利用低振動頻率顯微拉曼光譜儀即時監控結晶態1,1′-binaphthyl在同質異像體間變化動力學與拉曼成像分析

國 立 交 通 大 學

應用化學系碩士班

碩 士 論 文

利用低振動頻率顯微拉曼光譜儀即時監控結晶態

1,1

′-binaphthyl 在同質異像體間變化動力學與拉曼成像分析

In-situ Tracing of Transformation Dynamics and

Molecular Imaging of Crystalline 1,1′-Binaphthyl

by Low-Frequency Raman Microspectroscopy

研 究 生：張君輔

指導教授：重藤真介 博士

利用低振動頻率顯微拉曼光譜儀即時監控結晶態

1,1

′-binaphthyl 在同質異像體間變化動力學與拉曼成像分析

 

    研 究 生：張君輔                     Student：Chun-Fu Chang  

        指導教授：重藤真介 博士             Advisor：Dr. Shinsuke Shigeto  

 

 

國 立 交 通 大 學

應用化學系碩士班

碩 士 論 文

A Thesis

Submitted to M. S. Program, Department of Applied Chemistry College of Science

National Chiao Tung University in Partial Fulfillment of the Requirements

for the Degree of Master

in

M. S. Program, Department of Applied Chemistry July 2013

Hsinchu, Taiwan, Republic of China

 

利用低振動頻率顯微拉曼光譜儀即時監控結晶態

1,1

′-binaphthyl 在同質異像體間變化動力學與拉曼成像分析

學生: 張君輔 指導教授: 重藤真介 博士

國立交通大學應用化學系碩士班

摘要

除了 X-光繞射分析以外，是否有其他研究晶體結構的方法？低頻率拉曼光譜提供豐 富的晶體結構資訊以及分子位向。實驗室自行架設的低振動頻率拉曼光譜儀能觀測低至 9.8 cm-1_{的拉曼訊號。此外，本系統亦能在顯微鏡下進行空間解析量測並即時分析晶體結} 構。在此，作者研究不同晶形的 1,1’-binaphthyl 以展示這套新穎的分析方法。此聯芳化 合物在結晶態時因為兩個萘環間有不同的二面角而呈現兩種不同的晶形：類順式和類反 式（同質異像體）。當類順式被緩慢加熱時，會透過昇華和再結晶的機制逐漸轉變成類 反式。本文中，將即時監控此轉變過程並使用顯微拉曼成像法分析晶形轉變後類順式、 類反式晶體的分佈情形。 類順式和類反式在指紋區的拉曼光譜非常類似，唯一的區別指標是 510/532 cm-1_波 峰的強度比。然而，它們的光譜在低頻區截然不同。例如類順式最強的訊號在 100cm-1_； 類反式最強的訊號在 26 cm-1_{。是以此二低頻率特徵峰被用來區別類順式和類反式混合晶} 體。利用此二低頻率特徵峰建立的拉曼影像呈現了高度的選擇性與影像對比度。此外， 低頻率拉曼影像更在外表看似完美的單晶中發現了分子的不規則排列。諸此結果讓我們 意識到低頻率拉曼成像法，將會成為十分有用的晶體結構分析工具。

In-situ Tracing of Transformation Dynamics and

Molecular Imaging of Crystalline 1,1′-Binaphthyl

by Low-Frequency Raman Microspectroscopy

Student: Chun-Fu Chang Advisor: Dr. Shinsuke Shigeto

M. S. Program, Department of Applied Chemistry

National Chiao Tung University

Abstract (in English)

How can we understand crystal structure besides employing X-ray diffraction analysis? Low-frequency (LF) Raman spectra provide a wealth of precious information on crystal structure and orientation and can potentially be a complementary approach to the conventional X-ray diffraction method. However, observation of LF Raman spectra is very challenging due to severe interference of gigantic Rayleigh scattering. In this study, the author uses a laboratory-built LF Raman spectrometer that has the ability to observe Raman signals down to 9.8 cm-1_{. Because this apparatus is combined with a confocal microscope, it has}

become feasible to perform space-resolved measurements of crystals under the microscope for

in-situ crystal structure analysis. Here, the author demonstrates this novel approach by

studying crystal polymorphs of 1,1′-binaphthyl. This biaryl compound occurs as two crystal polymorphs with different dihedral angles between the two naphthyl groups: namely, the cisoid (C) and transoid (T) forms. As the C form is heated up, it gradually transforms to the T form via sublimation and subsequent recrystallization. In this thesis, the author presents tracing of this transformation dynamics and LF Raman imaging that visualizes the spatial distribution of the C and T forms after the transformation takes place.

The only effective way to distinguish between the C and T forms by high-frequency Raman bands would be to calculate the intensity ratio for a pair of marker bands (510/532 cm-1). However, the two forms show much more distinct spectral patterns in the LF region. For example, the C form shows the strongest band at 100 cm-1, whereas the T form shows the most intense band at 26 cm-1. Thus, these two LF bands were chosen as markers to selectively probe different crystal polymorphs that coexist in proximity. In sharp contrast with the intensity-ratio image, the LF Raman images at 100 and 26 cm-1 show markedly high contrast and clearly discriminate one crystal form from the other. Moreover, we observed inhomogeneous distribution patterns in the LF Raman images, suggesting different crystal

orientations within an apparent single crystal. Our results show that LF Raman spectroscopy and imaging should be a facile yet powerful addition to the toolbox for crystal structure analysis.

Acknowledgements

First, I want to thank Professor Shigeto for providing me a great environment and equipment to study this interesting topic. I also feel grateful for Professor Hamaguchi, who inspired me that “spectrum is a letter from molecule” and directed me how to read the letter. From Shimada san and Okajima san, I learned many knowledge on fundamental of spectroscopy, alignment of optics, data processing as well as programming. Without you, it is impossible to finish this project, I appreciate what you have taught me. Also, I feel thankful all Shigeto lab members, helping me work out the problems. Finally, I would like to give my sincerely appreciation to my parents who support me both economically and mentally.

Tables of Contents

Page

Abstract (in Chinese)...i

Abstract (in English)...ii

Acknowledgements...iv

Tables of Contents...v

List of Figures and Tables...vii

Chapter I. General Introduction...1

I-1. Motivation of This Study... 2

I-2. Raman Measurements of Low-Frequency Motions...3

I-3. Detection of Molecular Orientation...5

I-4. Organization of This Thesis...5

References for Chapter I...7

Chapter II. Laboratory-Built Fast Low-Frequency Raman Spectrometer...9

II-1. General Introduction...10

II-2. Fast Low-Frequency Raman Spectrometer Using a Single-Mode Ar+_{-ion Laser} (514.5 nm)...14

II-2-1. Apparatus...14

II-2-2. Low-Frequency Raman Measurement of L-Cystine...15

II-3. Fast Low-Frequency Raman Spectrometer Using a Single-Mode Solid-State Laser (532 nm)...17

II-3-1. Apparatus...18

II-3-2. Comparison of the Filter Transmittance and Examination of Low-Frequency Measurement Capability Using L-Cystine...20

II-3-3. Evaluation of the Spatial Resolution of the Confocal Raman Microspectrometer...22

II-3-4. Development of the Low-frequency Raman Mapping Program...24

Chapter III. In-situ Tracing of the Transformation Dynamics between Crystal Polymorphs of

1,1′-Binaphthyls...27

III-1. Introduction to 1,1′-binaphthyls...28

III-2. Preparation of the Cisoid and Transoid Form...32

III-3. Raman Spectra of the Two Polymorphs of 1,1′-binaphthyl...35

III-4. In-situ Tracing the Transformation between Polymorphs...39

III-5. Fitting Analysis...49

III-6. Discussion: Changes in Low-frequency Raman Bands Change During Transformation...55

III-7. Discussion: Which Low-Frequency Bands Is Attributed to the Torsional Motion?...59

