The use of microreactors and
optofluidic waveguides to
study photochemical reactions
MSc Chemistry - Literature study
By
Amber Jaspars
1
The use of microreactors and optofluidic
waveguides to study photochemical reactions
Literature project - MSc Chemistry – Analytical track
By
Amber Jaspars
UvA/VU
Supervised by:
Daily supervisor: I. Groeneveld First examiner: F. Ariese Second examiner: G. Somsen
2
Abstract
In this literature review, the use of different microreactors and optical waveguides for photochemical reactions is discussed. Photochemical reactions can be carried out in a batch reactor, but these batch reactors are not homogeneous, are hard to irradiate and take a long time before it is fully reacted. With a microreactor, these problems can be solved. Because of the small volume, mixing is faster and more efficient and the sample is more easily irradiated, creating a faster reaction. The microreactor can be irradiated sideways or along the length. For sideways irradiation, mostly chips with microchannels are covered with a glass plated. A light source above the chip irradiates the sample, the light penetrates through most of the sample and the mixture is easily irradiated.
The microreactor can also be irradiated along the length of the reactor, using a liquid core waveguide (LCW). The LCW is like a tube made from a reflective cladding material, the light is reflected within and guided through the LCW. The light can be guided either by total internal reflection. The light is guided if the cladding material has a smaller refractive index than the liquid core, meaning the cladding material should be carefully chosen. Another principle of light guiding is based on interference based waves, multiple layers of material are used as cladding, creating multiple reflections. This principle can be based on photonic bandgaps or on the Fabry–Pérot reflector.
3
Table of contents
1 Introduction ... 4
2 Photochemical reactions and photodegradation ... 5
3 Microreactors and waveguides ... 7
3.1 Micro photoreactors ... 8
3.2 Optofluidic waveguides ... 9
4 Total internal reflection based waveguides... 12
4.1 Teflon AF ... 14
4.1.1 Fabrication ... 15
4.1.2 Detection and Reaction use ... 18
4.2 Aerogel microchannels ... 21
4.2.1 Fabrication ... 22
4.2.2 Detection and Reaction use ... 24
5 Interference based waveguides ... 28
5.1 Bragg fibres ... 29
5.1.1 Fabrication ... 29
5.1.2 Detection and Reaction use ... 30
5.2 Photonic crystal fibres (PCF) ... 31
5.2.1 Fabrication ... 31
5.2.2 Detection use ... 34
5.2.3 reaction use ... 35
5.3 Anti-resonant reflecting optical waveguides (ARROW) ... 38
5.3.1 Fabrication ... 39
5.3.2 Detection and reaction use ... 41
6 Conclusions and outlook ... 42
4
1 Introduction
There is a growing interest in microreactors since the beginning of the 21st century. They consist of a solid support with channels of 10–1000 µm in width and depth [1][2]. The microreactors have several advantages over bulk or batch reactors, due to the small volumes and the narrow channels. For photochemical reaction, a light source for bulk reactors can be put on the outside of the reactor, but it would only irradiate the compounds near the edges, irradiating a small surface in a large volume. When using microreactors for photochemical reactions the main advantage is the surface-to-volume ratio [1][2].
For photochemical reactions in a microreactor, the sample can be irradiated in two ways. The first way is irradiating sideways by using a micro photoreactor. The micro photoreactor consists of channels in a solid plate covered with a transparent material. The light comes from above and due to the narrow channels the light penetrates through most of the reactor and the reaction mixture [1][3][4]. The other way of irradiating the microreactor is along the length by using a liquid core waveguide. A liquid core waveguide is a tube made of a reflective cladding material that reflects the light from within [1][5]. The maximized surface-to-volume ratio in the micro photoreactor enables highly efficient photochemistry, with an improved irradiation and better light penetration through the reaction volume [6]. Because of the better surface-to-volume, there is more light in a smaller volume, the molecules absorb the light easier and the reaction goes faster [7].
This study is for the TooCOLD project, the TooCOLD project has the goal to improve photodegradation research. They want to create an online two-dimensional liquid chromatography system, which is combined by a photodegradation microreactor. The goal of this study is to find what kind of micro photoreactors have been developed and which optofluidic devices or waveguides could or have been used for photochemical reactions. The questions would be if the found microreactors and waveguides could be coupled to a HPLC or other liquid separation methods and what the important parameters would be. This study will not be focused on photodegradation, but will take a broader view and look at photochemical reactions.
In the first chapter, the introduction with the goal of the research is given. The second chapter will give more information about the photochemical reactions, it will discuss the parameters that need to be considered. In chapter three more information will be given about the microreactors and waveguides. The general overview of the differences will be discussed together with the important parameters. Different waveguides with their principle, fabrication and the detection and reaction used will be discussed in chapter four and five. This study will be concluded and a future outlook will be given in chapter six.
5
2 Photochemical reactions and photodegradation
There is a growing interest in photochemical reaction in many areas, including medicine and chemical synthesis [8]. The photochemical reactions differ from the usually used thermal reactions. The photochemical reaction is activated mainly by the absorption of light, while the thermal reaction is mainly activated by the use of heat [9]. With the absorption of a photon, the molecule reaches an electronically excited state. In the excited state, the molecule possesses quite a high internal energy, the electronic structure and the nuclear configuration of the molecule will be different from the ground state [4]. These excited molecules can fall apart, change to new structures, combine with each other or other molecules, or transfer electrons, hydrogen atoms or protons. These photochemical reactions may result in products that cannot be formed by thermally driven reactions in the ground state. This allows photoreactions at very low temperatures, near-zero Kelvin [9].
There are many different photochemical reactions, mostly they are categorised based on the required conditions (homogeneous or heterogeneous) and not on their reaction type [1]. In homogeneous reactions all the reagents are in the same phase or solution [1], [3], [10], [11]. For homogeneous catalytic reactions, the catalyst is dissolved in the reaction solvent. Heterogeneous gas-liquid reactions are between two phases and acquire a gas supply, commonly Photooxygenations or photochlorinations [3][5][6]. However, gasses are difficult to handle in batch conditions. Filling a batch reactor with gas in unpressurized systems results in serious mass-transfer limitations. Although, in a pressurized system it has serious safety concerns when using toxic or reactive gases. A microreactor avoids these problems by using high amounts of gas in each part of the reactor, but it can also be easily pressurized in a safe manner [5]. For heterogeneous solid-liquid reactions, same as for homogeneous reactions where solid intermediates or products are formed, the microreactor is a challenge. The solid particles can interfere with the light, but can also clog in the small channels of the microreactor [5]. When using a heterogeneous catalyst, the catalyst can be immobilised within the reaction channel. Due to the large surface-to-volume ratio, there is a maximised interaction between the sample, the catalysts and the photons [1], [11]–[13]
The TooCOLD project is mainly focused on the photodegradation of products. The goal is to create an online two-dimensional liquid chromatography system, which is combined by a photodegradation microreactor. Photodegradation is beneficial for water and wastewater treatment, however, it is unwanted in many areas, for example, food products, packaging materials or cultural heritage like paintings [14]. Photodegradation is a photochemical reaction, where the exposure to light may cause significant degradation of many materials [15]. It may happen direct or indirect, by direct photodegradation the chemical absorbs the light itself. With indirect photodegradation, the light is absorbed by photosensitizers and the excited photosensitizers generate photoreactants, which can react with the compounds [16]. Most commonly the photosensitizer (like H2O2 or O3) generates singlet oxygen, the singlet oxygen reacts with for example food components, photooxidative degradation. This can influence flavour quality, nutritional quality and toxicity of food products [17]–[19]. The degradation rate depends on different parameters, like pH, temperature, concentration and purity of the chemicals and the concentration of the photosensitizer [20][21].
