國 立 交 通 大 學
應用化學系碩士班
碩 士 論 文
架設快速低振動頻率的拉曼光譜儀
並應用於即時監控結晶態1,1′-binaphthyl的熔化過程
Construction of a Fast Low-Frequency Raman Spectrometer
and Its Application to Real-Time Tracing of the Melting
Process of Crystalline 1,1′-binaphthyl
研 究 生: 王偲丞
指導教授: 重藤真介 博士
架設快速低振動頻率的拉曼光譜儀
並應用於即時監控結晶態1,1
′
-binaphthyl的熔化過程
研 究 生：王偲丞 Student：Szu-Cheng Wang
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
August 2011
Hsinchu, Taiwan, Republic of China
i
架設快速低振動頻率的拉曼光譜儀
並應用於即時監控結晶態1,1′-binaphthyl的熔化過程
學生: 王偲丞 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
研究凝態物質的分子間作用力和晶格振動可以藉由低頻率(小於 200 波
數)拉曼光譜學，但此技術的困難在於雷利散射比所要的拉曼訊號強上幾個
級數。市面上典型的 notch 濾片無法有效的過濾雷利散射，因此要得到 200
波數以下的拉曼訊號相當困難。雖然單通道偵測器結合三分光儀(triple
monochromator) 可以有效的分離雷利散射和拉曼散射並得到低頻率(小於
50
cm−1)且高訊雜比的拉曼光譜圖，但如果研究相變化這類型的動力學過程，
多通道的偵測器才有辦法以較短的曝光時間(小於 1 秒)完成測量。在本論
文中我們使用充滿碘蒸氣的玻璃槽做為消除雷利散射的濾片並結合多通道
的拉曼光譜儀，我們能夠以小於 1 秒的測量時間記錄 10 波數以下的史托克
斯和反史托克斯的拉曼訊號。接著將此低振動頻率拉曼系統應用到結晶態
1,1′-binaphthyl。這個分子在結晶態有兩個晶型: 類順式與類反式，並有著
不同的熔點。我們發現兩個晶型在低頻率範圍(−200–+200
cm−1)呈現截然不
同的拉曼光譜。為了瞭解低振動頻率訊號的來源，快速的加熱樣品並以 0.2
ii
秒的測量時間記錄每一張光譜圖，由此觀察兩個晶型的拉曼光譜隨溫度的
變化。除此之外，藉由史托克斯和反史托克斯譜帶強度的比率，每個數據
點的樣品溫度可以高精準度的被估計。將這兩組溫度變化的拉曼光譜數據
搭配理論計算的結果，我們推測類反式中的 26
cm−1峰和類順式中的 100
cm−1峰分別是碳碳單鍵的扭曲振動和平面外變形振動(分子內振動)。由於分子
間的作用力與熱膨脹造成低振動頻率波帶往低頻率位移有關，實驗得知類
反式的拉曼波帶位移的程度較類順式少，因此我們認為類反式的分子間作
用力較類順式強，而文獻中由單晶繞射實驗所得到的結果與我們的推測吻
合。
iii
Construction of a Fast Low-Frequency Raman
Spectrometer and Its Application to Real-Time Tracing
of the Melting Process of Crystalline 1,1′-binaphthyl
Student: Szu-Cheng Wang Advisor: Dr. Shinsuke Shigeto
M. S. Program, Department of Applied Chemistry
National Chiao Tung University
Abstract (in English)
Low-frequency (<200 cm−1) Raman spectroscopy enables us to investigate
intermolecular vibrations and lattice vibrations in condensed phase materials. However, low-frequency Raman signals can be almost completely buried under vast Rayleigh scattering. Typical commercial notch filters are not effective enough to eliminate Rayleigh scattering and do not allow us to reach the frequency region below 200 cm−1. Moreover, in order to keep track of dynamic processes such as phase transitions, multichannel detection rather than single-channel detection with a double/triple monochromator is required. In this work, by combining an I2 vapor-containing cell as a Rayleigh rejection filter with a multichannel
Raman spectrometer, we have successfully recorded Raman spectra down to ~10 cm−1 in both Stokes and anti-Stokes sides simultaneously. The constructed Raman spectrometer has then been applied to crystalline 1,1′-binaphthyl, which has two different crystal polymorphs, that is, the cisoid and transoid forms. They show quite distinct spectral features in the low-frequency region (−200–+200 cm−1). Real-time tracing of the melting process of two crystalline forms have also been conducted with rapid heating. A series of Raman spectra have been recorded every 0.2 sec, in which spectral changes of low-frequency bands are clearly observed. In addition, the sample temperature has been accurately estimated from the Stokes/anti-Stokes intensity ratio of the Raman bands. Base on the temperature-dependent Raman spectra and theoretical calculations, the 26 cm−1 band of the transoid form and the 100 cm−1 band of the cisoid form are assigned to the torsional vibration and out-of-plane deformation of the C–C single bond, respectively. Moreover, the peak position of each low-frequency band has been determined from the fitting. We found that peak shifts due to thermal expansion for the transoid form are smaller than those for the cisoid form. Thus, intermolecular interactions in the transoid form are stronger compared with those in the cisoid form. This finding is consistent with their X-ray crystal structures.
iv
Acknowledgments
My deepest gratitude goes first and foremost to Professor Shigeto, my supervisor, for providing me an opportunity for studies in this thesis. His continuing encouragements and excellent guidance help me a great deal. Without his consistent and illuminating instruction, this thesis could not have reached its present form.
I would like to express my heartfelt gratitude to Professor Hamaguchi, who led me into the world of Raman spectroscopy. I am also greatly indebted to Doctor Yabumoto, who have instructed and helped me a lot in the past two years.
I also owe my sincere gratitude to my friends (Tsung-wei, Je-hau, Jian-jung,) and all members in Shigeto group who gave me their help and time in listening to me and helping me work out my problems during the difficult course of the thesis. And I am deeply grateful to Mr. Sudhakar for his kind assistances in chemical syntheses.
Last my thanks would go to my beloved family for their loving considerations and great confidence in me all through these years. Without their sincere supports, I could not have accomplished this work.