III-8. Summary...61

References for Chapter III...62

Chapter IV. Low-Frequency Raman Imaging of 1,1′-binaphthyl Crystals...65

IV-1. Purpose of this Study...66

IV-2. Sample Preparation...67

IV-3. Low-Frequency Raman Imaging of the Cisoid Form Crystal...69

IV-3-1. Structure of the Cisoid Form Crystal...71

IV-4. Low-Frequency Raman Imaging of the Transoid Form Crystal Adjacent to Cisoid Form Crystal...74

IV-4-1. Domain Structures in the Transoid Crystal...78

IV-5. Polarized Low-Frequency Raman Imaging of the Transoid Form Crystal Adjacent to Cisoid Form Crystal...79

IV-5-1. Definition of Four Polarization Configurations...79

IV-5-2. Singular Value Decomposition and Band Fitting...80

IV-5-3. Polarized Raman Imaging of the Selected Low-Frequency Bands...83

IV-6. Summary...85

References for Chapter IV...87

List of Figures and Tables

Page

Figure I-1. Origin of Stokes and anti-Stokes Raman scattering...4  

Figure II-1. Elimination of Rayleigh scattering by the iodine vapor filter...10

Figure II-2. Calculated absorption lines of I2 in the visible frequency region accessible by (a) Ar+_{-ion (514.5 nm), (b) Nd:YAG (532.0 nm) lasers...11}

Figure II-3. Elimination of Rayleigh scattering by the BragGrateTM _{notch filter (BNF)...12}

Figure II-4. Transmittance spectra of the iodine vapor filter and combination of two BragGrateTM notch filters (BNFs)...13

Figure II-5. Schematic diagram of the low-frequency Raman spectrometer with 514.5 nm excitation...14

Figure II-6. Raman spectra of L-cystine...16

Figure II-7. Schematic diagram of the low-frequency Raman spectrometer (532 nm excitation) with 90° scattering setup...18

Figure II-8. Schematic diagram of the low-frequency confocal Raman spectrometer...19

Figure II-9. Transmittance spectra of different filters...21

Figure II-10. Observed Raman spectra of L-cystine taken with (a) three BNFs and (b) a combination of one BNF and the iodine vapor filter...22

Figure II-11. Evaluation of lateral (a) and axial (b) resolution of the low-frequency confocal Raman microspectrometer...23

Figure II-12. Schematic layout of the Raman imaging system...24

Figure II-13. (a) Developed user interface of the mapping program in LabVIEW 2012 (b) Stage movement during mapping...25

Figure III-1. Axial chirality of 1,1′-binaphthyl...28

Figure III-2.Schematic diagram of the two crystal forms of 1,1′-binaphthyl...30

Figure III-3. Column chromatography purification of 1,1′-binaphthyl...32

Figure III-4. Preparation of the cisoid and transoid forms...32

Figure III-5. Differential scanning calorimetry (DSC) measurements of cisoid form crystal with different heating rates...34

Figure III-6. Differential scanning calorimetry (DSC) measurements of transoid form crystal with different heating rates...34

Figure III-7. Raman spectrum of the cisoid powder at room temperature...37

Figure III-8. Raman spectrum of the transoid powder at room temperature...37

Figure III-9. Low-frequency Raman spectra of the two crystalline 1,1′-binaphthyl...38

Figure III-10. Laboratory-built heating cell...39

Figure III-11. Low-frequency Raman spectra at different temperatures recorded by heating the cisoid powder...41

Figure III-12. Raman spectra in the fingerprint region at different temperatures recorded by heating the cisoid powder...42

Figure III-13. Low-frequency Raman spectra at different temperatures recorded by heating the transoid powder...44

Figure III-14. Raman spectra of the fingerprint region at different temperatures recorded by heating the transoid powder...45

Figure III-15. The crystal structure of (a) the cisoid form and (b) the transoid form...46

Figure III-16. (a) Observed low-frequency Raman spectra of melted 1,1′-binaphthyl and (b) the corresponding reduced Raman spectra at different temperatures...48

Figure III-17. Experimental data in the low-frequency region and their fitted results...52

Figure III-18. Experimental data in the fingerprint region and their fitted results...54

Figure III-19. Peak position shift with respect to temperature of (a) the cisoid related bands and (b) the transoid related bands...56

Figure III-20. Low-frequency band intensity changes upon heating of (a) the cisoid related Bands, (b) the transoid related bands and (c) intramolecular vibration band at 300 cm-1...58

Figure III-21. (a) Reduced spectrum of the melted 1,1′-binaphthyl. (b) Raman spectrum of the cisoid form. (c) Raman spectrum of the transoid form. (d) Raman spectrum of 1,1′-binaphthyl dissolved in CCl4 (0.6 M) at room temperature...60

Figure IV-1. (a) Preparation of the cisoid form single crystal. (b) Bright-field optical image of a typical cisoid form single crystal obtained with the gas diffusion method...68

Figure IV-2. Bright-field optical image of a composite of the cisoid and transoid microcrystals...69

Figure IV-3. (a) Optical image of the cisoid form single crystal. (b) The fitting curve (blue line) can well reproduce the observed Raman spectrum...70

Figure IV-4. Low-frequency Raman imaging of the cisoid form single crystal and the optical

image (orange picture)...71

Figure IV-5. Averaged low-frequency Raman spectra of the cisoid crystal around the center (red

curve) and the left corner (blue curve) compared with the cisoid powder spectrum (green curve)...73

Figure IV-6. Raman imaging with the intensity ratio of the 100/110 cm-1 bands...74

Figure IV-7. Proposed heterogeneous crystal structure of the cisoid form crystal...74 Figure IV-8. (a) Optical image of cisoid/transoid crystal mixture.

(b)(c) Representative Raman spectra selected from the points labeled in panel (a) (red curves) and the fitting curves (blue curves)...76

Figure IV-9. Optical image (a) and Raman imaging of the cisoid/transoid crystal mixture

constructed of low-frequency Raman bands (b)–(f), intensity ratio (g)...78  

Figure IV-10. Comparison of the intensity variation of the 26 and 530 cm-1 bands along the

direction indicated by an orange arrow...78

Figure IV-11. (a) Raman imaging of 26/20 cm-1 intensity ratio.

(b) Low-frequency Raman spectra of the selected point...79

Figure IV-12. Schematic layout of the polarization configurations employed in the

low-frequency confocal Raman microspectrometer...81

Figure IV-13. Singular value of the first 100 spectral components (red dots). The dots in the

shaded region were retained to reproduce the matrix...82

Figure IV-14. Low-frequency Raman spectra of the transoid crystal before and after SVD...83 Figure IV-15. Representative fitting results (blue curves) of the cisoid (yellow curve), transoid

(red curve) and cisoid/transoid interface (green curves) spectra...84

Figure IV-16. Low-frequency Raman imaging of the cisoid/transoid crystal mixture performed

with four polarization configurations...85

Table I-1. Intensity ratios of anti-Stokes and Stokes Raman bands...4 Table III-1. Fitting parameters...51

Chapter I.

I-1. Motivation of This Study

There are many beautiful crystals exist in Mother Nature. Some are perfect single crystals, and some are polycrystals. How do molecules align in those crystals? How do the molecular interactions change during phase transitions such as melting? Investigating these interesting fundamental physical properties is enormously helpful for both materials and biological sciences. Although X-ray diffraction analysis can precisely determine the crystal structure, long data acquisition time makes real-time monitoring of phase transitions impossible. In this study, the author aims at tracing dynamic changes in intermolecular interactions with a short acquisition time (<1 s) as well as visualizing molecular orientations within organic crystals at the molecular level.

I-2. Raman Measurements of Low-Frequency Motions

Intermolecular forces of molecules in the condensed phase are reflected in intermolecular or collective motions. The collective motions of molecular ensembles move very slowly and are therefore usually observed in the low-frequency region (< 200 cm-1) of an optical spectrum. Measurements of Raman scattering in the low-frequency region have been utilized to discuss the optically active lattice vibration of crystals based on the important theory of the first-order lattice vibration Raman effect [1-3]. Likewise, the phonon modes of organic crystals, which are related to librational lattice vibrations, are observed below 150 cm-1 [4-6]. Recently, low-frequency

(SWNT) from the size-dependent ring breathing modes (RBM) in the range of 100–500 cm-1 [7-9].