6 The flow is another important parameter in the photochemical reactions, especially when minimizing them. The reactions can be performed without a flow. The reactor is filled with the reaction mixture before it is irradiated, after irradiation the product can be collected. A detector could measure the progress of the reaction. When performed under a continuous-flow the irradiation time is directly proportional to the continuous-flow rate of the reactor. This makes the optimization faster, because the irradiation time can be easily changed [1][5][11]. A detector could be installed at the outlet of the reactor, it would measure the speed of finding the equilibrium of the reaction rate. When the equilibrium is found, the detector would see if the equilibrium is unstable. A UV-VIS or a fluorescence detector could be used inline and measure the absorbance of the used light. Online many other detectors could be used for the same purpose, McQuitty et al. [22] used an online mass spectrometer to identify the reaction products.
Another question to ask when using a flow, is what kind of flow should be used. A hydrodynamic flow (shown in Figure 2.1 A) occurs because of a pressure difference between the inlet and outlet of the reactor channel. The HPLC uses a hydrodynamic flow, so the coupling of the two would be easy. The advantages of the hydrodynamic flow are that it may be used with any liquid and with devices constructed of any material. However, the hydrodynamic flow has a parabolic flow profile, the compounds in the middle of the flow will go faster than the particles close to the wall, these particles stay longer in the reactor and create a decrease in yield and selectivity. Another disadvantage from the hydrodynamic flow is the pressure involved, when the channels are too small the pressure will be too high [2]. An electrokinetic flow (shown in Figure 2.1 B) could also be used, it occurs due to a direct movement of ions in solution toward the electrode of opposite charge, as in capillary electrophoresis. The velocity profile of the electrokinetic flow is nearly flat across the channel, leading to a more homogeneous mixture and a higher yield compared to a hydrodynamic flow. Unfortunately, the electrokinetic flow needs ions, which can interfere with the reaction. Because of the ions, the electrokinetic flow is also limited with polar solvents and device materials that can handle the charges in the system [2]. The HPLC uses a hydrodynamic flow, so the coupling of the HPLC with a microreactor would be easy compared to the electrokinetic flow.
Figure 2.1 (A) hydrodynamic flow and (B) electrokinetic flow [2]
7
3 Microreactors and waveguides
Organic synthesis have mostly been proceeded in the round-bottom flask, with these traditional synthetic methods almost any organic compound can be fabricated. However, microreactors have received a lot of attention since the beginning of the 21st century. Microreactors, also known as microstructured reactors or micro channelled reactors, consist of a solid support with channels of 10–1000 µm in width and depth [1][2]. The microreactors have several advantages over these batch reactors, due to the narrow channels and the small volumes. The mixing in the microreactors is very homogenous and can happen really fast, using a static mixer in the microreactor makes the mixing happen in milliseconds. The stirring in classical reactors, for example, a round-bottom flask, is limited by its inhomogeneities due to its stirring mechanism. Due to the small size, heat is effectively dissipated. This prevents the hotspots typically found in batch reactors, contributing in a better selectivity and better control over the heat transfer. The small volumes decrease waste and increase safety. Because of the small volumes used in the microreactors, it allows the safe use of highly toxic or explosive reactants [2][13]. However, it is important that the device is minimized in adsorption to the wall, crystallization and clogging or it can ruin the experiments due to loss, pressure and scattering [3].
When using microreactors for photochemical reactions the large surface-to-volume ratio is a big advantage. In bulk reactors, the light can shine on the outside of the reactor, but you would only irradiate the compounds on the outside of the sample reactor, irradiating a small surface in a high volume. Another way is to place a lamp in the sample solution, but still only a part of the sample is irradiated, so again a small surface is irradiated in a high volume. When using a microreactor for the photochemical reactions, the sample can be irradiated in two ways. Irradiating sideways by using a micro photoreactor (see Figure 3.1 a) or by irradiating along the length using a waveguide for instance (see Figure 3.1 b) [1][5]. The maximized surface-to-volume ratio for the micro photoreactor enables highly efficient photochemistry, with an improved irradiation and better light penetration through the reaction volume [6]. Because there is a larger surface-to-volume ratio there is more light in a smaller volume, the molecules absorb the light easier and the reaction goes faster [7].
Figure 3.1 microreactors set up (a) irradiated sideways (micro photoreactor) [12] (b) irradiated along the length (optical waveguide) [6]
8
3.1 Micro photoreactors
Most micro photoreactors (shown in Figure 3.1 a) consists of channels in a chip covered with a transparent material. The sample is irradiated from a light source outside the channel, when the channel thickness is small it allows a high surface-to-volume ration. The light penetrates through most of the reactor and the reaction mixture can be easily irradiated [1][3][4]. For photochemical microreactors, many examples already exist in the literature. Many describe different reactions or they published a review comparing the reactions with the advantages of the microstructured photoreactors. Coyle and Oelgemöller [1] wrote a review about the different reactions with all the advantages on microreactors and Sambiagio and Noël [5] wrote one February 2020 about different micro photoreactors with a continuous-flow.
Micro photoreactors allow off-chip detection of the reaction, for example with a spectrometer at the outlet of the chip [1][3]. The detection could also be done on-chip, having an immediate detection before the outlet of the chip. This inline detector could also be on different spots in the channel, when using a continuous flow the reaction process could be measured [3]. However, these channels are really small, with really small pathlengths, making it hard to have a suitable inline detector. Instead of a detector, the micro photoreactor could also be coupled to a HPLC, but HPLC analysis can be relatively slow compared to most photochemical reactions [3]. A stop flow, loops or a really fast LC separation are needed to keep up with the flow of the micro photoreactor.
For homogenous reactions, where all reagents are dissolved in the liquid, many examples exist. With different junctions multiple liquids can be mixed before or in the microreactor itself. Commercially the serpentine channel type reactors are available but many researcher custom build their homogenous photoreactors. This gains them the flexibility for the length and depth, as well as to choose the materials for the solid substrate and the cover glass [1][3][10][11]. Heterogeneous gas-liquid reactions require a gas supply, where some kind of membrane could be used. Commercially is the falling film type reactor available, but they can also be custom build reactors [1][10][11].
The concentrations in the reaction mixture are also important, because in a reaction a high yield and a high amount of product are wanted. In the microreactors only small amount of starting materials are used, so only a small amount of reaction products can be produced. When increasing the concentration also the light absorption increases in the microreactor, due to the law of Lambert-Beer, see equation (1). This can affect the yield, because there are too many compounds and not all are excited for the photochemical reaction.
9
3.2 Optofluidic waveguides
In optofluidic microreactors or liquid core waveguides (LCW), the microreactor is a tube where the light irradiates along the length (see Figure 3.1 b). A reflective cladding material is used to reflect the light within the LCW and with the small volumes used, it creates a large surface-to-volume ratio. The light guiding can be achieved by two different principles [23]. Either by total internal reflection (TIR), where the light is guided if the cladding material has a smaller refractive index than the liquid core, meaning the cladding material should be carefully chosen. This will be discussed further in chapter 4 [24]. The other principle is based on interference based waves, where multiple layers of material are used as cladding, creating multiple reflections. This is further discussed in chapter 0 [25].