v
Tables of Contents
Page
Abstract (in Chinese) ... i
Abstract (in English) ... iii
Acknowledgments... 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 ... 2
I-3. Multichannel detection in Raman spectroscopy ... 3
I-4. Vapor filters for Rayleigh scattering elimination ... 5
I-5. Content of this thesis ... 7
Chapter II Construction of a Fast Multichannel Low-Frequency Raman Spectrometer ... 8
II-1. Introduction ... 9
II-2. Multichannel low-frequency Raman spectrometer using a single-mode Ar-ion laser (514.5 nm) and I2 vapor filter ... 12
II-2-1. Experimental setup ... 12
II-2-2. Results and Discussion ... 16
Chapter IIIReal-Time Tracing of the Melting Process of the Two Distinct Polymorphs of Crystalline 1,1′-Binaphthyl ... 27
III-1. Introduction ... 28
III-2. Experimental ... 31
III-3. Results ... 33
III-4. Fitting analysis ... 39
III-5. Discussion: Temperature change during the melting ... 44
III-6. Discussion: Changes in Low-frequency Raman bands change during heating ... 48
III-7. Discussion: Possible vibrational mode of 26 cm−1 band in the transoid form and 100 cm−1 band in cisoid form ... 52
vi
Chapter IV Conclusion ... 56 References ... 59
vii
List of Figures and Tables
Page
Figure I-1. Transmittance spectrum of a commercially available notch filter ... 4
Figure II-1. Laser gain curve and cavity modes of an Ar-ion laser. ... 9
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. Iodine vapor filter used in the present study. ... 12
Figure II-4. Temperature dependent of iodine vapor pressure in the gas–solid equilibrium. . 13
Figure II-5. Photographs of the I2 vapor filter were taken at different filter temperatures (a) 24 °C (room temperature) (b) 95 °C ... 14
Figure II-6. Schematic diagram of a single longitudinal mode Ar-ion laser. ... 15
Figure II-7. Schematic diagram of the multichannel low-frequency Raman spectrometer. ... 16
Figure II-8. Intensity correction of the Raman spectrum of CCl4 ... 18
Figure II-9. Raman spectra of L-cystine... 21
Figure II-10. L-cystine Raman spectra in the low-frequency region. ... 22
Figure II-11. Comparison of the Raman spectra of L-cystine measured with three different filters... 23
Figure II-12. Raman spectra of L-cystine obtained in the following two situations: (a) The laser light reflected by the glass capillary came into the I2 vapor filter (b) The reflected laser light was blocked by the 5 mm aperture ... 25
Figure II-13. Raman spectra of L-cystine measured by tuning the frequency of the single-mode Ar-ion laser. ... 26
Figure III-1. Axial chirality of 1,1′-binaphthyl. ... 28
Figure III-2. Schematic diagram of the two crystal forms of 1,1′-binaphthyl. ... 29
Figure III-3. Purification of commercially obtained 1,1′-binaphthyl and preparation of (a) cisoid form and (b) transoid form... 31
Figure III-4. Photograph of the heating apparatus ... 32
Figure III-5. Raman spectra of the two forms of crystalline 1,1′-binaphthyl. ... 33
Figure III-6. Low-frequency Raman spectra of two crystalline 1,1′-binaphthyl. ... 34
Figure III-7. Low-frequency Raman spectra of the transoid form with rapid heating. ... 37
Figure III-8. Low-frequency Raman spectra of the cisoid form with rapid heating. ... 38
Figure III-9. Experimental data (red closed circles) of the transoid form and their fitted results (blue solid lines). ... 42
Figure III-10. Experimental data (red closed circles) of the cisoid form and their fitted results (blue solid lines). ... 43
Figure III-11. Plot of the estimated temperature versus heating time for the cisoid form. ... 44 Figure III-12. Differential scanning calorimetry (DSC) measurements of cisoid form
viii
crystal with different heating rates. ... 45
Figure III-13. Plot of the estimated temperature versus heating time for the transoid form. . 46
Figure III-14. Differential scanning calorimetry (DSC) measurements of transoid form crystal with different heating rates. ... 47
Figure III-15. Plot of the peak positions of the seven low-frequency Raman bands against heating time. ... 49
Figure III-16. Plot of the peak positions of the seven low-frequency Raman bands against estimated temperature. (a) Transoid form and (b) cisoid form. ... 50
Figure III-17. The crystal structures of the two forms of 1,1′-binaphthyl. ... 51
Figure III-18. The intramolecular vibrations in gas phase 1,1′-binaphthyl molecule calculated from density functional theory. ... 53
Figure III-19. Raman spectra of aromatic compounds containing the torsional motion. ... 55
Table I-1. Absorbing atom/molecule and excitation source combinations... 5
Table III-1. Fitting parameters ... 40
1
Chapter I
2
I-1. Motivation of this study
How can we understand crystal structural changes upon heating from the molecular viewpoint? Differential scanning calorimetry (DSC), which is a widely used technique in thermodynamic studies, can trace the free energy changes of materials and provides us with macroscopic understandings of phase transitions such as melting or crystallization processes [1, 2]. For further understanding of dynamical and microscopic aspects of these phase transitions, elucidation of the loss of a crystal structure and associated changes in intermolecular interactions during the melting process is deemed necessary. The purpose of the present study is to construct a spectroscopic apparatus that enables us to rapidly (<1 sec) monitor changes in intermolecular interactions and thereby deepens our understanding of phase transition at the molecular level.
I-2. Raman measurements of low-frequency motions
Intermolecular forces in molecular assemblies are reflected in intermolecular and/or collective motions. The frequency of these motions is usually smaller than 200 cm−1 (6 THz or 20 meV) and is observed in the low-frequency region of an optical spectrum. Measurements of Raman scattering in the frequency region below 200 cm−1 have been used for investigating the intermolecular and/or collective motions of materials in the condensed phase. For example, the optically active lattice vibrations, which correspond to the optical phonon modes at k = 0 [3, 4], are observed below 500 cm−1 for alkali halides [5]. Similarly, the phonon modes of aromatic hydrocarbon crystals are observed below 150 cm−1, corresponding to librational lattice vibrations. The intermolecular vibrations of water are detected at 780, 450, 175, and 60 cm−1 due to hydrogen bonding interactions [6].
3
These low-frequency motions were conventionally studied with a scanning Raman spectrometer, which is composed of a large multiple monochromator and a single-channel detector such as a photomultiplier tube (PMT) [7]. More recently, the progress in ultrafast lasers has enabled the measurement of low-frequency Raman spectra in the time-domain by using the optical Kerr effect (OKE) [8, 9]. Both these spectroscopic techniques employ the single-channel detection for obtaining a spectrum. It needs to scan the grating of a monochromator or to change the time delay in a step-by-step fashion. The single-channel detection is not suitable for measuring a rapidly changing substance, because it cannot simultaneously detect a wide range (>1000 cm−1) of a spectrum and requires a long time for obtaining the whole spectrum.
The low-frequency motions, especially phonon modes in crystals, are also measured by far-IR (or THz) spectroscopy [10, 11], hyper-Raman spectroscopy [12-15], and neutron inelastic scattering [16]. Far-IR spectroscopy, both in the frequency-domain and the time-domain, employs the single-channel detection. Although hyper-Raman and neutron scattering can be detected by a multichannel detector, they need a very long acquisition time due to the small scattering cross sections. Thus, it is not easy to apply these techniques to the fast tracing of low-frequency motions.
I-3. Multichannel detection in Raman spectroscopy
There is no doubt that the use of an optical multichannel detector such as a charge coupled device (CCD) camera is one of the greatest innovations in Raman spectroscopy [17]. Multichannel detectors have more advantages compared with single-channel detectors in Raman spectroscopic measurements. They enable us to detect spatially dispersed signals simultaneously without scanning the grating of a spectrograph. A wide range of Raman
4
spectrum can be recorded in a short exposure time (<0.1 sec) with high spectral reproducibility. Furthermore, because the fluctuation of the excitation laser does not affect the observed spectra, the relative band intensities in the Raman spectra are highly reliable. If one can fully utilize the advantages of multichannel detection, Raman spectra of a short-lived transient species may be obtained.
