Characterization of Î±- and Î²-RDX Polymorphs in Crystalline Deposits

Subscriber access provided by the Henry Madden Library | California State University, Fresno

Article

Characterization of #- and #-RDX Polymorphs in Crystalline Deposits on Stainless Steel Substrates Amanda M Figueroa-Navedo, José L Ruiz-Caballero, Leonardo C Pacheco-Londoño, and Samuel P. Hernández-Rivera Cryst. Growth Des., Just Accepted Manuscript • DOI: 10.1021/acs.cgd.6b00078 • Publication Date (Web): 09 Jun 2016 Downloaded from http://pubs.acs.org on June 11, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Crystal Growth & Design is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

1

Characterization of α- and β-RDX Polymorphs in

2

Crystalline Deposits on Stainless Steel Substrates

3

Amanda M. Figueroa-Navedo, José L. Ruiz-Caballero, Leonardo C. Pacheco-Londoño and

4

Samuel P. Hernández-Rivera*

5

ALERT-II DHS Center of Excellence for Explosives Research

6

Department of Chemistry, University of Puerto Rico-Mayagüez, Mayagüez, PR, 00681 USA

7 8

ABSTRACT: The highly energetic material (HEM) hexahydro-1,3,5-trinitro-s-triazine, also

9

known as RDX, has two stable conformational polymorphs at room temperature: α-RDX

10

(molecular conformation of –NO2 groups: axial-axial-equatorial) and β-RDX (molecular

11

conformation of –NO2 groups: axial-axial-axial). Both polymorphs can be formed by deposition

12

on stainless steel substrates using spin coating methodology. α-RDX is the most stable crystal

13

form at room temperature and ambient pressure. However, β-RDX, which has been reported to

14

be difficult to obtain in bulk form at room temperature, was readily formed. Reflection-

15

absorption infrared spectroscopy measurements for RDX-coated stainless steel substrates

16

provided spectral markers that were used to distinguish between the conformational polymorphs

17

on large surface areas of the substrates. Raman microspectroscopy was employed to examine

18

small areas where the intensity was proportional to the height of the structures of RDX. Spectral

19

features were interpreted and classified by using principal component analysis (PCA). The

20

results from these spectral analyses provided good correlation with the values reported in the

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 28

21

literature. Conditions to generate predominantly β-RDX crystalline films as function of the spin

22

coating rotational speed on these substrates were obtained. PCA was also applied to predict

23

percentages of polymorphs present in experimental samples. Applications of the results obtained

24

suggest the modification of existing vibrational spectroscopy based spectral libraries for defense

25

and security applications. Understanding the effects of polymorphism in HEMs will result in the

26

attainment of higher confidence limits in the detection and identification of explosives,

27

especially at trace or near trace levels.

28 29

1. INTRODUCTION

30

The highly energetic material (HEM) studied in this work was the cyclic nitramine hexahydro-

31

1,3,5-trinitro-s-triazine, which is commonly known as cyclonite or Research Department

32

Explosive (RDX). This HEM is classified as a secondary explosive, and it is widely used in

33

military applications as one of the main ingredients of plastic bonded explosives (PBX: Semtex,

34

C2, C3, and C4), and other explosives formulations. RDX has been widely studied in its room

35

temperature stable crystalline α-polymorphic form (α-RDX). However, several studies discuss

36

the characterization of the β-metastable form of RDX (β-RDX) under standard growth

37

conditions, which involve drop-casting techniques at room temperature and atmospheric

38

pressure.1-5 The formation of β-RDX follows Ostwald’s step rule, where the least stable

39

polymorph is formed first due to a very small change in Gibbs free energy between both

40

crystalline forms.6 This rule states that kinetics takes a prominent role over thermodynamics and

41

that the kinetically favored species is formed first.7,8 Concentrations below the solubility

42

parameters increases the probability of obtaining the metastable form. Therefore, it has proven

43

difficult to obtain only β-RDX without simultaneous formation of α-RDX on the substrate of


2

Page 3 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


44

interest, because there is very little change in the rates of formation of these polymorphs.1

45

Nevertheless, the β-RDX polymorph was obtained with drop-cast crystallization studies by

46

Goldberg and Swift, who demonstrated consistent growth of β-RDX on glass substrates using

47

various solvents.1 Mercado et al. studied the characterization of RDX nanoparticles deposited by

48

an aerosol jet technique, where they also found the coexistence of room temperature phases of

49

RDX (β and α).9 Spectroscopic measurements of the polymorphic forms of RDX have been used

50

to follow its transition from α-RDX using temperature,10 concentration,11 mechanical

51

contact,2,3,5,12-14 and specific solvent crystallization.1,14 Polymorphism within energetic materials

52

(both primary and secondary) is a main cause of concern for trace detection and quantification

53

due to the change in physical properties inherent to polymorphism.