For homogenous reactions, where all reagents are dissolved in the liquid, both TIR and interference based waveguides work, because only one phase (the liquid phase) enters the tube. Multiple examples are already made and discussed in the next chapters. Most of them are commercially available, but the fabrication procedure to custom build the different waveguide is also given. For heterogeneous gas-liquid reactions a gas supply is needed, the cladding should not only be the reflecting material but also be gas permeable. This is easier with TIR based waveguide and might be difficult for the interference based waveguides because of the multiple cladding layers. Homogeneous catalytic reactions can proceed in both waveguide types, the catalyst is solved in the reaction mixture so it would not interfere with the light. For heterogeneous catalytic reactions, the interference based waveguides might have the advantage, because of the multiple layers starting with a catalytic layer might not worsen the loss that much. However, using a catalytic layer would change the refractive index of the cladding material and make total internal reflection way less possible.
The Lambert-Beers law calculates the absorbance (A), where ɛ is the molar extinction coefficient, L the optical pathlength, c the concentration and I0 and It the intensity of the light in the absence and the presence of the analyte.
𝐴 = 𝑙𝑜𝑔 (𝐼0
𝐼𝑡) = 𝜀𝐿𝑐 (1)
The concentration is an important parameter to think about, because with a high concentration and a high pathlength gives a high absorption. In the waveguide only small amount of starting materials are used, so only a small amount of reaction products can be produced. To increase the amount of reaction product the starting concentration needs to be increased, but this increases the absorbance. When using a cuvette (L = 1 cm), the wanted absorption is normally found between 0,2 and 1. When using the same concentration and molar extinction coefficient but you increase the length to a meter (L = 100 cm), the absorption increases by a 100 times and cannot be normally measured. With a high concentration and a long pathlength, the absorbance is too high. In this way, the light is already absorbed before it even reached the end of the waveguide and the yield is decreased. Therefore, small concentrations should be used [26].
10 To make sure the light penetrates through the entire waveguide the optical loss (dB/m) must be minimized. Typical the losses are created by Rayleigh scattering, OH absorption (~1310 and ~1550 nm) and the imperfection of the waveguide [27]. The optical losses are mostly given in attenuation (in dB) per meter. When using the waveguide as a microreactor, the waveguide would not be longer than a few meters and a loss around 2 dB/m is acceptable. Less would always be better but is not necessary. Having a loss of 1 dB/cm would be a big loss if you have a waveguide of one meter long. However, when using the waveguide in different lengths or for different purposes, the acceptable losses could be different.
The waveguides have already been used for increasing the optical pathlength of a spectrophotometer. According to Lambert-Beer (equation (1)) the sensitivity of spectrophotometry can be enhanced by increasing optical pathlengths. So at really low concentrations, increasing the length would still give a measurable absorbance [28][29]. The absorbance can also be measured during the photoreactions, as an inline detection. The absorbed light could be measured at the exit of the waveguide. With the absorption the stability of the reaction flow could be measured and the flow could be slowed down to increase the yield. When performing the reactions without a flow the reaction rate could be measured, because the sample is not moving the absorbance tells something about the entire reactor at that time [11]. The waveguides could also have online detection, where a detector is coupled at the end of the waveguide, this doesn’t have to be a spectrometer but can also be an MS [22]. Coiling the waveguide should for absorbance spectroscopy not be a problem, only if the light is guided through the waveguide. However, when the waveguide coils too much the angle of incidence can be too small and the light is not reflected in the waveguided and more losses can be found.
When increasing the pathlength, the chance to absorb light increases. When increasing the pathlength in fluorescence, you increase the absorbed excitation light, increasing the emitted fluorescence, which increases the sensitivity [30]. Fujiwara and Ito [31][32] were the first to execute fluorescence spectrometry in an LCW. They illuminated the waveguide axially, see Figure 3.2 a. The excitation light that does not excite a molecule, survives to the end of the waveguide and increases the fluorescence intensity signal. In the detector, the fluorescence signal needs to be separated from the excitation light to obtain the correct values. However, it will be hard to separate them when there is an overlap between the excitation and the emission wavelengths [30][33][34]. Another way to irradiated the LCW is by transverse illumination through the waveguide side-wall, shown in Figure 3.2 b. Less light goes into the waveguide, but the light that is not absorbed by the components passes through the tube, so almost no excitation light is found in the detector [30][33][34]. Dasgupta et al. [34] used a transverse illumination in an LCW for longer pathlength fluorescence spectrometry. They also changed the geometry by coiling the waveguide, but this causes a slight loss from the light source. It also increases the light that enters transverse, but propagate through the axial mode. However, the transverse illumination can only be used for TIR based waveguides. Due to the critical angle and a small angle of incidence the light is not reflected and goes in and out of the waveguide, but with interference based waveguides this is not always the case.
11
Figure 3.2 (a) axial and (b) transverse illuminated fluorescence in a liquid core waveguide [31][34]
The LCW can also be used to improve the sensitivity of Raman spectroscopy, normally the sensitivity is really low because of the rare inelastic scattering. After the laser excites the molecule a majority is Rayleigh scattering, which has no exchange of energy. A small part is inelastically scattered and has a change in energy, the difference in energy between the excitation and the scattered photons corresponds to a vibrational mode of the molecule, called Raman scatter [24][35]. When increasing the pathlength there will be more scattering, more Rayleigh scattering but also more inelastic scattering, this is also called fibre enhanced Raman spectroscopy [36][37]. This significantly improves the Raman signal [35], [38]–[41]. In Raman spectroscopy the illumination can also be done axially, Altkorn et al. [42] used 6 different axial geometries for an LCW. Consisting of forward and/or backward scattering with or without the used of a long pass filter or a mirror. The Raman intensity was measured over the cell length, giving the losses for the different geometries. Using a long bandpass filter or a mirror increases the intensity, the highest intensity was found using combined forward and backward scattering with a long pass filter. The illumination in Raman spectroscopy can also be done transverse, Holtz et al. [43] have successfully used an LCW for Raman spectroscopy with transverse illumination. Next to stokes and anti-stokes Raman, the LCW can also be used for resonance Raman spectroscopy [35][38].
12
4 Total internal reflection based waveguides
The first waveguides to be discussed are the total internal reflection (TIR) based waveguides. Snell’s law or the law of refraction, shown in Figure 4.1, provides the fundamental principle for a total internal reflection based optical waveguide [44]. The formula describes the relationship between the angle (𝜃) and the refractive index (n).
sin(θ2) sin(θ1)
=
n 1
n2 (2)
Figure 4.1 Snell’s law
When light reaches another medium two things can happen, either the light penetrates into the medium, it is (partly) reflected. When the second medium has a bigger refractive index than the first one (n1 < n2) or when the angle (θ1) is smaller than the critical angle, the light penetrates into the second medium. The light undergoes total internal reflection (TIR) when the second medium has a smaller refractive index (n1 > n2) and has an angle (θ1) equal to or bigger than the critical angle [24]. The critical angle can be calculated by (sin(θ2) = 1),:
θcritical = sin−1(
n2 (cladding)
n1 (core) ) (3)
For TIR-based waveguides, the cladding material needs to have a smaller refractive index than the refractive index of the liquid core [45]. The bigger the difference between n1 and n2, the smaller the critical angle will be. When the critical angle is small, the region for θ1 is bigger and more light will be reflected. Figure 4.2 shows a schematic side view of a liquid core waveguide, where the cladding material has a lower refractive index. The light follows the red arrows and is reflected at the wall, with an angle less than the critical angle so the light does not enter in the cladding material [45].