In order to apply the multichannel detection technique to low-frequency Raman spectroscopy, immense Rayleigh scattered light should be selectively reduced. Typically commercial notch filters have a broad Rayleigh band rejection of >200 cm−1 (figure I-1). It rejects not only Rayleigh scattering but also low-frequency Raman scattered signals.
A number of studies related to multichannel low-frequency Raman measurements with the use of narrow bandwidth or sharp cutoff Rayleigh scattering elimination filters have been reported recently. Using multiple holographic notch filters [18] and a chevron-type dielectric filter [19] enable Raman measurements down to 28 cm−1 and 20 cm−1, respectively. A
Figure I-1. Transmittance spectrum of a commercially available notch filter
5
multichannel Raman spectrometer with a zero-dispersion double grating filter could measured a Raman spectrum down to 5 cm−1 [20], but only the Stokes side of the spectrum was recorded.
II-4. Vapor filters for Rayleigh scattering elimination
Absorption bands of atomic or molecular vapor have been used for eliminating strong Rayleigh scattering (table I-1). The Rayleigh scattering elimination efficiency of these bands is as high as 10 O.D. with Doppler-limited narrow bandwidth. If a certain vapor absorption band has exactly the same wavelength as that of the excitation light, it can be regarded as an ultra “notch” filter. The use of a vapor filter was first reported by Rasetti in 1930 [21]. The author used mercury vapor for eliminating the Raman excitation light, which was the 253.7 nm resonance line of a mercury arc lamp. After the development of laser Raman spectroscopy, many other vapor filters were combined with various lasers for this purpose.
Table I-1. Absorbing atom/molecule and excitation source combinations Atom or
Molecule
Wavelength (nm)
Excitation source
Hg 253.7 Mercury arc [25], Frequency-doubled dye laser [36]
Pb 283.3 Frequency-doubled dye laser [22]
Cs 388.9
852.0
Frequency-doubled Alexandrite laser [22] Diode laser [23]
I2 514.5
532.0
Argon ion laser [24-31]
Frequency-doubled Nd:YAG laser[32-34]
Ba 553.7 Dye laser [22]
Na* 589.0 Dye laser [35]
K 769.9 Ti: sapphire laser [36]
Rb 780.0 Alexandrite laser, Ti: sapphire laser [37-39]
Ce 894.4 Ti: sapphire laser [38]
6
Devlin and co-workers used iodine vapor absorption bands for eliminating an Ar-ion laser line (λ = 514.5 nm) in 1971 [24]. Two electronic vibrational-rotational absorption lines of iodine vapor were selected [40]. One is the transition from the 12th rotation level of the zeroth vibrational level of the ground electronic state (X 1∑ ) to the 11throtational level of the 43rd vibrational level of the (B3∏ ) electronic excited state, i.e. 0-43 P(12). The other is 0-43 R(14) between the same electronic levels. For iodine vapor to work as a filter for the Rayleigh scattered argon-ion laser light, the laser must be single moded and the frequency of the single mode need to be accurately adjusted to 19429.82 cm−1. Single moding of the laser is necessary because of the extreme narrow width of the iodine rotational line (250–300 MHz or about 0.01 cm−1).
In the 1970s and 1980s, the iodine vapor filter was frequently used for low-frequency Raman or Brillouin scattering measurements, combined with a single-channel spectrometer such as a single/double monochromator or a Fabry-Pérot interferometer [28-34]. Wall and co-workers reported the use of an iodine vapor filter with multichannel detection [28-30]. They used a double monochromator converted to a polychromator and an intensified photodiode array for investigating surface enhanced Raman scattering from −60 to +60 cm−1. The iodine vapor filter has not been used frequently after the 1990s, because the vibrational and rotational structure of the I2 vapor absorption causes spiky artifacts appearing on the
observed spectrum. The elimination of these artifacts was troublesome when the filter was used with a scanning spectrometer [25]. In 2009, Okajima and Hamaguchi revived the iodine vapor filter and developed a novel low-frequency Raman spectrometer by combining the filter with a multichannel detector [31]. Before this revival, the I2 vapor filter has been used mainly
for measurements of filtered Rayleigh scattering in flow field imaging [32, 33] and airborne lidar [34] instead of low-frequency Raman measurements.
7
The use of alkali metal vapor was reported in the 1990s [35-39]. Unlike iodine vapor, metal vapor has no vibrational and rotational structures in the absorption lines and does not give extra spiky artifacts in the spectra. Thus, intensity correction of the observed spectrum became quite easy. The only but most serious experimental problem in using metal vapor filters is the existence of strong resonance fluorescence at the absorption wavelength. Quenching gas such as Ar or N2 should be added into the filter; otherwise the fluorescence
interferes with the low-frequency Raman spectrum to a great extent. However, the quenching gas makes the absorption lines broader, so elimination bandwidth of the metal vapor filters is typically larger than a few cm−1 [38].
I-5. Content of this thesis
The rest of this thesis is organized as follows. In Chapter II, I describe the multichannel low-frequency Raman spectrometer constructed in the present study. The working principle of this technique is mentioned in detail. In addition, the major components of the apparatus such as the iodine vapor filter, single-mode Ar-ion laser, and multichannel detector are described. The performance and advantages of the apparatus are also discussed in comparison with single-channel detection. In Chapter III, the application of the constructed low-frequency Raman system to the crystal polymorphs and melting process of 1,1′-binaphthyl is presented. The Raman spectra of the two crystal forms are compared especially for the spectral range below 200 cm−1, which provide useful information about intermolecular interactions between 1,1′-binaphthyl molecules in the crystal structure. We also discuss the relationship between the Raman spectra of the two crystal forms and their crystal structures. Finally, I conclude my master work in Chapter IV and discuss the future perspective of low-frequency Raman spectroscopy in order to improve the spectral quality for better elucidation of low-frequency Raman spectra.
8
Chapter II
Construction of a Fast Multichannel
Low-Frequency Raman Spectrometer
9
II-1. Introduction
The low-frequency region (<200 cm−1) of Raman spectra contains a wealth of information on intermolecular vibrations and lattice vibrations of molecular and ionic compounds in the condensed phase. In order to investigate the change of the crystal structure during the melting process, it is important to record a low-frequency Raman spectrum in a short measurement time (<1 sec). By using a multichannel detector such as a charge coupled device (CCD) camera, fast Raman measurements become possible. In the present study, we have constructed a fast multichannel Raman spectrometer, which can cover a wide spectral region (−200–1100 cm−1). An absorption band of iodine vapor, having as narrow as 0.03 cm−1 bandwidth, is selected to eliminate gigantic Rayleigh scattering.
As mentioned in Chapter I, the iodine vapor had been used with a single-mode Ar-ion
Two of the I2 vapor
absorption bands
Laser cavity modes Laser gain profile
Etalon loss curve ~500 MHz
(~0.02 cm-1)
Figure II-1. Laser gain curve and cavity modes of an Ar-ion laser.
One of the absorption bands of iodine vapor is extremely close to the laser gain maximum (1.5 GHz higher).
