combustion of energetic porous silicon composites containing different oxidizers

DOI: 10.1002/prep.201500108

Combustion of Energetic Porous Silicon CompositesContaining Different OxidizersAni Abraham,[a, b] Nicholas W. Piekiel,[a] Christopher J. Morris,*[a] and Edward L. Dreizin[b]

1 Introduction

Observations of highly exothermic reactions betweenporous silicon (PS) and nitric acid [1] or liquid oxygen [2] inthe early 1990’s and the incorporation of solid oxidizersinto PS pores reported in 2002 led to interest in PS compo-site energetic materials [3]. Initial demonstrations involvedalcohol solutions facilitating transport of gadolinium nitrateand other similar oxidizers into the PS pores [3,4]. Thatwork was followed by experimental surveys of severalother oxidizer candidates by Cl�ment et al. [5] and du Ples-sis [6,7] , including elemental sulfur, which may be melt castdirectly without a solvent.

One consistent result from the earlier studies is thatsodium perchlorate (NaClO4) performed very well in termsof the qualitative measures used, including optical emissionand acoustic intensities. Our group [8–11] and others [12–15] have since carried out extensive studies of the PS/NaClO4 system, quantitatively validating its performanceobserving highly tunable flame speeds from 1 m s�1 [16] toover 3000 m s�1 [17,18], and heats of reaction of up to22 kJ g�1 of PS [19]. The high performance of NaClO4 islikely due in part to its high solubility in alcohols [14]; e.g. ,in methanol, where it exceeds 400 g L�1 solvent at 25 8C[20]. The high solubility increases oxidizer filling of thepores when the solvent evaporates [21]. Although the PS/NaClO4 energetic materials perform well, the high solubilityof NaClO4 in both alcohols and water also correlates withits high hygroscopicity, making it difficult to use in manypractical situations. Furthermore, perchlorates may presentenvironmental and health hazards due to the long-term

stability of the chlorate ion and its tendency to mimiciodide ions in biological processes [22,23] .

Therefore, in this paper we explore alternative oxidizerswith potential benefits of increased moisture stability and/or perchlorate-free composition. Despite a recent study re-porting interesting combustion of free standing PS films inair without the use of any additional oxidizers [24] , wefocus on combustion of substrate-integrated PS films withsolid oxidizers including sulfur and several nitrates. Asubset of these oxidizers has been qualitatively exploredpreviously, including sulfur, calcium nitrate, and gadoliniumnitrate [5–7]. This paper is the first report of quantitativemeasurements for flame speeds and calorimetric combus-tion enthalpies using these oxidizers as well as magnesiumand manganese nitrate with PS. Iodine containing oxidizerswere also considered. We used thermodynamic calculationsto establish a baseline for the combustion enthalpy mea-sured. Comparisons of experiments and thermodynamiccalculations suggested that combustion involved a signifi-cant amount of hydrogen, which was produced by the SiHx

[a] A. Abraham, N. W. Piekiel, C. J. MorrisSensors and Electron Devices DirectorateU.S. Army Research LaboratoryAdelphi, MD 20783, USA*e-mail : [email protected]

[b] A. Abraham, E. L. DreizinNew Jersey Institute of TechnologyNewark, NJ, USA

Abstract : We present a quantitative investigation of com-bustion of on-chip porous silicon (PS) energetic materialsusing oxidizers with improved moisture stability and/orminimized environmental impact compared to sodium per-chlorate (NaClO4). Material properties of the PS films werecharacterized using gas adsorption porosimetry and profil-ometry to determine specific surface area, porosity, andetch depth. PS energetic composites were formed usingmelt-penetrated or solution-deposited oxidizers into thepores. Combustion was characterized by high speed imag-

ing and bomb calorimetry. The flame speeds quantified forPS/sulfur and PS/nitrate systems varied in the ranges of2.9–3.7 m s�1and 3.1–21 m s�1, respectively. The experimen-tal combustion enthalpies are reported for different oxidiz-er systems in both inert and oxidizing environments. Forthe PS/sulfur and the PS/nitrate systems, the experimentalheats of combustion were comparable to those calculatedfor the thermodynamic equilibrium and taking into accountan increased reactivity of PS due to the hydrogen terminat-ed silicon surface.

Keywords: Porous silicon · On-chip energetic materials · Combustion · Sulfur · Nitrate

DOI: 10.1002/prep.201500108 � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &1&

These are not the final page numbers! ��

Full Paper

passivating layer on the surface of PS; an artefact of the hy-drofluoric acid etch [2,8,25] .

2 Materials and Experimental Methods

2.1 Sample Fabrication

The authors would like to note that certain steps withinthe sample fabrication process can be hazardous withoututilizing proper facilities, procedures, and equipment. Inparticular, the hydrofluoric acid (HF) used during the etchprocess must only be used under an appropriate fumehood designed for use with acids, and with the proper per-sonal protective equipment.

The starting material for PS film fabrication was h100i ori-ented, double-side polished, boron doped, p-type siliconwafers of 525 mm thickness and 100 mm diameter with re-sistivity of 1–10 W cm. The wafers were obtained fromRogue Valley Microdevices (Medford, OR) with a doublesided 6000 � silicon nitride (Si3N4) layer deposited usinglow stress, low pressure chemical vapor deposition(LPCVD). For etching, the silicon nitride layer was complete-ly removed from the backside. A sputter deposited 1700 �platinum layer acted as a cathode for the electrochemicaletch.

A detailed description for fabrication of the PS strips(27 mm long � 3 mm wide) shown in Figure 1a for flamepropagation measurements is presented elsewhere [18,19].The etch solution consisted of hydrofluoric acid (HF, 49 % inH2O), ethanol (EtOH), and hydrogen peroxide (H2O2, 30 % inH2O) added at 2.4 % of the HF/ethanol bath volume. Thedesired macro-structure of PS film was obtained by selec-tively removing the silicon nitride layer from the front sideof the wafer with a photolithographic process to exposebare silicon. The HF : EtOH ratio of the etch solution was3 : 1 for all PS strips. An etching time of 5 min was used forwafers with resistivities of 1–10 W cm. Samples for bombcalorimetry shown in Figure 1b required a larger mass andtherefore were different from those for flame speed analy-sis. These consisted of 1–10 W cm wafers diced into 17 �17 mm chips and etched in the HF : EtOH for 30 min. Forthese samples, a longer etch time was required due toa larger surface area ratio of front-side silicon to back-sideplatinum, which affects the etch current [26].

