78
Microfluidic Preparation of Air- and Moisture-Sensitive Precursor Solutions and Semiconductor Nanocrystals by Parnian Saberi A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Graduate Department of Mechanical and Industrial Engineering University of Toronto © Copyright by Parnian Saberi 2016

Microfluidic Preparation of Air- and Moisture-Sensitive ... › bitstream › 1807 › ...Parnian Saberi Master of Applied Science Graduate Department of Mechanical and Industrial

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

  • Microfluidic Preparation of Air- and Moisture-Sensitive

    Precursor Solutions and Semiconductor Nanocrystals

    by

    Parnian Saberi

    A thesis submitted in conformity with the requirements

    for the degree of Master of Applied Science

    Graduate Department of Mechanical and Industrial Engineering

    University of Toronto

    © Copyright by Parnian Saberi 2016

  • ii

    Microfluidic Preparation of Air- and Moisture-Sensitive Precursor

    Solution and Semiconductor Nanocrystals Parnian Saberi

    Master of Applied Science

    Graduate Department of Mechanical and Industrial Engineering

    University of Toronto

    2016

    Abstract Syntheses of a wide range of air- and moisture-sensitive chemistries are conducted in

    dedicated, capital-intensive setups that are not amenable to automation and offer only limited

    process control and scalability. Here, we demonstrate a compact Schlenk-line-to go (SL2G)

    platform that overcomes these limitations. Case studies of semiconductor precursor solutions and

    colloidal nanomaterials are presented. The rapid and reproducible preparation of precursor

    solutions and NCs is achieved by a microfabricated sparger that promotes rapid mixing and

    temperature control. We expect the benchtop SL2G platform to enable air- and moisture-sensitive

    syntheses to be conducted independent of a fume hood or a glovebox and in a manner that is

    scalable, automatable, and compatible with flow chemistry.

  • iii

    Acknowledgments

    I am grateful for Prof. Axel Guenther for calming me down any time that the shear amount of work

    would test my patience level, and to help me find the path in this project. His perspective changed

    mine for the better. Dr. Yasser Hassan spent many hours in the lab explaining fundamentals of

    semiconductor nanocrystals to me, while being overloaded with other responsibilities. For this and

    many other lessons I learned from him, I thank him. As a groupmate, I enjoyed working with

    Shashi Malladi and I am grateful for her company, general consult and help on coding. Ruoxi

    Wang joined during the last two months of my work, and with her self-starter mindset,

    determination and hard work, she helped the project progress to the finish line, and I would like to

    thank her for travelling from the so-called heaven, California, to Toronto to work with me.

    Above all, the support of my family helped me throughout my work; my father for always

    believing in me, and my mother for being by my side during good and bad times. I hope to keep

    making you proud and hopeful that you did not go so wrong with one of your children! And lastly,

    I am thankful for Sahand Jafarian for being my second brain, ears to my mouth and for helping me

    survive tough times since too long ago.

  • iv

    Table of Contents

    Acknowledgments.......................................................................................................................... iii Table of Contents ........................................................................................................................... iv List of Tables ...................................................................................................................................v List of Figures ................................................................................................................................ vi List of Appendices ........................................................................................................................ vii List of Symbols ............................................................................................................................ viii List of Abbreviations .......................................................................................................................x List of Chemical Formulae ........................................................................................................... xii 1. Introduction .................................................................................................................................1 1.1. Semiconductor Nanocrystals ...............................................................................................1 1.2. Evolution of Synthetic Methods ..........................................................................................1 1.3. Hot Injection Synthesis ........................................................................................................2 1.4. Reaction Platforms ...............................................................................................................4 2. Experimental ...............................................................................................................................6

    2.1. Experimental Apparatus.......................................................................................................6 2.2. Materials and Methods .......................................................................................................12

    3. Results and Discussions ............................................................................................................20 4. Conclusions and Outlook ..........................................................................................................39

    4.1. Conclusions ........................................................................................................................39 4.2. Outlook ..............................................................................................................................39

    References ......................................................................................................................................41 Appendices .....................................................................................................................................46

    A Experimental Details ..........................................................................................................46 A.1 Device Design ....................................................................................................................46 A.2 Manifold Design ................................................................................................................50 A.3 Intensity Measurement .......................................................................................................56 B Microfabrication ................................................................................................................57 C Precursor Characterization .................................................................................................60 C.1 Te Precursor Characterization ...........................................................................................60 C.2 Cd Precursor Characterization ...........................................................................................65

  • v

    List of Tables Table 1. Preparation composition of Cd-OA precursor solution for CdTe NC Synthesis ............ 12 Table 2. Preparation composition of Te-TBP precursor solution for CdTe NC Synthesis........... 13 Table 3. Properties of CdTe NCs prepared at 200°C and 220°C .................................................. 29 Table 4. Time-dependent control properties of precursor and CdTe NCs in SL2G and SL. ........ 33 Table 5. Initial and Final Controller Parameters for Tsp=300°C in SL and SL2G. ....................... 34 Table 6. Peak list for Cd precursor and NC solutions................................................................... 37 Table 7. O-ring/Gland Design for Sealing Glass Cylinder ........................................................... 52 Table 8. PTFE GORE Sheet Gasket Specifications/Design ......................................................... 52

    Table A. 1. Precursor Preparation Device Design Specifications ................................................ 46 Table A. 2. Typical solution Properties of Cd-OA Mixture (Protocol 2)[57] .............................. 47 Table A. 3. Typical pressure calculations within the microfluidic device with 100 µm diameter holes. ............................................................................................................................................. 48

  • vi

    List of Figures Figure 1. Comparison of Schlenk Line (SL) and Schlenk Line-to-Go (SL2G) platforms.. ........... 7 Figure 2. Illustration of SL2G platform operation with MF device.. ............................................. 8 Figure 3. Experimental setup and control systems.. ....................................................................... 9 Figure 4. Schematic of detailed parts of SL2G manifold ............................................................. 11 Figure 5. Electrical diagram of temperature control system ......................................................... 15 Figure 6. Comparison of NMR preparation setups.. ..................................................................... 17 Figure 7. Time-dependent SL2G platform operation.. ................................................................. 21 Figure 8. Characterization of Te-TBP precursor solution prepared in SL2G platform. ............... 23 Figure 9. 31P NMR of phospine-based ligands and Te precursor for CdTe NC synthesis. ........... 24 Figure 10. Characterization of Cd-OA precursor solution prepared in SL2G platform. .............. 26 Figure 11. 1H NMR of HOA and Cd-OA precursor comparing removal of acidic proton. .......... 27 Figure 12. Characterization of prepared CdTe NCs in microfluidic platforms. ........................... 29 Figure 13. Profile of lattice plane distances of 5.3 nm CdTe NCs synthesized at 220°C............. 30 Figure 14. PXRD pattern of CdTe NCs prepared at 220°C. ......................................................... 30 Figure 15. Temperature profiles of precursor and NC preparations.. ........................................... 32 Figure 16. Process response behavior in Cd-OA preparation in SL.. ........................................... 35 Figure 17. Process response behavior in Cd-OA preparation in SL2G ........................................ 36 Figure 18. FTIR spectra of precursors and NC solutions.. ........................................................... 38 Figure A. 1. Device Configuration for Precursor Preparation.. .................................................... 49 Figure A. 2. Sealing Design of SL2G. .......................................................................................... 51 Figure A. 3. Drawing of top Al part of manifold. ......................................................................... 53 Figure A. 4. Drawing of bottom Al part of manifold.................................................................... 54 Figure A. 5. Drawing of insulating part of manifold. ................................................................... 55

    Figure B. 1. Microfabrication Process Flow Diagram .................................................................. 58 Figure B. 2. Side-wall angle analysis of microfluidic devices with DRIE protocol changes. ...... 59

    Figure C. 1. 31P NMR of Te-TBP and commercial TBPO. .......................................................... 60 Figure C. 2. MS results for centrifuged Te-TBP Sample. ............................................................ 61 Figure C. 3. MS results for magnified region of Figure 8b. ......................................................... 62 Figure C. 4. MS results for filtered Te-TBP Sample. ................................................................... 63 Figure C. 5. MS results for crude Te-TBP/TOP/ODE/OLA Sample. .......................................... 64 Figure C. 6. FTIR Comparison of Te ligand and precursors. ....................................................... 65 Figure C. 7. 113Cd NMR of Cd(ClO4)2 in comparison with previously-reported values. ............. 66 Figure C. 8. Superimposed 1H NMR of Cd-OA ( ), CdTe ( ), and HOA ( ). ............................ 66

  • vii

    List of Appendices

    A. Experimental Details………………………………………………………………………….46 B. Microfabrication………………………………………………………………………………57 C. Precursor Characterization……………………………………………………………………60

  • viii

    List of Symbols

    Symbols

    ∆𝐏𝐏𝐜𝐜𝐜𝐜𝐜𝐜 Capillary pressure (Pa)

    ∆𝐏𝐏𝐆𝐆 Gas pressure drop across the channel (Pa)

    0 Reference condition

    a Dimensionless parameter

    a, b, c Lattice parameters (Å)

    C Sutherland's constant for gaseous material (K)

    C=O Cabonyl group

    COO- Carboxylate group

    COOH Carboxylic acid

    de Equivalent bubble diameter (m)

    g Gravitational acceleration (m.s-2)

    H Device channel height (m)

    hl Liquid column height (m)

    i ith component

    I(t) Time-dependent intensity (arb.)