10
laser in the 1970–80s for eliminating strong Rayleigh scattering [25-30]. Two of the iodine vapor absorption bands lie very close (1.5 GHz higher) to the gain maximum of an Ar-ion laser (figure II-1). These two absorption bands have a broader bandwidth (~500 MHz) than one laser cavity mode (40 MHz). Hence, The Rayleigh scattered light can be removed by this I2 absorption band with high efficiency. Unfortunately, this excellent ability of I2 vapor to get
rid of Rayleigh scattering has rarely been used for Raman measurements because the superfluous artifacts caused by I2-vapor absorption structure appear in the observed Raman
spectra.
Not only the Argon ion laser but other single-mode laser can be combined with the I2
vapor filter, because the iodine vapor has many vibronic transitions in the visible region [41, 42]. For the measurements of filtered Rayleigh scattering in flow field imaging [32, 33] or airborne lidar [34], a single longitudinal mode frequency-doubled Nd:YAG laser is combined with the iodine vapor filter (figure II-2). Also, a large number of the iodine vapor absorption bands are used as the spectral standard. Some of these absorption bands are used for stabilizing the frequency of various single-mode lasers; both gas lasers such as Ar-ion (514.5 nm) [43], Kr-ion (568.2, 530.9, 520.8 nm) [44], He-Ne (632.8 nm) laser [45-47], and solid-state lasers such as a frequency-doubled Nd:YAG (532.0 nm) [48, 49]. The versatility of the iodine vapor filter has been revealed above.
In what follows, the constructed low-frequency Raman spectrometer, which is composed of a single-mode Ar-ion laser, an iodine vapor filter, a polychromator and a CCD camera, is fully discussed. In addition, the advantages of the multichannel detection, especially high wavenmber reproducibility (<1 cm−1) and short measurement time (<1 sec), are also shown.
11
★ ★
(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.
(Taken from Gregory S. Elliott and Thomas J. Beutner, Process in
Aerospace Science 35 799 (1999).)
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.
12
II-2. Multichannel low-frequency Raman spectrometer using a
single-mode Ar-ion laser (514.5 nm) and I
2vapor filter
The fast multichannel low-frequency Raman system used in the present study was originally developed by Okajima and Hamaguchi [31] and subsequently constructed as a new setup at NCTU by us. In this section, the combination of a multichannel Raman spectrometer and an I2vapor filter is described. A single-mode Argon ion laser is used as the excitation source, which is the same as the previous low-frequency Raman studies [25-30] with an I2 vapor filter. The
advantages of this technique are also discussed.
II-2-1. Experimental setup
Iodine vapor filter
Figure II-3 shows the cylindrical cell (purchased from Sacher Lasertechnik) we used. It is made of glass with a diameter of 2.5 cm and a length of 10 cm, and contains iodine solids. Few milligrams of iodine are placed inside the cell and sealed at a high vacuum level (<10−3 torr). The purity level of the natural elements used in manufacturing I2 vapor cell is higher
than 98%.
10 cm
2.5 cm
Figure II-3. Iodine vapor filter used in the present study. (Purchased from Sacher Lasertechnik )
13
The iodine inside the cell is in the gas-solid equilibrium, and its vapor pressure depends on the temperature. Figure II-4 shows the pressure of iodine vapor under the gas–solid equilibrium estimated from the following equation [50].
log ( ) = 3.36429 − (
..
)
(311.8≦T≦456) (II-1)where P is the vapor pressure of the I2 vapor filter (bar), T is the temperature inside the cell
(K). The vapor pressure increases almost exponentially with the temperature. A small fluctuation in the filter temperature causes a large change in the vapor pressure, which is proportional to the transmittance of the vapor filter. Therefore, the cell temperature should be highly stabilized during the experiment. In our I2 vapor filter, we used a rubber heater, which
wraps the whole glass cylinder in order to elevate the filter temperature (figure II-5). A chromel-alumel thermocouple placed between the rubber heater and filter is used to monitor the cell temperature. In each experiment, the cell temperature is kept at 95 °C and regulated within ±1 °C during the measurement.
14
Single-mode Ar-ion laser
A water-cooled Ar-ion laser is used as the excitation source (Beam Lok 2060 with Z-Lok option, Spectra Physics). Figure II-6 shows a schematic diagram of this laser. An etalon consists solely of two partial reflecting surfaces with parallel alignment to each other. When it is inserted within a laser cavity, the partial reflectors create multiple, overlapping beams, which are directed out of the cavity by an intentional small tilt of the etalon with respect to the laser beam. Because of the “destructive interference” of the reflected beams at particular frequencies, almost no light actually comes out from the cavity. However, once the off-peak mode beams are reflected out of the cavity, the single-mode laser line is generated.
Because the cavity length of the laser is 1.1 meters, the mode spacing of each adjacent cavity modes is about 136 MHz. The full width at half maximum (FWHM) of the single-mode laser line is less than 40 MHz, which is much narrower than that of the iodine vapor absorption (~1 GHz). In order to fine tune the single-mode laser line to the I2 vapor absorption band, the
single cavity longitudinal mode located at the etalon loss minimum is selectively operated (figure II-1). By slightly changing the etalon temperature, the etalon loss minimum can be
I
2solids
Rubber
heater
(a)
(b)
Glass wool
Figure II-5. Photographs of the I2 vapor filter were taken at different filter temperatures
(a) 24 °C (room temperature) (b) 95 °C
15
tuned, and one longitudinal mode hops to next mode (mode-hopping). Also, the cavity length can be changed by slightly adjusting the piezoelectric (PZT) stage of the output coupler, a minor shift of the laser frequency (<100 MHz). In the experiments, both the etalon temperature and the PZT stage of the output coupler are controlled so that the observed Rayleigh scattering of the sample is maximally reduced.
Apparatus
The constructed apparatus is schematically shown in figure II-7. The above-mentioned argon ion laser (514.5 nm) is used as the excitation source and the scattered light is collected at 90-degree by a camera lens (f = 50 mm, f/1.2, Nikon). The collimated light is passed through the I2 vapor filter, which filters out most of the elastically scattered light. The transmittance is
then focused onto the entrance slit of a polychromator (f = 500 mm, f/6.5, SP-2558, Princeton Instruments), and detected by a back-illuminated, deep-depletion, liquild-N2 cooled CCD
detector (Spec-10:100, Princeton Instruments) with 100×1340 pixels operating at −120 °C. The entrance slit width was typically set to 50 μm. A 1200 grooves/mm grating was used to cover a wide spectral range (>1300 cm−1) with a high spectral resolution of 2.7 cm−1.
16
The white light from a tungsten lamp is monitored before and after each measurement to correct the intensity profile of observed Raman spectra. Furthermore, the transmittance of the white light spectrum can be used to remove the superfluous I2-vapor absorption structure.
In order to introduce the white light into the filter with good position reproducibility, a flipper mirror and an aperture need to be placed in front of the I2 vapor filter.
