Implementation of Flow Manufacturing and Process Control in Nanoparticle Synthesis by the Wet Chemistry Method

MISSING IMAGE

Material Information

Title:
Implementation of Flow Manufacturing and Process Control in Nanoparticle Synthesis by the Wet Chemistry Method
Physical Description:
1 online resource (161 p.)
Language:
english
Creator:
Zhou, Jiaqing
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
Powers, Kevin W
Committee Members:
El-Shall, Hassan E
Moudgil, Brij M
Sigmund, Wolfgang M
Svoronos, Spyros

Subjects

Subjects / Keywords:
flowchemistry -- nanoparticle -- processcontrol -- qds -- stobersilica
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre:
Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Nano-particle manufacturing is a promising industry in the near future. Several methods are used for nano particle production. One method,called the wet chemistry technique, is widely used, but lacks reproducibility and scalability when batch processed. Possible solutions that avoid these problems are the flow synthesis system (FSS) and process control. However,despite their benefits, these methods are relatively new in the nano particle field. The combination of these two methods and their benefits shows potential in novel industrial-scale manufacturing of nano particles. In order to establish a system that monitors and controls the product quality, both online/inline measurements and sized map based/feedback process controls are introduced into the FSS. In order to study the efficacy of the process controls on particle properties such as size distribution, the Stober silica model was chosen to develop and test the FSS. Two types of process control were investigated in the Stober silica process. The size map based control was established by building an experimental database and using it to model the relationship between mean volume (M.V.) particle size and reagents’ concentration. The second method used feedback control with a PI controller. Its parameters were derived from the Cohen-Coon method. Two high-value colloidal products, dye doped silica and cadmium telluride quantum dots (CdTe QDs) were studied in the FSS as case studies. The synthesis of dye doped silica followed a modification of the Stober process to incorporate various fluorescent dyes into the product. Rubpy and Rhodamine 6G (R6G) dyes were physical adsorbed and fluorescein isothiocyanate (FITC) and 7-methoxycoumarin-3-carboxylic acid (MCA) were chemically bonded in doping the silica particles. CdTe QDs in the emission range of 500 - 800nm were synthesized hydrothermally by controlling the reaction temperature and the residence  time in the flow reactor. The effects of temperature, reagent concentration,and residence time on the emission spectrum were studied. The results indicated that higher concentrations of cadmium (Cd2+) ions and lower concentrations of N-acetylcysteine (NAC) produce QDs with a high quantum yield(QY) of 40- 60% in a much reduced reaction time compared to batch synthesis. The process control of the CdTe QDs relies on a proportional – integral (PI) controller. Both the Cohen-Coon and the Ziegler-Nichols tuning methods were used for the tuning parameters. The control algorithm was able to reach the desired emission wavelength in around 10 minutes with a precision of 2 nanometers (nm). Furthermore, a novel coating method for CdTe/CdS core/shell QDs was developed for the FSS using controlled degradation of sodium thiosulfate in an acidic environment. This resulted in a Type II quantum dot where the emission spectrum of the QDs was red shifted up to 70nm.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Jiaqing Zhou.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Powers, Kevin W.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044562:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 IMPLEMENTATION OF FLOW MANUFACTURING AND PROCESS CONTROL IN NANOPARTICLE SYNTHESIS BY THE WET CHEMISTRY METHOD By JIAQING ZHOU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Jiaqing Zhou