    Ifit Fitting grayscale intensity

    Io Incident light intensity

    L Device channel length (m)

    l Liquid component

    N Nth component

    PG,experimental Experimental inlet gas pressure (Pa)

    PG,min Minimum inlet gas pressure (Pa)

    PL Liquid column pressure (Pa)

    PL Photoluminescence spectroscopy

    P-O Phosphoryl group

    P-Te Phosphorus telluride group

    Q Liquid flow rate (m3.s-1)

    r Device hole radius (m)

  • ix

    R Fluidic resistance (kg.m-4.s-1)

    T Temperature (°C, K)

    t Time (min)

    T(t) Time-dependent temperature (°C)

    Tmin Minimum temperature (°C)

    trxn Reaction time (min)

    Tsp Temperature setpoint (°C)

    u∞ Terminal gas velocity (m.s-1)

    V Liquid volume (ml)

    VBN Viscosity Blending Number

    W Device channel width (m)

    x Mass fraction (g/g)

    γ surface tension (N.m-1)

    δ Chemical shift (ppm)

    θ Side-wall angle (°)

    λ Wavelength (Å)

    μ Dynamic viscosity (Pa.s)

    ν Kinematic viscosity (m2.s)

    νmixture Kinematic viscosity of mixture (cst)

    ρ Density (kg.m-3)

    τ Time constant (min)

    χ2 Chi-squared test statistics

  • x

    List of Abbreviations

    Abbreviations 113Cd NMR Cadmium NMR 1H NMR Proton NMR 31P NMR Phosphorus NMR

    Al Aluminum

    AOT Sodium di-2-ethylhexyl sulfosuccinate

    ASTM American Society for Testing and Materials

    Cd-OA Cadmium oleate

    CMOS Complementary metal–oxide–semiconductor

    DART Direct Analysis in Real Time

    DRIE Deep Reactive-Ion Etching

    FFKM Kalrez® perfluoroelastomer

    FTIR Fourier transform infrared spectroscopy

    FWHM Full width at half maximum (nm)

    HMDS Hexamethyldisilazane

    HOA Oleic acid

    HRTEM High resolution TEM

    ICDD International Centre for Diffraction Data

    ID Inner diameter (mm)

    LC Liquid chromatography

    MS Mass spectrometry

    NC Nanocrystal

    NMR Nuclear magnetic resonance spectroscopy

    OD Outer diameter (mm)

    ODE 1-Octadecene

    O-H Hydroxy group

    OLA Oleylamine

    P63mc 6mm hexagonal space group

    PEEK Polyether ether ketone

    PFA Perfluoroalkoxy

  • xi

    PID Proportional-integral-derivative

    PLQY Photoluminescence quantum yield

    PTFE Polytetrafluoroethylene

    PVC Polyvinyl chloride

    PXRD Powder x-ray diffraction

    QD Quantum dot

    r.t. Room temperature (°C)

    RTD Resistance termperature detector

    SEM Scanning Electron Microscopy

    Si Silicon

    SL Schlenk line

    SL2G Schlenk-line-to-go

    SS Stainless steel

    STP Standard temperature and pressure

    TBP Tri-n-butylphosphine

    TBPO Tri-n-butylphosphine oxide

    Te Tellurium

    TEM Transmission electron microscopy

    Te-TBP Tri-n-butylphosphine telluride

    Toluene-d8 Deuterated toluene

    TOP Tri-n-octylphosphine

    TOPO Tri-n-octylphosphine oxide

    UV-vis Ultra-violet-visible spectroscopy

    v/v Volume fraction

  • xii

    List of Chemical Formulae

    Chemical Formulae

    C12H27PTe Tri-n-butylphosphine telluride

    C18H34O2 Oleic acid

    C36H66CdO4 Cadmium oleate

    C3H9P Trimethyl phosphine

    Cd(CH3)2 Dimethylcadmium

    Cd(ClO4)2.6H2O Cadmium perchlorate hexahydrate

    Cd(SO4)2 Cadmium Sulfate

    Cd2+ Cadmium ion

    CdCl2 Cadmium chloride

    CdO Cadmium oxide

    CdS Cadmium sulfide

    CdSe Cadmium selenide

    CdTe Cadmium telluride

    CH2 Methylene group

    D2O Deuterium oxide

    H2S Hydrogen sulfide

    HgTe Mercury telluride

    KBr Potassium bromide

    ZnSe Zinc selenide

    Si(CH3)4 Tetramethylsilane

  • 1

    Chapter 1

    1. Introduction 1.1. Semiconductor Nanocrystals Nanocrystals (NCs) with an inorganic semiconductor core and an organic shelling surfactant layer

    exhibit interesting size-dependent properties within nanometer size ranges. These colloidal NCs

    differ from their bulk counterparts in their physicochemical properties, since their crystal size is in

    the order of the materials’ exciton Bohr radii, where electron transfer resembles molecular

    behavior as opposed to bulk.[1, 2] These NCs are also called quantum dots (QDs) due to their

    quantum confinement effect and 0-dimensional discrete energy levels. Quantum confinement

    effect can be translated to tunable electronic bandgaps between the occupied valence band and the

    unoccupied conduction band in semiconductor NCs.[1] Since introduction of quantum

    confinement in colloidal CdS NCs by Rossetti et al. in 1983, nanomaterial research has accelerated

    with considerable efforts on controllable synthetic routes, characterization, and study of

    fundamental properties of these materials.[3, 4] In addition, crystallinity of the inorganic core

    makes it possible to controllably manipulate surface energies, and to consequently change

    thermodynamic behaviors of the NCs.[1] Hence, precise control of the size, shape and composition

    of semiconductor NCs has attracted several researchers. The high level of interest in such materials

    is regarded to the tunable properties quantum confinement can result in. By increasing the size of

    the NC, the bandgap energy decreases, leading to an increase in the absorption and emission

    wavelengths. Therefore, by changing the size of particles, a wide range of spectral properties can

    be achieved to cover the entire visible color spectrum. This property has attributed to various

    applications, including biological imaging, detection and therapeutics, photovoltaics, sensing, and

    optoelectronics.

    1.2. Evolution of Synthetic Methods Although solid-state synthesis of nanoscale semiconductors was achieved in the 1970s using top-

    down approaches such as molecular beam epitaxy [5] and metal-organic chemical vapor

    deposition,[6] Kalyanasundaram et al. introduced the first bottom-up approach, arrested

    precipitation, to form soluble semiconductor colloids in aqueous solutions, overcoming photo-

    degradation in previous methods.[7] This method used slow injection of aqueous metal salts such

  • 2

    as Cd(SO4)2 into an alkaline solution containing chalcogenide salts such as (NH4)2S. After a

    double-displacement reaction resulting in CdS NCs, the extremely low solubility of the product in

    the aqueous solution led to precipitation, with styrene/maleic anhydride copolymer preventing

    agglomeration.[8, 9] However, this method exhibited low photoluminescence quantum yield

    (PLQY) of NCs due to charge trapping and photo-corrosion, as well as challenging size control

    due to Ostwald ripening. Another method to improve the size distribution of various NCs

    significantly was templated growth within structured media. In the original work by Meyer et al.,

    within a reversed micelle of sodium di-2-ethylhexyl sulfosuccinate (AOT), an organic iso-octane

    solvent was used containing CdCl2 and (NaPO3)6, into which CdS and H2S were introduced.[10]

    Therefore, CdS NCs were entrained within the micelle. Similar approaches using different micelle

    materials were implemented over the following few years, with sustained efforts to improve size

    distribution. Combining reagents in arrested precipitation and the micelle material from templated

    growth led to successful production of CdSe NCs by Steigerwald et al. In this method,

    Cd(ClO4)2.6H2O solution was emulsified in AOT reversed micelles, and organometallic precursor

    of bis(trimethylsilyl)selenide was injected into the solution.[11] With successful synthesis of

    narrow size-distribution CdSe NCs, subsequent research was followed on organometallic

    precursors solutions. Steigerwald et al. utilized tertiary phosphine chalcogenides for the first time

    as highly-reactive chalcogen sources, with triethylphoshpine telluride and Hg metal to yield HgTe

    NCs. Following this method, various NCs were synthesized benefiting from the instability of

    tertiary phosphine chalcogenides. It was later shown by Brennan et al. that branched phosphine

    complexes can be inserted in metal-carbon bond when metal alkyl compounds (eg. Cd(CH3)2) were

    used as the cation source.[12]

    1.3. Hot Injection Synthesis Inspired by the use of organometallic precursors in reversed micelles, Bawendi et al. used a melt

    of high boiling mixtures of tri-n-butylphosphine (TBP) and tri-n-butylphosphine oxide (TBPO),

    and annealed the intermediate nanoclusters leading to highly-luminescent NCs.[13] Consequently,

    a method referred to as hot injection was developed by Murray et al., in 1993, which ever since

    has proven to be the most controllable and reproducible approach for synthesis of metal

    chalcogenide NCs.[4, 8] This method included injection of organometallic precursor solutions of

    dimethyl cadmium (Cd(CH3)2) and selenium (Se) into a hot coordinating solvent and ligand (Tri-

    n-octylphosphine oxide (TOPO), and Tri-n-octylphosphine (TOP)),[14] where the oxide sites

  • 3

    would bond to Cd2+ ions, and the phosphine sites would bond to the chalcogenide, in order to

    create a neutrally-charged nanocrystal with an equal number of cations and anions. Organic ligands

    were used to passivate the surface of the nanocrystal and to control the size and shape, confirming

    Steigerwald’s hypothesis.[11, 15] In order to decrease safety risks of this toxic Cadmium salt and

    to eliminate thermodynamic limitations of the rapid injection approach, Peng and Peng suggested

    CdO, as the source of Cd2+ ion using a non-injection one-pot synthesis method, which has ever

    since become the most common approach for synthesis of Cd-based nanocrystals.[16, 17] Wu and

    Peng also synthesized high-quality CdS, CdSe and ZnSe nanocrystals with a non-coordinating

    solvent (1-octadecene (ODE) and ligand (oleic acid (HOA)).[18] By introduction of one or both

    of the reagents in the solvent at elevated temperatures, a burst of nuclei occurs followed by growth

    of NCs at lower temperatures. In order to control the size of NCs, the reaction is terminated by

    quenching. This method was developed to increase the potential for greener synthesis, scaling up

    and fundamental studies of the crystallization phenomenon. Despite several studies for other

    possible non-injection methods, the final nanocrystal quality is not comparable to abovementioned

    methods.[12, 18, 19] Therefore, this work has pursued based on adaptations of hot injection

    method for synthesis of CdTe NCs.[18, 20-22]

    In addition, since introduction of hot injection method by Murray et al.[14] more than two decades

    ago, several groups have investigated the synthesis, growth and characterization of semiconductor

    NCs,[15, 17, 19, 23-26] and more recently, precursor reaction mechanisms and kinetics.[27-31]