II-2-2. Results and Discussion
The intensity correction of the observed Raman spectra
To demonstrate how the intensity correction works, the Raman spectrum of carbon tetrachloride has been measured by using the apparatus in the present study. The observed and the intensity corrected Raman spectra are shown in figure II-8. The observed spectrum shows some superfluous artifacts due to I2-vapor absorption structure. The peak positions and
17
intensities of these artifacts precisely reproduce, indicating that they are not random noises but artifacts inherent to the I2 vapor filter. These unwanted artifacts can be removed simply
through dividing the observed spectrum by the white light spectrum.
It is noteworthy that the intensity and artifacts correction with scanning spectrometers is much more difficult than that with multichannel spectrometers because the wavenumber reproducibility of a scanning spectrometer is only about 1 cm−1. The sharp spikes caused by I2
vapor absorption appearing in the Raman spectrum and the white light spectrum require high spectral reproducibility for a rigorous correction. Thus, simultaneously detecting a wide spectral range (>1300 cm−1) of spectrum without scanning the grating of the spectrometer is desirable [25]. In the case of using multichannel spectrometers, which have high wavenumber reproducibility, the intensity and artifacts correction is simply done as mentioned above. The real power of the combination of I2 vapor filter and multichannel spectrometer is now
18
T=302 K
Figure II-8. Intensity correction of the Raman spectrum of CCl4.
(a) White light transmitted through the I2 vapor filter.
(b) Observed Raman spectrum.
(c) Corrected Raman spectrum (The inset is the fitting result for estimating the temperature, T = 302 K).
Laser power at the sample point was 4.5 mW, the spectral resolution was 2.7 cm−1, and the exposure time was (a) 50 sec and (b) 60 sec
19
It is possible to determine the sample temperature by measuring both the anti-Stokes and Stokes Raman spectra. The temperature at the sample point is estimated from the intensity ratio of the anti-Stokes and Stokes Raman lines using the equation [51] :
(anti-Stokes) (Stokes)
=
( ) ( )e
⁄ (II-2)where is the wavenumber of the laser line, is the vibrational frequency of a band of the solvent or sample, h is Planck’s constant, c is the speed of light, is Boltzmann constant, and T is absolute temperature of the sample. This equation is based on the canonical distribution; hence it is not applicable when anti-Stokes lines are very weak. It should be noted that this equation is valid only for the spectra obtained under off-resonance conditions. The temperature of CCl4 is estimated to be 302 K (the inset of figure II-8(c)), which is close
to room temperature (298 K). Such temperature estimation is quite reliable if and only if multichannel detection is employed because both anti-Stokes and Stokes regions are measured simultaneously and the resulting anti-Stokes/Stokes intensity ratio is not affected by the intensity fluctuation of the laser.
Performance check: Low-frequency Raman measurement of
L-cystine
In this section, we demonstrate the ability of our apparatus to measure low-frequency Raman spectra with a sufficiently exposure time by using L-cystine, which shows a Raman
band at ~10 cm−1. The Raman spectrum of micro-crystalline powder L-cystine is shown in figure II-9. L-cystine was purchased from Wako Pure Chemical Industries (99.0% pure) and
used without further purification. In order to obtain the corrected Raman spectrum of
L-cystine, the intensity correction procedure mentioned previously was used. Due to high
Rayleigh scattering elimination efficiency of the I2 vapor filter, both the Stokes and
20
gap (−5 to +5 cm−1). The most intense Raman band of L-cystine at 498 cm−1 has nearly the
same area intensity as that of the remaining Rayleigh scattering band at 0 cm−1.
The constructed spectrometer enables us to record a wide spectral range (1300 cm−1), including not only lattice vibrational modes but also intramolecular vibrational modes simultaneously. The low-frequency region of the Raman spectra of the L-cystine is shown in
figure II-10. The ±9.8 cm−1 band, which is usually used as a test of the performance of a low-frequency spectrometer, can be clearly observed. Because the present spectrometer employs a multichannel detection, it is not necessary to scan the spectrograph during the Raman measurement. Hence, the measurement time is, in principle, determined only by the exposure and read-out times of the CCD camera. The CCD read-out time in our case is 0.14 sec. As shown in figure II-10, a 0.2-sec exposure time is enough to measure a high S/N low-frequency Raman spectrum of L-cystine. Thus, it is possible to record a Raman spectrum
21 (a) Observed spectrum (b) Filter transmittance (c) Corrected spectrum
Figure II-9. Raman spectra of L-cystine.
(a) Observed Raman spectrum of L-cystine. (b) Transmittance spectrum of I2 vapor filter.
(c) Intensity corrected Raman spectrum of L-cystine (= a/b).
The measurement was done with 2.7 cm−1 spectral resolution, 10 sec exposure, and 70 mW laser power.
22
(a) 0.2 sec
(b) 1.0 sec
(c) 10 sec
Figure II-10. L-cystine Raman spectra in the low-frequency region.
The measurement was done with different exposure times: (a) 0.2 sec
(b) 1.0 sec (c) 10 sec
23
Comparison of I
2vapor filter with commercial notch filters
The Raman spectra of L-cystine measured with three different filters are compared in figure II-11. As mentioned in Chapter I, typical notch filters having a broad Rayleigh rejection band (>200 cm−1) can be seen in the blue line spectrum. Though the green line spectrum detects Raman bands below 100 cm-1, the Raman bands at 67 and 78 cm−1 give incorrect relative intensity. A possible reason is that these two Raman bands were detected at the edge of the Rayleigh rejection bandwidth. The red line spectrum measured with the iodine vapor filter showing ±9.8 cm−1 bands of L-cystine has already been discussed before. The excellent Rayleigh scattering elimination efficiency of the iodine vapor filter has been demonstrated here.
Figure II-11. Comparison of the Raman spectra of L-cystine measured with three different filters. Laser power at the sample point was 20 mW, the spectral resolution was 2.7 cm−1, and the exposure time was 1 sec.
24
Artifacts due to the resonant fluorescence of iodine vapor
Although no fluorescence quencher is added into the I2 vapor filter, resonant
fluorescence of I2 vapor makes no appreciable interference in the Raman spectrum of
L-cystine as shown in figure II-9. This result contrasts with metal vapor filters such as
potassium vapor-containing cell, in which resonant fluorescence is so strong that the addition of a quenching gas is necessary [36].
Fluorescence of iodine vapor is observed in the spectrum only when the strong laser light directly comes into the I2 vapor filter. It occasionally happens when the laser light is
reflected by the wall of a glass capillary and enters the collecting optics. Figure II-12(a) shows the Raman spectrum of L-cystine measured under such conditions. Extra bands other
than the Raman bands of L-cystine are observed in this spectrum. The band at −13 cm−1,
which is close to the Rayleigh light, is due to spontaneous emission of the Ar-ion laser. Other bands (212, 424, 637, and 844 cm−1) result from the resonant fluorescence of the iodine vapor [52]. To reduce these annoying signals, the strong reflected laser light should be blocked. In a 90-degree scattering system, the removal of the reflected light is not so difficult. By placing an aperture in front of the I2 vapor filter (figure II-7), the removal of the reflected light can be
achieved easily. If a back or forward scattering geometry is employed for the measurement, quenching the fluorescence of the iodine vapor becomes much important. As a result, the fluorescence quencher may be required to be added to the filter similarly to the case of the metal vapor filters.