2.2 Porous Silicon Characterization

After the fabrication of PS films, material properties werecharacterized using gas adsorption porosimetry. The PSstrip was cleaved and tested using a Micromeritcs Tristar II3020 surface area analyzer to determine the surface area,pore size, and pore volume. The system uses Brunauer-Emmett-Teller (BET) theory [27,28] to determine surfacearea and Barrett-Joyner-Halenda (BJH) theory [27,29] to de-termine pore size and pore volume. The mass of the PSwas determined using a gravimetric technique, with massrecorded before and after the removal of PS via a reactionwith sodium hydroxide (NaOH). Due to the destructivenature of this measurement, additional samples were pre-pared for every batch of etch solution for material charac-terization experiments. The PS mass for these samplesvaried from 1.6–1.8 mg for strips for the flame speed studyand 9.1–18.3 mg for the chips for bomb calorimetry. Wehave previously showed [19] that the variation in materialproperties of samples from the same wafer is between 1 %and 3 %, and therefore the samples used for combustionwere assumed to have the same material properties assamples from the same wafer that were used for materialcharacterization.

Due to the variation in the mass from wafer to wafer, ni-trogen adsorption porosimetry was conducted for eachwafer used in this study to determine the PS material prop-erties. Additionally, the porosity was determined using thefollowing Equation (1):

P ¼ 100vv þ 1=1ð Þ ð1Þ

where P is the porosity (in %), v is specific volume of thepores found from porosimetry, and 1 is the bulk density ofsilicon (2.33 g cm�3 [27]). The etch depth of PS strips wasdetermined from profilometry using an Ambios Technology,Inc. XP-2 High Resolution Surface Profiler after the removalof PS with NaOH. The resolution of this device was 10 nm.

2.3 Pore Loading

The energetic composites were formulated by impregnat-ing the PS film with various oxidizers. Previous efforts fo-cused on characterizing PS with sodium perchlorate(NaClO4), where a nearly saturated 3.2 m methanol solutionof NaClO4 was drop cast on top of the porous structure[18,19,26,30] . The oxidizer was then allowed to seep intothe pores and dry, resulting in an energetic system. Usingthe NaClO4 system as a reference, the present study quanti-tatively investigates the reactivity of PS film with otherviable oxidizers. The list of oxidizers is presented in Table 1along with the pore loading technique used for each oxi-dizer. With various oxidizers, different pore loading tech-niques were necessary to achieve stable energetic compo-sites. With the drop cast pore loading technique, different

Figure 1. (a) PS strip (27 mm long � 3 mm wide) used for flamepropagation measurements; (b) square PS chip (17 mm � 17 mm)used for bomb calorimetry measurements.

&2& www.pep.wiley-vch.de � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/prep.201500108

�� These are not the final page numbers!

Full Paper A. Abraham, N. W. Piekiel, C. J. Morris, E. L. Dreizin

www.pep.wiley-vch.de

compatible solvents were used depending on the oxidizer.For sulfur, due to its low melting point, 119.6 8C [31], andstability in the liquid state, melting sulfur powder on top ofthe PS strip (Smelted) was used as a pore loading technique.After melting, sulfur on top of the PS was cooled to roomtemperature.

All PS/oxidizer materials after pore loading became sensi-tive to ignition stimuli. Although we did not quantify thesensitivities, it should be noted that several samples ignitedwhen being handled using non-grounded tools (i.e. metaltweezers). Electro-static sparks or friction likely caused igni-tion. PS based composites should be handled with extremecare after pore loading. Specifically, PS loaded with oxidiz-ers, NaClO4 and Mn(NO3)2, were observed to be most sensi-tive to handle. Samples with Mg(NO3)2, Ca(NO3)2 and sulfurwere less sensitive.

For bomb calorimetry samples, the oxidizer content wasmeasured for melted or drop cast cases. The averageamount of sulfur melted on top of the PS chip was 0.077�0.012 g. For nitrate-based oxidizers, 40–50 mL of solutionwas drop cast on top of the PS chip to fully cover the topsurface of the chip and dried inside a nitrogen flowing dry-box for at least 30 min.

Gravimetric measurements were conducted to estimatethe oxidizer mass which fills the pores from drop castingsulfur and nitrates as described elsewhere [19]. In an at-tempt to account for excess dry oxidizer on the surface ofeach chip, surface oxidizer was removed from the insensi-tive (non-porous) Si3N4 regions on the top of the chip bya cotton swab, leaving excess oxidizer above the sensitivePS region. The removed oxidizer was weighed, and the sur-face area that was covered by it was measured. Assumingthat the surface oxidizer above the active PS area, whichwas not wiped, was covered by the same mass per unitarea of excess oxidizer, the mass of excess dry oxidizerabove PS was estimated. By subtracting this estimate fromthe total amount of oxidizer left on the sample, the massof oxidizer in the pores was calculated.

2.4 Characterization of Porous Silicon Energetic Composite

After oxidizing the PS strips, the ignition and combustionevents were captured by a Photron FASTCAM SA5 highspeed camera at 100,000 frames per second inside an en-closed optically transparent box. All experiments were con-ducted in a nitrogen flowing dry-box with at least three re-peats, except one experiment performed in air. The highspeed video was processed using open source software,Tracker Video Analysis and Modeling Tool, to track theflame front during combustion. The average flame speedwas calculated for steady flame fronts using a linear fit ofthe distance traveled by the luminous flame front as a func-tion of time. The standard deviation was calculated for allexperiments with at least three or more flame speed meas-urements for all cases except for one. Flame speeds weresuccessfully measured for energetic composites involvingsulfur and nitrate-based oxidizers, but not for iodine-basedoxidizers. All iodine-containing oxidizers had low solubilityin the respective solvents, and the PS strips with iodine-based oxidizers did not self-propagate for the concentra-tions shown in Table 1. Because of their low reactivity,these materials were not characterized further.