    The dependence of precursor conversion on concentration, solvent type and ligands has been

    established in many colloidal synthesis reactions.[15, 29, 32] However, a more complete

    fundamental understanding of precursor behavior is still limited by contradictory results and the

    quest to expand the range of candidate chemistries available for the preparation of air-sensitive

    materials. For example, conflicting literature reports on the effect of alkylamine ligands have been

    attributed to a lack of mechanistic understanding of reactions.[30, 33, 34] In addition, the

    continued discovery and improvement of air-sensitive precursor chemistries will remain a

    necessary step to further improve consistent control over NC size and shapes at high yield, at

    reduced cost and in environmentally-benign manners. As recently suggested for the case of metal

    chalcogenide nanocrystal synthesis,[31] new chalcogen sources promise more controlled and

    tunable conversion kinetics. Increasingly, limitations associated either with current precursor

    reactions or preparatory approaches become bottlenecks in preparing semiconductor NCs with

    more consistently defined size, shape, and charge transfer properties.[29]

  • 4

    1.4. Reaction Platforms Since the introduction of microfluidics-based synthesis of NCs in 2003,[35] several researchers

    have gravitated towards flowable approaches to produce highly-controllable nanoparticles with

    precursor solutions that were prepared ex situ.[36-47] In spite of the increased control over

    nucleation and growth stages and the fidelity gained by employing flow chemical formats, the

    upstream precursor preparation process so far remained unchanged, relying on a Schlenk line (SL)

    apparatus introduced by Wilhelm Schlenk more than one century ago.[48] He invented a set of

    glassware and Teflon valves which could create inert and vacuum environments efficiently and

    within one platform. In addition to SLs, gloveboxes have found some popularity due to elimination

    of handling dangerous glassware within an inert environment. Gloveboxes are continuously

    controlled to remove any air and moisture from the closed apparatus. These methods have become

    the standard method for handling air- and moisture-sensitive materials since 1913.[4] Nowadays,

    depending on the available laboratory space, sensitivity of the material to exposure to air, and the

    ease of handleability, said infrastructures find popularity. Both of these capital-intensive

    approaches limit operation outside of closed environments, while enabling chemical synthesis of

    materials with unprecedented qualities. The head space of SLs is connected to vacuum and inert

    gas lines to ensure an inert and moisture-free gas atmosphere. The preparation of a precursor

    solution begins by adding the starting materials to a flask. By increasing the temperature of the

    continuously-stirred solution above the boiling point of water and by applying a vacuum below

    300 mTorr, water vapor is removed. A subsequent further increase in the solution temperature

    results in a color change that is indicative of the formation of ionic bonds and the dissociation of

    solid starting materials. The reaction is thermally quenched and the prepared precursor solutions

    are collected under inert conditions. The use of (SL) increasingly becomes the bottleneck that is

    in part responsible for undesirable batch-to-batch variations, inflexibility and lack of scalability

    due to the dependence on fume hoods, mass and heat transport limitations and lack of in-line

    characterization. Removing this problem will therefore be a necessary requirement to

    comprehensively study air-sensitive precursor conversion reactions and demonstrate the flow

    chemical preparation of NCs all the way from the starting materials. Inspired by this Peng’s

    approach, we have developed a platform to portably and controllably perform air-sensitive

    processes.[16, 17] We combine the favorable heat and mass transfer associated with bubble

    columns with the controlled fluid delivery of microfluidic devices in a platform that enables rapid

  • 5

    and reproducible preparation of precursor complexes. We hope to enable study of nanocrystal

    synthesis without inherent limitations of Schlenk line techniques, towards more convergent

    literature results, greener chemicals and conditions for nanocrystal synthesis. Schlenk Line-to-Go

    enables rapid and controllable preparation of air-sensitive solutions.

  • 6

    Chapter 2

    2. Experimental The rapid preparation of of moisture- and air-sensitive NC precursor complexes is presented in the

    Schlenk Line-to-Go (SL2G) platform with a considerably smaller footprint, a high degree of

    repeatability, and amenability to automation and scale-out. Dynamic monitoring of the absorbance

    of the solution indicated complex formation. The chemical integrity of the precursor solutions was

    validated using 1H, 31P and 113Cd NMR and Mass Spectrometry (MS). To demonstrate the in situ

    preparation of NCs using adapted hot injection method, we selected CdTe NCs as a case study

    with availability of well-documented protocols for the preparation of cadmium-oleate (Cd-OA)

    and tri-n-butylphosphine telluride (Te-TBP) precursor solutions.[14, 18, 20-22, 27, 49] NCs were

    characterized based on their UV-vis and photoluminescence (PL) spectra, Transmission Electron

    Microscopy (TEM) and Powder X-ray Diffraction (PXRD). The details of experimental apparatus

    and assembly, materials and methods for characterization are discussed further.

    2.1. Experimental Apparatus Briefly, all materials were mixed via bubbling for at least 5 minutes at 130°C to ensure oxygen

    and water vapor were removed from the system. The excess fumes were drained through the top

    vent line. The liquid was heated above 200°C to increase solubility of solids, and to ensure

    complex formation. Change of solution turbidity and optical intensity indicated reaction

    termination point. Consequently, waste was drained, and the final product was pumped back into

    the chip and collection vial. Figure 1 schematically compares the preparation of precursor

    solutions between the SL and the SL2G approaches. To represent unit operations, Te-TBP

    precursor was selected as the green solution shown in the following figures. As shown in Figure

    1a, with an estimated volume of 1.8 L, the SL2G platform possesses nearly 2 orders of magnitude

    smaller of a volume compared to a simplified SL with an estimated volume of 82.5 L. In addition,

    based on nearly-identical Te-TBP preparation experiments in both systems, Figure 1b shows a

    reduction of experiment time by more than two-fold. Both systems were used for preparation of

    10 ml solution at the same temperatures. With a similar approach of addition of starting materials,

    dissolution/complex formation and quenching, both Cd-OA and Te-TBP precursors were

    prepared, followed by synthesis of CdTe NCs in SL and SL2G.

  • 7

    The cross-sectional image of the SL2G platform is shown in Figure 2a, where the starting

    materials are sequentially introduced into the vertical glass cylinder from above, via fluid inlets

    provided within a customized cap. The latter encloses the top of the cylinder and accommodates a

    thermocouple for measurement of the time-dependent temperature of the precursor solution, T(t).

    On the bottom, the cylinder is enclosed via a temperature-controlled mantle with embedded

    cartridge heaters and provision for perfusion of cold water, and a silicon-based microfluidic device.

    The microfluidic device contains at its bottom side two microchannel networks that are capped

    with an anodically-bonded glass layer (see Appendix A for details). After all of the starting

    materials are introduced into the continuously bubble-perfused cylinder and water vapor is

    removed by increasing the solution temperature above 100°C, the temperature is further increased

    to completely dissolve all solids. The solution’s change in color and absorbance as assessed by a

    time-dependent intensity measurement from the side (camera 1) is indicative of the desired

    complexes to be formed and allows to objectively determine when this step is completed. Upon

    completion of the precursor preparation, the solution temperature is lowered by perfusing cooling

    water through the mantle. The precursor is then collected into a nitrogen-purged vial by opening

    the outlet and increasing the pressure applied to the nitrogen-filled head space of the column. A

    Figure 1. Comparison of Schlenk Line (SL) and Schlenk Line-to-Go (SL2G) platforms. (a) Schematic of SL and SL2G platforms with 2 orders of magnitude reduced volume in the latter case. (b) Schematic comparison of precursor preparation methods using SL (top) and SL2G (bottom) approaches. Symbols indicate (1a) solvent and (1b) solid addition, (2) air/moisture removal via heating above 100°C, (3) solid dissolution and complex formation, (4) cooling and collection.

  • 8

    detailed instrumentation diagram and the assembled SL2G are illustrated in Figure 3 for

    preparation of Te-TBP precursor. The process is monitored using cameras 1 and/or 2, the PID

    temperature and pressure controllers to ensure leak-tight and controlled operation, while cycling

    temperatures. To highlight the portable nature of the SL2G platform as well as its ability to prepare

    air- and moisture-sensitive precursors outside of a fume hood, the experimental setup was

    successfully operated within a secondary containment (commercially available metal lantern) as

    shown in the photograph Figure 2c.

    Figure 2. Illustration of SL2G platform operation with MF device. (a) Schematic illustration of sparged reservoir employed for precursor preparation in SL2G platform with thermally-conducting layers (gray), insulating layer (nude), and precursor liquid in column (green), cameras for two fields of view and bottom image of microfluidic device Scale bar: 10 mm. (b) Bright field photograph of microfluidic device obtained from camera 2. Scale bar: 1.5 mm. (c) Photograph of SL2G platform operated within portable secondary containment.

  • 9

    Figure 3. Experimental setup and control systems. (a) P&ID control system with gas ( ), liquid ( ), starting materials ( ) and (b) photograph of experimental setup used for precursor preparation, (c) Schematic of SL2G setup for Te precursor preparation.

  • 10

    Experimental Setup: The apparatus shown in Figure 3c, consisted of a thermal insulation layer

    and two thermal conduction layers to help dissipate and conduct heat, respectively. A 2.25 cm2

    square microfluidic device was embedded between an aluminum and a ceramic layer, and

    compression-sealed using a PTFE sheet gasket (McMaster-Carr). A pyrex glass cylinder (OD: 14

    m, ID: 11.6 mm, Wale Apparatus) was sandwiched between both conduction layers for liquid

    holdup, and was sealed form both top and bottom using Paroflour (FFKM) O-rings (2-011/FF200

    and 2-015/FF200, Parker). Starting materials were introduced through the top conduction layer,

    and inert gas was injected through the bottom conduction layer, bubbled through the microfluidic

    device into the bubble column. Liquid temperature was controlled with a PID temperature

    controller, with an immersed Transition Joint thermocouple in liquid and a cartridge heater in the

    bottom Aluminum layer.