25
Raman spectral change caused by mode-hopping
During a long hour measurement, the selected laser cavity mode sometimes hops to another cavity mode (mode-hopping), because of the slight change of the etalon temperature. Once the mode-hopping occurs, the laser-line elimination of the iodine vapor filter become inefficient, and the observed spectrum drastically change in the low-frequency region. Figure II-13 shows the L-cystine Raman spectra using different cavity modes for measurement.
When the mode-hopping occurs by only 0.3 GHz away from the optimum cavity mode, the observation of low-frequency Raman bands below 30 cm−1 is found to be very difficult.
Ar
I
2I
2I
2I
2(a)
(b)
Figure II-12. Raman spectra of L-cystine obtained in the following two situations:
(a) The laser light reflected by the glass capillary came into the I2 vapor
filter.
(b) The reflected laser light was blocked by the 5 mm aperture.
The mark“Ar” stands for spontaneous emission of the Ar-ion laser,“I2”
26
Hence, in order to avoid mode-hopping, it is crucial to keep the environmental temperature stable.
In particular, when the laser is operated for over five hour, it is necessary to confirm that the selected cavity mode exactly coincides with the iodine absorption band before starting the measurement. It can be easily confirmed by examining whether the Rayleigh scattering intensity still remains minimum or not.
(a) 1.91 GHz higher than the gain maximum (b) 1.77 GHz higher (c) 1.64 GHz higher (d) 1.50 GHz higher
Figure II-13. Raman spectra of L-cystine measured by tuning the frequency of the
single-mode Ar-ion laser.
(a)1.91 GHz, (b) 1.77 GHz, (c) 1.64 GHz, and (d) 1.50 GHz higher than the laser gain maximum.
27
Chapter III
Real-Time Tracing of the Melting Process
of the Two Distinct Polymorphs of
28
III-1. Introduction
1,1′-binaphthyl and its derivatives represent a special class of biaryl molecules. It is well-known for their application as chiral recognition receptors and chiral catalysts such as 2,2'-bis(diphenylphosphino)-1,1'-binaphthyl (BINAP) [53-55]. Biphenyl is the simplest biaryl molecule, in which two aromatic rings are linked via a C–C single bond. Its rotation barrier around the phenyl–phenyl bond in the gas phase was found to be ~1.4 kcal/mol [56]. 1,1′-binaphthyl, in which two naphthalene moieties are linked via a C–C single bond, has a substantially large increased rotation barrier of 23.5 kcal/mol (∆ ) [57]. The large rotation 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. This fact allows the isolation of the optically active 1,1′-binaphthyl enantiomers. The dissymmetry of 1,1′-binaphthyl is molecular in nature, and enantiomeric interconversion is made possible simply by rotation about the interannular bond
(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)
29
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 [58]. He found that racemic 1,1′-binaphthyl underwent spontaneous resolution to generate the optically active R or S enantiomer when this compound crystallized from the melt.
The chiral 1,1′-binaphthyl molecule contains a chiral axis other than 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). The designations are based on the same Cahn–Ingold–Prelog priority rules used for tetrahedral stereocenters. In the crystalline state, 1,1′-binaphthyl is known to exist as at least two forms [59]: the cisoid configuration or lower melting point (148 °C) form and the transoid configuration or higher melting point (161 °C) form. Brown et al. [60] 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 [61] confirmed the dihedral angle θ to be 68.6° (figure III-2). Badar et al. [59] suggested, on the evidence of infrared spectroscopy, that the high melting point form adopts a transoid configuration, which was later confirmed by X-ray analysis [62] to have a dihedral angle of 103.1°. The cisoid form, whose structure was determined by Kerr and Robertson [61], consists of racemic monoclinic crystals (space group C2/c). The lattice unit (Z = 4) of this
θ = 68.6° m.p. 148 ℃ θ = 103.1° m.p. 161 ℃ Cisoid configuration Transoid configuration Figure III-2. Schematic diagram of the two crystal forms of 1,1′-binaphthyl.
30
achiral crystal is comprised of two R enantiomers and two S enantiomers. On the other hand, Kress et al. [62] reported the crystal structure of the transoid form as an optical active compound which belongs to the tetragonal system of space group P42121 with either R or S
enantiomer in a unit cell (Z = 4).
The molecular structural change of the ground and excited states of 1,1′-binaphthyl in the solution phase has also been studied by absorption spectroscopy [63, 64], fluorescence spectroscopy [65, 66], and Raman spectroscopy [66-68]. The molecule has also been the subject of many theoretical studies which have provided a wide range of estimates of the dihedral angle adopted by the 1,1′-binaphthyl molecule in its most stable conformations [69-75].
Raman spectroscopic studies on the two crystal forms of 1,1′-binaphthyl was first reported by Lacey et al [67] in 1986. However, their Raman spectra in the low-frequency region (<150 cm−1) have not been reported yet. They showed a wealth of information about lattice vibrations of the crystal polymorphs. In addition, low-frequency Raman spectroscopy is sensitive to some intramolecular vibrations which have large amplitude, such as the C–C torsional motion of crystalline biphenyl [76]. Since the Raman intensity is proportional to the thermal average of the vibrational amplitude, large vibrational amplitudes will lead to large Raman intensities. Therefore, in the present study we applied the constructed low-frequency Raman spectrometer to 1,1′-binaphthyl, which has C–C torsional vibrations as well. The melting process of its two crystal forms has also been studied. We aim to understand the interplay between the intermolecular interactions and crystal structures of the two distinct polymorphs of 1,1′-binaphthyl (i.e., the cisoid and transoid forms) based on time-resolved low-frequency Raman spectra.
31
III-2. Experimental
Sample preparation
Solid 1,1′-binaphthyl of 98% purity, which turned out to be dominated by the cisoid form, was purchased from Tokyo Chemistry Industry Co., LTD. It contains unknown impurity which causes fluorescence appearing in the Raman spectrum with 514.5 nm excitation. We carried out crystallization with a procedure similar to that previously reported [58, 77, 78]. A schematic diagram of the crystallization procedure for obtaining two crystal forms of 1,1′-binaphthyl is shown in figure III-3. We placed a container with 500 mg of the sample in a bath of silicone oil and heated it to 180 °C. The melt was kept in this temperature for 20 min, ensuring that the whole sample had racemized completely. We then cooled the melt to 153 °C and allowed it to crystallize. In the first round (figure III-3 (a)), the transoid form was found
180 ℃ 153 ℃ 180 ℃ 153 ℃ A B C D A (a) (b) Liquid phase B C D Stir Stir Heating Heating Cooling Cisoid Transoid Transoid + impurity 98% pure sample Cooling
Figure III-3. Purification of commercially obtained 1,1′-binaphthyl and preparation of (a) cisoid form and (b) transoid form.