In the past decades, low-frequency Raman spectra were measured with a scanning Raman spectrometer, which is composed of two or three monochromators and a single-channel detector such as photomultiplier tube (PMT) [10]. This kind of setup takes a long time to slowly rotate the grating in order to acquire a Raman spectrum and it can record only one side of anti-Stokes or Stokes Raman signals. Nowadays, thanks to the improvement of laser technology and development of ultra-narrow band rejection filters as well as multi-channel detector, low-frequency Raman measurements can be realized with a short exposure time (< 0.1 s) and a wide spectral coverage [11]. Thus, it is now possible to monitor dynamic behaviors of vibrational motions associated with phase transitions in the condensed phase by low-frequency Raman spectroscopy [12].

The great advantage of the ability to measure both Stokes and anti-Stokes Raman spectra simultaneously is that the intensity ratio of the Stokes and anti-Stokes Raman bands of a given mode, IAS/IS, can be used to determine “molecular temperature” as follows [13]:

𝐼(anti-Stokes)

𝐼(Stokes)

=

(𝜈

_!

   −    𝜈

_!

)

!

(𝜈

_!

   +    𝜈

_!

)

!

𝑒

!!!!! !!!

where 𝜈_! is the wavenumber of the laser line, 𝜈_! is the vibrational frequency of a band of the sample, h is Planck’s constant, c is the speed of light, 𝑘! is the Boltzmann constant, and T is

absolute temperature of the sample (or more specifically molecules). This equation stems from the fact that Stokes Raman scattering comes from the excitation of molecules in the vibrational ground state (v = 0 → 1), while anti-Stokes Raman scattering comes from the vibrational excited

state (v = 1 → 0), as shown in Figure I-1. Therefore, IAS/IS is proportional to the population

difference between vibrational excited and ground states by assuming Boltzmann distribution. It is worth noting that around room temperature (~300 K), the IAS/IS can be barely detected for

Raman bands above 800 cm-1 due to the weak anti-Stokes Raman signals (Table I-1). However, the IAS/IS has a large enough value in the low-frequency region (< 200 cm-1) for temperature

estimation using Equation I-1, which makes Stokes/anti-Stokes ratio work as a “molecular thermometer”. Raman Shift (cm-1₎ Temperature (K) 100 300 500 700 900 1100 1300 1500 200 0.056 0.383 0.562 0.663 0.726 0.769 0.801 0.825 400 0.003 0.147 0.316 0.439 0.528 0.593 0.642 0.681 800 0.000 0.022 0.100 0.193 0.278 0.351 0.413 0.464 1600 0.000 0.000 0.010 0.037 0.077 0.123 0.170 0.216 3200 0.000 0.000 0.000 0.001 0.006 0.015 0.029 0.046

Table I-1. Intensity ratios of anti-Stokes and Stokes Raman bands. [14]

 

I-3. Importance of Detecting of Molecular Orientation

Molecular orientation represents how constituent molecules align three-dimensionally. It is a very important physical property that affects the packing of molecules in crysyals, mechanical properties of polymers [15, 16], device performance of organic light-emitting diodes (OLEDs) [17], membrance bio-physics [18, 19], etc. X-ray diffraction has been commonly used to precisely determine the molecular alignment in a single crystal. However, X-ray diffraction is not a panacea for crystal orientation when the target crystal has domain structures or twinning as well as when high-quality singles crystals are unavailable. Some fluorescent dyes help to visualize molecular orientaitons in biological samples, but they are applicable only to certain functional groups [18]. Coherent anti-Stokes Raman scattering (CARS) [20] and polarized Raman measurements [21] have been employed for studying molecular orientation as well. Unfortunately, intramolecular vibrations are not informative enough toward molecular ensemble structures. Thus, there has been an increasing demand for an alternative method for detecting molecular orientation and the method we present in this thesis, low-frequency Raman spectroscopy, is one of the promising candidates.

I-4. Organization of This Thesis

The rest of this thesis is organized as follow. In Chapter II, the author introduces a laboratory-built low-frequency Raman microspectromester, including its working principles and

to in-situ study of the transformation of crystal polymorphs of 1,1′-binaphthyl. By discussing the low-frequency Raman bands together with high-frequency Raman bands, the information about crystal structures is obtained. In Chapter IV, the author demonstrates low-frequency Raman microspectroscopy as a novel approach to crystal structure analysis by using 1,1′-binaphthyl crystals. Constructing Raman images of low-frquency bands can selectively visualize the crystal polymorphs. Moreover, detection of molecular orientation distributions within the crystal by calculating intensity ratio of low-frequency Raman bands will be discussed as well.

References for Chapter I.

[1] M. Born and K. Huang, Dynamical theory of crystal lattices: Oxford : Clarendon Press, 1954.

[2] R. Loudon, "The Raman effect in crystals," Advances in Physics, vol. 50, pp. 813-864, 2001.

[3] S. P. S. Porto, P. A. Fleury, and T. C. Damen, "Raman Spectra of TiO2, MgF2, ZnF2,

FeF2, and MnF2," Physical Review, vol. 154, pp. 522-526, 1967.

[4] H. Takeuchi, S. Suzuku, A. J. Dianoux, and G. Allen, "Low frequency vibrations in crystalline biphenyl: Model Calculations and Raman and Neutron Spectra," Chemical

Physics, vol. 55, pp. 153-162, 1981.

[5] K. Krebs, S. Sandroni, and G. Zerbi, "Low-Frequency Vibrations of Crystalline Biphenyl," The Journal of Chemical Phycis, vol. 40, pp. 3502-3506, 1964.

[6] M. Suzuki, T. Yokoyama and M. Ito, "Polarized Raman Spectra of Naphthalene and Anthracene Single Crystals," Spectrochimica Acta Part A: Molecular Spectroscopy, vol. 24, pp. 1091-1107, 1968.

[7] A. M. Rao, E. Richter, S. Bandow, B. Chase, P. C. Eklund, K. A. Williams, S. Fang, K. R. Subbaswamy, M. Menon, A. Thess, R. E. Smalley, G. Dresselhaus, M. S. Dresselhaus, " Diameter-Selective Raman Scattering from Vibrational Modes in Carbon Nanotubes,"

Science, vol. 275, pp. 187-191, 1997.

[8] Zhonghua Yu and L. E. Brus, "(n, m) Structural Assignments and Chirality Dpendence in Single-Wall Carbon Nanotube Raman Scattering," Journal of Physical Chemistry B, vol. 105, pp. 6831-6837, 2001.

[9] M. S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, "Raman Spectroscopy of Carbon

Nanotubes," Physics Reports, vol. 409, pp. 47-99, 2005.

[10] J. Loader, Basic laser Raman spectroscopy. London: Sadtler Research Laboratories, 1970.

[11] H. Okajima and H. Hamaguchi, "Fast Low Frequency (Down to 10 cm-1) Multichannel Raman Spectroscopy Using an Iodine Vapor Filter," Applied Spectroscopy, vol. 63, pp. 958-960, 2009.

[12] H. Okajima1 and H. O. Hamaguchi, "Unusually Long trans/gauche Conformational Equilibration Time during the Melting Process of BmimCl, a Prototype Ionic Liquid,"

[13] J. R. Ferraro, K. Nakamoto, and C. W. Brown, Introductory Raman Spectroscopy. New York: Oxford Academic Press, 2003.

[14] 濱口宏夫．平川暁子, ラマン分光法, 学会出版センター, 1988.

[15] W. Retting, "The Effect of Molecular Orientationon the Mechanical Properties of

Rubber-Modified Polystyrene," Pure & Applied Chemistry, vol.50, pp. 1725-1762, 1978. [16] S. A. Arvidson, K. C. Wong, R. E. Gorga and S. A. Khan, "Structure, Molecular

Orientation, and Resultant Mechanical Properties in Core/Sheath Poly(lactic acid)/Polypropylene Composites," Polymer, vol. 53, pp. 719-800, 2012.