PAGE 3

3 To my wife, Xingyu Z hao, who support s me with endless love

PAGE 4

4 ACKNOWLEDGMENTS Five years has passed since I started my PhD study in University of Florida. As the first experiment of living far away from home, I feel so fortunate to study in such peaceful but energetic campus. I appreciate all the help and guide that I received in th ese years. First, I would like to acknowledge my advisor, Dr. Kevin Powers, for his patient guidance and great support. His enthusiasm for science and technique always encouraged me to pursue the knowledge behind superficialities My sincere thanks go to the other members of my PhD supervisory committee, Dr. Hassan El Shall, Dr. Brij Moudgil, Dr. Wolfgang Sigmund and especially Dr. Spyros Svoronos for their invaluable discussion and suggestion s I am grateful to my research group members and all the staff at the Particle Engineering Research Center (PERC) for their assistant in these years. Many thanks to Dr. Ajoy Saha Dr. Megan Hahn and Dr. Parvesh Sharma for their knowledge and experience about quantum dot synthesis, to Dr. Gill Brubaker and Gary Scheif fele for their guidance and suggestion s on the instruments and particle characterization, to Dr. Kerry Siebein from Major Analytical Instrumentation Center (MAIC) for her assistance with TEM and SEM, to Jim from Chemical Engineering for the assistance of m achining and to Paul Carpinone for his help in all aspects of my research. I would like to acknowledge the C enter for Particulate & S urfactant S ystems (CPaSS) and National Science Foundation for financial support and friendly research environment. I would like to acknowledge the support I have received from my parents and wife throughout my academic career with the reliable and warm affection which always rele ieves my pressure.

PAGE 5

5 Finally, I especially thank my uncle, Renliang Xu, who illuminated the wa y towards my dream.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATION S ................................ ................................ ........................... 1 3 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 Promising Material s : Nano particles ................................ ................................ ....... 16 The Emerging market for Nano particles ................................ ................................ 16 Synthesis of Nano particles ................................ ................................ .................... 17 The Advantages of Wet Chemistry Processes ................................ ........................ 18 Current Barriers for Commercialization of Nanotechnology ................................ .... 18 The Potential Solution: Flow Chemistry ................................ ................................ .. 19 Gap Analysis and State ment of Problem ................................ ................................ 20 2 BACKGROUND ................................ ................................ ................................ ...... 2 2 Flow Synthesis and Process C ntrol ................................ ................................ ........ 2 2 Silica Synthesis and Application s ................................ ................................ ............ 2 5 Mechanism ................................ ................................ ................................ ...... 2 5 Reproducibility ................................ ................................ ................................ 2 6 Application of Silica in the Flow/Micro System ................................ ................ 2 7 Dye Doped Silica ................................ ................................ ............................ 2 7 Flow Synthesis plus Feedback Control for CdTe Nano particle s ............................ 2 8 Core Shell QD s ................................ ................................ ............................... 3 0 QD s in the FSS ................................ ................................ ............................... 3 0 3 STOBER SILI CA SYNTHESIS BY FLOW MANUFACTURING WITH P ROCESS CONTROL ................................ ................................ ................................ .............. 3 1 Stober Silica Particles Made by Batch Synthesis ................................ .................... 3 1 The Assembly of the Flow Synthesis System ................................ ......................... 3 2 Materials ................................ ................................ ................................ ................. 3 3 Characterization ................................ ................................ ................................ ...... 3 3 Online detectors ................................ ................................ .............................. 3 4 DelsaNano ................................ ................................ ............................... 3 4 Nanotrac (Microtr ac Inc.) ................................ ................................ ......... 3 5 Compara ison between batch and flow synthesized Stober silica particles ..... 3 6

PAGE 7

7 Optimization of the Flow System ................................ ................................ ............ 3 6 Sufficient Reaction time ................................ ................................ .................. 3 6 Heating effect ................................ ................................ ................................ .. 3 7 Stability and accuracy ................................ ................................ ..................... 38 Sedimentation in tube ................................ ................................ ..................... 38 Process Co ntrol ................................ ................................ ................................ ...... 4 0 Size map based Control ................................ ................................ .................. 4 0 Map the Size Range ................................ ................................ ................ 4 0 Control A lgorithm ................................ ................................ ..................... 4 2 Feedback Control ................................ ................................ ............................ 4 3 Methods ................................ ................................ ................................ ... 45 Results and Discussi on ................................ ................................ ............ 4 5 Dye Doped Silica ................................ ................................ ................................ .... 5 0 4 HYDROTHERMAL QUANTUM DOT SYNTHESIS IN FSS AND PROCESS CONTROL ................................ ................................ ................................ .............. 88 Conversion from Batch to FSS ................................ ................................ ............... 88 Instrument and D esign ................................ ................................ ............................ 89 Results and D iscussion ................................ ................................ ........................... 9 0 Effect of reagent concentration on QDs ................................ .......................... 9 1 Effect of reaction temperature on QDs ................................ ............................ 9 3 Effect of residence time ................................ ................................ ................... 9 4 XRD characterization of CdTe QDs ................................ ................................ 9 5 TEM characterization of CdTe QDs synthesized at 180C .............................. 9 6 Thermal Control ................................ ................................ ................................ ...... 9 6 Process C on trol ................................ ................................ ................................ .... 10 0 Graphical process identification from step responses ................................ ... 10 1 Cohen Coon tuning method ................................ ................................ .......... 10 3 Z iegler Nichols tuning method ................................ ................................ ...... 10 5 Core shell QD in FSS ................................ ................................ ........................... 10 6 Materials and method ................................ ................................ ................... 1 07 Results and Discussion ................................ ................................ ................. 1 08 5 CONCLUSION AND FUTUR E WORK ................................ ................................ .. 1 49 Summary ................................ ................................ ................................ .............. 149 Conclusion ................................ ................................ ................................ ............ 150 F uture W ork ................................ ................................ ................................ .......... 15 0 LIST OF REFERENCES ................................ ................................ ............................. 15 2 BIOG RAPHICAL SKETCH ................................ ................................ .......................... 16 2