C
32
to be contaminated by the unknown impurity at the bottom of the container (marked as B in the figure). It was confirmed by our Raman apparatus that the product at the bottom gives stronger fluorescence compared with the sample before purification. Additionally, the pure cisoid form was observed as a result of the evaporation and recrystallization of 1,1′-binaphthyl from the vapor in the top part of the reactor (marked as C). This behavior is consistent with the vapor-solid process reported previously, where a polymorph transformation was observed in this compound [62]. In the second round (figure III-3 (b)), we used the pure cisoid form of 1,1′-binaphthyl obtained in the first round as the starting material to prepare the pure transoid form. The prepartion method was the same as in the first round except for using a magnetic stir-bar in the container. Sainz-Díaz et al. reported that more transoid form formed if the reaction went under the condition of stirring [77]. They found that chiral symmetry breaking can be induced by stirring the melt as it crystallizes. For each crystal form, a small (~1 mm) piece of the pure crystal was sealed in a 1.2-mm glass capillary for rapid heating Raman measurement discussed below.
Heating apparatus
As shown in figure III-4, the hot air rework system (No.FR-802, HAKKO) with the power of 570 W was used for heating the samples from room temperature to the melting points rapidly. Its set temperature can be changed from 100 °C to 500 °C with an accuracy of ±4 °C. In the case of 1,1′-binaphthyl, the melting points of its cisoid form and tranoid form are 148 °C and 161 °C, respectively. The heating apparatus enables us to melt each crystalline form within 10 sec. Low-frequency
2.5 mm nozzle
Figure III-4. Photograph of the heating apparatus.
33
Raman spectral change during the melting was continuously measured every 0.2 sec.
III-3. Results
Raman spectra of the two distinct polymorphs of 1,1′-binaphthyl
The Raman spectra of two purified crystals were measured at room temperature as shown in figure III-5. For a comparison sake, the Raman spectrum of the commercially obtained 1,1′-binaphthyl was also recorded (the inset of figure) clearly seen from figure III-5
98% pure sample
Figure III-5. Raman spectra of the two forms of crystalline 1,1′-binaphthyl.
The inset is the Raman spectrum of commercially obtained 1,1′-binaphthyl. The spectra were measured with 30 mW laser power, 2.7 cm−1 spectral resolution, and 0.2 sec exposure.
34
that strong fluorescence is observed for the commercial crystal with the spectral pattern being almost the same as the cisoid form, whereas no appreciable fluorescence is detected for both cisoid and transoid forms after purification. In addition, the spectral pattern in the region of >150 cm−1 of the two crystal forms is consistent with the literature [67, 68]. The Raman intensities in this region appear to be similar. It should be noted that the bands in the higher-frequency region is much weaker in intensity than the bands in the low-frequency region <150 cm−1, which indicates that these low-frequency vibrational motions accompany large polarizability changes than the intramolecular vibrations.
Figure III-6 shows the Raman spectra of the two polymorphs of crystalline 1,1′-binaphthyl in the −200–+200 cm−1 region. The Raman bands below 200 cm−1 both in Stokes and anti-Stokes sides have been recorded simultaneously. For the transoid form, the
35
±26 cm−1 bands, which are very close to the Rayleigh scattered peak, give the strongest intensity in the whole spectrum. On the other hand, the most intense band for the cisoid form is the doublet around 105 cm-1, which may arise from a splitting of one band into two peaked at 100 and 110 cm−1. The three Raman signals in the range of 46 to 90 cm−1 appearing in both crystal forms give similar peak positions. However, their band intensity in the transoid form is quite strong compared with that of the cisoid form. Such detailed information on the low-frequency Raman spectra of the crystal polymorphs of 1,1′-binaphthyl has been obtained for the first time.
It is noteworthy that there are two small peak shoulders in the transoid form near 20 cm−1 and 44 cm−1, which may be due to the lattice symmetry. These shoulders are so small that they can hardly be separately recorded. Thus, in order to investigate these shoulders appearing in the present data, high spectral resolution or polarized Raman measurements might be required in the future.
Raman spectral change during the rapid heating
Figure III-7 shows the low-frequency Raman spectra of a rapidly heated crystal of the transoid 1,1′-binaphthyl. Although each spectrum was recorded with as short as 0.2 sec exposure time, a high signal to noise ratio has been achieved. One of the intramolecular vibrations located at 295 cm−1 has been used for normalizing all the spectra. The spectral resolution for this measurement (~2.7 cm−1) is not high enough to separate the 20 cm−1 shoulder band from the band at 26 cm−1. As the temperature goes up, the low-frequency Raman bands gradually lose their band shape and are hardly distinguishable 8 second after heating. Moreover, the central band at 0 cm−1 gets broad enedand the S/N ratio of the spectra become lower at 7.6 sec. It may indicate that the liquid phase of 1,1′-binaphthyl appears at this moment. If we take a closer look at the two bands at 55 cm−1 and 66 cm−1, they are
36
getting closer to each other as the temperature increases. However, the spectral pattern remains the same for the 79 cm−1 band before 7.4 sec. It is likely that these two bands are associated with quite similar vibrational motions and hence coupled to one another.
Tracing of the melting process of the cisoid form of 1,1′-binaphthyl has also been done and the result is shown in figure III-8. Normalization procedure was applied to all the spectra as mentioned before. Similarly to the previous experiment, the crystals melt at 6 sec judging from the 0 cm−1 band broadening and the reduced S/N ratio. We observe that the relative intensity of the two peaks at 100 and 110 cm−1 (strong doublet band) varies with increasing the temperature. It seems that the 100 cm−1 band intensity decreases upon melting. It is also clear that the 75 cm−1 band disappears soon while the bands at 57 and 65 cm−1 remain nearly unchanged. It is in contrast with the result in the transoid form.
37
Figure III-7. Low-frequency Raman spectra of the transoid form with rapid heating. Each spectrum was measured with an exposure time of 0.2 sec, and the laser power and spectral resolution were 36 mW and 2.7 cm−1, respectively.
38
Figure III-8. Low-frequency Raman spectra of the cisoid form with rapid heating. Each spectrum was measured with an exposure time of 0.2 sec, and the laser power and spectral resolution were 31 mW and 2.7 cm−1, respectively.
39
III-4. Fitting analysis
Although there are a number of low-frequency Raman bands below 200 cm−1 in both crystal forms of 1,1′-binaphthyl, both the Stokes and anti-Stokes sides can be fitted well assuming Lorentzian band shapes. The sample temperature as well as the peak position of each low-frequency Raman band was estimated by means of the fitting. Details of the procedure are described below.
Lorentz function was used for fitting each low-frequency band including the two peak shoulders for the transoid form. Since the Stokes and anti-Stokes sides of each Raman band have the same peak position and band shape, the Lorentz function used for the fitting should be symmetrical with respect to the 0 cm−1 Raman shift, like the following function.
( ) ∆
+
( ) ∆where is the Raman shift, is the peak position, is the band intensity, and ∆ is the bandwidth. This function has two peak positions at + (Stokes side) and − (anti-Stokes side). If is zero, equation III-1 reduces to one single Lorentz function located at 0 cm-1, representing the Rayleigh scattering.
