Effect of relative humidity and absorbed water on the ethyl centralite consumption in nitrocellulose-based propellants

The influence of relative humidity on the consumption of ethyl centralite during the degradation of a nitrocellulose-based propellant was investigated. For such energetic material, long term safe storage – maybe decades – is required and various relative humidity conditions can be found. The investigations of chemical integrity based on accelerated ageing by heating and stability tests are not sufficient to characterize the degradation in real conditions. A sample of propellant was aged in different conditions of relative humidity and the residual stabilizer was quantified by HPLC analyses. The stabilizer consumption at ambient temperature was predicted according to the linearized Arrhenius model. We observed a substantial difference in ethyl centralite consumption by comparing the values obtained for distinct humidity conditions. A possible explanation was proposed. Our results allow one to propose an optimum range of relative humidity for long term storage.


Introduction
Nitrocellulose (NC) is an important cellulose derivative with a broad set of applications for everyday life, including varnishes, lacquer coatings, inks, resins, and adhesives. Besides these uses, NC is also used in the manufacturing of solid propellants since it is an energetic material. Recent studies obtained other energetic materials based on other cellulose derivatives (Tarchoun et al. 2021;2022a, b). The use of NC as one of the main components of most solid propellants of ammunition of weapons and rocket-motors, is related to its ease of manufacture from mixtures of concentrated HNO 3 , concentrated H 2 SO 4 , and H 2 O. Distinct physical and chemical properties are found depending on the manufacturing process, raw materials, degree of nitration, purity, and final water content (Camera et al. 1982). Its polymeric chemical structure allows obtaining a moldable mass that is suitable for extruding the sophisticated geometries used in propellants. These propellants are manufactured with NC, organic solvents, plasticizers, stabilizers, and other additives related to burning and flash control. The use of stabilizers is crucial to retard the spontaneous self-degradation to which NC is subject to, known as ageing, and allows the shelf life of propellants to reach few decades (Lindblom 2002). The degradation process, which also occurs with nitroglycerine and other nitric acid esters, consists mostly in the autocatalytic exothermic release of the nitric oxides by homolytic cleavage or hydrolysis (Guo et al. 2009). In parallel, NC chains can be broken, producing lighter compounds (Defanti et al. 2020). The stabilizers react with the decomposition products and convert them into relatively more stable compounds, thus preventing the autocatalytic process (Lindblom 2004) and reducing the global decomposition kinetics. Although several new stabilizers are being investigated (Tarchoun et al. 2022a, b;Cherif et al. 2020a, b), ethyl centralite (EC) (1,3-diethyl-1,3-diphenylurea) is a good representative for stabilizers of industrial interest (Meyer et al. 2007) for cost issues and because it reduces the propellant toxicity without compromising ballistic performance, when compared to other formulations (Mendonça-Filho et al. 2019;Rodrigues et al. 2018;De Klerk 2015;Trache and Tarchoun 2018).
NC-based propellants have about 1% of intrinsic water as a consequence of the manufacturing process. After manufacturing, the water content may increase substantially as a result of storage and handling conditions (Chin et al. 2007). A high humidity level is a critical safety issue because it affects, together with temperature, the propellant decomposition process (Druet and Asselin 1988). For this reason, both humidity and temperature should be controlled during storage, especially for long-term. However, the monitoring procedures used nowadays consider practically only the temperature effect on the decomposition, with no regard to the humidity (Eerligh et al. 1998).
Ageing of propellants can be investigated in laboratory by using accelerating methods based on heating to temperatures higher than ambient and extrapolating the data to predict the chemical stability of the propellant at ambient condition (Vogelsanger 2004). As far as we were able to verify, literature presents only a few studies about the effect of the relative humidity over the chemical decomposition of the propellant during ageing (Eerligh et al. 1998;McDonald 2011;Zhao et al. 2019). Focusing on other applications of NC, a recent study emphasized the risks associated the ethanol content (Huang et al. 2021). However, even if NC is known to be a slightly hygroscopic material (Brill and Gonger 1997) no systematic study on the kinetics of NC hydrolysis was found. Furthermore, the few studies on other nitrate esters hydrolysis suggest two main mechanisms: the acidic and the alkaline routes (Camera et al. 1982;Christodoulatos et al. 2001). So far, the available works about the effect of the humidity on the chemical stability of NC focused on the activation energy of the decomposition process and they indicated that an increase in the moisture content decreases the activation energy. This phenomenon was related to an exchange of the initiation mechanism of decomposition, from a bond cleavage to an elimination reaction in the presence of water (Guo et al. 2009). In consequence, dry NC will behave distinctly than NC previously containing humidity, when submitted to accelerated ageing.
Several methods can be used in order to determine stability and help to predict shelf life of propellants, including tests which measures the heat production, such as DTA, DSC, TG, and HFC, the evolution of gases by the Vacuum Stability test, the production of nitrogen oxides via Bergmann-Junk, Able, Vieille, and Methyl Violet tests, the mass loss with the Dutch test, and the amount of remaining stabilizer (Vogelsanger 2004). This work investigated more in-depth the effect of humidity in the consumption of ethyl centralite stabilizer in a NC-based propellant under controlled humidity conditions. A simple method to control the ambient humidity was proposed. Artificial ageing was performed at three temperatures: 60 °C, 70 °C, and 80 °C. For each temperature, five values of relative humidity were chosen: 0%, 25%, 50%, 75%, and 100%. This combination of temperatures and relative humidity values could clarify how ambient humidity affects the EC amount after ageing.