PAGE 8

8 LIST OF TABLES Table page 3 1 Several formula s for different size of Stober silica ................................ .............. 56 3 2 Experiment al set up for mapping size range ................................ ....................... 56 3 3 Detailed experiment for covering silica size range ................................ .............. 56 3 4 Step change of flow rate and resulting particle size ................................ ............ 57 3 5 K, data. ................................ .................. 58 4 1 ................................ ............................. 115 4 2 Preliminary batch test of coating with sodium thiosulfate ................................ 116 4 3 Residenc e time and temperature effect on CdS coating ................................ .. 117

PAGE 9

9 LIST OF FIGURE S Figure page 3 1 Piston pump from Syrris Co ................................ ................................ ................ 5 4 3 2 Sketch of flow system ................................ ................................ ......................... 5 5 3 3 Detection of bi dispersed Stober silica particle by multiple techniques ............... 5 6 3 4 Calibration of DelsaNano by LS13320 ................................ ................................ 59 3 5 Sketch of online DelsaNano and its dilution system and flow chart for the DelsaNano online detector system ................................ ................................ ..... 6 0 3 6 Particle size distribution of Stober silica measured by the Coulter LS13320 ...... 6 1 3 7 The relationship between p mean size of batch made Stober silica ................................ ............................... 6 2 3 8 SEM picture of batch made Stober s ilica. ................................ ........................... 6 3 3 9 The relationship of settling distance in 90min with M.V. particle size for the Stober silica suspension. ................................ ................................ .................... 6 4 3 10 Stability of FSS. The residence time was controlled at 30 minutes .................... 6 5 3 11 Repeat experiments about the flow rate changed in 30min tube reactor ............ 6 6 3 12 Gradually decrease of p article size during the long term operation of FSS without ultrasonication. ................................ ................................ ....................... 6 7 3 13 Cross section of PTFE tubing showing the sedimentation of silica particle on the tube wall ................................ ................................ ................................ ....... 6 8 3 14 Stability test on FSS with ultrasonicator. ................................ ............................. 69 3 15 Tri axial diagram of particle size map. ................................ ................................ 70 3 16 Three dimensional graph of particle size map ................................ .................... 71 3 17 Flow chart of the size map based control. ................................ .......................... 72 3 18 Size map based on size map control method. ................................ .................... 7 3 3 19 Ammonia step up data, with a 2 period moving average, ammonia flow rate increase from 0.15mL/min to 0.17mL/min. ................................ ......................... 7 4