The intensities of the Stokes and anti-Stokes bands are not the same, hence an important factor must be included in equation III-1. As we mentioned in Chapter II, the anti-Stokes/Stokes intensity ratio is related to the Boltzmann distribution in the vibrational energy levels (equation II-2). Considering the Stoke and anti-Stokes intensity difference results from the thermal effect, it can be compensated by multiplying the symmetrized Lorentz function (equation III-1) and the Bose-Einstein factor ( ( )) [79] :
40
( ) = 1 + exp −
Where c is the light velocity, h is the Planck’s constant, kB is the Boltzmann constant, and T is
the sample temperature. Note that the extra fitting parameter required is only the temperature (T). In the fitting, the temperature parameter of all the low-frequency Raman bands is assumed to be the same.
Table III-1 summarizes the fitting parameters. Symmetrized Lorentz function convoluted with the Bose-Einstein factor was applied to fit the seven low-frequency Raman bands of the cisoid form as well as those of the transoid form in both Stokes and anti-Stokes sides. Twenty-three independent parameters are determined simultaneously for each fitting.
The fitted results of the Raman spectra of the two crystalline forms of 1,1′-binaphthyl are shown in figures III-9 and III-10, they were done under the unnormalized condition. Whether the spectra are normalized or not would not affect the determination of the sample temperature because it is obtained from the relative intensities of the Stokes and anti-Stokes bands rather than from their absolute intensities. The fitting was done in the spectral region below 200 cm−1 except for the Rayleigh gap region (−15–+15 cm−1), but was not possible for the spectra recorded after 7 sec in both cases of the two crystal forms. Judging from the results as shown in figures III-9 and III-10, the low-frequency Raman bands were fitted well in both crystal forms before 7 sec. Though we observe that the Raman band intensity dropped a lot at 6 sec for the cisoid form, the line intensity as well as line shape of these low-frequency bands still can be reproduced fairly well by the fitting. In addition, it is noteworthy that the
Baseline Peak position (i = 1–7) Bandwidth ∆ (i = 1–7) Intensity (i = 1–7) Temperature Table III-1. (III-2)
41
shoulder peaks at 20 and 44 cm−1 in the Raman spectra of the transoid form can also be fitted well.
Based on these high quality fitting results, the sample temperature can be estimated to a high accuracy. Moreover, the peak position of each low-frequency Raman band has also been determined by the fitting. Thus, the sample temperature change and peak position shift of the low-frequency Rama bands will be the main topic in the following sections. And the spectral differences between the cisoid and transoid forms during the heating and their possible explanations will be discussed as well.
42
Figure III-9. Experimental data (red closed circles) of the transoid form and their fitted results (blue solid lines).
43
Figure III-10. Experimental data (red closed circles) of the cisoid form and their fitted results (blue solid lines).
44
III-5. Discussion: Temperature change during the melting
Temperature change of the cisoid form upon heating
How the temperature changes with the heating time in the case of the cisoid conformer of 1,1′-binaphthyl is shown in Figure III-11. Each temperature was estimated from the Raman spectra shown in figure III-8. The estimated temperature before starting heating (<0 sec) is about 315 K with a fluctuation of ±6 K. It is 17 °C higher than the room temperature (298 K) because of the high laser power used. After the heating starts, the sample temperature increases rapidly to over 486 K in 7 sec. By differential scanning calorimetry (DSC) measurement, the melting point of the crystalline 1,1′-binaphthyl can be obtained (figure
Figure III-11. Plot of the estimated temperature versus heating time for the cisoid form. The estimation error is about ±6 K, which is evaluated from the fluctuation before heating (<0 sec).
45
III-12). Figure III-12 clearly demonstrates that the DSC band (146 °C) shifts to higher temperature as the heating rate used increases. Because our heating apparatus is capable of increasing the temperature to 700 K within 30 sec, the crystal may melt at the temperature which is higher than the reported in the literature (421 K).
In the DSC result, not only the 146 °C band but also a much stronger band at 160 °C is observed when a heating rate of 5 °C/min is used. In order to confirm that our cisoid form of crystalline 1,1′-binaphthyl is pure, we performed the DSC measurement with different heating rates. We found that the relative intensity of these two bands was changed if we increase the heating rate. This means that the melting of the cisoid form takes place concomitantly with its transformation into the transoid form. This phenomenon was also reported in the literature as a solid-state process [77]. It may be related to the temperature drop around 5.6 to 6 sec shown in figure III-10. The central peak became broaden exactly at the same moment as already described in figure III-8. It is due to the fact that a rapid transformation from the melting cisoid to the transoid crystal results in an exothermic recrystalization process.
Figure III-12. Differential scanning calorimetry (DSC) measurements of cisoid form crystal with different heating rates.
46
Temperature change of transoid form upon heating
Figure III-13 shows a plot of the estimated temperature versus the heating time for the transoid form. The estimated temperature before heating is about 325 K, which is 10 °C higher than that of the cisoid form. The difference may be caused by different laser power (5 mW) and/or focal point of the laser on the sample. The same tendency as shown in the cisoid form is observed, the sample temperature increases rapidly to over 482 K within 7 sec heating.
Figure III-13. Plot of the estimated temperature versus heating time for the transoid form. The estimation error is about ±8 K, which is evaluated from the
47
The transoid form of 1,1′-binaphthyl, which was prepared by the procedure as described before, are highly pure. Only one DSC band is observed (figure III-14), and the DSC band (161 °C) also shifts to higher temperature as the heating rate increases. Hence, the melting point of the transoid form observed in the Raman experiment should be higher than 161 °C (434 K). However, unlike the case of the cisoid form, there is no obvious temperature drop for the transoid form (figure III-13). And the transoid conformer will not transform back to the cisoid form during heating. It is confirmed by the DSC measurement as shown in figure III-14.
Figure III-14. Differential scanning calorimetry (DSC) measurements of transoid form crystal with different heating rates.
48
III-6. Discussion:
Changes in Low-frequency Raman bands change during heating
Peak position shift caused by the temperature
Figure III-15 display the changes in peak position of the low-frequency bands for the cisoid and transoid forms. The peak shifts of intramolecular vibrations such as 510 cm−1 and 683 cm−1 bands are no more than 1 cm−1. Therefore, the shifts of the low-frequency Raman bands, which are larger than 1 cm−1 should give some special physical meanings.
A plot of peak position of low-frequency Raman bands versus temperature is shown in figure III-16. The selected range of the data points is from 0 sec to the point before melting, which are 7.4 sec for the transoid form and 5.6 sec for the cisoid form. Thus, the thermal behavior of the two crystal forms before melting can be compared. The figure seems to show linear dependence; therefore we can obtain peak positions corresponding to each temperature by simply fitting the data to a line function. The peak position at room temperature (298 K) and before melting can be derived based on the fitting function obtained. Table III-2 lists the peak shift of each low-frequency Raman band upon heating. It is known that the peak shift is related to thermal expansion of a crystal structure. We found that the Raman bands shift to a
Transoid form band ( ) 20 26 44 55 66 79 95
Peak shift (cm−1)
〔 (T=489)− (T=298)〕 −1.1 −1.3 −6.2 −2.9 −3.3 −2.1 −3.8
Cisoid form band ( ) 37 57 65 75 100 110 175
Peak shift (cm−1)
〔 (T=459)− (T=298)〕 −4.6 −5.1 −5 −5.3 −6.8 −5.6 −4