54

Various studies on the polymorphism of RDX show that there are five phases: α, β, γ, δ, and

55

ε.2-5,13-18 Of these five forms, only α-RDX and β-RDX exist at room temperature and ambient

56

pressure. In the α-RDX crystalline form, the orientations involving two of the –NO2 groups are

57

in axial positions and the third group is in an equatorial position (AAE), imparting CS symmetry

58

for the molecular conformer when in its static equilibrium position. However, in the β-RDX

59

polymorph, the orientations of the three –NO2 groups are all axial (AAA), leading to a higher

60

C3V symmetry for its equilibrium conformation. Regarding crystal morphology, Hultgren first

61

described α-RDX crystals as orthorhombic.19 In contrast, McCrone described the β-polymorph as

62

needle-like.15 Using IR vibrational spectroscopy, Karpowicz demonstrated that RDX assumes an

63

(AAA) conformation in both liquid solution and its vapor phase, but that the room temperature

64

bulk solid phase assumed the (AAE) conformation in crystalline form.16 More recent studies by

65

Torres et al. described the β-polymorph as structures resembling islands as well as scattered


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 28

66

particles.11 They described the α-RDX polymorph as well defined crystals. In contrast, Goldberg

67

and Swift described both α/β-RDX polymorphs as assuming a variety of different morphologies.1

68

To investigate these apparent contradictions, it is important to use basic aspects of nucleation

69

theories. An ideal thin film growth on a surface involves smooth, layered growth throughout the

70

surface. This two-dimensional (2D) growth is called Frank-van der Merwe mode. The growth of

71

3D islands throughout the surface is called Volmer-Weber mode. There is a third type of growth,

72

called Stranski-Krastanov mode, that implies a wetting monolayer below the 3D structures.20

73

One possible explanation for the formation of Volmer-Weber mode structures originates from

74

mass transfer limitations, where a major challenge in shearing solution is the prevention of

75

crystal defects, which often plays a major role in void formation and dendritic growth.7,8 The

76

formation of islands with defined or void centers provides indication of abrupt growth.21

77

Evidence of possible Stranski-Krastanov nucleation has been reported by atomic force

78

microscopy (AFM) measurements for dihexylterthiophene (DH3T), which exhibited island-like

79

structures.22 However, this evidence to support this type of growth mechanism has been highly

80

debated.21-23 AFM measurements are required to characterize the morphological aspects of the

81

films prepared where a general idea of the polymorphs within the substrate is known. However,

82

AFM measurements can provide evidence of Stranski-Krastanov growth if a monolayer or thin

83

film is present.17 Furthermore, Raman spectra should confirm polymorphism in small areas of

84

the structures generated. In contrast, Fourier transform infrared (FT-IR) spectroscopy should

85

allow assessment of larger sample areas. Therefore, RDX conformational polymorphs within

86

these crystal structures can be discovered by FT-IR measurements and confirmed by Raman

87

spectra. In this work, the area studied by FT-IR was > 3,000 times larger than the area studied by

88

Raman microspectroscopy. Thermal Raman microspectroscopy was used to study drug


4

Page 5 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


89

polymorphism and is commonly used in the pharmaceutical industry.24,25 An important aspect of

90

this study involved the methodology developed for samples preparation since these were needed

91

to address the required morphological AFM evidence for the presence of thin films. Various

92

thermal inkjet methodologies have provided major advances in the control of morphology and

93

polymorphism within thin films.20,26

94

In this study, a methodology for growing thin crystalline films by spin coating27 was

95

developed, where the α/β-RDX polymorph distribution was studied as a function of angular

96

velocity (ω) of deposition coating at room temperature and atmospheric pressure. The

97

explanation for this behavior lies in the theories of spin coating, as well as the theory of creating

98

a metastable polymorph. Spin coating is governed by the magnitude of shear stress (τ) applied to