[17] D. Yokoyama, "Molecular Orientation in Small-Molecule Organic Light-Emitting Diodes," Journal of Materials Chemistry, vol. 21, pp. 19187-19202, 2011.

[18] K. Godula, M. L. Umbell, D. Rabuka, Z. Botyanszki,C. R. Bertozzi and R. Parthasarathy, "Control of the Molecular Orientation of Membrane-Anchored Biomimetic

Glycopolymers," Journal of the American Chemical Society, vol. 131,pp. 10263-10268, 2009.

[19] K. Godula, D. Rabuka, K. T. Nam and C. R. Bertozzi, "Synthesis and Microcontact Printing of Dual End-Functionalized Mucin-like Glycopolymers for Microarray

Applications," Angewandte Chemie International Edition, vol. 48, pp. 4973-4976, 2009. [20] M. Zimmerley, R. Younger, T. Valenton, D. C. Oertel, J. L. Ward and E. O. Potma,

"Molecular Orientation in Dry and Hydrated Cellulose Fibers: A Coherent Anti-Stokes Raman Scattering Microscopy Study," Journal of Physical Chemistry B, vol. 114, pp. 10200-10208, 2010.,

[21] L. M. Bellan and H. G. Craighead, "Molecular Orientation in Individual Electrospun Nanofibers Measured via Polarized Raman Spectroscopy," Polymer, vol. 49, pp. 3125-3129, 2008.

Chapter II.

Laboratory-Built

II-1. General Introduction

To observe low-frequency Raman signals with the conventional Raman spectrometer is not an easy task because the excitation laser output is often very broad and low-frequency Raman signals are buried under immense Rayleigh scattering. In addition, commercial notch filters usually have a broad band-rejection width, which eliminates not only Rayleigh scattering but low-frequency Raman signals. A fast multichannel low-frequency Raman spectrometer using an iodine (I2) vapor filter was first developed by Okajima and Hamaguchi at Tokyo [1] and

reconstructed by us at National Chiao Tung University [2]. The principle of this apparatus is to use a single-mode laser which has a very narrow bandwidth of output wavelength (≤0.001 cm-1) coupled with an iodine vapor filter [3] as an ultra-narrow notch filter. Thus, it does not block the low-frequency Raman signals. Iodine vapor has a numerous number of absorption bands due to electronic transitions [4] and their bandwidths are as narrow as 0.03 cm-1. If one of these narrow and intense absorption bands of iodine can efficiently absorb light happens to absorb light exactly at the Raman excitation wavelength work, then the absorption band can function as a high-performance notch filter that removes elastic Rayleigh scattering with minimum blocking of inelastic Raman scattered light (see Figure II-1). Figure II-2 shows the two absorption lines of iodine vapor that are suitable for 514.5 and 532 nm excitation lasers.

Figure II-1. Gaseous iodine molecules can efficiently absorb unwanted Rayleigh scattering and

only the wavelength-shifted Raman signals pass through it.

Stokes ! Raman Scattering anti-Stokes Raman Scattering Rayleigh ! Scattering Stokes ! Raman Scattering anti-Stokes Raman Scattering Iodine Vapor Filter!

★   ★  

(a)

(b)

Figure II-2. Calculated absorption lines of I2 in the visible frequency region accessible

by (a) Ar+_{-ion (514.5 nm), (b) Nd:YAG (532.0 nm) lasers [5]. The 19429.82 cm}−1 _and

1889.90 cm−1 absorption lines (marked by stars) are used for laser line elimination of the single-mode Ar+-ion laser and Nd:YAG laser, respectively.

Recently, OptiGrate launched a new line of optical notch filters based on Bragg’s diffraction law, named BragGrateTM notch filters (BNFs) [4]. By precisely calculating the angle between the incident light and filter surface, BNFs is capable of reflecting a certain wavelength of light with a bandwidth as narrow as 10 cm-1. We used either a home-built iodine vapor filter or BNFs with single-mode lasers and achieved observation of Raman signals down to 9.8 cm-1.

Figure II-3. The BragGrateTM notch filter can reflect a given wavelength of light (Rayleigh

scattering) and other wavelengths (Raman scattering) pass through it.

The iodine vapor filter and BNFs have both pros and cons. For example, the iodine vapor filter can more efficiently eliminate Rayleigh scattering (OD≥6) and has a larger aperture (diameter = 5 cm) than BNFs. The larger aperture has a practical advantage because it is easier for alignment and suitable for use in both conventional 90° scattering setup and confocal Raman microscope system. Despite these advantages, the Raman spectra acquired with the iodine vapor filter are suffered from the artifacts due to the rotational and vibrational absorption of iodine molecules. Correction of the artifacts is required every time after experiments. In contrast with the iodine vapor filter, the BNFs are free from the annoying artifacts and have higher transmittance for Raman signals. However, the optical density of one BNF is three orders of

BragGrateTM _{Notch Filter} Rayleigh Scattering Stokes Raman Scattering anti-Stokes Raman Scattering Stokes Raman Scattering anti-Stokes Raman Scattering

scattering elimination efficiency as the iodine vapor filter. Moreover, the elimination efficiency of BNF is very sensitive to the incident angle. Therefore, the alignment of the BNFs is critical and at present only compatible with the confocal Raman system. Figure II-4 shows typical transmittance spectra of the iodine vapor filter and BNFs. The absorption band of the iodine vapor filter used to eliminate Rayleigh scattering cannot be observed in this spectrum because the bandwidth (0.03 cm-1) is too narrow to be distinguished compared with the spectral resolution (2.7 cm-1).

Figure II-4. Transmittance spectra of the iodine vapor filter operated at 60 °C (solid line) and

combination of two BragGrateTM notch filters (dotted line).

In this study, we used two fast low-frequency Raman spectrometers equipped with different lasers, that is, a single-mode Ar+-ion gas laser (514.5 nm) and a single-mode Nd-YVO4

solid-state laser (532 nm), and details of the two spectrometers are described below.

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 T ra ns mi tt an ce 565 560 555 550 545 540 535 530 Wavelength (nm)

II-2. Fast Low-Frequency Raman Spectrometer Using a

Single-Mode Ar

+

-ion Laser (514.5 nm)

II-2-1. Apparatus

Figure II-5. Schematic diagram of the low-frequency Raman spectrometer with 514.5 nm

excitation.

A schematic diagram of the 514.5 nm apparatus is shown in Figure II-5. An argon ion laser (BeamLok 2060, Spectra Physics) with a 514.5 nm output wavelength was used as the excitation source and focused on the sample cell. The scattered light was collected at 90° by a camera lens (f = 50 mm, f/1.2, Nikon), then focused on a 5 mm aperture to block strong reflected light from the sample cell. The collimated light passed through the iodine vapor filter operating at 95 °C, which filtered out almost all Rayleigh scattering. The transmitted light was further

Single-mode Ar ion laser (514.5 nm) Laser line filter

Polarizer Camera lens f/1.2 Sample Polarizer Iodine vapor filter Spectrometer CCD detector Tungsten lamp Camera lens Flipper mirror 5 mm aperture λ/2 plate lens f = 100 mm Depolarizer

focused onto the entrance slit (50 µm) of a spectrometer (f = 500 mm, f/6.5, SP-2558, Princeton Instruments) and dispersed by a 1200 g/mm grating, which covers a wide spectral range (>1300 cm-1) with a high spectral resolution of 2.7 cm-1. The dispersed light was finally detected by a back-illuminated, liquid-N2 cooled CCD detector (Spec 10:100 B, Princeton Instruments) with

100×1340 pixels operating at −120 °C.

White light from a tungsten lamp was monitored before and after each measurement to correct the rotational-vibrational absorption of the iodine vapor. In order to direct the white light into the filter with good position reproducibility, a motorized flipper mirror and an aperture needed to be placed in front of the iodine vapor filter.