Bomb calorimetry measurements were conducted witha Parr Semimicro Calorimeter (Parr 6725) and a Parr Calori-metric Thermometer (Parr 6772). The calibration of the cal-orimeter was completed using benzoic acid pellets. Theconstant volume specific heat capacity of the bomb calo-rimeter (Cv) was determined to be 2.1�0.14 kJ K�1. Usingthe PS chip (17 mm � 17 mm), the energetic composite wasprepared with both sulfur and nitrate-based oxidizers forthe bomb calorimetry measurement. The samples wereplaced inside the bomb cell pressurized to 2.026 � 105 Pawith either dry nitrogen or oxygen. An electrically heatedfuse wire was used to ignite the sample inside the sealedcell. The consumed fuse wire length during each measure-ment was determined and the energy contribution fromthe burned fuse wire was subtracted from the total heatmeasured. The average combustion enthalpy and standarddeviation after subtraction was calculated for all energetic

Table 1. Commercially available oxidizers used in this study.

Chemical Name Chemical formula Supplier Purity Pore loading technique

Sulfur, powder, �100 mesh S Sigma-Aldrich Reagent grade Melted on top of PS; Drop cast CS2 solution;1.9 M

Gadolinium nitrate hydrate Gd(NO3)3 · xH2O (x�6) Alfa Aesar 99.9 % Drop cast methanol solution; 0.8 M & 2.7 MMagnesium nitrate hexahydrate Mg(NO3)2 · 6H2O Alfa Aesar 98 % Drop cast ethanol solution; 0.8 MCalcium nitrate tetrahydrate Ca(NO3)2 · 4H2O Alfa Aesar 99 % Drop cast ethanol solution; 1 MManganese nitrate tetrahydrate Mn(NO3)2 · 4H2O Alfa Aesar 98 % Drop cast ethanol solution; 1 MPotassium periodate KIO4 Alfa Aesar 99 % Drop cast methanol solution; conc. <0.1 MSodium metaperiodate NaIO4 Alfa Aesar 98 % Drop cast methanol solution; conc. <0.2 MIodopentoxide I2O5 Sigma-Aldrich �98 % Drop cast methanol solution; conc. <0.2 M;

Drop cast acetone/distilled water mixturesolution; conc. <0.6 M

DOI: 10.1002/prep.201500108 www.pep.wiley-vch.de � 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &3&


Energetic Porous Silicon Composites


composites. The measurement uncertainty also includedthe error associated with the calibrated Cv value of the cal-orimeter. For specific nitrates (containing Mg, Ca, or Mn),multiple experiments were not conducted in nitrogen.Therefore, error bars shown in the results are based onlyon the calibration error. Samples without multiple experi-mental runs were difficult to reproduce for a variety of rea-sons. For Mn(NO3)2 · 4H2O samples, preparing and loadingthem inside the bomb cell was difficult due to their highsensitivity. Therefore, only one experiment was conductedin both nitrogen and oxygen. In the case ofMg(NO3)2 · 6H2O, and Ca(NO3)2 · 4H2O, a larger mass was re-quired to register a successful signal in nitrogen, and there-fore two 17 � 17 mm samples were placed in the calorime-ter. These samples were less sensitive; however, extremecare was taken to load the bomb cell with two samples.Additionally, for the Sdrop,cast sample, the fuse wire ignitedthe energetic composite but a signal was not registered inthe nitrogen environment even with the use of two PSchips. Although, sulfur containing samples were less sensi-tive than perchlorates and nitrates, they were still sensitiveto handle. We did not attempt an experiment with morethan two PS chips due to this reason for these samples.

3 Results and Discussion

3.1 Characteristics of PS

The gas adsorption porosimetry results for the PS strips forflame speed experiments are reported in Table 2. The poresize, pore volume, and surface area of these samples variedin the range of 27.6–31.8 �, 0.73–1.12 cm3 g�1, and 733–983 m2 g�1, respectively. The etch depth measured usingprofilometry (representing the maximum depth the oxidizercould penetrate) ranged from 24 to 26 mm. The porosity re-ported in Table 2 was calculated using Equation (1) for allsamples. PS chips for bomb calorimetry were also charac-terized using gas adsorption porosimetry, where pore size,pore volume, and surface area ranged from 32.4–34.0 �,0.81–1.08 cm3 g�1 and 647–911 m2 g�1, respectively.

In addition to porosity parameters, SiH2 present on thesurface of PS may have affected the combustion enthalpy

significantly. It is well known that the etching of siliconwith HF removes the native oxide layer and passivates thesurface with a hydrogen terminated layer in the form ofSiHx (x = 1,2,3) [32,33] , where it can be assumed that theaverage x = 2, as for SiH2. Surface functionalization of PSand the presence of SiH2 can affect combustion and, in par-ticular, results of the bomb calorimetry measurements.Therefore, we estimated the mole ratio of silicon hydriderelative to the bulk PS (bulk PS = porous silicon + hydrogenterminated surface), n(SiH2 :PS), using the following Equa-tion (2),

n SiH2 : PSð Þ ¼ SA 1atom

NA

� �MWSi

1 g of PS

� �ð2Þ

where SA is the specific BET surface area (647–911 m2 g�1 ofPS) for the PS chip (17 mm � 17 mm bomb calorimetry sam-ples), 1atom is the atom surface density (7.5 � 1018 atoms m�2

of PS), NA is the Avogadro constant, and MWSi is the molec-ular weight of silicon. n(SiH2:PS) was estimated to rangefrom 0.23 to 0.32, due to the variation in surface area be-tween wafers.