13 Many solvents have low refractive indexes, most aqueous solutions have a refractive index around 1,33, so the cladding material needs to have a really low index of refraction to realize TIR. For common solvents like methanol and acetonitrile their refractive index are also relatively low, 1,33 and 1,34. Solvents with higher refractive indexes are for example chloroform (n = 1,45) or toluene (n = 1,49), compared to glass (n = 1,46) is this still low [47]. In this chapter two materials are discussed which have a refractive index lower than water, Teflon AF (RI of 1,29-1,33) and aerogels (RI = 1, as air ).
14
4.1 Teflon AF
In 1989 the fluoroplastic material was announced by DuPont, a material with the requirements for a TIR based LCW [48]. Teflon AF [chemical name: fluorinated (ethylenic-cyclo oxyaliphatic substituted ethylenic) copolymer] is a family of amorphous copolymer of 2,2-bistrifluoromethyl-4,5-difluoro-1,3-dioxol (PDD) and tetrafluoroethylene (TFE), the structure is shown in Figure 4.3 structure of Teflon AFFigure 4.3 [48][49]. The difference between Teflon AF varieties (for example AF1600 and AF2400) is in the relative amount of the dioxol monomer (TFE: PDD) in the basic polymer chain. The labelling of Teflon AF refers to the glass transition temperature. As the relative concentration of PDD increases the glass transition temperature increases, so the transition temperature of Teflon AF1600 is 160⁰C and for AF2400 the transition temperature is 240⁰C [48][49].
Figure 4.3 structure of Teflon AF [49]
Teflon AF has many advantages, next to outstanding chemical, thermal, and surface properties it also has unique electrical and solubility characteristics. But the major advantage of Teflon AF is their low refractive index, making them very suitable as low-index claddings for waveguide applications [49].
Figure 4.4 shows the index of refraction for three different grades of Teflon AF (AF1300, AF1601 and AF2400) by different wavelengths. It shows that Teflon AF2400 has a lower refractive index, and AF1300 the highest, so the refractive index increases when the PDD concentration decreases. After 200 nm the index values are fairly stable (1,29-1,33), it would be recommended to use wavelengths longer than 200 nm [49].
15
4.1.1 Fabrication
Teflon AF waveguides can be fabricated in multiple ways: a capillary made of Teflon or a capillary coated internally or externally with Teflon AF. The first one to be discussed is a liquid core waveguide composed solely of Teflon AF (often called type I, shown in Figure 4.5) [24] [50]. The light travelling through the liquid core can interact with the Teflon AF walls. Most of the light will be reflected back through the liquid core, but some can be transmitted in the Teflon AF. At the Teflon AF/air interface the transmitted light can be transmitted to the air and lost or reflected back in the Teflon AF and transmitted in the liquid core waveguide [50]. The waveguides are formed by melting Teflon pellets into a preform and drawing capillaries in a similar way glass capillaries are made [25].
Figure 4.5 schematic drawing of a type I Teflon AF waveguide
Altkorn et al. [51] used Teflon AF 2400 capillary tubing. The waveguides were drawn from preform, these preforms were really expensive at that time ($10 000/kg). The LCW had an outer diameter of 525 µm and an inner diameter of 250 µm with a variation of 8%. The attenuation at 632,8 nm was tested over different waveguide lengths, see Figure 4.6 (a). The loss can be found in the slope of the line, the solid line shows the least-squares fit and indicates a loss of 1,9 dB/m from 0-0,45 meters and 1,4 dB/m from 0,55-0,75 meter. Between 0,45 and 0,55 meter the loss was high (5,9 dB/m), but this was probably caused by scattering and imperfection. They calculated with two different waveguide lengths (0,05 m and 0,95 m), the losses over a wavelength range of 400 and 800nm, see Figure 4.6 (b) (solid is water and dashed is methanol). The loss is lowest with wavelengths around 600 nm for water (3dB/m) and methanol (2dB/m).
Figure 4.6 (a) Attenuation versus length in a Type I Teflon AF 2400 capillary filled with methanol, solid lines show the least-square fit to the data (b) loss spectra of water (solid) and methanol (dashed) in a Teflon AF 2400 capillary [51]
16 One of the advantages of Teflon AF is its flexibility. It can be easily coiled, which makes it easier to use longer lengths without using a lot of space [24]. Another advantage is the porous structure of Teflon AF, which makes it very gas permeable. This makes it possible to use the Teflon AF waveguide as a gas sensor [24]. An example is given by Dasgupta et al. [28], who used two different Teflon AF devices for gas sensing. Their first device was made from commercially available Teflon AF 2400 tubes. The waveguides were surrounded by glass to keep the gas inside the device. The lengths of Teflon ranged between 15 and 30 cm. For the second device they bought a 24 cm long silica capillary externally coated with Teflon AF. Again the waveguide was surrounded by glass, but they removed the silica by pumping an aqueous solution of 20% HF through the capillary. The two devices were used as sensors for CO2 in water samples, with an increase of CO2 there was a decrease in pH (using NaHCO3). With the use of an indicator (phenol red), the colour in the liquid core changed and a difference in absorption could be measured. Because the colour change of the indicator is extremely fast it did not interfere with the response. Wang et al. [52] also used a Teflon AF 2400 waveguide (inner diameter of 550 µm, an outer diameter of 625 µm and 12-21 cm long) to measure CO2, but they used gas samples. The amount of CO2 caused a decrease or increase of the pH (using Na2CO3) of the filling medium that is reflected in turn by the optical behaviour of the indicator (phenol red), creating more or less absorption in the waveguide. Milani and Dasgupta [53] used the same setup as Dasgupta et al. [28], but used the LCW to measure nitrous acid and nitrogen dioxide.
Teflon AF can also be coated on the inside of glass, silica or PDMS. Creating the same effect as Teflon AF alone, but surrounded by an extra protection layer of tubing. The light travelling through the liquid core can interact with the Teflon AF walls, the light will be reflected back through the liquid core, but some of the light can be transmitted in the Teflon AF. By interaction from Teflon AF to waveguide there is no reflection, because the refractive index of the extra cladding is bigger than the refractive index of Teflon AF. The light will be reflected in the cladding/air interface and travel back to the liquid core. A way to coat the inside of the waveguide with Teflon AF is by using a wafer with channels, as used in the micro photoreactors. Manor et al. [30] used soda-lime glass and Pyrex, Datta et al. [54] used silicon and Wu and Gong. [55] and Cho et al. [56] used polydimethylsiloxane (PDMS). But the adhesion Teflon AF is difficult. Manor et al. [30] used a monolayer fluorocarbon-based solvent which was spun coated on the wafer. It provides low surface energy to oxide surfaces, therefore, allowing better adhesion of Teflon AF. Datta et al. [54] used thin fluorocarbon films formed by plasma-enhanced-chemical vapour deposition using inductively-coupled high density plasma methods.