II-2-2. Low-Frequency Raman Measurement of

L

-Cystine

In this section, we demonstrate the ability of the apparatus to clearly detect the Raman bands at 9.8 cm-1 in both Stokes and anti-Stokes region. L-Cystine was chosen to be a standard

sample due to its many low-frequency Raman bands. The Raman spectrum of micro-crystalline powder L-cystine is shown in Figure II-6. L-Cystine was purchased from Wako Pure Chemical

Industries (99% pure) and used without further purification. As can be seen from the inset of Figure II-5(c), the lowest Raman band of L-cystine at 9.8 cm-1 is clearly observed. Moreover, the

most intense Raman band of L-cystine located at 498 cm-1 has nearly the same area intensity as

that of the remaining Rayleigh scattering band at 0 cm-1, which can help us to check the Rayleigh elimination efficiency of the iodine vapor filter.

Figure II-6. Raman spectra of L-cystine.

(a) Observed Raman spectrum of L-cystine.

(b) Transmittance spectrum of the iodine vapor filter. (c) Intensity corrected Raman spectrum of L-cystine (= a/b).

The measurement was done with 2.7 cm-1 spectral resolution, 1 s exposure time, and 68 mW laser power. (a) Observed spectrum (b) Filter transmittance (c) Corrected spectrum Ra ma n In te ns it y 1000 800 600 400 200 0 -200 Raman Shift (cm-1) 0.5 0.4 0.3 0.2 0.1 0.0 T ra ns mi tt an ce Ra ma n In te ns it y 50 0 -50 Raman Shift

II-3. Fast Low-Frequency Raman Spectrometer Using a

Single-Mode Solid-State Laser (532 nm)

Nowadays, diode-pumped solid-state (DPSS) lasers have been widely used in Raman measurements to replace gas lasers because they have higher output power, better stability and smaller size. Here, we used a DPSS laser with 532 nm output wavelength. The iodine vapor filter can eliminate a laser line at 532 nm more efficiently than the argon ion laser line (514.5 nm). This high elimination efficiency at 532 nm allows us to decrease the operation temperature of the iodine vapor filter and increase the transmittance of the Raman signals. In addition, the operating frequency of the DPSS laser is monitored by a wavelength meter, and fixed exactly to the absorption band of the iodine vapor filter. Together, the system with the DPSS laser (532 nm) has better stability and higher efficiency than that with the argon ion laser (514.5 nm).

In this system, besides the conventional 90° scattering setup discussed in the last section, an optical microscope and a 2-axis motorized stage were introduced to construct a confocal low-frequency Raman microspectrometer, which enables us to investigate the sample under the microscope and acquire its Raman images. For the confocal Raman microscope system, both iodine vapor filter and BragGrateTM notch filters can be utilized.

II-3-1. Apparatus

Figure II-7. Schematic diagram of the low-frequency Raman spectrometer (532 nm excitation)

with 90° scattering setup.

Low-Frequency Raman Spectrometer with 90° Scattering Setup (532 nm)

A schematic diagram of the 532-nm apparatus is shown in Figure II-7. A frequency-doubled Nd:YVO4 laser (532 nm) based on a ring cavity (Verdi-V5, Coherent) was

used as the excitation light source. A small portion (<5%) of the output was reflected to a high-resolution wavelength meter (WS-7, HighFinesse). The frequency of the laser line was monitored by the wavelength meter with 60-MHz accuracy and maintained at 18789.902 cm-1, which coincides with the frequency of an absorption line of iodine vapor (Figure II-2 (b)). This apparatus is similar to the low-frequency Raman spectrometer with the argon ion laser introduced in section II-2. An advantage of the low-frequency Raman spectrometer described here is that the iodine vapor has better Rayleigh elimination efficiency at 532 nm than at 514.5 nm. As a result, the operation temperature of the iodine vapor filter can be decreased to 67 °C,

Nd#YVO4((532(nm)( Single#mode(laser Lens f = 62.9 mm Flipping mirror Polarizer Beam( sampler( Lens f"= 100 mm Tunable aperture φ = 0.2~6.0 mm Sample Lens f"= 100 mm Camera lens F/1.2, f = 50 mm Camera lens Flipping mirror Standard white light Analyzer Isolator CCD Fiber coupler (To wavelength meter) Verdi-V5 COHERENT Iodine vapor filter

and the transmittance of the Raman signals increases accordingly. The scattered light from the sample was collected by a camera lens (f = 50 mm, f/1.2, Nikon). The collimated light was passed through the iodine vapor filter, which filtered out almost all Rayleigh scattering. The transmitted light was dispersed by the spectrometer (f = 500 mm, f/6.5, SP-2558, Princeton Instruments) and detected by a back-illuminated, liquid-N2 cooled CCD detector (Spec 10:400 B,

Princeton Instruments) with 400×1340 pixels operating at −110 °C.

Figure II-8. Schematic diagram of the low-frequency confocal Raman spectrometer

Confocal Low-Frequency Raman Microspectrometer (532 nm)

The system is schematically shown in Figure II-8. The same laser as in the 90° scattering setup was used. In order to make full use of the numeric aperture of the objective lens, the

Nd-YVO4 (532 nm) Single-mode laser Lens f = 80 mm Objective f = 16.56 mm Pinhole φ = 25 µm Microscope Expand 4.83X Damper Objective f = 180 mm Pinhole φ = 75 µm Lens f = 170 mm Halogen lamp Condenser Sample Motorized X-Y stage Objective 20X, NA=0.45, air Dichroic mirror Optical image Rayleigh + Raman Incident laser Long-pass filter Camera Polarizer Beam sampler Beam sampler Movable mirror Lens f = 100 mm Iodine vapor filter Isolator Spectrometer CCD Fiber coupler (To wavelength meter) Verdi-V5 COHERENT BragGrate™ Notch Filter

diameter of the laser beam was magnified from 2.25 mm to 10.87 mm by a pair of lenses. The expanded laser beam was directed to an inverted optical microscope (ix71, Olympus) and focused onto the sample by a long working distance objective (20X, NA=0.45). The sample was placed on a 2-axis motorized stage (Sigma Koki) with a minimum of 0.1 µm step of movement. The back-scattered light was collected by the same objective, and then focused by a second objective (f=180 mm) onto a pinhole (diameter = 75 µm). With this confocal configuration, the axial resolution at the sample point was increased. The well-collimated light was passed though the iodine vapor filter or BragGrateTM notch filters to reject almost all Rayleigh scattering. The transmitted light was dispersed and detected by the same spectrometer and CCD detector as in the 90° scattering setup.

II-3-2. Comparison of the Filter Transmittance and Examination of

Low-Frequency Measurement Capability Using

L

-Cystine

In the confocal Raman spectrometer system, both the iodine vapor filter and BragGrateTM notch filters (BNFs) can be utilized as Rayleigh rejection filters. Compared to the iodine vapor filter, the advantage of using BNFs is not only that they can be free from the artifacts but also that the higher transmittance allows more Raman signals to pass through. The transmittance spectra of different filter sets are shown in Figure II-9. Which kind of filter set should be used during the measurement depends on the exposure time. Generally speaking, if the exposure time is short (<1 s), two BNFs can reach the best S/N ratio because of the high transmittance (>60 %).

can reach the best Rayleigh elimination efficiency.

Figure II-9. Transmittance spectra of (a) two BNFs, (b) three BNFs, (c) iodine vapor filter

(at 60 °C), and (d) one BNF + iodine vapor filter (at 60 °C).

As described in section II-2-2, L-cystine was employed as a standard sample to check the

low-frequency measurement capability of the system. Figure II-10 shows the observed Raman spectra. In both spectra taken with either iodine vapor filter or BNFs, the ±9.8 cm-1 Raman band can be seen, indicating a similar low-frequency measurement capability of the BNFs to the iodine vapor filter. It is noteworthy that the observed spectrum taken with the iodine vapor filter was interfered with artifacts on the background. The band intensities are also incorrect due to the absorption of the iodine molecules. In contrast, the observed spectrum taken with three BNFs does not have these problems, so we can directly look at the spectrum without any background

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 T ra ns mi tt an ce 1000 800 600 400 200 0 -200 Raman Shift (cm-1) (a) (b) (c) (d)

Figure II-10. Observed Raman spectra of L-cystine taken with (a) three BNFs and

(b) a combination of one BNF and the iodine vapor filter.