3.2 Pore Loading

The energetic composite can be further understood by de-termining the amount of oxidizer present inside the pores.To estimate this amount, a gravimetric technique presentedin Ref. [19] for NaClO4 was used for the drop cast poreloading of 1.8 m and 0.8–1 m concentrated sulfur and ni-trate-based oxidizers, respectively. The measured oxidizermass inside the pores ranged from 0.56 to 0.89 mg, de-pending on the drop cast oxidizer. The equivalence ratiosderived from this measured mass, Fm, are reported inTable 3, and show the energetic systems to be fuel-rich. Ad-ditionally for drop cast samples, equivalence ratios, Fc,were calculated for samples with 70 % porosity, assumingthe concentrated solution initially occupies the entireporous volume, before the evaporation of the solvent. Thiscalculation was expected to produce an upper bound forthe amount of oxidizer deposited in the pores. For NaClO4,the reported values of Fc and Fm are comparable to each

Table 2. Material properties of PS strips for flame speed characterization, etched for 5 min in 3 : 1 (HF : Et) etch solution; flame speed meas-urements using various oxidizers.

Pore loading technique/Experimental envi-ronment

Surface area[m2 g�1]

Pore volume[cm3 g�1]

Pore size[�]

Porosity[%]

Etch depth[mm]

Flame speed[m s�1]

Smelted/air 845 0.98 31.8 70 26 2.9Smelted/N2 808 0.73 27.6 63 24 3.0 �0.1Sdrop cast/N2 983 1.12 31.8 72 25 3.7 �0.12.7 M Gd(NO3)3 (drop cast)/N2 845 0.98 31.8 70 26 3.10.8 M Gd(NO3)3 (drop cast)/N2 845 0.98 31.8 70 26 10.1 �2.20.8 M Mg(NO3)2 (drop cast)/N2 733 0.76 29.9 64 25 4.4 �1.41 M Ca(NO3)2 (drop cast)/N2 777 0.76 28.9 64 25 9.8 �1.91 M Mn(NO3)2 (drop cast)/N2 777 0.76 28.9 64 25 21.0 �7.0





other [19] . However, in this work, Fc was 1.5–3 timeshigher than Fm, indicating that the measured amount ofoxidizer in the pores was larger than the calculated value.Although porosity values in Table 2 actually ranged from63–72 %, Fc was still higher than Fm over this range foreach sample. This apparently unrealistic result may be ex-plained by different morphology of oxidizers deposited ontop of the porous surface as compared to that of bulk Si.Therefore, it seems that the gravimetric technique using re-moval of oxidizer from insensitive surfaces of crystalline sili-con is not well suited for drop cast sulfur and nitrate basedoxidizers. A better estimate of mass inside the PS after poreloading remains to be the calculated Fc values presentedin Table 3, and in any case it appears that very fuel-rich en-ergetic composites were formed.

Since it was not possible to use the gravimetric tech-nique for the Smelted case, the equivalence ratio as a functionof porosity was calculated assuming that all the availablevolume inside the porous structure was occupied by sulfur,creating a fully dense system (Sfully dense). For example, a 70 %porous system filled with sulfur (density of 2.069 g cm�3

[31]) is expected to reach equivalence ratio of 1.1, near stoi-chiometric conditions (equivalence ratio of unity) as shownin Table 3.

3.3 Flame Propagation

3.3.1 PS/Sulfur

A built-in gold bridge wire [9,17–19] was used to ignite thesample in a nitrogen atmosphere and simultaneously trig-ger the image capture process. Average measured flamespeeds for different sets of samples are shown in Table 2.Samples with sulfur powder melted on top of the strip(Smelted) exhibited similar flame speeds as drop cast sulfur(Sdrop cast) samples at 3 m s�1. An additional flame speed ex-periment was repeated in air to ensure both the stabilityand reactivity of the Smelted system under ambient condi-tions. A comparable flame speed of 2.9 m s�1 was observedin air ; however, the flame appearance was very differentfrom that in nitrogen. Figure 2a and b show the combus-tion events of Smelted tested in nitrogen and air, respectively.

The flame in air flared much more aggressively and thecombustion products emitted a higher intensity light com-pared to the flame in nitrogen for both Smelted and Sdrop cast.We suspect a considerable secondary reaction in the pres-ence of air occurred in the gas phase, which caused the in-tensified flame characteristics observed in air. The combus-tion products from the initial Si/S reaction are expected tofurther react in air to form various silicon and sulfur oxides.It is surprising that both Smelted and Sdrop cast with apparentlydifferent pore loading and equivalence ratio as indicated inTable 3, showed comparable flame speeds that were muchlower than the PS/NaClO4 system [17–19]. It is also interest-ing that the violent gas phase reaction occurring in air didnot alter the flame speed. Note that the pattern of the gasphase flame suggests a relatively low speed of gas motionand substantial effect of natural convection. Thus, pro-duced gas species likely did not cause a local pressure in-crease or substantial convective heat transfer in the direc-tion of flame propagation. Possibly, the flame speed in thiscase was controlled by heat transfer through a compositeSi/S structure obtained in PS, although further work is de-sired to elucidate the flame propagation mechanism.

Table 3. Gravimetrically measured amount of oxidizer inside the pores along with measured and calculated equivalence ratios.

Pore loading assumption Pore loading technique Measured massinside the pores[mg]

Eq. ratio usingmeasured mass(Fm)

Calculated eq.ratio using as-sumption (Fc)

70 % porous sample fully dense with oxidizer Smelted – – 1.1(1) 70 % porous sample filled with oxidizer solution Sdrop,cast 0.89 3.8 4.8(2) Only residual oxidizer after evaporation of solventis expected to be loaded into the pores

0.8 M Gd(NO3)3 (drop cast) 0.67 10.1 15.6

0.8 M Mg(NO3)2 (drop cast) 0.67 8.5 24.61 M Ca(NO3)2 (drop cast) 0.70 7.5 20.51 M Mn(NO3)2 (drop cast) 0.56 10.1 19.9

Figure 2. Combustion event of (a) smelted tested in nitrogen, and(b) smelted tested in air.