Dress et al. [57] tested the influence of the layer thickness on the optical properties of the LCW. For different layer thicknesses they calculated the reflectivity of Teflon AF 2400 as a function of the angle. The reflectivity is defined as the ratio of the reflected to the incident intensity, or also known as the transmission. Figure 4.7 summarizes all data in a three-dimensional plot. The critical angle is measured at 75,41⁰ with respect to the normal. A detailed calculation demands a minimum value of the Teflon AF2400 coating of about 5 mm to obtain total internal reflection. To get a smooth coating Datta et al. [54] recommends to
17 bake the Teflon AF above its transition temperature, this softens the polymer and produces a smooth surface.
Figure 4.7 Three-dimensional plot of the reflectivity spectrum of the layer system for different thicknesses of the Teflon AF 2400 inner coating in the LCW [57]
Another type of Teflon AF waveguide is a silica or quartz capillary coated on the outside with Teflon AF to create the LCW effect, this is often called a type II LCW, Figure 4.8 [24]. The waveguide made of glass or quartz is surrounded by Teflon AF. Because the refractive index of the cladding material is bigger than the refractive index of the liquid core the total internal reflection does not happen at the liquid/glass interface. For this reason, the light travels both in the liquid core and in the glass itself. The light is effectively trapped in the waveguide wall, because its refractive index is much larger than the ones from the liquid and the Teflon AF. The light that does exit the waveguide wall will propagate within the liquid core and again enter the tube wall. The light is trapped in the waveguide walls but it also involves a large number of traverses of the water/quartz interface [50].
Figure 4.8 schematic drawing of an externally coated Teflon AF waveguide
Altkorn et al. [58] used a fused-silica tubing (530 µm inner diameter and 630 µm outer diameter) coated externally with a thin (~15 µm) layer of Teflon AF, but they do not describe which type of Teflon AF. The light is coupled into the waveguide by an optical fibre. The attenuation at 632,8nm was tested over different waveguide lengths, see Figure 4.9 (a).
18 Inherent waveguide loss, given by the slopes of these lines, is approximately 1,6 dB/m for water. They also tested the losses between 400 and 800 nm, see Figure 4.9 (b), at two different waveguide lengths (0,05 m and 0,95 m). The loss is lowest with wavelengths shorter than 700 nm and is lowest around 600 nm being 1 dB/m.
Figure 4.9 (a) Attenation versus length in a Type II silica capillary coated with Teflon AF 2400 at 632,8 nm filled with water (b) loss versus wavelength for water filled waveguide [58]
When comparing the type I and the type II, there are not many differences, the losses for both are almost the same, as are the used diameter and length of the LCW. Probably the main reason why the type I Teflon AF waveguide is more used it the travel path of the light. In the type II Teflon AF waveguides the light propagates mostly through the glass walls, but in type I the light mostly propagates through the core. When the light only needs to travel to the other side of the waveguide this does not matter, but when the light needs to interact with the liquid in the core. Another advantage of the type I Teflon AF waveguides is the gas permeability, which makes it possible to use the LCW for heterogeneous gas-liquid reactions. These require a gas supply, the type I Teflon AF waveguide can homogeneously add gas to the reaction without creating bubbles. The optical losses in both waveguides are acceptable when using them for photochemical reactions, because it is not to be expected to use waveguides longer than maximal a few meters.
4.1.2 Detection and Reaction use
As said in chapter 0 the LCW can be used as a longer pathlength in absorption spectroscopy. The Teflon AF waveguide is used a lot for trace analysis, for example in water and marine chemistry [24][59][60]. Many other general applications have also been described, for example, Cheng et al. [29] used the LCW to create a longer pathlength for the determination of organophosphorus pesticides in vegetables and fruits. The LCW can also improve fluorescence detection, it improves the light–fluorophore interaction and fluorescence collection efficiency and improves the detection limit [61]. The LCW significantly improves
19 the Raman signal by the propagation of Raman scattered light along the length of the waveguide [37]–[41].
The Teflon AF liquid core waveguide can also be used as a microreactor. Its optical properties are good for photoreactions and the gas permeability of type I Teflon LCW makes it perfect for heterogeneous gas-liquid reactions. However, the reactions executed in the Teflon AF LCW are not photochemical reactions. The reactions make use of the gas-permeable Teflon AF type I and some use the reflective properties as detection of the reaction. The gas permeability of type I Teflon LCW, is a good advantage because of the possibilities of photooxidation and other gas-liquid photochemical reactions.
O’Brien, Ley and others of their group [62]–[65] did multiple studies on the use of Teflon AF for flow reactions for processing of gas-liquid reactions. They used the set-up from Figure 4.10 (a) to prove their concept that Teflon AF works as microporous gas-permeable material that can afford a practically simple method of achieving efficient gas−liquid reactions [62]. They had a tubing of Teflon AF-2400 passing through a chamber of ozone. The permeation of O3 across the Teflon was demonstrated by bleaching a 0,1 M flow stream of 1,1-diphenylethene in methanol. After this, they built the set-up shown in Figure 4.10 (b) where they did many experiments with different gas-liquid reactions.
Figure 4.10 (a) proof of concept reactor [65] (b) schematic of the tube-in-tube reactor configuration [64]
Ponce et al. [66] used a type I LCW made of Teflon AF 2400 tube as a microreactor. Figure 4.11 (a) shows their schematic setup. They used a custom-made T-fluidic cell to connect the liquid and optic pathway. The Teflon AF was used as a waveguide but also as a membrane for a gas-liquid reaction. The liquid filled Teflon AF waveguide allowed the gas (N2, O2) to transfer through the Teflon AF in the liquid core filled with an aqueous solution of methylene blue, sodium hydroxide and glucose at pH 12,75.
20 When no oxygen was present, the solution turns colourless, meaning that methylene blue has been reduced to leuco-methylene blue by glucose. If oxygen is transported to the solution, it will cause the oxidation of methylene blue and create a blue colour. All these steps are schematically illustrated in Figure 4.11 (b). Ponce et al. [67] also demonstrated that their LCW membrane reactor is suitable for the studies of kinetics of gas/liquid reactions. They studied the redox kinetics of vanadium substituted heteropoly acids.
Figure 4.11 (a) Schematic liquid core waveguide membrane setup of in situ sensing and gas permeation using a Teflon AF tube [66] [68] (b) Schematic illustration of the methylene blue redox cycle.
However, these reactions are not photochemical reactions, they showed that it would be possible to do the photoreactions in the Teflon AF waveguides. The waveguide could be filled with a photodegradable solution and the light could be guided through the solution. As mentioned in chapters 2 and 3.2 the flow matters, when doing the reaction without a flow the transmission could be measured at the end of the waveguide and the reaction could be followed. The disadvantage of not using a flow are the small reaction volumes that can be used. When using a flow for the reaction, the speed determines the yield of the reaction.
21
4.2 Aerogel microchannels
Another possible material for TIR is aerogel, where the aerogel is used as a cladding material. Aerogel is a nanostructured material, consisting of a crosslinked polymers with a solid structure and a large number of air-filled pores. These pores create a high porosity, which is sometimes called “solid air”, creating refractive indexes close to one (air). This makes them perfect for total internal reflection based waveguides [23], [69]–[72].