    Experimental Assembly: Standard PEEK and SS adaptors, PFA and SS tubing were used and

    installed on both Aluminum layers of the manifold initially (IDEX Health and Science). As shown

    in Figure 4a, after lubrication of O-rings (2, 4), they were placed in both top and bottom Aluminum

    layers (1, 5). The glass tube (4) was diamond-filed and lubricated to be sealed by O-rings. The

    microfluidic device (7) and laser-cut PTFE sheet gasket (6) were placed in the bottom Aluminum

    layer (5) with through fluidic connections aligned. The quartz disc (8, McMaster-Carr) was placed

    in the glass-mica ceramic layer (9, McMaster-Carr) and inverted to be attached to (5) by

    compression sealing. After insertion of (3) into (4), (1) was pressed into (3). After addition of an

    O-ring to (9), the pyrex petri-dish (10, vwr) was placed as a secondary container in case of

    apparatus leakage. After the apparatus was tested and proven to be leak-tight, this peace was

    removed to facilitate bottom-view imaging of the microfluidic device. Details of microfluidic

    device and sealing designs with drawings of individual parts of the SL2G manifold are illustrated

    in Appendices A. In addition, microfabrication of the device is discussed in Appendix B.

  • 11

    Figure 4. Schematic of detailed parts of SL2G manifold; (a) Exploded cross-sectional view with (1) Top conducting layer, (2) Top O-ring, (3) Pyrex Column, (4) Bottom O-ring, (5) Bottom Conducting layer, (6) PTFE sheet gasket, (7) Microfluidic device, (8) Quartz disc

  • 12

    2.2. Materials and Methods This section includes materials and their compositions, control approaches as well as

    characterization methods for both precursor and NC syntheses.

    Chemicals: All reagents were used as received without further purification. Te (99.8%), CdO

    (99.99+%), tri-n-octylphosphine (TOP, 90%), tri-n-butylphosphine (TBP, 97%), Tri-n-

    butylphosphine oxide (TBPO, 95%), oleic acid (HOA, 90%), oleylamine (OLA, 70%), 1-

    octadecene (ODE, 90%) were purchased from Aldrich. Toluene-d8 (99.5%) was obtained from

    Cambridge Isotope Laboratories Inc., and hexanes (ACS grade), chloroform (ACS grade), acetone

    (ACS grade) and isopropanol (ACS Reagent grade, 99.5+%) were purchased from Caledon

    Laboratory Chemicals. Ethanol (anhydrous, 100%) was purchased from Commercial Alcohols.

    Precursor Compositions: Precursors were prepared using two protocols within SL (Protocol 1)

    and SL2G (Protocol 2) platforms, followed by preparation of CdTe NCs in a segmented-flow

    reactor, and SL2G, respectively. In Protocol 2, fixed molar concentrations from Protocol 1 were

    used for reduced volume of 5 ml. Protocol 3 was prepared considering CdO:Te (2 mmol:4 mmol)

    as compared with other literature.[27, 30, 50] This method was simplified to Protocol 4 in order to

    study precursor coordination without interfering effects from organic ligands and solvents. Typical

    values for all protocols are listed in Table 1and Table 2.

    Table 1. Preparation composition of Cd-OA precursor solution for CdTe NC Synthesis

    CdO ODE HOA Total

    Protocol 1

    [18, 20-22] Molar Concentration (M) 0.031 2.403 0.731

    Volume (ml) 0.006 10.000 3.000 13.006

    Protocol 2 Molar Concentration (M) 0.031 2.403 0.731

    Volume (ml) 0.002 3.844 1.153 5.000

    Protocol 3 Molar Concentration (M) 0.341 1.813 1.313

    Volume (ml) 0.032 3.500 2.500 6.032

    Protocol 4 Molar Concentration (M) 0.678 0.000 3.135

    Volume (ml) 0.032 0.000 3.000 3.032

  • 13

    Table 2. Preparation composition of Te-TBP precursor solution for CdTe NC Synthesis

    Te ODE TOP TBP OLA Total

    Protocol 1

    [18, 20-22]

    Molar Concentration (M) 0.058 1.102 0.659 0.470 0.714

    Volume (ml) 0.010 3.000 2.500 1.000 2.000 8.510

    Protocol 2 Molar Concentration (M) 0.058 1.102 0.659 0.470 0.714

    Volume (ml) 0.006 1.763 1.469 0.588 1.175 5.000

    Protocol 3 Molar Concentration (M) 0.660 0.771 0.553 0.658 0.999

    Volume (ml) 0.082 1.500 1.500 1.000 2.000 6.082

    Protocol 4 Molar Concentration (M) 1.314 0.000 0.000 3.896 0.000

    Volume (ml) 0.083 0.000 0.000 3.000 0.000 3.083

    CdTe NC Synthesis via Segmented-Flow Reactor: Using Protocol 1, precursors were transferred

    to glass syringes, and were run via two automated syringe pumps (high pressure OEM syringe

    pump module, Harvard Apparatus). Syringes were attached to PFA tubing (1/16” ID, IDEX Health

    and Sciences LLC.) and fluid was introduced to an Aluminum manifold. The design of the

    microfluidic device and manifold was achieved by Dr. Milad Abolhasani during his PhD. studies

    at Guenther laboratory. With an inherent separation of the cold and hot regions on the device, use

    of hot injection was made possible through halo etches in the device. Therefore, the device was

    not at a uniform temperature, preventing pre-mature nucleation and growth. Cd-OA and Te-TBP

    precursors were introduced in the hot and cold regions, respectively, and Te-TBP was injected into

    Cd-OA at high temperature. By introducing an inert gas in the T-junction of the device, the reaction

    medium was divided into trains of liquid parcels separated by gas bubbles. Gas pressure was

    controlled using a digital pressure regulator (Type 3410, full scale: 0-15 psig, accuracy of 0.5 %

    full scale, Marsh Bellofram). Temperature of the reaction (hot) region was controlled using a PID

    controller, with the help of a machined Aluminum heating jacket attached to the reactor via a high

    temperature epoxy (OmegaBOND, Omega Engineering Inc.). The RTD temperature sensor was

    inserted into the custom-made slit, and the cartridge heaters were inserted into through-holes of

    the heating jacket. The reaction was monitored with a high-speed CMOS camera (pco 1200hs,

    PCO), where image-based gas velocity was calculated and the product was characterized similar

    to the SL2G platform. Residence times were calculated based on known reactor volume, liquid/gas

    length ratio and gas velocity.

  • 14

    SL2G-prepared CdTe NC Experimental Conditions and Characterization: Based on Protocol 2

    and 3, Te precursor was initially prepared, followed by Cd precursor. Te precursor was then

    injected into Cd precursor at high temperatures and while bubbling. Gas pressure was controlled

    similar to the segmented flow reactor with a range of 0.1-0.8 psig (see above). An apparatus was

    dedicated to each operation in order to prevent contamination and crosstalk between precursors

    causing undesired nucleation or growth. In order to characterize the system in real-time,

    temperature measurements from the PID controller was interfaced with camera 1 to capture

    images. Later, a MATLAB code was used for image-based intensity analysis, and temporal

    behavior of both variables (T(t) and I(t)) is shown. In addition, after preparation of precursors,

    samples of free ligands and solvents, precursors, and eventually NCs were characterized for

    chemical integrity and quality. A variation of NMR (1H, 31P and 113Cd), MS, FTIR, UV-vis/PL,

    TEM, and PXRD were utilized. The details of characterization methods will be discussed in this

    section.

    Temperature Control and Optimization: Using a Platinum PID temperature controller combined

    with a solid-state relay (CN16DPT-304-C24, G3NA-205B, Omega), the diagram shown in Figure

    5 was set-up for SL2G platform, and controller parameters were configured via the computer

    interface. Later, similar configuration was used for SL, where the CSH cartridge heater was

    replaced with a heat mantle (180W, Glas-Col). Both setups were initially autotuned for optimum

    performance with different timeouts, followed by manipulation of P, I, D parameters and output

    modes. A compromise between minimum overshoot percentage and rise time was selected for

    further experiments.

    Image-based Intensity Measurements and Analysis: With a possibility to image the apparatus

    from two views using cameras 1 and 2 (see Figure 2a), a high-speed CMOS camera (pco 1200hs,

    PCO) was used. An image of the microfluidic device is shown in Figure 2b in order to monitor

    fluid behavior within the channels and ports of the device (view from camera 2). In order to

    perform intensity measurement over the entire length of experiment, framerate of the camera was

    reduced drastically to 0.94 fps (corresponding to maximum 1.5-hour experiment) and view from

    camera 1 was used with the side-view of the bubble column as the region of interest. The apparatus

    was enclosed within two perpendicular white foam slabs to prevent environmental intensity

    artifacts and reflections from the glass cylinder. A white light illuminator (Doal 75, Microscan

    Systems Inc.) was used either in-line with the camera lens or under the apparatus to illuminate the

  • 15

    solution in the column without any background shadows (see Figure 3b). Temperature and images

    were recorded simultaneously throughout the experiment. The MATLAB code (see Appendix

    A.3.) was used to measure grayscale intensity and plot it as a function of time. It needs to be

    mentioned that the intensity measurement in this code was relative, and the value was normalized

    with respect to the global maximum and minimum. Spikes in the data correspond to noise in the

    images captured caused by inevitable change of background or reflected light from the cylinder.