3.3.2 PS/Nitrates

The average flame speeds for PS/nitrate-based energeticcomposites are reported in Table 2, ranging from 4.4–21.0 m s�1 for �1 m nitrate-based solution. In the case ofGd(NO3)3, a reduced flame speed was observed fora sample with a higher concentration obtained froma 2.7 m solution. We suspect that due to the higher viscosi-ty solution, the pore loading was reduced, causing a de-creased flame speed. When utilizing Gd(NO3)3 in solution,variation in oxidizer concentration produced visible differ-ences in solution viscosity. The 2.7 m solution createda thick layer of Gd(NO3)3 over the top of the PS film afterdrop casting and drying. This behavior was in contrast tothe energetic system involving sodium perchlorate [19] ,where a noticeable difference in viscosity was not seen fora nearly saturated 3.2 m solution of NaClO4 compared to0.8 m NaClO4 solution. Additionally, the pore loading in-creased four times when drop casting higher concentra-tions of NaClO4. We suspect that due to a higher viscositysolution in Gd(NO3)3, oxidizer penetration into the poreswas limited, and likely hindered observed flame speeds.

Figure 3 shows combustion event snapshots for all ni-trates used in this study. A self-propagating reaction wasobserved for these composites with flame speeds slightlygreater than for the sulfur containing energetic composites.Qualitatively, greater amounts of ejected gas phase speciesobserved in Figure 3 correlated with higher flame speeds inTable 2. This observation was supported by chemical equi-librium calculations performed with Cheetah 7.0 [34] underconstant pressure for PS/sulfur, PS/Ca(NO3)2, and PS/NaClO4.Other nitrates were not available in Cheetah 7.0. Resultspredicted that PS/NaClO4 produces 25 % more gaseousproducts than PS/Ca(NO3)2, and gas production from PS/Ca(NO3)2 was ca. 6 � that of PS/sulfur. Note that based onthe shape and structure of the luminous zones above thePS surface, it is apparent that the ejected gas speciesmoved faster than those observed in Figure 2b, for PS/sulfur combustion in air. Thus, it may be that the gas spe-cies released by decomposing nitrates accelerated convec-tive heat transfer and thus caused greater flame speeds.Because all flame speeds observed in these experimentsare much lower than PS/NaClO4 system [17–19], the effectof convective heat transfer for nitrates appears weaker than

for sodium perchlorate. This conclusion is consistent witha lower expected total gas release by the decomposing ni-trates. Similarly, conductive heat transfer may increase inthe PS/nitrates due to changes in the combustion tempera-tures compared to PS/sulfur. This could result in increasingflame speeds for PS/nitrates compared to PS/sulfur. There-fore, it is likely that both convective and conductive heattransfer mechanisms affect flame speeds for PS/nitrates.The dominating mechanism may be difficult to identify.

3.4 Bomb Calorimetry

3.4.1 PS/Sulfur

Experimental results shown in Figure 4 are compared tocalculations using NASA CEA thermodynamic equilibriumcode [35] for the PS/sulfur combustion in nitrogen. A con-stant volume configuration was selected in CEA, where thecalculations considered the experimental bomb cell volumeand initial pressure. Using these parameters, the amount ofnitrogen filling the bomb cell volume was calculated usingideal gas law. The initial amount of nitrogen which filledthe bomb cell was kept constant for all experiments andfor respective CEA calculations. Initially, a CEA calculationwas performed with reactants involving Si/S/N2 to comparewith the experimental results. Additionally, two other calcu-lations were performed using the reactants SiH2/Si/S/N2, ac-counting for possible functionalization of the PS surface asdiscussed in part 3.1. Because the heat of formation offunctionalized PS is not well defined in the literature, theonly available gas phase SiH2 species from the CEA librarywas selected for calculations. Although the use of gasphase species may result in overestimation of the heat ofcombustion, the calculations will nevertheless provide in-sight into how functionalization affects combustion. ForSiH2/Si/S/N2, the upper and lower range of n(SiH2:PS) ob-

Figure 3. Combustion event snapshots of PS/nitrate systemsduring flame speed measurements.

Figure 4. Combustion enthalpy for Si/S/N2 and SiH2/Si/S/N2 systemfrom CEA compared to bomb calorimetry experiment of Smelted inN2.





tained from Equation (2) were used to calculate the com-bustion enthalpy. The mole ratios of S:Si and/or S:(Si + SiH2)were varied in Figure 4 to account for the range betweenthe sulfur limited case, Sfully dense (mole ratio�1.40), and theexcess sulfur case, Smelted (mole ratio�3.65). The calcula-tions did not converge for the Si/S/N2 at the lowest S:Siratio, corresponding to its Sfully dense case. Although the com-bustion enthalpies for the considered compositions [Si/S/N2, and SiH2/Si/S/N2 with n(SiH2:PS) = 0.23 and n(SiH2 :PS) =0.32] show a maximum for each case, the variation in pre-dicted reaction enthalpy was relatively small. The maximumcalculated combustion enthalpies for the compositionswith Smelted were 1.2 kJ g�1 of PS for Si/S/N2, 3.7 kJ g�1 of PSfor SiH2/Si/S/N2 with n(SiH2 :PS) = 0.23 and 4.7 kJ g�1 of PSfor SiH2/Si/S/N2 with n(SiH2:PS) = 0.32. When comparingmajor product species predicted by CEA during combus-tion, silicon sulfide was the most dominant in all threecases. In the cases of SiH2/Si/S/N2, other major species suchas H2S and H2 were present. The calculations did not pre-dict significant amounts of silicon nitride species.

