wt% of high energy fuel Al was mixed with GO, the detonation velocity (VoD) was calculated to be 3836 m s À1 , close to that of a blasting agent. EXPLO-5 could not calculate VoDs of our CNMs in a single phase due to their very high oxygen deciency (<À68%), indicating that our CNMs are fuel-rich EMs, which need to be combined with oxidants such as ammonium perchlorate (AP). In a subsequent calculation, when 40 wt% of AP was added to our CNMs, the calculated VoDs for GO-TAG-Cu(II) and GO-TAG-Cu(II) are 7815 and 7723 m s À1 , respectively. These VoD parameters are lower than that of 1,3,5-trinitro-1,3,5triazacyclohexane (RDX, 8750 m s À1 at a density of 1.76 g cm À3 ). The density of GO-TAG-Cu(II)/Cu(I) (3.14 g cm À3 ) is even higher than that of its parent TAG-Cu(II) complex. It is also signicantly higher than that of pristine GO and GO-TAG-Cu(II), due to the largely reduced oxygen functional groups on the surface of GO. More importantly, as highly thermostable EMs, our CNMs have much higher VoDs than the most commonly used highly thermostable insensitive EMs, including HNS (7000 m s À1 ) and TATB (7350 m s À1 ). Our CNMs are also different in their sensitivity to impact due to the change in the amount of oxygen functional groups. GO-TAG-Cu(II)/Cu(I) is more sensitive than GO-TAG-Cu(II), but both of them could be clearly classied as insensitive EMs with excellent impact sensitivity, which is much lower than that of TAG$HCl and the TAG-Cu(I) complex (Table 1). We believe that our CNMs have a potential to be used as energetic catalysts for solid rocket propellants due to the presence of copper. Moreover, because GO could be fully consumed by oxidation during high temperature combustion of propellants with improved oxygen balance, our CNMs can contribute to the fuel energy density. In fact, the functionalized GO was found to be a promising catalyst for the combustion of nitromethane, 38 and the presence of nucleophilic oxygen-containing functionalities on the surfaces and edges of our CNMs could be also chemically active sites with catalytic behavior for the combustion of solid propellants. 39

Conclusions
The EDC/NHS coupling agent can be used for the functionalization of GO with the TAG moiety, and further coordination reaction with metal ions leads to the formation of highly thermostable CNMs. These CNMs are very oxygen decient (<À68%) and should be combined with oxidizers for applications such as explosive charges and as combustion catalysts for solid propellants. Our CNMs GO-TAG-Cu(II) and GO-TAG-Cu(II)/ Cu(I) have VoDs of 7815 and 7723 m s À1 , respectively, when 40% of AP was added as the oxidant. More importantly, our CNMs are very insensitive (I m > 81 J) and thermostable up to more than 450 C, which is much higher than that of TATB and HNS.