99

the deposited solution at increasing ω values.27 Most of the solution is instantly ejected during

100

this process, where the air flow on top of the substrate applies the τ, which increases by

101

increasing ω according to the following equation.27

102

, = / = [ ] / π / /

π

/ /

(1)

103

where τ is represented in planar polar coordinates (r, z), υ is the viscosity of air, f is the rotational

104

speed or frequency (revolutions/min, rpm), and µ represents the dry airflow as controlled by the

105

spin-coater. The aforementioned equation states that shearing stress depends on viscosity and

106

surface tension of the solvent, as well as the rotational speed applied. This equation agrees with

107

Stranski-Krastanov growth, since it relies on the formation of a “solid skin” on the surface due to

108

an increased solvent evaporation process with the spin coating method. In addition, Capes and

109

Cameron discussed the parameters to form a metastable polymorph in lieu of the room stable

110

form of paracetamol: seeding of a saturated solution, obtaining the metastable form from the

111

melting process, and isolating and drying the crystals removing as much remaining solvent as


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 28

112

possible.28 Therefore, a qualitative hypothesis for this method would explain that the particle

113

size and height of the film decreases while increasing f, which can also imply that lower sample

114

concentration would be present. In addition, identification of remnant solvent traces would

115

support the Stranski-Krastanov theory.

116

Principal component analysis (PCA) is one of the most frequently used multivariate analysis

117

techniques for chemometrics in a broad range of industrial and research applications.29 PCA is

118

predominantly used for evaluating sources of variation within a data set. These variations may be

119

easily distinguished in the original data or they may not be as obvious to the analyst. The main

120

purpose of applying PCA in the analysis of FT-IR spectra is to classify the changes in the

121

vibrational signatures of the compounds of interest in an organized manner, via a graphical

122

representation called scores plot or principal components (PCs) plot.29 The changes obtained

123

could be due to shifts in maximum wavenumber location, a decrease in relative intensity of the

124

bands, or the emergence or disappearance of signals, among others. The first PC contains the

125

highest variance and the following PCs decrease in percentage of the variance captured. This

126

variance can be examined using an orthogonal projection of the original data in the form of PCs,

127

when these are plotted as scores.30 A simple example to determine if the PCs do not express

128

correlation is the capture of spectral noise or any other type of interference.31 Consequently,

129

these factors can be visualized in the scores plot or PC plot (PC1 vs. PC2, for example), where a

130

noisy PC spectrum may indicate the presence of interferences from ambient conditions (carbon

131

dioxide, water vapor, or contaminants) in FT-IR spectroscopic data. In this case, preprocessing

132

routines can be applied to the data prior to performing the multivariate analyses. If little or no

133

spectral noise were present, preprocessing options would be unnecessary, resulting in a faster for

134

the spectral analysis in obtaining good results and the corresponding scores plots. These plots are


6

Page 7 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


135

important because they allow the analyst to extract additional information from these data and

136

formulate tools to explain why certain properties extracted as classes from the original spectra

137

are alike or different. When dealing with vibrational spectra, the PCs can be compared to a

138

“ground truth” by adding a reference spectrum of the sample to the original spectra used in the

139

PCA regression to determine where the variations originate.

140 141

2. EXPERIMENTAL

142

2.1 Materials. The solvents used in this work: acetone (CH3COCH3, 98% w/w), methanol

143

(HPLC grade), and isopropyl alcohol (IPA, 99% w/w), were purchased from Sigma-Aldrich

144

Chemical Co. (Milwaukee, WI, USA) and used without further purification. Water used was

145

doubly distilled and deionized (18.0 M ohms) to render it equivalent to HPLC grade water. RDX

146

was synthesized according to the Bachmann method with minor modifications.31 Stainless steel

147

(SS) substrates (1.0 x 1.0 in2) were purchased from Stainless Supply, Inc. (Monroe, NC, USA).