Theoretical fundamentals
Decomposition mechanisms of nitrocellulose Literature indicates that the decomposition of nitrate esters occurs by two main pathways: thermolysis and hydrolysis (Chin et al. 2007). However, the mechanism usually mentioned for the initial phase of decomposition is thermolysis, with the breaking of the O-NO 2 bond (155 kJ/mol), as it is weaker than C-ONO 2 (238 kJ/mol) (Lindblom 2004;Kimura 1988). Chemiluminescence analysis indicate that radicals are formed and they can react to form a peroxide and nitric oxide radical, which can react with oxygen and form nitrogen dioxide (Kimura 1989).
Hydrolysis is commonly considered dependent on the initial phase of thermolysis. The NO 2 formed in the initiation of thermolysis reacts with H 2 O producing nitric acid, causing hydrolysis (Eerligh et al. 1998). However, some results presented in the literature propose that hydrolysis can start independently of thermolysis due to the presence of acidity (Brill and Gongwer 1997). Due to the scarcity of studies specifically related to the hydrolysis of NC, this mechanism will be addressed with the available information about general nitrate esters. The hydrolysis of a nitrate ester in acidic medium involves the formation of nitric acid. The acid formed triggers other reactions, with the formation of nitrous acid and causing the autocatalytic process.
Three previous works investigated the influence of relative humidity (RH) on the decomposition of nitrate esters (Eerligh et al. 1998;McDonald 2011;Zhao et al. 2019). In the first, heat flow calorimetry (HFC) was used to study the behavior of propellants of distinct compositions in the presence of a mixture of glycerol and water in different proportions. The results showed that higher values of relative humidity increased the heat production. This increase was exclusively attributed to an acceleration effect on the decomposition of the propellant, due to a reduction in the activation energy, without changing the reactions responsible for the heat produced (Eerligh et al. 1998).
In the second study (McDonald 2011), the ageing of a propellant containing RDX and NG was investigated under controlled humidity. Tests performed up to 102 days at 60 °C indicated that the highest and the lowest stabilizer consumption were found for RH = 100% and RH = 0%, respectively. However, the relation of stabilizer consumption and relative humidity was found to be nonlinear since the consumption was lower for RH = 30% than for 6%. The lowest consumption recorded was found when RH = 32%. These results indicate that although the existing ageing assessment methods consider only the temperature, humidity has a substantial influence in the degradation velocity of NC.
In the third study (Zhao et al. 2019), the effect of relative humidity and temperature on the shelf life was investigated using hygrothermal ageing of two NC-based propellants stabilized with diphenyl amine (DPA). Although this study predicted the propellants shelf life, no discussion was presented about how the relative humidity affected the ageing, and also only few values of higher RH (50-80%) were tested.