The measurement was done with 2.7 cm-1 spectral resolution, 10 s exposure, time and 1 mW laser power.

II-3-3. Evaluation of the Spatial Resolution of the Confocal Raman

Microspectrometer

The spatial resolution of the confocal system is governed by an objective lens and the confocal pinhole. The lateral (XY) resolution was determined by scanning the focused laser beam horizontally with a step of 0.2 µm across a spherical polystyrene bead with 1 µm diameter. The Raman band intensity of polystyrene at 1003 cm-1 was used to estimate the lateral resolution. Figure II-11 (a) displays the band intensity of polystyrene at 1003 cm-1 versus the stage-scanned distance. The beam profile was assumed to be a Gaussian function and its width obtained by fitting was 0.925 µm. The lateral resolution is defined as the full-width at half-maximum (FWHM) of the Gaussian width, which is equal to 2 ln2×Gaussian  width. Therefore, the calculated lateral resolution was 1.5 ± 0.1 µm.

(a) (b) 200 150 100 50 0 -50 -100 -150 -200 Raman Shift (cm-1) Ra ma n In te ns it y

Likewise, the axial resolution was evaluated by scanning the focused laser beam vertically with a step of 5 µm across interfaces between air and glass and between glass and tetrachloroethylene. The Raman band intensity of tetrachloroethylene at 236 cm-1 was employed for evaluation of the axial resolution. The Raman intensity change at the two interfaces can be fitted by a Heaviside step function H(x) convoluted with a Gaussian function G(x) (Equation II-1) [7]. 𝑓 𝑥, 𝑁, 𝑐, 𝐵 =  𝑁 𝐻 𝑎! _{𝐺 𝑎}!_{− 𝑎 𝑑𝑎}! ₌  𝑁 2 1 + 𝑒𝑟𝑓 𝑥 − 𝑎 2𝐵 ! !! + 𝑐      (𝐈𝐈 − 𝟏) where 𝑁 is a normalization constant, 𝑎 is the onset of the step function, 𝐵 is the width of the Gaussian function, 𝑐 represents a constant offset, and 𝑒𝑟𝑓 denotes the error function. The fitting result is shown in Figure II-11 (b). The obtained Gaussian width was 19.2 µm. The axial resolution is defined as the FWHM of the Gaussian width. Consequently, the calculated axial resolution was 32 ± 1.3 µm.

Figure II-11. Evaluation of lateral (a) and axial (b) resolution of the low-frequency confocal

Raman microspectrometer. Red dots represent observed Raman intensities and blue curves represent fitted lines.

Ra ma n In te ns it y 9 8 7 6 5 4 3 2 1 0 X Distance (um) Fitting Coefficients y0 =25.613 ± 34.4 A =2237.6 ± 103 x0 =4.5689 ± 0.0337 width =0.92516 ± 0.0517 (a) (b) Ra ma n In te ns it y 250 200 150 100 50 Depth (um) Fitting Coefficients N =16374 ± 150 a =121.02 ± 0.566 B =19.217 ± 0.768 c =263.88 ± 107

II-3-4. Development of the Low-frequency Raman Mapping

Program

Since the low-frequency confocal Raman microspectrometer system is equipped with a 2-axis motorized stage and an optical microscope, it is possible to control the movement of the stage by a computer and acquire spectra synchronously (Figure II-12).

Our low-frequency Raman mapping program was developed on LabVIEW 2012 (National Instruments). This program sends commands to the stage controller (SHOT-702, Sigma Koki) then moves the motorized stage with a minimum of 0.1 µm step. The stage was programmed to move in a zigzag way, as depicted in Figure II-13 (b). After moving each step, the stage stops for a while to wait for CCD exposure. At this moment, the program sends a transistor-transistor-logic (TTL) signal to the CCD controller (ST-133, Princeton Instruments) and triggers CCD exposure. Acquisition of Raman spectra is done with Winspec/32 software provided by Princeton Instruments. Consequently, it is necessary to run the LabVIEW program and Winspec/32 software simultaneously when doing Raman mapping.

Stage Controller PC CCD Controller CCD Detector Motorized Stage

   

Figure II-13. (a) Developed user interface of the mapping program in LabVIEW 2012.

(b) Stage movement during mapping.

(a)

References for Chapter II.

[1] H. Okajima and H. Hamaguchi, "Fast Low Frequency (Down to 10 cm-1) Multichannel Raman Spectroscopy Using an Iodine Vapor Filter," Applied Spectroscopy, vol. 63, pp. 958-960, 2009.

[2] Szu-Cheng Wang, "Construction of a Fast Low-Frequency Raman Spectrometer and Its Application to Real-Time Tracing of Melting Process of Crystalline 1,1’-binaphthyl,"

National Chiao Tung University, Master Thesis, 2011.

[3] G. E. Devlin, J. L. Davis, L. Chase, and S. Geschwind, "Absorption of Unshifted Scattered Light by a Molecular I2 Filter in Brillouin and Raman Scattering," Applied

Physics Letters, vol. 19, pp. 138-141, 1971.

[4] W. L. Peticolas, "Inelastic Light Scattering and the Raman Effect," Annual Review of

Physical Chemistry, vol. 23, pp. 93-116, 1972.

[5] Gregory S. Elliott and Thomas J. Beutner, "Molecular Filter Based Planar Doppler Velocimetry," Process in Aerospace Science, vol. 35, pp. 799-845, 1999.

[6] Alexei L. Glebov, Oleksiy Mokhun, Alexandra Rapaport, Sébastien Vergnole, Vadim Smirnov, Leonid B. Glebov, "Volume Bragg Gratings as Ultra-Narrow and Multiband Optical Filters," Proceedings of SPIE, vol. 8428, pp. 84280C-1-84280C-11, 2012.

[7] V. V. Pully, and C. Otto, "The Intensity of 1602 cm-1 Band in Human Cells is Related to Mitochondrial Activity," Journal of Raman Spectroscopy, vol. 40, pp. 473-475, 2009.

Chapter III.

In-situ Tracing of the Transformation Dynamics

III-1. Introduction to 1,1′-binaphthyls

1,1′-Binaphthyl is composed of two naphthyl groups linked via a C-C single bond at 1,1′-positions, which represents a special class of biaryl molecules. The molecule is well known for its applications as chiral recognition receptors and chiral catalysts such as 2,2′-bis(diphenylphosphino)-1,1′-binaphthyl (BINAP) [1-3]. The rotational barrier along the C-C single bond is 23.5 kcal/mol (∆  𝐺!_{) [4], which is much larger than the rotational barrier of}

biphenyl around the phenyl-phenyl bond in the gas phase (~1.4 kcal/mol) [5]. The large rotational barrier is attributed mainly to the repulsion of the hydrogen atoms at the 8 and 8′ positions. Besides, 1,1′-binaphthyl no longer possesses a plane of symmetry that biphenyl has in its perpendicular conformation, resulting in axial chirality of the molecule. This fact allows the isolation of the optically active 1,1′-binaphthyl enantiomers.

(S)·(＋)·1, 1′-binaphthyl (R)·(－)·1, 1′-binaphthyl H H H H Sa configuration Ra configuration Interconversion (b)

Figure III-1. Axial chirality of 1,1′-binaphthyl.

View along this direction (a)

The asymmetry of 1,1′-binaphthyl is molecular in nature, and enantiomeric interconversion is made possible simply by rotation about the interannular bond instead of by any bond-breaking process (Figure III-1 (a)). The racemization half-life of the enantiomers was found to be 14.5 min at 50 °C. Chiral 1,1′-binaphthyl was discovered by Pincock et al. in 1971 [6]. He found that racemic 1,1′-binaphthyl undergoes spontaneous resolution to generate the optically active R or S enantiomer when this compound crystallizes from the melt. The chiral 1,1′-binaphthyl molecule contains a chiral axis instead of a chiral center. The enantiomers of axially chiral compounds are usually given the stereochemical labels Ra and Sa as shown in

Figure III-1 (b).