    Figure 5. Electrical diagram of temperature control system

  • 16

    NMR Sample Preparation and Characterization: In 5 mm standard NMR tubes, two methods for

    preparation of air-free samples were used in order to substitute J Young NMR tubes. In both

    systems, at least 3 degas/purge cycles were completed using a vacuum pump and flowing nitrogen

    within a double-manifold Schlenk line. Method 1 is illustrated in Figure 6a, where a tightly-fitted

    combination of Tygon 33, Nylon 6, PVC and PEEK tubes and fittings and a PEEK four-way valve

    were used. It was ensured that all connections to the sample and toluene-d8 were through PEEK

    and pneumatic air connections in order to minimize contamination or damage. The Tygon tube

    was connected to the Schlenk line from one end, and through a pneumatic air tube, an adaptor and

    PEEK tubing was connected to port #1. Port #2 was connected to PEEK tubing and a flexible PVC

    tube to seal around the NMR tube. With a pressure relief/vent connection to port #3, all samples

    were introduced through port #4. After initial degassing and purging of the setup, the NMR tube

    were filled with 1 ml of toluene-d8, degassed and purged with nitrogen multiple times, followed

    by adding test sample, mixing, degassing and purging multiple times. Method 2 (see Figure 6b)

    included use of a Stainless Steel needle (6” long 18 G, Aldrich), NMR septum (Kimble™

    Kontes™, Fischer) attached to the NMR tube and sealed using Parafilm tape (Aldrich). The Tygon

    tube was connected to the Schlenk line, and the apparatus was degassed and purged with nitrogen

    initially, with a vent needle with the same diameter. The samples were added in sequence under

    nitrogen flow, degassed, and the process was repeated at least three times. TBP samples were

    prepared in both systems to study their applicability based on the degree of oxidation as a result of

    exposure to air. With 31P NMR, it was determined that Method 2 was more reproducible, with no

    detectable source of air contamination (formation of TBPO), and therefore was selected for further

    NMR studies. Confirmation of oxide formation was concluded through 31P NMR of commercial

    TBPO as shown in Figure C. 1. As shown in Figure 6c, Method 1 exhibits a peak corresponding

    to TBPO at 42.65 ppm, while Method 2 indicates both peaks associated with TBP. All NMR

    characterizations were performed using Agilent DD2 600MHz and O500MHz spectrometers at

    room temperature and in toluene-d8. 31P NMR spectra were referenced to trimethyl phosphine

    (C3H9P), acquired at 242.81 and 202.35 MHz, with a relaxation delay of 0.1 s and 1H decoupling. 113Cd NMR was referenced to dimethyl cadmium (Cd(CH3)2) at 110.87 MHz, with relaxation delay

    of 0.2 s, and 1H decoupling. 1H NMR was referenced to tetramethylsilane (Si(CH3)4) at 599.82

    MHz, with relaxation delay of 1.0 s and 13C decoupling.

  • 17

    Figure 6. Comparison of NMR preparation setups. (a) Method 1 with ports to (1) SL, (2) NMR tube, (3) Vent, and (4) Sample injection. (b) Method 2 while being purged with nitrogen, (c) 31P NMR results of TBP samples prepared by (a) and (b), with magnified inset showing oxide peaks at 52.98, 42.65, 28.34 ppm for Method 1 ( ), and 52.67, 41.45, 27.72 pm for Methods 2 ( ).

  • 18

    MS Sample Preparation and Characterization: In order to remove undissolved Te solid Te-TBP,

    crude sample was characterized in by centrifugation and filtration. For the former method, as-

    prepared samples of Te-TBP precursor were collected in nitrogen-filled centrifuge tubes,

    centrifuged at 6000 rpm for 5 min, and the supernatant was collected and transferred into nitrogen-

    filled vials, sealed with silicone septa. In the latter, as-prepared sample was filtered through a 0.2

    μm filter and transferred into a similar vial. Positive Direct Analysis in Real Time (DART)

    Ionization was used in ambient environment at 250°C, where the compound mass was calculated.

    With raw data of both methods shown in Appendix C, centrifuged sample was deemed less

    contaminated.

    Fourier Transform IR (FTIR) Sample Preparation and Characterization: A Perkin Elmer is50

    ATR spectrometer was used equipped with KBr crystal, where crude samples were deposited and

    analyzed within two sets of ranges between 550-4000 cm-1 and 100-1800 cm-1, with a number of

    scans of 8. Lower range of samples was aimed for characterization of phosphorus groups, while

    the higher range focused on organic groups in Cd-based samples. Additional spectrum for Te-

    based samples is shown in Appendix C.

    Spectroscopy Sample Preparation and Characterization: Using a Cary Eclipse and Varian 50 Bio

    fluorescence and UV-vis spectrophotometers (Agilent Technologies), samples were dissolved in

    n-hexane or toluene and studied for size estimation and quality, quantum confinement effect with

    respect to different residence times and precursor preparation approaches.

    TEM Sample Preparation and Characterization: CdTe samples were dissolved in equal volumes

    of isopropanol and chloroform in centrifuge tubes (Falcon), mixed with a vortex mixer, and

    centrifuged for 5 min at 6000 rpm. Supernatant was removed and the sample was re-dissolved in

    the same mixture, and centrifuged at least twice more. The samples were prepared by drop casting

    the nanoparticles in acetone and isopropanol solutions on a carbon/formvar coated 400-mesh TEM

    copper grid. (Prod #: 01824, Ted Pella, USA). Microscopy analysis was carried out using the

    Hitachi HF-3300 TEM/STEM/SEM at 300 kV.

    XRD Sample Preparation and Characterization: 2.5 ml of CdTe samples was dissolved in 7.5 ml

    of methanol, centrifuged for 2 × 5 min at 6000 rpm, after which the supernatant was drained, the

    sample was re-dissolved in toluene and filled to 10 ml with methanol, and the process was repeated

    7 times. The final sample was dried under nitrogen pressure. PXRD was performed using a Rigaku

  • 19

    MiniFlex 600 diffractometer, with a graphite monochromatized Cu-Kα radiation (λ=1.5406 Å).

    Sample was loaded on zero-background Silicon single crystal, with a scanning interval of 10°< 2θ

    < 80° and a scan speed of 0.4°/min (see Figure 12b). In addition, in order to study the effect of

    temperature on crystal structure, after centrifugation of the sample for 5 times with above

    conditions, 1 ml of the solution was inserted in a nitrogen-filled glass vial, while being heated to

    250°C for 2 hours (see Figure 14).

  • 20

    Chapter 3

    3. Results and Discussions Precursor Real-Time Analysis: Figure 7a shows time-dependent measurements of T(t) and I(t)

    associated with Cd-OA precursor preparation in the SL2G platform during steps that resemble the

    ones schematically shown in Figure 7b (bottom). The temperature was controlled using a PID

    controller during each of the two steps in the heating process. Cooling was achieved using cold

    water circulation. Bright field images of the purged precursor solution were continuously captured

    using camera 1 (frame rate 0.94fps). During step (1a), i.e. for times 0

  • 21

    Figure 7. Time-dependent SL2G platform operation. (a) Measured time-dependent changes in solution temperature and visible light transmission during Cd-OA precursor preparation. (b) Bright field photographs (camera 1) of column after addition of CdO powder to HOA (top), and after formation of Cd-OA complex and cooling (bottom). (c) Photograph of SL2G platform operated within portable secondary containment. Scale bars: 10 mm.

  • 22

    Te-TBP Characterization: In order to examine coordination of Te precursor using SL2G, the

    simplified protocol was used with Te being dissolved and bound to TBP. Protocols 2,3 and 4 were

    used for this precursor as shown in “Materials and Methods”. Figure 8 illustrates 31P NMR and

    MS results of Te-TBP precursor prepared at 280°C. Figure 8a compares 31P NMR (242.81 MHz,

    toluene-d8, r.t.) resonances of TBP (top) and Te-TBP (bottom), with a downfield shift and line

    width broadening of the resonance from -32.51 ppm to -16.24 ppm. This 16.27 ppm downfield

    shift indicates the exchange between TBP and Te in close agreement with previously-reported

    values.[27, 28, 30] The difference between aforementioned and reported values of -31 ppm and -

    13.2 ppm by Anderson et al. and Liu et al. is attributed to characterization conditions, such as the

    deuterated solvent.[27, 28] Therefore, our results confirm the coordination of the phosphorus atom

    where the electron density decreases when Te bonds to phosphorus, leading to deshielding and

    broadening of the resonance peak.[51] After centrifugation of Te-TBP precursor, Te-TBP was

    studied using DART positive ion MS at 250°C, resulting in confirmation of Te pattern and

    fundamental structural formula of C12H27PTe, with abundance of 130Te. In Figure 8b, MS results

    validate the formation of the Te-TBP complex by detection of the tellurium intensity pattern, with

    the most pronounced peak of 130Te at 333.1 m/z. In addition, bands have been observed at 219.2

    m/z and 437.4 m/z (see Figure C. 2) that correspond to [TBPO+H]+ and [2TBPO+H]+ and are

    attributed to the MS measurement being performed at high temperature and under non-inert

    conditions. Also, results of Te-TBP MS from preparation by filtration are shown in Figure C. 4,

    with assigned peaks for 203.2, 219.2. 333.1, 421.4, 437.4 m/z, 551.3 m/z corresponding to

    [TBP+H]+, [TBPO+H]+, [Te-TBP+H]+, [TBP+TBPO+H]+, [2TBPO+H]+, [Te-TBP+TBP+H]+.

    The calculated and found masses of 333.09908 and 333.09995 with 12C121H2831P1130Te chemical

    formula is the same for both preparation methods confirming the chemical nature of the Te-TBP

    sample. However, the higher number of peaks found in Figure C. 4 are associated to exposure to

    air and filter material while processing the sample. Therefore, further analysis is pursued using

    centrifugation of Te-TBP precursors.

    To characterize the crude Te precursor for CdTe NC synthesis, Te-TBP/TOP/OLA/ODE was

    prepared and characterized using DART MS at 400°C due to the added solutes and solvent. As

    expected, various peaks were observed, which might exhibit interfering effects as well (see Figure

    C. 5). The highest intensities were attributed to [TBP+H]+, [TOP+H]+ and [2TBPO+H]+. A

    magnified region of the spectrum is shown confirming existence of Te-TBP complex, while other

    complexes were not all characterized due to complexity of the sample components. Liquid

  • 23

    chromatography (LC) is suggested for future work to separate components and characterize

    individual compound patterns.

    In order to further verify coordination of the phosphine-based ligand to Te for preparation of CdTe

    NCs, samples of TBP, TOP, a mixture of TBP/TOP/OLA/ODE and the prepared Te-

    TBP/TOP/OLA/ODE (Te precursor for CdTe synthesis) were characterized using 31P NMR and

    are shown in Figure 9. Dashed lines indicate defined peaks corresponding to the vicinity of free

    ligands of TBP and TOP and bound Te-TBP and Te-TOP as well as TBPO/TOPO. The common

    peak of phosphorus vicinity is dominant at -32.62 ppm for all free ligand mixtures (a, b, c). The

    resonance peaks of TBP in (a) and TOP in (b) are at -18.88ppm and at -19.56ppm. These peaks

    can be determined similarly in the mixture of TBP/TOP/OLA/ODE (c) with -18.83 ppm and -

    19.56 ppm corresponding to TBP and TOP, respectively.