The experimental combustion enthalpy obtained frombomb calorimetry under nitrogen was 5.2�0.6 kJ g�1 of PS.The error bar is based on standard deviation of both theerror from multiple measurements as well as the error asso-ciated with the calibrated Cv value of the calorimeter. Notethat the reported experimental value is normalized by themeasured PS mass. The mass of the PS was determinedusing a gravimetric technique, with mass recorded beforeand after the removal of PS via a reaction with sodium hy-droxide (NaOH). Since this was a destructive process, it wasnot possible to test the mass of the PS chips used in theactual bomb calorimeter measurements. Instead, the mea-sured mass of one of the 17 mm � 17 mm PS chip fromeach wafer was assumed to be representative of everysample from that wafer, and was used to normalize thetotal heat release observed during the bomb calorimetrymeasurement conducted with samples from the samewafer. The error in defining the mass of PS used in theactual test was not quantified and not taken into account.Therefore, the error reported for the experimental values isunder-estimated. Additionally, the bulk silicon and platinumwere considered to be inert, while the total heat releasewas solely considered from the PS reaction. However, ifeven a small percentage of bulk silicon reacted with the ox-idizer, the combustion enthalpy per unit mass would de-crease because of the additional higher density fuel. Evenconsidering the above caveats, comparisons of CEA calcula-tions and the experimental values suggest that the likelypresence of SiH2 species discussed in part 3.1 led to a signif-icant increase in the measured reaction enthalpy comparedto that expected for PS with an unmodified surface.Indeed, only the combustion enthalpy calculated for the0.32SiH2/Si/S/N2 system with the highest possible value ofn(SiH2 :PS) was close to the experimental value.

3.4.2 PS/Nitrates

The combustion enthalpy of each energetic composite pre-pared using the nitrate-based oxidizer was tested insidethe bomb calorimeter under a nitrogen environment. Com-parisons of measurements with theoretical heats of com-bustion were of interest. The thermodynamic calculationsfor the PS/nitrates using NASA CEA code could not be con-ducted because the CEA library does not contain thesechemicals. However, the enthalpies of formation for anhy-drous form of nitrates and NaClO4 were obtained from theliterature [36] for all but Gd(NO3)3 to estimate the expectedheat of reaction based on the following reactions (R1) and(R2). In reaction (R1), M represents the anions (Mg, Ca orMn) and n represents the amount of hydrogen termination(23–32 % of SiH2).

2:5nSiH2 þ 2:5ð1�nÞSiþMðNO3Þ2 !2:5SiO2 þMOþ 2:5nH2 þ N2

ðR1Þ

nSiH2 þ ð1�nÞSiþ ð0:5þ 2:5nÞNaClO4 !SiO2 þ ð0:5þ 2:5nÞNaClþ nH2O

ðR2Þ

According to the pore loading analysis conducted in Sec-tion 3.1, the composition of the reactive energetic compo-site depends on how the oxidizer is loaded into the poresof the PS film. For example, in the case of nitrates andNaClO4, only about 20–30 wt % of the oxidizer drop cast onthe surface is expected to fill the pores, while the residualoxidizer sits on top of the chip [19]. This residual amount isnot usually expected to participate in the reaction due toa relatively long diffusion distance between the PS and re-sidual oxidizer. However, this may not be the case for thecombustion event taking place inside a pressurized con-stant volume cell in the bomb calorimeter, when the timeof reaction could be extended. To evaluate the amount ofoxidizer participating in the reaction with PS, the experi-mental results were compared with three calculations. Foreach calculation, a different oxidizer-limited assumptionshown below was used to calculate the expected energyrelease for each system, while also including the SiH2 termi-nated surface.

Oxidizer-limited assumptions:(i) All of the drop cast oxidizer participated in the reac-

tion with n(SiH2:PS) = 3.2(ii) All of the drop cast oxidizer participated in the reac-

tion with n(SiH2:PS) = 0(iii) Only the pore loaded oxidizer participated in the re-

action with n(SiH2:PS) = 3.2Results of these calculations are shown in Figure 5. The

error bars include experimental error as well as calibrationerror. Only the experimental value for Gd(NO3)3 is shownbecause the heat of formation value for Gd(NO3)3 could notbe obtained to calculate its combustion enthalpy. The mea-sured energy release best correlates with the calculated





heat of reaction for Ca(NO3)2, Mn(NO3)2, and NaClO4 whenconsidering an oxidizer limited case with all of the dropcast amount participating in reaction 1 and 2. Consistentwith the result from the PS/sulfur system, it appears thatthe PS surface functionalization with SiH2 had a substantialeffect on the combustion enthalpy. In the case of Mg(NO3)2,the experimental value falls in between the assumptions (ii)and (iii). This may be due to the fact that not all of the oxi-dizer reacted.

For all systems, the experimental enthalpies exceededtheir calculated values obtained assuming that only the oxi-dizer within the pores was involved in the reaction with PS.This was a surprising result and suggests that in bomb cal-orimetry experiments, excess oxidizer on the sample sur-face contributes to its reaction – a concept that was previ-ously neglected [19,37] . The exact amount of oxidizer par-ticipating in the reaction with PS is still poorly defined,however. A gravimetric technique described in Refs. [19,37]is not sufficiently versatile to be used for all oxidizers. Abetter technique must be developed in the future to quan-titatively correlate the amount of the pore loaded oxidizerwith the heat release observed.

An additional experiment in oxygen was conducted forall composites and the results are shown in Figure 6. Con-sistently for all energetic systems, higher heat of combus-tion values were observed in oxygen compared to nitro-gen, qualitatively confirming that fuel-rich PS energeticcomposites were, in fact, formed after loading the poreswith the oxidizer. The combustion enthalpies calculated as-suming that all of the drop cast oxidizer and excess oxygenare available to fully oxidize the PS correlated well with ex-perimental results. Consistently with previous calculations,possible functionalization of the PS surface with SiH2 affect-ed the combustion enthalpy substantially. Calculations forPS/sulfur (not shown in Figure 6) predicted nearly completeoxidation of both Si and S. Resulting reaction enthalpy cal-culated from CEA exceeded substantially that observed in

experiments, in which PS reacted with sulfur filled in pores,forming SiS2 shielded from ambient oxygen by the excesslayer of sulfur on the surface. The calculations were notperformed for the case of Gd(NO3)3, due to the lack of heatof formation value for Gd(NO3)3. A direct comparison be-tween Figure 5 and Figure 6 can be used to roughly assessthe amount of PS left unreacted in the experiments con-ducted in nitrogen. In the presence of oxygen, this unreact-ed PS appears to be fullly oxidized, so that the measuredenthalpies approach their respective theoretical limits.