In the crystalline state, 1,1′-binaphthyl is known to occur as two polymorphs [7]: the cisoid configuration or lower melting point (148 °C) form and the transoid configuration or higher melting point (161 °C) form. Brown et al. [8] were the first to recognize the fundamental difference between the two forms. They found that the molecules of the lower melting point form take a cisoid configuration. X-ray diffraction [9] confirmed the dihedral angle θ to be 68.6° (Figure III-2). Badar et al. [7] suggested, on the basis of their infrared spectroscopic results, that the high melting point form adopts a transoid configuration, which was later confirmed by X-ray analysis [10] to have a dihedral angle of 103.1°. The cisoid form, whose structure was determined by Kerr and Robertson [9], consists of racemic monoclinic crystals (space group

C2/c). A unit cell of this achiral crystal is comprised of two R enantiomers and two S

transoid form as an optically active compound which belongs to the tetragonal system of space group P42121 with either R or S enantiomer in a unit cell (Z = 4).

Extensive studies on the structural change of the ground and excited states of 1,1′-binaphthyl in the solution phase have been done by absorption spectroscopy [11, 12], fluorescence spectroscopy [13, 14], and Raman spectroscopy [14-16]. Also there are some theoretical studies, in which the dihedral angle adopted by the 1,1′-binaphthyl molecule in its most stable conformation was estimated [17-23].

The first report on the two crystal forms of 1,1′-binaphthyl in the viewpoint of Raman spectroscopy was provided by Lacey et al. [15]. However, the Raman spectra of the cisoid and transoid form in the low-frequency region (＜150 cm-1) have not been reported yet. In this study, we are interested to see how different are the spectral patterns in the low-frequency region of the different crystal structures of 1,1′-binaphthyl. Then we can use these differences to monitor the transformation dynamics between the cisoid and transoid forms. Moreover, the low-frequency Raman imaging technique can be applied to differentiate the two crystal forms on a molecular

θ = 68.6°

m.p. 148°C

θ = 103.1°

m.p. 161°C Cisoid

configuration configuration Transoid

Figure III-2. Schematic diagram of the two crystal forms of 1,1′-binaphthyl.

basis and to reveal spatial distributions of molecular orientations inside the crystal, both of which can hardly be realized with any other existing methods.

III-2. Preparation of the Cisoid and Transoid Form

1,1′-Binaphthyl powder with 98% purity, which is dominated by the cisoid form, was

purchased from Tokyo Chemistry Industry Co., LTD. Unknown impurities may cause strong fluorescence in the Raman spectrum measured with 514.5 and 532 nm excitations. In order to get rid of possible fluorescent impurities, column chromatography was used to purify the commercially obtained 1,1′-binaphthyl (Figure III-3).

Figure III-3. Column chromatography purification of 1,1′-binaphthyl.

Figure III-4. Preparation of the cisoid and transoid forms. 98% 1,1’-binaphthyl

in benzene

Stationary phase: silica gel Mobile phase: 99.5% n-hexane

Impurities Pure 1,1’-binaphthyl Cisoid Transoid Pure 1,1’-binaphthyl Pure 1,1’-binaphthyl Liquid phase Cisoid Transoid

180°C

Heating

_150°C

Preparation of the cisoid and transoid crystal was followed the similar procedure reported previously [6, 24, 25, 26], figure III-4 visually describe how we prepared them. A small amount of the pure 1,1’-binaphthyl (0.35 g) was placed in a 10 ml round-bottom flask and heated up to 180 °C in an oil bath. The melt was kept in the temperature for 15 minutes to completely

racemize the sample, then we cooled down the melt to 150 °C and allowed it to crystallize. After 4 hours, the sample was cooled down to room temperature. The transparent, flake-shaped cisoid crystals was found at the top of the reactor, as the result of evaporation and recrystallization of 1,1’-binaphthyl. This process is consistent with the previous literature, which claimed the transformation of these polymorphs follows vapor-solid transformation [10]. On the other hand, the white transoid solids were observed to stay at the bottom of the reactor.

To examine the purity of the cisoid and transoid crystals so obtained, differential scanning calorimetry (DSC) was applied to measure the melting point of the two crystal forms. Figure III-5 shows the DSC curves of the cisoid crystal obtained with different heating rates. The 146 °C band was observed, which represents the melting point of the cisoid form. In addition, the 160 °C band was also observed, which indicates the melting point of the transoid form. With a faster heating rate (45 °C/min), the cisoid form predominates. However, with a slower heating rate (5 °C/min), the DSC curve is dominated by the transoid form. This result implies that to observe the transformation between the cisoid and transoid forms, a slower heating rate should be used. The DSC curves of the transoid form measured with different heating rates are shown in Figure III-6. This time, the transoid form dominates in all the cases. Thus we would be unable to observe the transformation from the transoid crystal.

Figure III-5. Differential scanning calorimetry (DSC) measurements of cisoid form crystal with

different heating rates [26].

Figure III-6. Differential scanning calorimetry (DSC) measurements of transoid form crystal

III-3. Raman Spectra of the Two Polymorphs of 1,1′-Binaphthyl

The Raman spectra of the two purified crystal powder samples measured at room temperature with 514.5 nm laser excitation and 1 second exposure time are shown in Figures III-7 and III-8. The most notable difference in the high-frequency region (200 – 1100 cm-1) between the two spectra is concerned with different intensity ratios of pairs of Raman bands centered at 460, 520, 1025, and 1588 cm-1 [15]. The difference in the intensity ratio of a pair of the bands can be explained in terms of the weakly coupled model of the two naphthyl groups [27]. Due to the coupling, each of the bands splits into a doublet in 1,1′-binaphthyl by ~21 cm-1, and the intensity ratios can reflect how these two naphthyl groups interact with each other with different dihedral angles (68.6° and 103.1°). Among those pairs of bands, the 510/532 cm-1 pair shows the strongest intensity; thus they are selected to be the pair of marker bands to distinguish the cisoid and transoid forms in the high-frequency region. Compared with the high-frequency region, the low-frequency region (<200 cm-1) shows significantly strong signal intensities. The reason for this is that the low-frequency motions usually originate from collected motions or lattice vibrations, which give large polarizability changes that result in strong Raman signals. Figure III-9 shows the low-frequency (−200 to +200 cm-1) Raman spectra of the two polymorphs of crystalline 1,1′-binaphthyl. Decent low-frequency Raman spectra in both Stokes and anti-Stokes regions have been obtained simultaneously within 1 s. This is the first observation, to our knowledge, of the low-frequency Raman spectra of the crystal polymorphs of 1,1′-binaphthyl. In the transoid form, the 26 cm-1 band gives the strongest intensity in the whole

which may arise from a symmetry splitting of one band into a doublet at 100 and 110 cm-1. Detailed band assignments are yet to be done, and high resolution polarized Raman measurements are required in the future.

Figure III-7. Raman spectrum of the cisoid powder at room temperature.

Figure III-8. Raman spectrum of the transoid powder at room temperature.

Ra ma n In te ns it y 1000 800 600 400 200 0 -200 Raman Shift (cm-1) Ra ma n In te ns it y 600 550 500 450 400 Raman Shift (cm-1) Expand 5X 510 532 468 450 Ra ma n In te ns it y 1000 800 600 400 200 0 -200 RamanShif (cm-1) Ra ma n In te ns it y 600 550 500 450 400 Raman Shift (cm-1) Expand 8X 510 532 441 465

Figure III-9. Low-frequency Raman spectra of the two crystalline 1,1′-binaphthyl. 200 150 100 50 0 -50 -100 -150 -200 Raman Shift (cm-1) Ra ma n In te ns it y 37 56 65 75 100 110 175 20 26 43 55 66 79 95

Transoid

Form

Cisoid

Form

III-4. In-situ Tracing the Transformation between Polymorphs

Heating Apparatus

As demonstrated by the DSC measurements of the cisoid crystal (Figure III-5), we now know that when the cisoid form is heated, it transforms to the transoid form. In order to measure the Raman spectra during the transformation, we designed a heating cell.