    As can be seen in Figure 9, the resonances all shift downfield and broaden, and (1), (2) and (3)

    indicate the shift of resonance peaks with respect to coordination of Te to phosphine sites in Te-

    TBP/TOP/OLA/ODE mixture (d). With resonance peaks at -24.43 ppm, -7.74 ppm and -8.85 ppm,

    sample (d) exhibits downfield shifts in resonance peaks of (1) 8.19 ppm, (2) 11.09 ppm and (3)

    10.71 ppm, all of which are consistent in direction and magnitude. Therefore, coordination of Te

    Figure 8. Characterization of Te-TBP precursor solution prepared in SL2G platform (Protocol 4). (a) 31P NMR (242.81 MHz, toluene-d8, δTBP, δTe-TBP, ppm): -32.51, -16.24. (b) MS of Te-TBP using positive polarity DART Ionization at 250°C.

  • 24

    to both TBP and TOP is concluded. Small peaks for oxidation are present within 52.66 ppm and

    27.72 ppm for all samples and visible in the magnified inset, in agreement with previously-reported

    values for TOPO and TBPO impurities.[27, 28]

    Figure 9. 31P NMR of phospine-based ligands and Te precursor for CdTe NC synthesis (Protocol 3). (A) and commercial TBPO indicating partial oxidation of the precursor. (a) 31P NMR TBP (202.35 MHz, toluene-d8, δ, ppm): -19.56, -32.64, -37.74. (b) 31P NMR spectrum of TOP (202.35 MHz, toluene-d8, δ, ppm): 52.78, 41.39, 27.74, -18.88, -32.58, -37.70. (c) 31P NMR spectrum of TBP/TOP/OLA/ODE (202.35 MHz, toluene-d8, δ, ppm): -18.83, -19.56, -32.62. (d) 31P NMR spectrum of Te-TBP/TOP/OLA/ODE (202.35 MHz, toluene-d8, δ, ppm): -7.74, -8.85, -24.43.

  • 25

    Cd-OA Characterization: As mentioned in “Materials and Methods”, crude Cd-OA precursor for

    CdTe NC synthesis included ODE as the non-coordinating solvent (Protocol 2 and 3). However,

    similar to Te-TBP precursor, in order to study the direct behavior of complex formation and the

    quality of the precursor, the simplified protocol was used with CdO being dissolved and bound to

    HOA (Protocol 4). Figure 10 illustrates 1H and 113Cd NMR results of the crude Cd-OA precursor

    prepared at 200°C, both of which demonstrate coordination of the acidic proton and the Cd2+ in

    the precursor, indicative of complex formation and in agreement with previous literature.[28, 30,

    31] According to 1H NMR (599.82 MHz, toluene-d8, r.t.) of HOA and Cd-OA solutions in Figure

    10a, the resonance peak of the carboxylic acid proton (-COOH) is shifted upfield from 12.43 ppm

    in HOA (top) to 11.36 ppm in Cd-OA (bottom), and the intensity of the resonance decreases by

    49% due to near-stoichiometric consumption of two HOA molecules by one CdO.[28, 30] Since

    excess amount of HOA was used in the reaction, the acidic proton is detected in the precursor as

    well as the solvent (HOA) (2 mmol CdO: 16 mmol HOA). Excess acid was successfully removed

    and the sample was characterized further (see Figure 11). Broadening of the peak line width

    indicates coordination of the HOA to Cd2+ resulting in a total of 66.24 protons, which is within

    0.36% of the Cd-OA chemical formula (C36H66CdO4). This value is consistent with the total

    number of protons in HOA, which was determined to be 33.91 within 0.26% of the chemical

    formula of HOA (C18H34O2). These values were derived from integration of the 1H NMR data with

    toluene-d8 internal lock, and normalization with respect to the ligand vinyl resonances. The loss of

    two protons corresponds to the reaction between CdO and HOA towards formation of water.

    However, lack of a resonance attributed to water indicates successful removal of moisture using

    SL2G. The small peak around 0.21 ppm corresponds to silicone impurity from the liner of the vial

    caps, as shown in the magnified inset of Figure 10a.[31] As suggested by Figure 10b, 113Cd NMR

    of Cd-OA in toluene-d8 indicates a resonance peak at -703.06 ppm, referenced at 0 ppm using

    Cd(CH3)2. This result is in agreement with the previously-reported values with a difference of

    references between Cd(CH3)2 and Cd(ClO4)2, relatively.[30, 50] García-Rodríguez and Liu

    reported chemical shifts corresponding to coordination of carboxyl Cd2+ at -9.1 and -28.1 ppm in

    absence and at -48.7 ppm in presence of excess HOA, with a reference at -641.5 ppm.[30, 50] In

    addition, to compare our direct results of 113Cd NMR with abovementioned references, Cd(ClO4)2

    was used as a second reference for 113Cd NMR, with the peak at 644.95 ppm (see Figure C. 7.),

    leading to a 58.11 ppm chemical shift of Cd peak for Figure 10b, coherent with previous reports.

  • 26

    Using 1H NMR, it was confirmed that the excess amount of HOA was removed from Cd-OA

    sample by dissolving in 5 ml of acetone, centrifuging at 6000 rpm for 5 min, repeating the

    procedure for 6 times, and drying the white precipitate under nitrogen overnight. The sample was

    then dissolved in 1 ml toluene-d8 and transferred to a NMR tube with the characterization shown

    in Figure 11. The calculated number of protons in HOA and Cd-OA are 34.79 and 65.71, matching

    empirical formulas of HOA (C18H34O2) and Cd-OA (C36H66CdO4), respectively. Evident in the

    magnified region, there is no resonance corresponding to the acidic proton indicating successful

    removal of HOA from the sample.

    Figure 10. Characterization of Cd-OA precursor solution prepared in SL2G platform (Protocol 4) (a) HOA: 1H NMR (599.82 MHz, toluene-d8, ppm, δ): 12.43 (s, 1H; OH), 5.39, 2.00-2.08, 1.44-1.47, 1.13-1.36, 0.88-0.90; Cd-OA: 1H NMR (599.82 MHz, Toluene-d8, δ): 5.40-5.45, 1.53-1.55, 1.19-1.38, 0.89-0.92. Δ is attributed to silicone impurity from vial caps at 0.20-0.22 ppm, as shown in the magnified inset. (b) 113Cd NMR (110.87 MHz, toluene-d8, δCd-OA, ppm): -703.06, referenced to CdCl2 in D2O at 0 ppm.

  • 27

    CdTe NC Characterization: As mentioned in “Materials and Methods”, both available platforms

    of the segmented-flow microreactor and SL2G were used in order to prepare CdTe NCs. The

    segmented-flow microreactor was primarily utilized for preparation of CdTe NC cores, and CdSe

    shells in a two-pass approach. Therefore, precursors were prepared in conventional SLs (Protocol

    1), transferred to syringes and pumped into the reactor manifold. The relevant spectral results of

    CdTe core synthesis in this platform are compared with the second platform, where SL2G was

    used to prepare CdTe NCs in situ following precursor preparation. CdTe NC preparation was

    carried out according to previously-reported procedures (Protocol 2).[24, 27, 30, 52] As mentioned

    earlier, Cd precursor was prepared from dissolution of CdO (0.15 mmol) in a mixture of ODE (3.9

    ml) and HOA (1.2 ml). Te precursor prepared from dissolution of Te (0.3 mmol) in a mixture of

    ODE (1.8 ml), TBP (1.0 ml), TOP (1.5 ml) and OLA (1.2 ml). The Te precursor was injected into

    the Cd-OA precursor at 220°C. NC nucleation and growth took place during periods of 2.0 min

    (A) and 0.8 min (B), which was accompanied by a gradual increase in the visible light absorbance

    by the solution: from optically-clear to black. The normalized UV-vis and PL (n-hexane) spectra

    for both prepared NCs in situ from precursors in our SL2G platform were compared with CdTe

    NCs from conventionally-prepared precursors at 200°C ((C), see Figure 12a). The detailed reaction

    Figure 11. 1H NMR of HOA and Cd-OA precursor comparing removal of acidic proton (Protocol 4). 1H NMR HOA (599.82 MHz, toluene-d8, δ, ppm): 12.43 (s), 5.41, 2.06-2.09, 1.14-1.46, 0.90. 1H NMR Cd-OA (499.88 MHz, toluene-d8, δ, ppm): 5.47, 2.08, 1.29-1.56, 0.93.

  • 28

    conditions and results for NC synthesis are listed in Table 3. Identical Cd and Te precursors were

    prepared at 240°C and 280°C in a SL, respectively. They were then fed into a silicon-based

    microfluidic device at 70 µl min-1, for Cd-OA/ODE, and 5 µl min-1, for Te-TBP/TOP/OLA/ODE.

    The mixture was heated to 200°C and segmented by nitrogen bubbles at 1.0 psi. The absorption

    and emission wavelengths of CdTe NCs red-shift from 683 and 714 nm in (A) to 637 and 662 nm

    in (B), and to 598 and 625 nm in (C), respectively. Accordingly, the average NC diameters were

    predicted based on reported fitting functions as 5.3 nm in (A), 4.1 nm in (B) and 3.6 nm in (C).[53]

    The inset in Figure 12a represents a high resolution TEM (HRTEM) image of CdTe NCs with an

    average NC diameter of 5.3 nm. Using HRTEM images from condition (A), an averaged lattice

    spacing of 0.25 nm was determined by ImageJ software, where grayscale intensity of image pixels

    was plotted within a distance of 5 nm. The distances between each two consecutive peaks and

    valleys were determined and averaged (see Figure 13).