4 Conclusions

Energetic composites using PS film with various oxidizerswere prepared and their performance was analyzed quanti-tatively. In particular, moisture stable energetic compositeswere prepared using PS film and sulfur. PS/sulfur systemswere similarly readily ignitable and combustible in bothinert and oxidizing gases. Similar flame speeds around3 m s�1 were observed for the composites prepared usingdifferent pore loading techniques, Smelted and Sdrop cast. Theflame speed was also unaffected by the presence of an ex-ternal oxidizing gas. The reaction heat was increased dueto functionalization of PS surface with SiH2, taking part incombustion. The PS/sulfur system is a potential alternativeto NaClO4 for applications requiring moisture stable andperchlorate-free energetic composites. The flame speed forPS/sulfur was much lower than for an earlier characterizedPS/NaClO4 system. A detailed study, outside the scope ofpresent work, is needed to understand how more rapidcombustion speeds might be achieved for PS/sulfur com-bustion.

Fuel-rich composites with minimal pore filling were alsoprepared with nitrate-based oxidizers. These energetic

Figure 5. Experimental combustion enthalpy in nitro-gen for 3 : 1etch solution ratio (HF : ethanol). The top of each bar color indi-cates the total heat calculated using each oxidizer-limited assump-tion.

Figure 6. Experimental combustion enthalpy in oxygen for 3 : 1etch solution ratio (HF : ethanol). The error bar includes experimen-tal error as well as calibration error.





composites were ignited in nitrogen gas. The flame speedsfor these oxidizers with PS were higher than for PS/sulfur,but still considerably lower than those for PS/NaClO4. Theseoxidizers may be viable for applications requiring lowerburn speeds and more controlled reactions. The combus-tion studies using the bomb calorimeter showed all sys-tems have higher combustion enthalpies in oxygen com-pared to nitrogen, confirming the starting composition tobe fuel-rich. More interestingly, the thermodynamic esti-mates correlated better with experimental reaction enthal-pies when the surface terminated hydrogen and excess oxi-dizer, in addition to the oxidizer loaded in the pores, wereaccounted for in the reaction. Determining the exactamount of the oxidizer reacting with PS remains a chal-lenge.

Symbols and Abbreviations

MWSi Molecular weight of silicon1 Silicon bulk density (2.33 g cm3)1atom Silicon atom surface density

(7.5 � 1018 atoms m�2 of PS)NA Avogadro constantn(SiH2 :PS) Ratio of terminated hydrogen toFc Calculated equivalence ratioFm Measured equivalence ratioP Porosity [%]PS Porous SiliconSA Specific surface areaSdrop cast Sulfur drop cast on top of PS chipSfully dense Pore volume occupied by sulfurSmelted Sulfur melted on top of the PS chipv Specific volume of pores

Acknowledgments

The authors would like to thank Cory R. Knick for assistance withbomb calorimetry experiments, and Brian Isaacson for assistance insample fabrication.

References

[1] P. McCord, S. L. Yau, A. J. Bard, Chemiluminescence of Ano-dized and Etched Silicon: Evidence for a Luminescent Silox-ene-Like Layer on Porous Silicon, Science 1992, 257, 68–69.

[2] D. Kovalev, V. Y. Timoshenko, N. K�nzner, E. Gross, F. Koch,Strong Explosive Interaction of Hydrogenated Porous Siliconwith Oxygen at Cryogenic Temperatures, Phys. Rev. Lett. 2001,87, 683011 – 683014.

[3] F. V. Mikulec, J. D. Kirtland, M. J. Sailor, Explosive Nanocrystal-line Porous Silicon and its Use in Atomic Emission Spectrosco-py, Adv. Mater. 2002, 14, 38 – 41.

[4] S. K. Lazarouk, A. V. Dolbik, P. V. Jaguiro, V. A. Labunov, V. E.Borisenko, Fast Exothermic Processes in Porous Silicon, Semi-conductors 2005, 39, 881 – 883.

[5] D. Cl�ment, J. Diener, E. Gross, N. K�nzner, V. Y. Timoshenko,D. Kovalev, Highly Explosive Nanosilicon-based Composite Ma-terials, Phys. Status Solidi A 2005, 202, 1357 – 1364.

[6] M. d. Plessis, Nanoporous Silicon Explosive Devices, Mater. Sci.Eng. B 2008, 147, 226 – 229.

[7] M. d. Plessis, Properties of Porous Silicon Nano-Explosive Devi-ces, Sens. Actuators A 2007, 135, 666 – 674.

[8] C. R. Becker, L. J. Currano, W. A. Churaman, C. R. Stoldt, Ther-mal Analysis of the Exothermic Reaction between GalvanicPorous Silicon and Sodium Perchlorate, ACS Appl. Mater. Inter-faces 2010, 2, 2998 – 3003.

[9] W. Churaman, L. Currano, C. Becker, Initiation and ReactionTuning of Nanoporous Energetic Silicon, J. Phys. Chem. Solids2010, 71, 69 – 74.

[10] L. J. Currano, W. A. Churaman, Energetic Nanoporous SiliconDevices, J. Microelectromech. Syst. 2009, 18, 799 – 807.

[11] W. Churaman, L. Currano, A. K. Singh, U. S. Rai, M. Dubey, P.Amirtharaj, P. C. Ray, Understanding the High Energetic Behav-ior of Nano-Energetic Porous Silicon, Chem. Phys. Lett. 2008,464, 198 – 201.

[12] V. S. Parimi, S. A. Tadigadapa, R. A. Yetter, Control of Nanoener-getics through Organized Microstructures, J. Micromech. Mi-croeng. 2012, 22.

[13] S. Wang, R. Shen, Y. Ye, Y. Hu, An Investigation into the Fabri-cation and Combustion Performance of Porous Silicon Nano-energetic Array Chips, Nanotechnology 2012, 23.

[14] E. C. Koch, D. Cl�ment, Special Materials in Pyrotechnics: VI.Silicon – An Old Fuel with New Perspectives, PropellantsExplos. Pyrotech. 2007, 32, 205–212.