Figure III-10 draws the design of the laboratory-built heating cell for the 90° scattering setup Raman spectrometer. It is a closed system and can be raised its temperature by heating coils inside the cell. The sample sealed in a capillary is inserted from the side port, and a thermocouple is placed beside the capillary to monitor the surrounding temperature. The laser beam is introduced from the bottom window and focused on the sample. The scattered Raman signals can be collected from the large optical window.

Figure III-10. Laboratory-built heating cell.

Incident Laser Sample (in capillary) Heating Controller Thermocouple Optical Window

Raman Spectral Change Observed When the Cisoid Powder Was Heated

Figure III-11 shows the low-frequency (−200 to +200 cm-1) Raman spectra of slowly heated cisoid powder at different temperatures. The measurement was started from room temperature, and the sample was heated up with a rate of 3 °C/min. When the temperature reached to the desired value, we waited another 1 minute to ensure that thermal equilibrium was reached and then acquired the Raman spectrum. As the temperature increases, all Raman bands shift to lower frequencies. At 126 °C, the Raman band located at 26 cm-1, which is a low-frequency marker band for the transoid form starts to appear. In other words, the transformation from the cisoid to transoid form takes place at this temperature. After 138 °C, the cisoid was totally transformed into the transoid form. The transoid form melts at 156 °C, which is in agreement with the literature [7].

Compared to the low-frequency region, the spectral pattern in the higher frequency region (Figure III-12) does not change significantly. The only difference is the intensity ratios of three pairs of Raman bands: 450/468 cm-1, 510/532 cm-1 and 1010/1035 cm-1. From these results, we can conclude that the low-frequency Raman bands are more sensitive to the structural change, exhibiting much more notable differences than the high-frequency Raman bands do.

Figure III-11. Low-frequency Raman spectra at different temperatures recorded by heating the

cisoid powder. Laser power, 55 mW (514.5 nm excitation); exposure time, 1 s; spectral resolution, 2.7 cm-1; and heating rate, 3 °C/min. The temperatures shown on the right are those indicated by the thermocouple.

Ra

ma

n

In

te

ns

it

y

200

100

0

-100

-200

Raman Shift (cm

-1

)

24°C 68°C 88°C 98°C 110°C 120°C 124°C 126°C 128°C 131°C 133°C 134°C 135°C 136°C 137°C 138°C 139°C 140°C 141°C 142°C 143°C 144°C 145°C 146°C 147°C 148°C 149°C 150°C 151°C 152°C 153°C 154°C 156°C

Figure III-12. Raman spectra in the fingerprint region at different temperatures recorded by

heating the cisoid powder. Laser power, 55 mW (514.5 nm excitation); exposure time, 1 s; spectral resolution, 2.7 cm-1_{; and heating rate, 3 °C/min. The}

temperatures shown on the right are those indicated by the thermocouple.

Ra

ma

n

In

te

ns

it

y

1000

750

500

250

Raman Shift (cm

-1

)

24°C 68°C 88°C 98°C 110°C 120°C 124°C 126°C 128°C 131°C 133°C 134°C 135°C 136°C 137°C 138°C 139°C 140°C 141°C 142°C 143°C 144°C 145°C 146°C 147°C 148°C 149°C 150°C 151°C 152°C 153°C 154°C 156°C

Raman Spectral Change Observed When the Transoid Powder Is Heated

To examine the reversibility of the transformation between the cisoid and transoid forms, similar heating measurements were performed on the transoid powder. However, we did not observe any transformation of the transoid form. The low-frequency Raman spectra at different temperatures displayed in Figure III-13 show no spectral pattern change. The low-frequency Raman bands just keep showing a decrease in intensity and a slight shift to lower frequency upon heating. After 154 °C, almost all low-frequency Raman bands disappear, indicating the transoid form melts. Also the Raman spectra in the higher frequency region (Figure III-14) do not show any significant change, including the three pairs of Raman bands.

This result suggests that the transformation is irreversible. It only happens with slower heating of the cisoid form. Our result also indicates that the transoid form is a more thermally stable structure. This interpretation is consistent with the X-ray diffraction analysis of the cisoid and transoid single crystals [24]. Figure III-15 depicts the crystal structure inside a unit cell of the cisoid and transoid crystals. In both cases, the intermolecular interactions are mainly from van der Waals force, acting between the H atoms and π electrons of the aromatic ring. In the transoid form, the molecules stack parallel to each other via the CH- π interactions, while the CH- π interactions in the cisoid form are expected to be small within each layer. Thus, the intermolecular interactions within a unit cell of the transoid form are stronger than those in the cisoid form.

Figure III-13. Low-frequency Raman spectra at different temperatures recorded by heating the

transoid powder. Laser power, 55 mW (514.5 nm excitation); exposure time, 1 s; spectral resolution, 2.7 cm-1; and heating rate, 3 °C/min. The temperatures shown on the right are those indicated by the thermocouple.

Ra

ma

n

In

te

ns

it

y

200

100

0

-100

-200

Raman Shift (cm

-1

)

24ºC 68ºC 88ºC 98ºC 108ºC 118ºC 123ºC 128ºC 133ºC 136ºC 138ºC 140ºC 143ºC 146ºC 148ºC 149ºC 150ºC 151ºC 152ºC 153ºC 154ºC

Figure III-14. Raman spectra of the fingerprint region at different temperatures recorded by

heating the transoid powder. Laser power, 55 mW (514.5 nm excitation);

exposure time, 1 s; spectral resolution, 2.7 cm-1; and heating rate, 3 °C/min. The temperatures shown on the right are those indicated by the thermocouple.

Ra

ma

n

In

te

ns

it

y

1000

750

500

250

Raman Shift (cm

-1

)

24ºC 68ºC 88ºC 98ºC 108ºC 118ºC 123ºC 128ºC 133ºC 136ºC 138ºC 140ºC 143ºC 146ºC 148ºC 149ºC 150ºC 151ºC 152ºC 153ºC 154ºC

Figure III-15. The crystal structure of (a) the cisoid form and (b) the transoid form. [24]

Raman Spectra of Melted 1,1′-Binaphthyl

We kept recording the Raman spectra after the melting point of the transoid form (158 °C) was reached. The observed Raman spectra after melting are shown in Figure III-16 (a). To get rid of the broad Rayleigh wing, the following reduce equation (Equation III-1) was employed [28].

𝐼

_𝑟𝑒𝑑

𝜐 = 1 + 𝑒

!!!!!!! !!

× 𝜐

_!

− 𝜐

!!

_×𝐼

𝑜𝑏𝑠

𝜐

 

where c is the speed of light, h is Planck’s constant, kB is the Boltzmann constant, T is the sample

temperature, 𝜈_! is the absolute wavenumber of the incident laser, 𝜈 is the Raman shift, and

𝐼

_𝑜𝑏𝑠

𝜐

is the observed Raman intensity.

The reduced Raman spectra at different temperatures (Figure III-16 (b)) clearly show that the 20 and 26 cm-1 bands of the transoid form disappear upon melting. The absence of these bands in the liquid state suggests that they are most likely assigned to lattice vibrations. In contrast, a broad feature from 40 to 140 cm-1 still remains even at 173 °C, which is much higher than the melting point. Based on this observation, the three low-frequency Raman bands at 55, 66, and 79 cm-1 may be due to intramolecular vibrations. With the present data alone, it is hard to tell whether the other two shoulders located at 43 and 95 cm-1 are lattice vibrations or intramolecular vibrations. Accurate calculations of Raman spectra of molecular crystals, which are highly challenging at the present stage of theoretical development, will help clarify the band assignments.