Aerogels consist of three categories: inorganic, organic, and inorganic-organic hybrids. The inorganic aerogels were the first to be studied and are most widely used. They are formed from metal alkoxides, like alumina (Al2O3), titania (TiO2), or zirconia (ZrO2), but mostly used is silica (SiO2) [23], [69], [73]–[76]. Organic aerogels consist of connected colloidal particles or polymeric chains, like Resorcinol–formaldehyde and melamine–formaldehyde aerogels. Carbon aerogels are obtained by pyrolysis of organic aerogels [23], [73]–[75], [77], [78]. Hybrid aerogels are synthesized by building organic polymers within an inorganic aerogel matrix [23][73][74][79][80].
Aerogels have many advantages due to their high porosity and the structural nature of their microscopic and macroscopic features. Aerogels have an ultralow thermal conductivity, ultralow dielectric constant, ultralow sound speed, high specific surface area, ultrawide adjustable ranges of the density and most important they have an ultralow refractive index [75]. The refractive index of the aerogel is almost one, being so low it will almost always be lower than the refractive index of any liquid. Figure 4.12 (a) shows a liquid core waveguide in aerogel cladding. The aerogel consists of connected secondary particles, with pores filled with air. The secondary particles are formed from primary particles (Figure 4.12 (b)), the air in the aerogel pockets is responsible for guiding light through the core. To create the LCW the aerogel-based optofluidic waveguides exists of microchannels inside the aerogel. The channels are filled with a suitable liquid as liquid core, as long as the liquid doesn’t penetrate the aerogel network and stays in its channel, any type of liquid can be used [23][71].
22
4.2.1 Fabrication
Aerogels are synthesized using a sol-gel process, which is a method for producing solid materials from small molecules. The typical steps of the synthesis are shown in Figure 4.13 and can be divided into 3 general steps. The first step is the gel preparation: where the sol is formed from reactants with the addition of a catalyst (b). The condensation of the sol leads to the formation of alcogel, where the sol particles are linked (c). The second step is the aging of the gel (d), where the aging process strengthens the gel. The last step is the drying of the gel, where the gel is freed of the pore liquid (e). To prevent the collapse of the gel structure, drying has to take place under special conditions and is mostly done with supercritical CO2 [23], [69], [70], [74]–[76], [81].
Figure 4.13 Preparation of monolithic aerogels by the sol-gel method [23]
Some types of aerogels are hydrophilic, because of the hydroxyl groups on the surface of the secondary particles [76]. When filling the channels with water, the water enters the porous microchannel system of the aerogels. The hydroxyl groups can take part in hydrogen bonding with water and together with high capillary stress this leads to erosion which leads to collapse of the structure and the waveguide. Therefore the channel walls should be made hydrophobic to prevent the penetration of water. With the presence of a non-polar hydrophobic group they can meet the stability requirements for long-term use when using an aqueous core. These surface modifications can either be performed after the channel forming, during the aging step or after the drying step [23]. Eris et al. [70] used hexamethyldisilazane (HMDS) solved in supercritical CO2 to turn the hydrophilic silica aerogels into hydrophobic ones. Özbakır et al. [46] treated silica aerogels with (HMDS) vapour, the hydrophobic nature was tested during experiments over several weeks and water did not penetrate into the pores.
There are several possibilities to fabricate channels in aerogel blocks. One of the ways is by forming the aerogel around a capillary and after formation withdrawing the capillary from the block. Xiao et al. [71] created a silica aerogel block around a glass capillary and after the drying step it was withdrawn from the block. They filled the channel with water and used a
23 fibre to guide a 635 nm laser into the channel. The light was well guided and the waveguide loss was measured for a 16 mm long channel filled with water, the total loss was 1,5 dB/cm. This is much more than the Teflon AF, which were in dB/m. The withdrawal generally leads to breakage due to the attachment of the capillary to the aerogel network. There are limited shapes and sizes for the channels with this technique [23].
Furthermore, preformed shapes can also be used and the aerogels are formed around it. When the preformed shapes are soluble in the supercritical CO2 (from the drying step) and not in the solvents used in the aging steps, the preform can be easily removed. However, there are not that many polymers that are soluble in supercritical CO2 and insoluble in the other solvents used [23]. Eris et al. [70] created a fibre from trifluoropropyl POSS in a U-shape. Before gelation, they placed U-shaped trifluoropropyl POSS fibre inside the sol solution. Because the trifluoropropyl POSS solves in supercritical CO2, it was removed during the drying step. Figure 4.14 (a) shows that the U form is clearly visible at the top of the aerogel block. In Figure 4.14 (b) the channel is still filled with air and due to the absence of water, the light is not guided. Figure 4.14 (c) shows that when the channel is filled with water, the light is nicely guided through the channel and creates an LCW.
Figure 4.14 (a) aerogel block with an empty U-shape channel (b) laser light focused from the top of the channel before it was filled with water (c) laser light focused from the top on the right end of the water filled channel [70]
Another possibility is to fabricate the aerogels and to create channels with a laser. Yalızay et al. [69][72] used femtosecond laser ablation to create channels in a hydrophobic silica aerogel. The ultrafast laser has a better focus, creating a better precision and a smoother channel. After the formation of the channel, they filled it with ethylene glycol instead of water, because water was evaporating relatively fast due to open-ended channels and ethylene glycol is less volatile than water. A laser at 632,8 nm was coupled into the channel through an optical fibre. The propagation loss of the optical waveguide was then calculated as 9,9 dB/cm. This high loss is mainly due to increased roughness of the channel surfaces produced by the laser, leading to higher scattering losses in the waveguides.
A manual drilling technique is also a possibility [23]. Özbakır et al. [46] synthesised hydrophobic silica aerogels and created microchannels by manual drilling using a drill bit. A hand drill bit was put to the aerogel surface and moved slowly into the sample (rotating continuously) to prevent stress build up, as shown in Figure 4.15 (a). A straight channel with a diameter of ~2,1 mm was opened. An L-shaped channel was formed by creating a horizontal channel and another channel starting from the top, see Figure 4.15 (b), creating an L-shaped channel shown in Figure 4.15 (c).
24
Figure 4.15 (a) channel formation in aerogel by manual drilling (b) L-shaped channel formation in aerogel by manual drilling (c) Side-view of inclined L-shaped channel in monolithic aerogel [46]
The laser was coupled into the horizontal part of the channel by an optical fibre, when the light was coupled into an empty channel, no guiding of light to the opposite end of the channel was observed. As can be seen in Figure 4.16 (a), the light is scattered strongly. When filling the channel with water, see Figure 4.16 (b), the light was guided and delivered to the opposite end of the channel. A propagation loss of 1,45 dB/cm was found.
Figure 4.16 (a) light coupled into an empty channel (b) light propagation in the water filled channel [46]
The losses in the aerogel waveguides are large, in dB/cm. This is probably because the channels are not smooth, causing the light not to be in the right angle for TIR and leaving the waveguide. Likewise, the not so smooth channels create more light scattering, causing more losses.
4.2.2 Detection and Reaction use
Aerogels guide light by the principle of total internal reflection, which works to create a longer pathlength. Özbakır et al. [46], Eris et al. [70] and Xiao et al. [71] showed that the light can be guided through the waveguide channel. Therefore, the aerogel-based optofluidic waveguides could be used for applications including light-driven detection or identification and quantification of particular chemical compounds [46]. However, none has done experiments other than proving it to be possible to use the aerogels for light-driven detection.