148

2.2 Substrate Preparation. The substrates were cleaned by sonication in acetone

149

followed by rinsing with distilled water and IPA. Once the substrates were cleaned, vibrational

150

chemical maps using Raman microspectroscopy (InVia, 532 nm; Renishaw, Inc., Hoffman

151

Estates, IL, USA) were obtained to determine the degree of cleanliness of the substrates’

152

surfaces. The chemical maps of the substrates showed that these evidenced: (a) no presence of

153

RDX; and (b) no solvent remaining after the drying process. The RDX coatings on hydrophobic

154

SS substrates (RDX/SS) were prepared from concentrated stock solutions of approximately 0.2

155

M RDX in acetone (working solution), according to the methodology discussed by Hikal where

156

homogeneity on quartz substrates was established.32 This solvent choice did not necessarily

157

guaranteed that homogeneity would persist on the hydrophobic SS substrates. Various previously


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 28

158

reported RDX deposition methods have used metallic substrates for samples and standards

159

preparation as well as for fundamental studies.33-35 Dilute solutions of 49 ppm (mg L-1) were

160

prepared from the working solution to evaluate the reported findings on the influence of analyte

161

concentration on the occurrence of polymorphism in RDX.7

162

2.3 Crystal Growth by Spin Coating. A model 6800 spin coater system (Specialty

163

Coating Systems, Inc., Indianapolis, IN, USA) was used to develop the methodology for RDX

164

deposition on SS substrates. Depositions on the substrates were made at 20 °C and for f values

165

from 3 k to 8 k rpm at 1 k rpm increments. The ramp increase time, dwell time, and ramp

166

decrease time were set at 2, 60, and 5 s, respectively. The interior of the spin coater chamber was

167

purged from dust particles by filtered dry air. Six substrates and replicas were prepared by

168

depositing 100 µL of the working RDX solution on various SS substrates by spin coating using

169

as f was gradually increased.

170

2.4 Instrumentation. FT-IR spectra were recorded on an evacuable benchtop

171

interferometer (IFS-66 v/s, Bruker Optics, Billerica, MA, USA) equipped with a DTGS detector

172

and a KBr beamsplitter with the aid of a reflection accessory (A-513/Q, Bruker Optics) in which

173

the angle of incidence could be varied from: 13° to 85°. FT-IR spectral data were acquired and

174

analyzed using OPUS™ software suite (v. 7.2.139.1294, Bruker Optics). The optimum incidence

175

angle chosen for RAIRS depends on the type of metal where the analyte is adsorbed.36 The

176

grazing angle is the value of the incidence angle where the intensity of reflected light increases

177

the most with respect to a phase change (δ) when using polarized light parallel and perpendicular

178

to the substrate.36 All measurements were made at an angle of incidence of 80°, which although

179

it is not the optimum incidence angle for reflectance-absorption infrared spectroscopy (RAIS)

180

measurements as seen in SI-1, it was a much more convenient and reproducible value. The


8

Page 9 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


181

reference spectra which included the β-RDX spectra used for the last section of this study were

182

also carried out at 80° incidence.35 The interferograms were collected at a 4 cm-1 nominal

183

spectral resolution. Sample areas of 1.0 x 1.0 in2 were analyzed in RAIS mode in the spectral

184

range 400–1600 cm-1 with 32 background and sample scans and a scanner velocity set to 4.0

185

kHz.

186

Raman spectra of the various crystal structures formed on the SS surface were obtained using

187

a Raman microspectrometer (RM2000; Renishaw, Inc.) equipped with a Leica DM/LM

188

microscope (100× objective) in the 100–3200 cm-1 region. The spectral parameters included 10 s

189

scans with 8 accumulations per measurement. A 632 nm laser (Melles-Griot, IDEX Corp.,

190

Carlsbad, CA, USA) was employed as the excitation source with a power of 0.23 mW reaching

191

the sample. The laser beam diameter was measured at 5.2 µm with the 100× objective. Mapping

192

measurements were made in an model Xplora Raman microspectrometer (Horiba Jobin-Yvon

193

SAS, Edison, NJ, USA) using a CCD detector, 100× objective and a 638 nm solid state diode as

194

the laser excitation source with 1.10 mW of laser power at the sample. These spectral parameters

195

included 10 s scans with 3 accumulations and 9 µm of linear scan area with steps of 2 µm.

196

An AFM (model CP-II; Veeco Instruments, Inc., Plainview, NJ, USA) was employed to

197

characterize the morphology and topology of the coatings prepared. The cantilever was used in

198

non-contact mode with a natural frequency of 75.5 kHz. The imaging software, Image

199

Processing and Data Analysis Software, v. 2.1.15 (Veeco Instruments, Inc.) provided with this

200

equipment was employed for roughness calculations.