[15] D. Cl�ment, D. Kovalev, Nanosilicon-Based Explosives, in: SiliconNanocrystals : Fundamentals, Synthesis and Applications (Eds. : L.Pavesi, R. Turan), Wiley-VCH, Weinheim, 2010, p. 537 – 554.

[16] A. Plummer, V. Kuznetsov, T. Joyner, J. Shapter, N. H. Voelcker,The Burning Rate of Energetic Films of Nanostructured PorousSilicon, Small 2011, 7, 3392 – 3398.

[17] N. W. Piekiel, C. J. Morris, L. J. Currano, D. M. Lunking, B. Isaac-son, W. A. Churaman, Enhancement of on-chip Combustionvia Nanoporous Silicon Microchannels, Combust. Flame. 2014,161, 1417 – 1424.

[18] C. R. Becker, S. Apperson, C. J. Morris, S. Gangopadhyay, L. J.Currano, W. A. Churaman, C. R. Stoldt, Galvanic Porous SiliconComposites for High-Velocity Nanoenergetics, Nano Lett.2011, 11, 803 – 807.

[19] N. W. Piekiel, C. J. Morris, W. A. Churaman, M. E. Cunningham,D. M. Lunking, L. J. Currano, Combustion and Material Charac-terization of Highly Tunable On-Chip Energetic Porous Silicon,Propellants Explos. Pyrotech. 2015, 40, 16 – 26.

[20] H. H. Willard, G. Frederick Smith, The Perchlorates of the Alkaliand Alkaline Earth Metals and Ammonium. Their Solubility inWater and other Solvents, J. Am. Chem. Soc. 1923, 45, 286–297.

[21] M. Du Plessis, A Decade of Porous Silicon as Nano-ExplosiveMaterial, Propellants Explos. Pyrotech. 2014, 39, 348 – 364.

[22] K. Sellers, K. Weeks, W. R. Alsop, S. R. Clough, M. Hoyt, B. Pugh,J. Robb, Perchlorate: Environmental Problems and Solutions,Taylor & Francis, London, 2006.

[23] J. A. Conkling, C. Mocella, Chemistry of Pyrotechnics: Basic Prin-ciples and Theory, Taylor & Francis, London, 1985.

[24] Y. Ohkura, J. M. Weisse, L. Cai, X. Zheng, Flash Ignition of Free-standing Porous Silicon Films: Effects of Film Thickness andPorosity, Nano Lett. 2013, 13, 5528–5533.

[25] A. Loni, L. T. Canham, Exothermic Phenomena and HazardousGas Release during Thermal Oxidation of Mesoporous SiliconPowders, J. Appl. Phys. 2013, 113.





[26] N. W. Piekiel, W. A. Churaman, C. J. Morris, L. J. Currano, Com-bustion and Material Characterization of Porous Silicon Nano-energetics, in: Proceedings of the IEEE International Conferenceon Micro Electro Mechanical Systems (MEMS), Taipeh, Taiwan,January 20–24, 2013, pp. 449–452.

[27] M. Ruike, M. Houzouji, A. Motohashi, N. Murase, A. Kinoshita,K. Kaneko, Pore Structure of Porous Silicon Formed on a Light-ly Doped Crystal Silicon, Langmuir 1996, 12, 4828 – 4831.

[28] S. Brunauer, P. H. Emmett, E. Teller, Adsorption of Gases in Mul-timolecular Layers, J. Am. Chem. Soc. 1938, 60, 309 – 319.

[29] E. P. Barrett, L. G. Joyner, P. P. Halenda, The Determination ofPore Volume and Area Distributions in Porous Substances. I.Computations from Nitrogen Isotherms, J. Am. Chem. Soc.1951, 73, 373–380.

[30] N. W. Piekiel, W. A. Churaman, D. M. Lunking, B. Isaacson, C. J.Morris, Characterization of Patterened Galvanic Porous Siliconfor On-Chip Combustion, in: 52nd AIAA Aerospace SciencesMeeting – AIAA Science and Technology Forum and Exposition,National Harbor, MD, USA, January 13–17, 2014.

[31] B. Meyer, Elemental Sulfur, Chem. Rev. 1976, 76, 367 – 388.[32] A. Grosman, C. Ortega, Chemical Composition of ‘Fresh’ Porous

Silicon, in: Properties of Porous Silicon (Eds. : L. Canham), Insti-

tution of Engineering and Technology, London, 1997,pp. 145 – 153.

[33] R. B. Heimann, M. B. Ives, N. S. McIntyre, The Effect of Deposi-tion Mechanism on the Composition of Surface Films on Sili-con, Thin Solid Films 1984, 112, 329 – 348.

[34] S. Bastea, L. E. Fried, K. R. Glaesemann, W. M. Howard, I.-F. W.Kuo, P. C. Souers, P. A. Vitello, Cheetah 7.0, Lawrence LivermoreNational Laboratory, Livermore, CA, USA, 2012.

[35] B. J. McBride, S. Gordon, Computer Program for Calculation ofComplex Chemical Equilibrium Compositions and Applications II.Users Manual and Program Description, NASA RP 1311, Nation-al Aeronautics and Space Administration Lewis ResearchCenter, Cleveland, OH, USA, 1996.

[36] CRC Handbook of Chemistry and Physics, 92nd ed. (Ed.: W. M.Haynes), CRC Press, Boca Raton, 2011.

[37] V. S. Parimi, A. Bermffldez Lozda, S. A. Tadigadapa, R. A. Yetter,Reactive Wave Propagation in Energetic Porous Silicon Com-posites, Combust. Flame. 2014, 161, 2991 – 2999.

Received: May 1, 2015Revised: August 21, 2015

Published online: && &&, 0000





FULL PAPERS

A. Abraham, N. W. Piekiel, C. J. Morris,*E. L. Dreizin

&& –&&

Combustion of Energetic PorousSilicon Composites ContainingDifferent Oxidizers





combustion of energetic porous silicon composites containing different oxidizers

Documents