    As shown in Figure 12b, PXRD patterns ( ) of these NCs exhibit prominently wurtzite crystal

    structure, with lattice constants of a=b=4.5911(11) Å and c=7.575(2) Å. The overlapped refined

    pattern ( ) shows agreement of the observed and theoretical plots using Rietveld refinement

    method with χ2 of 2.84, and the broadening and attenuation of (102) and (103) peaks are associated

    with structural defects such as stacking faults.[14] Diffraction peaks are located at 22.44°, 23.57°,

    25.38°, 32.73°, 39.31°, 42.47°, 45.69°, 46.31°, 47.33° and 52.00°. The corresponding diffraction

    planes of bulk CdTe crystals with space group of P63mc are shown ( ), and agree well with

    literature values of 22.34°, 23.47°, 25.28°, 32.63°, 39.21°, 42.37°, 45.59°, 46.21°, 47.23° and

    51.90° according to the ICDD database (DB card number 01-079-3179). These values are in

    agreement with CdTe NCs prepared for PXRD via annealing indicating wurtzite structure (see

    Figure 14). PXRD pattern for this method exhibit wurtzite crystalline structure prominently with

    lattice constants of a=b=4.5698(12) Å and c=7.516(3) Å. The overlapped refined pattern ( ) shows

    agreement of the observed and ideal plots using Rietveld refinement method with χ2 of 2.28, and

    similar stacking faults.

  • 29

    Table 3. Properties of CdTe NCs prepared at 200°C and 220°C

    (A) (B) (C)

    Precursor Preparation Protocol (see Table 1and Table 2) 2 3 1

    Reaction Temperature (°C) 220 220 200

    Nitrogen Gas Pressure (psi) 0.2 0.3 1.0

    Residence Time (min) 2.0 0.8 1.6

    Absorption Peak Wavelength (nm) 683 637 598

    Emission Peak Wavelength (nm) 714 662 625

    FWHM (nm) 46 46 44

    Predicted NC size (nm) 5.3 4.1 3.6

    Figure 12. Characterization of prepared CdTe NCs in microfluidic platforms (a) Normalized UV-vis (solid line; A,B,C) and PL spectra (dashed line; A’,B’,C’) of CdTe NCs, with Te- and Cd-precursors, and CdTe NCs at 220°C all prepared in SL2G platform (A, A’, B, B’); conventionally-prepared Te- and Cd-precursors and downstream microfluidic CdTe synthesis at 200°C (C,C’,( )). CdTe NCs prepared at 220°C have residence times of 0.8 ( ) and 2.0 min ( ). Inset: HRTEM image of CdTe NCs with an average diameter of 5.3 nm (A). (b) PXRD pattern of CdTe NCs prepared at 220°C ( ) with Retvield refinement ( ) indicating χ2 of 2.84. Bulk PXRD pattern of CdTe ( ) is shown with observed planes.

  • 30

    Figure 13. PXRD pattern of CdTe NCs prepared at 220°C ( ) with Retvield refinement ( ). Bulk PXRD pattern of CdTe ( ) is shown with observed planes.

    Figure 14. Profile of lattice plane distances of 5.3 nm CdTe NCs synthesized at 220°C and 2.0 min residence time, with an average lattice spacing of 0.25 nm.

  • 31

    Precursor and NCs Thermal Behavior: Another set of experiments was designed to observe

    temperature and intensity changes with subsequent reactions. In order to prevent unreacted solids

    adhering to the column wall, material addition sequence was changed. It was ensured that after

    addition of solids, the remainder of the liquid was added with unchanged molar concentration.

    Results of a sequence of precursor and NC preparation are presented in Figure 15 with controller

    output values and temperature changes. Each step of the reaction is observed in the system

    response, and has been coordinated in the same fashion as shown in Figure 7a. Based on the step

    response assumption of the control system, thermal behavior of the above system is shown in

    Table 4. In addition, temperature profiles with intensity measurement shown in Figure 16 and

    Figure 17 followed thermal behavior shown in Table 4d.

    Based on Figure 15a, ODE was first injected, bubbled and heated to 130°C, followed by addition

    of CdO and OA at t=13, 28 min, respectively, with observed temperature drops of 51 and 29°C.

    Set point of the temperature controller was changed to 200°C at t=35 min, and the reaction was

    terminated at 5=39 min with visual determination of steady-state clarity of the Cd-OA precursor.

    The setup was then cooled using water circulation. The kink in the cooling region is due to

    withdrawal of the sample using a syringe from the head space, rather than head pressurization.

    Both Cd-OA and Te-TBP/TOP/OLA/ODE sample exhibited this complication, where in order to

    preserve sample quality, it was quickly withdrawn and transferred to a nitrogen-filled vial. For Te-

    TBP/TOP/OLA/ODE precursor shown in Figure 15b, after initial addition, mixing and heating of

    ODE and TOP, Te was added at t=17 min, TBP and OLA were added at t=25 min. Temperature

    drops of 76 and 18°C were observed, respectively, with a quick rise as soon as the controller was

    engaged.Set point of the temperature controller was changed to 280°C at 30 min, and the reaction

    was terminated at t=55 min with visual determination of steady-state clarity of the Te precursor.

    Finally, for CdTe synthesis, 3ml of as-prepared Cd-OA precursor was bubbled and sequentially

    heated to 130, 200 and 220°C, until addition of 3 ml of Te precursor at t=14.73 min. trxn was

    determined starting with this point and running until t=15.53 min, resulting in 48 s (0.8 min)

    residence time reported earlier in the text. Reaction was then terminated, cooled via water

    circulation, and the sample was collected in an inert vial using head pressurization. The change in

    cooling slope is regarded to the quick pumping of CdTe, resulting in nitrogen-filled glass cylinder

    at t=18 min.

  • 32

    Figure 15. Temperature profiles of precursor and NC preparations. (a) Cd-OA at 200°C. (b) Te-TBP/TOP/OLA/ODE at 280°C. (c) CdTe at 220°C.

  • 33

    Table 4. Time-dependent control properties of precursor and CdTe NCs in SL2G and SL. (a) Cd-OA @ 200°C

    (i) Heating from

    Tmin to 130°C

    (ii) Heating from

    130°C to 200°C

    (iii) Cooling from

    200°C to Tmin

    Time constant (min) 2.22 2.53 3.49

    Total required time (min) 11.09 12.65 17.47

    Percentage Overshoot (%) 7.58 -4.30 N/A

    (b) Te-TBP-TOP-OLA-ODE @ 280°C

    (i) Heating from

    Tmin to 130°C

    (ii) Heating from

    130°C to 200°C

    (iii) Cooling

    from 200°C to

    Tmin

    Time constant (min) 1.98 4.67 3.67

    Total required time (min) 9.92 23.36 18.33

    Percentage Overshoot (%) 13.23 2.94 N/A

    (c) CdTe @ 220°C

    (i) Heating

    from Tmin to

    130°C

    (ii) Heating

    from 130°C

    to 200°C

    (iii) Heating

    from 200°C

    to 220°C

    (iv) Cooling

    from 200°C

    to Tmin

    Time constant (min) 1.81 1.40 0.99 3.17

    Total required time (min) 9.03 6.99 4.94 15.87

    Percentage Overshoot (%) 12.42 10.56 30.42 N/A

    (d) Cd-OA in Conventional SL

    (a) Heating from

    Tmin to 130°C

    (b) Heating from

    130°C to 200°C

    (c) Cooling from

    200°C to Tmin

    Time constant (min) 3.39 1.06 6.33

    Total required time (min) 16.97 5.28 31.65

    Percentage Overshoot (%) 65.18 -62.38 N/A

    An apparent advantage of the SL2G over SL is the enhanced heat transfer and temperature control,

    as seen by heating time requirements and overshoot (Table 4a and d). Temperature plots of Cd-

    OA preparation are shown in Figure 16 and Figure 17 for optimized performance of SL and SL2G,

    respectively.

  • 34

    Temperature Control:

    For both conditions of SL and SL2G, an output mode of “ON-OFF” was chosen with a deadband

    of 2 s for all conditions after other possibilities were ruled out. During controller calibration,

    maximum overshoots of 1.3% and 3.4% were achieved for SL (P:4.3, I:0.1, D:76.1) and SL2G

    (P:7.0, I:0.0, D:70.0) for Tsp of 300°C respectively. However, in operation of both systems at their

    optimum conditions and sequential heating, overshoot was higher, with maximum observed

    overshoots of 65.18% for Sl and 13.23% for SL2G. The overshoot percentage is within acceptable

    range (

  • 35

    Intensity Measurements:

    As can be seen in Figure 16, for the case of conventional SL Cd-OA preparation, intensity of the

    liquid was measured as before, and for an interval of 0 to 18 min of the experiment, the intensity

    curve was fitted with a sigmoidal function with least squares fit, as shown below, with 95%

    confidence interval. The response behavior was categorized as first-order-plus-delay.

    Ifit = 4.4615 × 10−2 +3.3894 × 10−1

    1 + exp �− t − 8.33066.4377 × 10−1�

    Figure 16. Process response behavior in Cd-OA preparation in SL. (a) Temperature and scaled intensity plots with respect to time evolution. (b) Magnified region of intensity plot in (a) for maximum intensity change corresponding to complex formation. Sigmoidal function fit indicates first-order-plug-delay behavior of the solution.

  • 36

    As can be seen in Figure 17, for the case of Cd-OA preparation in SL2G, intensity of the liquid

    was measured as before, and for an interval of 20 to 38 min of the experiment, the intensity curve

    was fitted with a sigmoidal function with least squares fit, as shown below, with 95% confidence

    interval. Similar to SL, the response behavior was categorized as first-order-plus-delay.

    Ifit = 4.8440 × 10−2 +6.6390 × 10−1

    1 + exp �− t − 7.7409 × 101

    1.007 × 101 �

    Figure 17. Process response behavior in Cd-OA preparation in SL2G. (a) Temperature and scaled intensity plots with respect to time evolution. (b) Magnified region of intensity plot in (a) for maximum intensity change corresponding to complex formation. Sigmoidal function fit indicates first-order-plug-delay behavior of the solution

  • 37

    FTIR Results for Functional Groups in Precursors and NCs: To identify functional groups

    involved in precursors and NCs, FTIR was performed, with a list of assigned peaks shown in Table

    6, followed by stacked spectra shown in Figure 18.