Stabilizer depletion and accelerated ageing
Stabilizer consumption is influenced by the propellant decomposition as the stabilizer captures the NOx generated by the degradation of NC, and generates other derivatives, that depending on its chemical structure may still have some ability to absorb more NOx. For ethyl centralite, degradation mechanisms can be found in the literature (Curtis 1987;Volk 1976). The mechanisms of stabilizer consumption are complex and may not always be caused by the capture of gases from the NC degradation, although it is certainly influenced by it. EC can also be consumed via acidic hydrolysis. Nevertheless, the remaining amount of stabilizer is one of the important metrics used in standardized propellant stability protocols (Vogelsanger 2004;Trache and Tarchoun 2018).
Energetic materials are artificially aged by applying temperatures higher than room temperature to increase the rate of decomposition, assuming the reaction mechanism is the same found in the ambient conditions (Bégin and Kaminska 2002). Arrhenius equation describes the dependence of the reaction rate on temperature. Its linearized version is given by where k is the rate constant, A is the pre-exponential factor, E a is the activation energy, R is the ideal gas constant and T is the absolute temperature. Linear regression of the experimental data obtained at the high temperatures allows to predict the behavior for lower temperatures (Vogelsanger 2004).
The kinetics of stabilizer consumption is usually considered to be first order. However, first order models are related to very simple descriptions of the reaction mechanism, especially when reaching low stabilizer contents. A combination of first and zero order is a preferable approach when the first order model fails (Bohn 2009;Bohn 2009). The kinetic models for the zero order, first order, and combined zero-first orders are, respectively, given by where S(t) is the stabilizer content at instant t and k 0 and k 1 are the rate constants for zero and first order. Below 100 °C, the mechanism is expected to be the same, by hypothesis, for all values of relative humidity and all temperatures studied (Brill and Gongwer 1997;Kimura 1989;Pfeil et al. 1985;Volk and Wunsch 1985;Sovizi et al. 2009). In this sense, only the rate constant will change, being the reaction order the same.

Sample preparation
A sample of commercial single base propellant manufactured in October 2019 was used in this study. The high content of NC was especially relevant for this investigation in other to reduce any possible interference from other components in the formulation. The original sample contained above 95% w/w NC, 1.42% w/w of ethyl centralite, 1% w/w water, being the composition completed with graphite coating and burning additives. The propellant grains were shaped as porous discs with diameter of 0.9 mm (Fig. 1).
The temperatures of 60, 70, and 80 °C, and five values of relative humidity of 0, 25, 50, 75 and 100%, were used to simulate the storage conditions of the propellant. According to the literature, the test periods equivalent to 10 years at normal temperature are 90, 30 and 10.5 days, for the three selected temperatures (Volk and Wunsch 1985;NATO 2008), respectively. Each sample contained 5 g of propellant, and they were aged within test tubes with a solution of glycerol/water in the bottom to provide the relative humidity, as no specific method for ageing of NC in controlled humidity atmosphere was found in the literature (Eerligh et al. 1998;Forney and Brandl (2) 2018). The condition of RH = 0% was produced by keeping silica gel in the same ambient as the sample. The composition of each water/glycerol solution and the corresponding relative humidity are presented in Table 1.
Glass wool was used as porous material to separate the solution from the sample that was placed in the top of the tube (Fig. 2). The tubes were not closed tightly, in order to allow the release of gases and therefore to maintain ambient pressure in the tubes. The samples were heated using a classical stability test block with electrical heating and digital temperature control. Test tubes were not completely filled in order to promote an oxygen content similar to that found in ammunition (De Klerk and Boers 2003).
For each temperature, the stabilizer consumption was evaluated for three artificial ageing periods according to Table 2. The corresponding natural ageing periods can be found in AOP-48-Edition 2  The water absorbed by the samples were measured by weighting them right after being removed from the heating block, and again after being stored for drying in a desiccator with silica gel for at least 30 days. The weight difference was considered to be absorbed water. The use of a desiccator aimed at achieving a very low humidity without submitting the sample to heating, which could favors additional ageing. Afterwards they were prepared for the chromatographic analyses in order to determine the residual stabilizer content. The effectiveness of this procedure in recovering the total stabilizer amount was confirmed with a preliminary calibration with a known sample to confirm the results.