201 202

3. RESULTS AND DISCUSSION

203

3.1 FT-IR RAIS Spectra. Representative RAIS spectra of thin RDX films on highly

204

polished SS substrates prepared by spin coating at increasing f values are shown in Figure 1. The


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 28

205

main differences observed between the two RDX phases present were the spectral markers

206

related to the position of one of the –NO2 groups. The prominent feature corresponding to the

207

asymmetric –NO2 stretch in β-RDX is the broad peak that appears in the 1592-1600 cm-1 region.

208

However, for α-RDX, this prominent peak unfolds into additional peaks, 1535 cm-1 (tentatively

209

assigned to νas NO2eq asymmetric stretch in α-RDX), 1581 and 1582 cm-1 (tentatively assigned as

210

νas NO2ax asymmetric stretch in α-RDX; 1600 cm-1 in β-RDX). These vibrational modes

211

assignments are based on the values reported by Millar, et al.2, Dreger and Gupta4, and Infante-

212

Castillo, et al.10,37 Other prominent vibrational modes at 1279 cm-1 (tentatively assigned as νN-

213

N-O2 stretch in α-RDX; 1270 cm-1 β-RDX), 1390 cm-1 (tentatively assigned as γ-CH2 out of

214

plane bending in α-RDX; 1377 cm-1 in β-RDX), 1535 cm-1 (tentatively assigned as νas NO2eq

215

asymmetric stretch in α-RDX) are also present. These spectra were compared to a reference

216

standard obtained from a deposition of the 0.2 M RDX solution that formed only the α-

217

polymorph (labeled “Alpha” in Figure 1). Further evidence of the coexistence of α/β-RDX

218

phases was investigated by Raman microspectroscopy. However, the differences in FT-IR

219

spectra were evaluated by PCA.

220 221

Figure 1. RAIS spectra of thin RDX films on SS substrates as a function of f show the

222

appearance of the –NO2 equatorial peak at higher rotational frequencies (marked by red arrows).

223


10

Page 11 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


224

3.2 PCA of RAIS Spectra. The RAIS spectral region of 1500–1650 cm-1 was used for

225

PCA regression analysis, since this region is highly important in the spectroscopic differentiation

226

between the two polymorphs. Figure 2 displays the scores plot for baseline corrected RAIS

227

spectra. First derivative preprocessing was applied to the spectral data to eliminate variations

228

within the spectra for shifts in the baseline. The figure shows the spectral variation of the RDX

229

deposited at each f, where a general increase in f is correlated with PC2 in the multivariate

230

analysis. PC2 also reflects a general clustering at low (3 k-5 k rpm) and high (6 k-8 k rpm) f

231

values. This is indicative of a major change detected by FT-IR that differentiates these two broad

232

classification parameters.

233 234

Figure 2. PCA scores plot for RAIS spectra of RDX deposited on SS substrates showing a

235

marked difference between low f values (3 k, 4 k, and 5 k) and high f values (6 k, 7 k, and 8 k).

236


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 28

237

The spectra obtained for depositions at 7 k rpm and 8 k rpm point to the direction of

238

coexistence of α/β-RDX. This was correlated with the PCA loadings plot using vibrational

239

modes at 1039 cm-1 (tentatively assigned to νN-C-N), 1390 cm-1 (tentatively assigned as γ-CH2

240

out of plane bending in α-RDX), 1437 cm-1 (βCH2 in-plane bending), and 1535 cm-1 (νas NO2eq

241

asymmetric stretch in α-RDX). The spectra reveal similar peaks at 1270 cm-1, with no

242

broadening of the peak (νN-N-O2 stretch). For 6 k rpm, intense bands were located at 1326 cm-1

243

(νN-N-O2 stretch) and 1446 cm-1 (tentatively assigned to βCH2 in-plane bending) compared to

244

other f values.6 PC1 correlates with the increasing intensity of the axial –NO2 vibrational mode,

245

which is characteristic of the presence of β-RDX.

246 247

Figure 3. Comparison of RAIS spectra of α/β-RDX polymorphs with PC1 and PC2 of the

248

loadings plot for the PCA model in the asymmetric stretch spectral region: 1500 to 1650 cm-1.