25 The aerogels are gas permittable due to the open porous network. The pores are open, so the aerogels allow gasses to diffuse from the surroundings in the liquid core and the other way around [23][76][82]. Like Teflon AF, they could be used for gas sensing or heterogeneous gas-liquid reactions.
The aerogel based waveguides could also be used as a micro photoreactor, Özbakır et al. [6] demonstrated an aerogel-based micro photoreactor for the degradation of methylene blue (MB). They used the same fabrication as before [6], a hydrophobic silica based aerogel with an L-shaped channel (1,1 cm long horizontal channel and 3,7 cm long inclined channel). Figure 4.17 shows a schematic overview of the experimental set-up that was used for the photodegradation of MB dye in an aqueous solution, an L-shaped waveguide was used. Using a T-connector a fibre was connected to the channel entrance to connect the laser to the reaction, the other port of the T-connector was connected to a syringe to fill the channel with the aqueous MB solution. Due to the hydrophobic walls, the MB solution did not penetrate into the pores of the aerogel. The MB might have been adsorbed on the surface of the aerogel, influencing the concentration and the results. The aerogel channel was filled with MB solution and kept in the dark for 2 hours, samples were collected at different times and the concentration was measured by a nanodrop spectrophotometer. The concentrations were constant, indicating that MB was not adsorbed at the channel surface.
Figure 4.17 Schematics of experimental set-up used for photochemical reactions, the waveguide is pictured linear but the L-shaped waveguide was used [6]
26 A laser with wavelength 388 nm was coupled into the aerogel waveguide filled with MB solution. Because of the guiding properties, the MB could be degraded along the full length of the channel. The amount of degradation is then depending on the exposure time and the laser power. Figure 4.18 shows the decrease in MB concentration by a longer exposure time and a constant power. The photodegradation rate was highest at the entrance, due to absorption and scattering. These experiments were executed using no liquid flow, but it should be possible to operate with a continuous flow, using a suitable pump. The difference in photodegradation rate would then be solved and bigger amounts of MB could be degraded.
Figure 4.18 The concentration of aqueous MB solution in an aerogel channel under irradiation at 388 nm at different times with a varying power of the incident light [6]
Özbakır et al. [83] also used the photo microreactor for the photocatalytic degradation of phenol over titania. The silica-titania aerogels were fabricated by adding anatase titania powder in the sol during the sol preparation in different weight percentage ranging from 1,7 wt% to 50 wt%. Afterwards, they were aged, dried and transformed from hydrophilic into hydrophobic using HMDS. For their experiments they used the same set up as Figure 4.17, just the waveguide contains titania particles now, see Figure 4.19.
Figure 4.19 aerogel waveguide with catalytic titania [83]
Because the phenol may be adsorbed on the channel wall, the channels were filled and kept in the dark for 60 minutes, Figure 4.20 (a) shows that the concentration of phenol remained almost constant. The photodegradation of phenol was also investigated without titania particles at 366 nm in a silica aerogel waveguide. Figure 4.20 (b) shows that the concentration remained constant which indicated that there is no photodegradation, just the photocatalytic degradation of phenol. Figure 4.20 (c) shows a decrease in phenol
27 concentration over time within the first 10 minutes, after 10 minutes the reactor system henceforth operated at steady state at constant flow.
Figure 4.20 (a) concentration of phenol in the dark over time (b) concentration of phenol in silica aerogels without titania at a laser wavelength of 366 nm over time (c) concentration of phenol in silica-titania aerogels at a laser wavelength of 366 nm
over time [83]
Compared to the Teflon AF waveguides the aerogel has a much better refractive index, creating a smaller critical angle. Another advantage from the aerogel over Teflon AF is the possibility to use heterogeneous catalysts. However, the losses in the aerogel are too high (~dB/cm) and the channels can only be a few centimetres long. These dimensions and losses need to be improved before the aerogel could be used as microreactor of photochemical reactions.
28
5 Interference based waveguides
Another approach for light guiding is the use of interferences-based waveguides. This means the interference of two waves forming a wave with a higher, lower or the same amplitude. The waves can be correlated or coherent with each other, either because they come from the same source or because they have the same or nearly the same frequency, resulting in constructive (Figure 5.1 left) or destructive (Figure 5.1 right) interference [45].
Figure 5.1 interference of waves, (left) constructive and (right) destructive [Wikipedia]
In interference-based waveguides, multiple layers of dielectric materials are used as the waveguide cladding. These layers create multiple reflections that can interfere constructively or destructively. Figure 5.2 shows that the light is partially reflected at each interface of the structured cladding material. The figure shows only a small amount of the reflected waves, but for the reflected power all the reflected waves need to be added up. The idea is that an almost perfect reflection in the liquid core can be achieved even if the core has a lower refractive index than all of the cladding layer materials [23][25][45].
Figure 5.2 Schematic view of multiple interfering reflections [45]
The interference waveguides can be based on the photonic bandgap effect. Photonic bandgap materials or photonic crystals are dielectric materials with a periodic structure. The photonic bandgap represents the forbidden range where certain wavelengths cannot be transmitted through the material [84]–[86]. The photonic crystal can be layered in different dimensions, see Figure 5.3. The first dimension (Bragg fibres) and second dimension (photonic crystal fibres) are discussed below. The other interference based waveguides that will be discussed are Anti-resonant reflecting optical waveguides (ARROW).
29
5.1 Bragg fibres
The first interference based waveguide to be discussed is the Bragg fibre, also known as the one dimensional photonic crystal. The Bragg fibres contain multiple dielectric layers of low (n1) and high (n2) refractive indexes around the hollow core, shown in Figure 5.4. The light is guided due to the photonic bandgap [88]–[91].
Figure 5.4 A schematic of a Bragg fibre with a liquid core. ncore < n0 < n1 [92]
5.1.1 Fabrication
Bragg fibres rely on the layered dielectric cladding and are mostly used for air filled mediums [25]. The Bragg fibres are mostly made by placing multiple layers of film alternately on each other, the films are rolled into a hollow multilayer tube that acts as a preform. The preform is then placed in a draw tower to draw a long fibre with thin films [25][93][94]. Low losses (~ dB/km) were found with the Bragg waveguides [95]. A few years ago they filled the fibres with a liquid and tried to guide the light in a liquid medium, like water. Qu and Skorobogatiy [88] demonstrate a liquid-core Bragg fibre. The Bragg fibre was 40 cm long and had a large core (~0,8 mm) surrounded by an alternating polymethyl methacrylate and polystyrene. The core was filled with NaCl solutions with different weight concentration. The refractive indexes of the salt solutions were between 1,33 and 1,38. The transmission spectra of the NaCl solutions that were analysed by a spectrometer are shown in Figure 5.5. Observed is a blueshift in transmission spectra when the RI of the analyte increases.
30
5.1.2 Detection and Reaction use
Filling the Bragg fibres with a liquid is a relatively new phenomenon and first happened less than 10 years ago. Only a few Bragg fibres filled with liquids have been published, there were all about fabrication and how to fill the fibres with a liquid. Little is known about using the Bragg fibres as a longer pathlength or as a reaction vessel, but that does not make it impossible. However, little is known about its use, guidance and loss when filled with liquid, so I would not recommend it.