Determination of residual stabilizer by high performance liquid chromatography
The stabilizer consumption was performed by the determination of the propellant residual stabilizer via HPLC (Trache and Trachoun 2018; Trache and Khimeche 2013). The stabilizer was extracted from the samples with acetonitrile (ACN) leaving behind the NC, to prevent the obstruction of the column. An amount of 1 g of the sample was solubilized by ultrasound agitation for 4 h in a 50 mL volumetric flask, with a solution of 60% v/v ACN HPLC grade in 40% v/v deionized water, similarly to Wilker et al. (2007). The NC was settled by letting the suspension rest for 1 h. An aliquot of 1 mL was taken from the supernatant and diluted to 25 mL in another volumetric flask with the same ACN:H 2 O solution from the extraction.
The HPLC analysis was carried out in an Agilent 1260 Infinity Liquid Chromatograph with automatic injector. An UV detector at 250 nm and an Agilent Poroshell 120 EC-C18 4.6 × 50 mm and particle size 2.7 µm column were also used. The mobile phase consisted of a solution of 70% v/v ACN and 30% v/v deionized water in a flow of 0.7 mL min -1 at 35 °C. A volume of 1.5 mL of the solution were transferred to a vial and 10 µL and were injected into the equipment in triplicate. The retention time with this configuration was lower than 5 min for the ethyl centralite.
The chromatographic reference sample for ethyl centralite was provided by Sigma-Aldrich (ID: 372889) and had purity higher than 99%. Three solutions of 50 mL were prepared with 80 mg, 154 mg and 310 mg of EC and completed with ACN HPLC grade. 1 mL of this solution was transferred to a 100 mL flask and diluted with a solution of ACN:H 2 O of 60:40 in volume. The resulting concentrations were, respectively, 800 ng·µL -1 , 1540 ng·µL -1 and 3100 ng·µL -1 . The retention time of 1.797 min was used to plot the calibration curve.

Results and discussion
Moisture after artificial ageing at 60 °C, 70 °C, and 80 °C The values of the moisture content were computed for each sample and for the three temperatures investigated (60, 70, and 80 °C). The results indicated the effectiveness of controlling moisture absorption as a function of the relative humidity inside the test tube during ageing. In addition, ANOVA analysis indicated that there was not a significant difference in the water absorption by comparing the results found for the three temperatures. Figure 3 shows data for average values of the absorbed water in the samples  versus RH (%). A monotonic third-degree polynomial fit was used to describe the behavior found.