249


12

Page 13 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


250

3.3 PCA of Raman Spectra. Raman microspectroscopy is often described as a

251

complementary tool to RAIRS mode FT-IR spectroscopy because it is able to focus on particles

252

and structures on the micron scale with higher spatial resolution. This technique was employed

253

to examine the polymorphism of RDX within the various crystal structures detected. Raman

254

mapping measurements were taken in steps of 2 nm to examine and confirm the presence of each

255

polymorph on the substrates. Fiber-like structures, as well as islands and crystals (prisms), were

256

examined for Raman activity under ×100 magnification. The region studied was 100–3200 cm-1.

257

A wide variation in intensities is noted in each of the colored coded spectra shown in Figure 4.

258

These intensities were directly dependent on the topology of the structure examined. The

259

variations are illustrated in Figures SI-2 and SI-3 of the Supporting Information section. Spectral

260

data were preprocessed with baseline correction to avoid affecting the PCs. Figure 4 shows three

261

regions that can be used to distinguish the polymorphs: Region A: the ring breathing modes at

262

800–900 cm-1: the most prominent spectral feature in this region corresponds to the ring-

263

breathing mode for the β-polymorph was observed at 881 cm−1. The corresponding mode for the

264

α-polymorph was located near at 866 cm−1.10,11,37 Region B: asymmetric –N-O2 stretch region at

265

1600 cm-1, and Region C: CH symmetric stretch region at 2800–3050 cm-1.6 Notably, the

266

spectrum at 7 k rpm falls within the region of the α-RDX polymorph in Figure 4. In addition,

267

another spectrum of the sample prepared at 7 k rpm has to be assigned as belonging to the β

268

conformer of RDX.


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 28

269 270

Figure 4. Raman spectra of RDX crystals deposited on SS substrates at various f values.

271 272

Examples of RDX polymorph identification by Raman spectra are shown in Figure 5. One

273

representative spectra per rpm is shown in normalized intensity values. Alpha and beta RDX

274

spectra were found at 7 k rpm and representative spectra of both polymorphs formed at 7 k are

275

shown. The α-RDX form can be recognized for the samples prepared at f values of 3 k (red), 6 k

276

(violet), and 7 k alpha (orange). The β-RDX form was present at depositions made at f values of

277

4 k (navy blue), 5 k (green), 7 k beta (lighter blue), and 8 k (sky blue). The most noticeable

278

region for differentiating between the two polymorphs is the C-H stretch region from 2850 to

279

3100 cm-1, where α-RDX deposits show three peaks at 2953, 3009, and 3081 cm-1, while β-RDX

280

shows two at 2996 and 3084 cm-1. The ring-breathing mode located at 881 cm-1 for the β-RDX

281

polymorph (886 cm-1 for the α-RDX polymorph) agree precisely with literature value.1,9-11,37

282

Furthermore, the –N-O asymmetric stretching mode shows a broad peak at 1600 cm-1 for the β-

283

form, indicating the presence of slightly non-totally equivalent axial nitro group substituents.1, 9-

284

11,37

These vibrational modes have been extensively documented in the literature, and there is


14

Page 15 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


285

good agreement between our results and those of Infante-Castillo et al. and others.1-3,9-11,18,37

286

Figure 5 provides evidence of the presence of the solvent used for depositing the RDX samples:

287

acetone. Evidence of the presence of acetone in α-RDX spectra in Figure 5, was also based on

288

the –CH vibrational marker at 2926 cm-1.38 Vibrational markers from the solvent (acetone) were

289

found. These can be appreciated in Figure 5 after normalizing intensity values in the secondary

290

“y” axis. Notice that an increased presence of solvent at 2927 cm-1 is evidence of a solvent-

291

mediated transformation from β-RDX to α-RDX.

292 293

Figure 5. Raman spectra (638 nm) for RDX deposits examined in the CH-stretch region showing

294

two peaks for β-RDX (4 k and 6 k)) and three peaks for the α-RDX polymorph (3 k and 7 k).

295

Acetone Raman shift signature is clearly shown at 2927 cm-1.

296 297

Raman and IR spectra were then added to the same matrix for analysis by direct

298

standardization in the region from 1500 to 1600 cm-1. To account for intensity variations within

299

the spectra, centroid scaling, standard normal variate, and first derivative were applied as

300

preprocessing steps to the modified matrix. The scores plot obtained after model modification is


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 28

301

shown in Figure 6, where the IR and Raman spectra are correlated for f values (3 k and 4 k) and

302

for high f values (5 k, 6 k, 7 k, and 8 k) with respect to PC1 for RDX. The β-RDX reference

303

spectra at a concentration of 49 ppm was added for comparison of the –NO2 vibrational mode in

304

RAIS. When compared to the reference IR spectra for α/β-RDX, low f values favors the β-RDX

305

polymorph.