    Table 6. Peak list for Cd precursor and NC solutions Peak (cm-1) Stretch Sample

    2500-3500 O-H (broad) [54] HOA

    2921 Asymmetric CH2 [54] HOA, Cd-OA, CdTe, TBP, Te-TBP

    2853 Symmetric CH2 [54] HOA, Cd-OA, CdTe, TBP, Te-TBP

    1707-1716 C=O [55] HOA, Cd-OA, CdTe

    1554-1569 Asymmetric COO- [29] CdTe, Cd-OA

    1169 P-O [56] TBP, Te-TBP

    459 P-Te [56] Te-TBP

    Figure 18 compares FTIR bands of Cd NCs (i, vii), free ligands (iii, iv) and precursors (ii, v, vi).

    In Figure 18a, the wavenumber range of 550-4000cm-1 was focused on to determine respective

    organic functional groups pertaining to Cd-related samples. It was observed that both (ii) and (iii)

    exhibit sharp peaks at 2853 cm-1 and 2921 cm-1, corresponding to symmetric and antisymmetric

    CH2, respectively.[54] The broad peak between 2500 and 3500cm-1 corresponds to the O-H stretch

    of HOA (iii).[54] CdTe (i), Cd-OA (ii) and HOA (iii) experience a peak at 1707cm-1, 1712 cm-1

    and 1716 cm-1 associated with C=O stretch. In addition, samples (i) and (ii) both exhibit a broad

    peak within 1554 cm-1 and 1569 cm-1 to symmetric COO- stretch in formation of oleate and NC.

    This observation agrees with a study by Wu et al., indicating HOA is chemisorbed as a carboxylate

    onto the CdTe NCs.[54] As shown in Figure C. 6, similar sharp peaks between all spectra at

    2921cm-1 and 2851cm-1 correspond to asymmetric and symmetric CH2. Furthermore, P-O stretch

    from TBP (A.iv) and Te-TBP (A.v) causes a relatively small but sharp peak at 1169cm-1,[55]

    indicating deposited samples TBP and Te-TBP were oxidized leading to TBPO formation due to

    exposure to the atmosphere during characterization. In order to study phosphorus coordination for

    Te-related samples, the lower range of 100-1800 cm-1 was selected. Although the band at 459 cm-

    1 is not sharp, it is attributed to the double bond of Te-P according to a similar value of 472 cm-1

    reported by Zingaro et al..[56] This peak is less pronounced for more complex samples such as

    (vi) and (vii), possibly due to typical interfering effects of other functional groups below 1000 cm-

    1.

  • 38

    Figure 18. FTIR spectra of precursors and NC solutions. (a) Comparison of CdTe NCs (i), Cd-OA precursor (ii) and free HOA (iii), with enlarged area indicating coordination of O-H band to form carboxylate. (b) Comparison of free TBP (iv) and Te-TBP (v), Te-TBP/TOP/OLA/ODE (vi), and CdTe NCs (vii), with similar behavior except for possible band for P-Te coordination in vii.

  • 39

    Chapter 4

    4. Conclusions and Outlook 4.1. Conclusions This work was inspired by colloidal synthesis of semiconductor NCs, and focused on preparation

    of air- and moisture-sensitive precursor solutions for such NCs. In summary, starting with Chapter

    1, semiconductor NCs, their synthesis methods and precursor solutions were introduced. Chapter

    2 focused on the experimental aspect of the introduced platform, Schlenk Line-To-Go (SL2G),

    materials and methods used for preparation and characterization of Cd, Te precursors and CdTe

    NCs. Finally, results and discussions on the prepared materials and the capabilities of SL2G was

    presented in Chapter 3. It was shown that SL2G allows the rapid and controllable preparation of

    air- and moisture-sensitive precursor solutions as well as the in situ preparation of semiconductor

    NCs. The SL2G platform is expected to be highly scalable as it requires 2 orders of magnitude less

    footprint compared with a conventional SL: In a standard 8’ fume hood only up to three SL systems

    can be installed and concurrently operated, compared with, based on our estimates, up to 600 SL2G

    platforms. The SL2G platform provides for improved process control via in situ temperature and

    turbidity measurements and reduces preparation times >1.5 fold for the fastest conventional

    syntheses, and up to 15 fold for samples which require overnight preparation. This advantage is

    due to the enhanced mixing and heat transfer behavior. We envision SL2G platforms to enable

    highly scalable and fully automated approaches for the discovery and consistent preparation of

    new precursor chemistries. Their seamless integration with flow chemical synthesis may help

    overcome a current bottleneck in the broad adaptation of flow chemical approaches in preparing

    colloidal nanostructures with high fidelity.

    4.2. Outlook With our introduction of a miniaturized, fast and reproducible platform for air- and moisture-

    sensitive materials, a great amount of potential is envisioned for versatile applications of SL2G.

    These can include study of mechanistic behaviors of known chemistries, synthesis of new

    materials, as well as automation and portable operation for scalable material synthesis. Some of

    the recommended next steps are followed for future work using SL2G. A suggestion for operating

    this platform under vacuum is application of negative relative pressure from inlet gas pressure to

  • 40

    the vent line, where a feedback control can be implemented. It is envisioned that based on an

    image-based bubble velocimetry, gas velocity and gas pressure can be determined at the sparging

    holes, followed by head space pressure readout. Dynamic behavior of gas flow considering

    capillary forces and temperature-dependence need to be taken into account. In addition, another

    suggestion for the system is to utilize the chip-based imaging possibility to determine bubble

    frequency, and hence, gas velocity. Similar approach can be taken to both automate the system

    and to operate at variable pressure levels. In order to confirm these suggestions, a series of pressure

    and velocity control platforms are to be developed.

    In order to characterize more complex precursors, separation processes such as Liquid

    Chromatography are recommended, whereby free ligands and coordinated precursors can be

    characterized without interfering results or the risk of degradation. In addition, intensity

    measurements of the liquid column will benefit from calorimetric analyses in order to determine

    reaction termination point based on the available library of desired product colors. Also, to better

    define the control system, both temperature and intensity measurements are to be taken as second-

    order systems, where damping parameters also play an important role. Therefore, by understanding

    the solution intensity/color change based on temperature change, mass transfer through bubbling

    and the required times, SL2G platform can be fully-automated.

  • 41

    References 1. Alivisatos, A.P., Perspectives on the physical chemistry of semiconductor nanocrystals.

    The Journal of Physical Chemistry, 1996. 100(31): p. 13226-13239.

    2. Murray, C.B., C. Kagan, and M. Bawendi, Synthesis and characterization of monodisperse

    nanocrystals and close-packed nanocrystal assemblies. Annual Review of Materials Science,

    2000. 30(1): p. 545-610.

    3. Rossetti, R., S. Nakahara, and L.E. Brus, Quantum size effects in the redox potentials,

    resonance Raman spectra, and electronic spectra of CdS crystallites in aqueous solution. The

    Journal of Chemical Physics, 1983. 79(2): p. 1086-1088.

    4. de Mello Donegá, C., P. Liljeroth, and D. Vanmaekelbergh, Physicochemical Evaluation

    of the Hot‐Injection Method, a Synthesis Route for Monodisperse Nanocrystals. Small, 2005. 1(12): p. 1152-1162.

    5. Cho, A. and J. Arthur, Molecular beam epitaxy. Progress in solid state chemistry, 1975.

    10: p. 157-191.

    6. Kohl, P.A., S.N. Frank, and A.J. Bard, Semiconductor Electrodes XI. Behavior of n‐and p‐Type Single Crystal Semconductors Covered with Thin Films. Journal of The Electrochemical

    Society, 1977. 124(2): p. 225-229.

    7. Kalyanasundaram, K., et al., Cleavage of Water by Visible‐Light Irradiation of Colloidal CdS Solutions; Inhibition of Photocorrosion by RuO2. Angewandte Chemie International Edition

    in English, 1981. 20(11): p. 987-988.

    8. Rossetti, R., et al., Size effects in the excited electronic states of small colloidal CdS

    crystallites. The Journal of Chemical Physics, 1984. 80(9): p. 4464-4469.

    9. Evans, C.M., et al., Review of the synthesis and properties of colloidal quantum dots: the

    evolving role of coordinating surface ligands. Journal of Coordination Chemistry, 2012. 65(13):

    p. 2391-2414.

    10. Meyer, M., et al., Photosensitized charge separation and hydrogen production in reversed

    micelle entrapped platinized colloidal cadmium sulphide. Journal of the Chemical Society,

    Chemical Communications, 1984(2): p. 90-91.

    11. Steigerwald, M.L., et al., Surface derivatization and isolation of semiconductor cluster

    molecules. Journal of the American Chemical Society, 1988. 110(10): p. 3046-3050.

    12. Brennan, J., et al., Bulk and nanostructure group II-VI compounds from molecular

    organometallic precursors. Chemistry of Materials, 1990. 2(4): p. 403-409.

  • 42

    13. Bawendi, M., et al., X‐ray structural characterization of larger CdSe semiconductor clusters. The Journal of Chemical Physics, 1989. 91(11): p. 7282-7290.

    14. Murray, C., D.J. Norris, and M.G. Bawendi, Synthesis and characterization of nearly

    monodisperse CdE (E= sulfur, selenium, tellurium) semiconductor nanocrystallites. Journal of the

    American Chemical Society, 1993. 115(19): p. 8706-8715.

    15. Peng, X., et al., Shape control of CdSe nanocrystals. Nature, 2000. 404(6773): p. 59-61.

    16. Peng, Z.A. and X. Peng, Mechanisms of the shape evolution of CdSe nanocrystals. Journal

    of the American Chemical Society, 2001. 123(7): p. 1389-1395.

    17. Peng, Z.A. and X. Peng, Formation of high-quality CdTe, CdSe, and CdS nanocrystals

    using CdO as precursor. Journal of the American Chemical Society, 2001. 123(1): p. 183-184.

    18. Yu, W.W. and X. Peng, Formation of high‐quality CdS and other II–VI semiconductor nanocrystals in noncoordinating solvents: tunable reactivity of monomers. Angewandte Chemie

    International Edition, 2002. 41(13): p. 2368-2371.

    19. Cao, Y.C. and J. Wang, One-pot synthesis of high-quality zinc-blende CdS nanocrystals.

    Journal of the American Chemical Society, 2004. 126(44): p. 14336-14337.

    20. Jasieniak, J., et al., Solution-proces