Residual stabilizer content
The initial concentration of EC was determined for the unaged sample in order to establish a reference for the stabilizer content. The result obtained was (1.415 ± 0.020) % w/w for a 95% confidence interval. The data obtained for the remaining ethyl centralite content (% w/w) after the artificial ageing at the three chosen temperatures and five values of relative humidity are presented in Fig. 4. The results show narrow confidence intervals for averages. The previous data for the stabilizer consumption presented (Fig. 4)  For each temperature the stabilizer consumption was evaluated at RH = 0, 25, 50, 75, and 100%. Errors bars indicate the confidence intervals (95%) for averages of the three runs of each sample constant k for each value of RH and temperature. The first order model (Eq. (3)) fitted better to the data in comparison to the zero order (Eq. (2)) and the combined zero-first order (Eq. (4)) models. The behavior of the rate of stabilizer consumption calculated for the first order kinetics as function of the absorbed water is presented in Fig. 5. For RH = 0%, the decomposition rate was very low. However, it significantly increased when RH was increased to 25%, followed by lower rate at RH = 50% and again an increase at RH = 75%. The rate constants found for RH = 100% at the three temperatures are higher than the values found for RH = 0%, but they are not the maximum values.
The behavior found can be explained for each RH value. Firstly, when RH = 0%, the residual free acids from the manufacturing process would be practically inactive and would not attack the nitrate ester groups of the NC. When RH = 25%, the small water content absorbed would create a higher concentrated acid in the sample which would react with the NC considering the acidic hydrolysis mechanism for nitrate esters. For RH = 50%, this acid would become diluted and the rate of decomposition would be slower than that of RH = 25%. When RH = 75%, the high amount of water favors NC hydrolysis initiated by an elimination reaction (Guo et al. 2009) combined with a weak acidic hydrolysis, increasing the consumption rate of EC. For RH = 100%, the effects of acidic hydrolysis are lowered leading to the predominance of elimination reactions. It can be also said that the hydrolysis is not just a secondary process that depends on the thermolysis. It plays a significant role in the decomposition, especially at higher temperatures, confirming previous results (Brill and Gongwer 1997).
From the rate constants computed assuming a first order model for 60, 70 and 80 °C, it was possible to find E a and A of Arrhenius equation (Eq. (1)) for the five values of relative humidity investigated. In all cases, the regression led to R 2 > 0.98. With these values, the rate constant k was extrapolated down to 30 °C for each case and the first order kinetics halflife, t 1/2 (≡ k -1 ·log (2)), at this temperature was computed. The expected remaining stabilizer content (% w/w) after 10 years of ageing at 30 °C was computed using the kinetic parameters for the RH values investigated. The results are presented in Table 3.
In general, chemical stability standards are established for a storage period of at least 10 years (De Klerk and Boers 2003), and therefore this period was used as a reference to evaluate the relative humidity influence in the stabilizer consumption. It was found that RH = 25% led to the faster rates of the decomposition process. By selecting 50% of the stabilizer initial content as a control parameter and using linear interpolation between the RH values, the interval of

Conclusions
This work presented an efficient method to investigate the effect of relative humidity on the stabilizer consumption in a NC-based propellant manufactured with EC (1.42% w/w). Samples of the single base propellants were conditioned in a test tube apparatus which provide a controlled relative humidity condition. Preliminary results indicated that the propellant studied was not a hydrophobic system. The higher the relative humidity, higher the water absorption by the samples. HPLC results were used to determine kinetic constants of EC consumption along the propellant ageing under five values of relative humidity from 0 to 100%. The kinetics fitted better a first order model. These data were used to predict the rate constant of decomposition and the remaining stabilizer content after 10 years at 30 °C, which is a typical operational condition, and reproduces long term storage. At this temperature, higher decomposition rates are expected to be found for 20.2% < RH < 30.4%. In consequence, the results indicate that the relative humidity is an important control parameter to be considered for the prediction of propellant chemical integrity and storage management. Our results did not confirm neither the prediction of RH = 100% as leading to the fastest decomposition rate (Eerligh et al. 1998), nor to RH ~ 0% or RH ~ 40% having the higher rate parameters (McDonald 2011).
The results suggest two processes in the propellant degradation by hydrolysis. The first comprehends influence of the concentration of the free acids in the sample, which seems to be more relevant for 0% ≤ RH ≤ 50%. This was observed experimentally by an inactivation of the hydrolysis at RH = 0%, an increase in the reaction rate for RH = 25%, when there will be high concentration of free acids, and the dilution of those acids at RH = 50%, leading to a decrease of the reaction rate. The second process involves the increase of the decomposition rate due to the increase moisture content absorbed by the sample, which provides more water available for the pure hydrolysis reaction, when RH ≥ 75%, initiated by an elimination reaction (Guo et al. 2009) combined with a weak acidic hydrolysis, increasing the decomposition rate. For RH = 100%, the effects of acidic hydrolysis are lowered leading to the predominance of elimination reactions and no enhance of pure hydrolysis.
Since this study was conducted in the early stage of the decomposition, future works may be carried considering larger ageing periods in order to evaluate the order of reaction and to confirm the profile of stabilizer consumption versus relative humidity.