306 307

Figure 6. Correlation between IR and Raman spectra. PC1 favors grouping β-RDX at low f

308

values.

309 310

3.4 AFM. The purpose of characterizing the RDX crystal structures formed on SS substrates

311

by AFM was to study the surface topology and morphology for both polymorphs. The Raman

312

spectra shown in Figures 4 and 5 show marked intensity fluctuations, before normalization. High

313

intensities were obtained for the prism structures near the center of the substrates at 5 k rpm. The


16

Page 17 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


314

next most intense Raman spectra also represented prism structures, which were near the center

315

and the corner of the substrate at 8 k rpm. As seen in Figure 7, the height of these crystal

316

structures was as high as 550 nm following deposition at 8 k rpm on a 10x10 µm2 region.

317 318

Figure 7. Surface structures formed at 8 k rpm: (a) topological AFM imaging; (b) height profile

319

along white line on the surface; and (c) white light micrograph of RDX/SS with both prisms and

320

fibers.

321 322

Fiber like structures in the shape of dendrites were found for the RDX deposits at 3 k, 4 k,

323

and 5 k rpm (Supporting Information). The RDX structures with lower intensity found on the

324

substrates were fibers, which were found on the RDX/SS samples formed at 3 k rpm (Supporting

325

Information). The characterization of the fibers deposited at 3 k rpm is shown in Figure 8, where

326

the white line along the length of the dendrite (longitudinally across one of the fibers) shows the


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 28

327

surface roughness. Surface roughness was calculated using the filter kernel for a Gaussian filter,

328

where the average roughness (Ra) and root-mean-square roughness (Rq) were calculated by:

329

330

=

& =

∑ % ∑ % , ! − #$

∑ ∑ ( ) ' * * , ! #$

(2)

(3)

331

where +,,-! represents surface topography, +./0 is the average surface roughness and the 1, 2

332

values denote the pixels in directions x and y. The maximum default pixel values for the

333

software were 34 = 35 = 256.

334

Interestingly, the observations for alternating Raman intensities corresponded to the height

335

and roughness of the structures analyzed. The prismatic structures provided stronger Raman

336

intensities than the fiber structures. The average roughness for the fibers was 16.6 nm in the

337

longitudinal direction. A transversal line was also traced along the surface of the crystal

338

structures. The positions of the fibers are evident from the height profiles shown in Figure 8,

339

where the average surface roughness was 33.8 nm.

340

One possible explanation for the presence of polymorphs after the spin coating process at a

341

given f value may be the mechanical influence of nucleation during the coating process. This has

342

been discussed in detail for the pharmaceutical applications of crystal growth and polymorphism

343

induced by pressure and temperature.24,25 In addition, solvent choice and stress by mechanical

344

conversion are also known to induce polymorphism.1,2,11,24 Recent studies have classified the

345

crystalline structure of α-RDX as orthorhombic.39 However, this study established that the β-

346

polymorph predominantly forms on the smooth SS surface at low f values as well as specific

347

solubility parameters at room temperature and ambient pressure using an optimized spin coating


18

Page 19 of 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


348

methodology. With the information gathered from the scores for PCA using FT-IR, the

349

percentage of α-RDX and β-RDX was calculated using the following equation:

350

% α−RDX =

351

where the term “nk” represents the score value for the spectra selected for evaluation at each f

352

value, “a” represents the average of the scores for α-RDX spectra, and “b” the average of the

353

scores for β-RDX spectra.38 The results are listed in Table 1, where shear stress (ττ) is expressed

354

as a ratio with respect to the f value of 3 k rpm. A clear tendency to increase the percentage of α-

355

RDX polymorph present on the substrate as f increased was found. This trend may be influenced

356

by the solvent choice (acetone). There is evidence of coexistence and rate of crystallization in

357

drop casting techniques due to solvent-mediation.1,7,8,20

358

Table 1. Spectral tests for coexistence of α- and β-RDX polymorphs with increase in f value

;

Characterization of Î±- and Î²-RDX Polymorphs in Crystalline Deposits

Recommend Documents