31
5.2 Photonic crystal fibres (PCF)
Photonic crystal fibres are optical fibres based on the properties of photonic crystals, also known as the two dimensional photonic crystals. The cladding consists of microscopic hollow capillaries along the length of the entire fibre [8]. The holes of the first PCF were too small to expect a photonic bandgap and they had a solid core. The guiding principle was based on modified total internal reflection, where the holes create a low refractive index. When using bigger holes or a hollow core the light is guided by the photonic bandgap principle due to the photonic crystal [84][96]. Here the photonic crystal cladding forbid the light with a certain range of frequencies to propagate, the light within these bandgaps is confined and guided in the core [97].
5.2.1 Fabrication
PCFs are commonly made of silica, but glass and polymers are also used [25]. Their fabrication is based on the ‘stack-and-draw’ method, see Figure 5.6 [8][98][96]. First, the capillaries and rods need to be drawn with specific diameters and cut into short lengths, they have a very strong impact on the transmission characteristics of the fibre. The capillaries and rods are stacked in a larger diameter glass tube, also called a jacket tube, forming a preform. The preform is pulled down into a cane under pressure to prevent holes between the capillaries. Finally, the canes are drawn into a fibre with a specific diameter using a drawing tower, a typical fibre diameter is 40 to 100 µm [98][99]. This is already smaller than the core of the TIR based waveguides, but this is just the fibre diameter, not even the holes. It is important to remember that even smaller volumes go into the PCFs.
Figure 5.6 schematic of the 'stack-and-draw' method and some examples: Solid-core photonic crystal fibres (a); suspended-core photonic crystal fibres (b); hollow-core photonic bandgap fibre (c) and Kagome photonic crystal fibre (d)
32 As mentioned before the core of the PCF can be solid and hollow, the first one to be discussed is the solid-core PCF (SC-PCF). The SC-PCF has a solid core surrounded by periodic cladding channels, see Figure 5.6 (a). The hollow channels reduce the refractive index of the cladding below the refractive index of the core, creating a guidance mechanism based on total internal reflection [8]. Kurokawa et al [27] created a SC-PCF to improve the fibre bandwidth for internet traffic. They created a 25 km long silica SC- PCFs with 60 holes. The hole diameter was 7,6 µm and the outer diameter of the fibre was 125 µm. Figure 5.7 shows a loss spectrum of the fabricated PCF with the lowest loss, a loss of 0,18 dB/km was found in the 1550 nm wavelength region. The big loss at 1380 nm is caused by the OH absorption. This is really low, which is pretty impressive compared to the waveguides discussed above. On the other hand, it is not necessary, to have such low losses or long waveguides when using them as a microreactor. When using the PCF for sensing or detecting applications the holes can be filled with liquid. The light is guided in the solid core, but interacts with the samples by an evanescent field that penetrates into the holes. The amount of evanescent field penetration into the holes can be controlled by parameters such as core size, hole diameter and the distance between the hollow channels [8][100]. This principle might work, but the question would still be of how much light would go into the channels by evanescent field penetration and would it also go into the second row of channels? This seems a little bit unlikely, but it would also be really hard to fill the extremely small channels, use a flow and keep a practical pressure.
Figure 5.7 Loss spectrum of solid-core PCF with the lowest losses [27]
Because of the large flow resistance in the micro cladding holes in SC-PCF, the suspended-core PCF was fabricated. It consists of a solid suspended-core held in the air by three nanowebs, shown in Figure 5.6 (b) [8]. Webb [101] fabricated multiple suspended-core PCFs. Instead of stacking, they drilled holes into a silica preform, but the drawing inside a jacked process was the same. The outer diameter of the fibres was 125 µm, the core variated from 0,8 to 1,8 µm and the holes were around 8 µm in diameter. In each case, they drew several tens of meters fibre to prove there was no change in structure, but they used a meter of fibre for the experiments. They found a loss of 0,29 dB/m at 1550 nm, the relatively high loss was mainly due to scattering because of the rough surface followed by the drilling. This is an acceptable loss when using the suspended-core PCF as a microreactor for photochemical reactions. For sensing applications the holes are filled with the sample, the amount of evanescent field that extends into the holes can be changed by changing the core diameter [8][102]. The suspended-core PCF works ideal for broadband transmission [103]. The channels are already a little bit bigger than the SC-PCF and would be better for the flow resistance. However, the evidenced field penetration would still be questionable, but there
33 is only one layer of channels around the core and the core is smaller. The chance of the light to be in the channels is bigger but also is the loss bigger. The loss is still acceptable when using waveguides of a few meters. The reaction may take a little bit longer because the light is mostly in the solid core and sometimes in the liquid channels. However, it would be acceptable/neglectable because of the small volumes being used.
The other type of PCF is the hollow-core photonic crystal fibre (HC-PCF), it has a core with a low refractive index. It can be fabricated in different structures, mostly by the “stack and draw” method, where the hollow core is created by leaving out some capillaries at the core, usually seven [99]. The first type of HC-PCF, known as hollow-core photonic bandgap fibre (HC-PBF) is shown in Figure 5.6 (c). It consists of a central hollow core surrounded by a honeycomb lattice of hollow channels. The light is guided by a two-dimensional photonic bandgap, because of the photonic crystal [8][104]. Due to these photonic bandgaps, in the HC-PBF the wavelength of the light has to be matched to a specific spectral bandgap, otherwise, high losses are to be found. By filling the hollow cores completely with a liquid media, the bandgap can shift to lower wavelengths [36]. Like the SC-PCF, the HC-PBF can provide extremely low transmission losses, as low as 1,2 dB/km [105]. Which is again pretty impressive, but not completely necessary when not using more than a few meters for the microreactor. Because the HC-PBF only reflects the light at certain wavelength ranges, it is not very useful when using multiple wavelengths.
Another type of HC-PCF is the kagome PCF, which is named after its particular lattice arrangement in the cladding structure, see Figure 5.6 (d) [8]. An important property of kagome HC-PCF is the wide transmission bandwidth, covering from the UV to the near-infrared, this fibre is, therefore, the preferred choice for broadband spectral measurements [106][107]. But the kagome PCF has the disadvantage of a higher transmission loss, 1 dB/m [108]. This is much higher than the HC-PBF, but the advantage is that the light reflects in a bigger range of wavelengths. This loss is still acceptable when using it as a microreactor because maximal a few meters of waveguide would be used. Williams et al. [109] used a 30 cm long kagome PCF with a core diameter of 19 µm and a loss of ~1 dB/m. They used the kagome PCF for the photoisomerization of azobenzene. In 10 seconds the reaction process was completed, whereas using a 1 cm cuvette, almost three orders of magnitude more power was needed.
When filling the HC-PCF with a liquid sample, the sample must be pumped into the core of the fibre. The fibre can be filled in two different ways, selectively or completely filled. When only filling the core, the cladding holes are air-filled. The refractive index of the core exceeds the average refractive index of the fibre cladding, creating a total internal reflection and a broadband guiding waveguide. Sometimes the holes are thermally collapsed by a fusion splicer to prevent them from being filled [110][111][112]. When completely filling the HC-PCF, the core is filled with the sample. The holes are homogeneously filled with a liquid whose index of refraction is lower than the refractive index glass [97][113]. However, only filling the core is proved to be more suitable than filling the airholes and the core, because light can be more easily confined in the core region [114].