PAGE 10

10 3 20 Ammonia step down data, with a 2 period moving average, ammonia flow rate decreas ed from 0.15mL/min to 0.13mL/min. ................................ ...................... 7 5 3 21 Water step up data, with a 2 period moving average, water flow rate increased from 0.15 mL/min to 0.17mL/min. ................................ ................................ ........ 7 6 3 22 Water step down data, with a 4 period moving average, water flow rate decreased from 0.15mL/min to 0.13mL/min. ................................ ...................... 7 7 3 23 Bode Plot created using AAS_ECH4323NP. The p lot showing Bode Stability lines for our transfer function. ................................ ................................ ............. 7 8 3 24 GM and PM from Bode plot: ................................ ................................ ............... 79 3 25 Simulation of simple step change of ammonia flow rate. ................................ .... 8 0 3 26 Simulation of feedback control with set point changed from 240 to 300. ............ 8 1 3 27 Simulation of feedback control with discrete (stepped) flow rate. ....................... 8 2 3 28 Simulation of feedback control with discrete flow rate and noise of data. ........... 8 3 3 29 Feedback control in the flow synthesis system including a set point change at time zero. ................................ ................................ ................................ ............ 8 4 3 30 Flow chart of feedback control algorithm ................................ ............................ 8 5 3 31 Dye doped silica samples prepared by FSS. ................................ ...................... 8 6 4 1 Flow system for QD synthesis ................................ ................................ .......... 11 0 4 2 ...... 11 1 4 3 Normaliz ed emission spectra for QDs synthesized at different temperatures and r max and temperature ................................ ............... 11 2 4 4 Normalized emission spectra for QDs synthesized with different residence time and r max and residence time ................................ ... 1 1 3 4 5 The calculated QD average radius as the function of residence time and the plot of cube of average QD radius as a function of residence time. .................. 11 4 4 6 Images of QDs prepared via continuous flow ................................ ................... 11 5 4 7 XRD patterns of the CdTe QD by flow synthesis at different residence time. ... 11 6 4 8 TEM image of QD produced under 180 C with a residence time of 3.5 seconds. ................................ ................................ ................................ ........... 11 7

PAGE 11

11 4 9 Sketch of heating system ................................ ................................ ................. 11 8 4 10 The temperature response of new heating system with increase set point ...... 1 19 4 11 The temperature response of new heating system with increase set point ...... 12 0 4 12 The temperature response of new heating system with decrease set point ..... 12 1 4 13 Step change data for heating system from MV 40 to 35 ................................ ... 12 2 4 14 QD emission wavelength disturbed by temperature deviation .......................... 12 3 4 15 Performance of heating system with PI controller (Kc= 3.6, =13.15, set point at 120 / 135 / 145) ................................ ................................ .................... 12 4 4 16 Performance of heating system with PI controller (Kc= 1.8, =13.15, set point at 120) ................................ ................................ ................................ ...... 12 7 4 17 Performance of heating system with PI controller (Kc= 0.9, =13.15, set point at 130) ................................ ................................ ................................ ...... 12 8 4 18 T he performance of heating system with on off controller and its effect on stabilizing the QD emission wavelength. ................................ .......................... 1 29 4 19 The potential relationship between flow rate, reaction time and emission wavelength ................................ ................................ ................................ ....... 13 1 4 20 Step change from 0.5 to 0.6mL/min, 1.5 to 1.6 mL/min, 2.5 to 2.7 mL/min, 3.5 to 3.8 mL/min ................................ ................................ ................................ .... 13 2 4 21 step change data. ................................ ... 13 4 4 22 C C method Kc and ................................ ................................ .................... 13 5 4 23 C C method tuning (Feedback control for Stainless steel tubing with c=(a)1, (b) 0.5, (c) 0.25). ................................ ................................ ................................ .... 13 6 4 24 The weight of Kc and in tuning program for c=0.25(a), 0.5(b) and 1.0 (c). .. 1 39 4 25 Z N method Kc ................................ ................................ ............................ 14 1 4 26 Z N method tuning (530nm(a), 580nm(b), 637nm(c). ................................ ....... 14 2 4 27 The weight of Kc and in tuning program for set point=530. ......................... 14 5 4 28 Red shift of emission wavelength from the coating of CdS shell at 120 C ...... 14 6

PAGE 12

12 LIST OF ABBREVIATION S ACF Autocorrelation function APTS 3 Aminopropyltriethoxysilane C C Cohen Coon Cy Cyanine DI Deionized DLS Dynamic light scattering DDS Dye doped silica FITC Fluoresc ein isothiocyanate FSS Flow synthesis system IR Infra red LD Laser diffraction M V Mean volume MV Manipulated variable PI Proportional Integral PL Photon luminescence QD Quantum dot QY Q uantum yield SD Standard deviation SS Stainless steel TEOS Tetraethyl orthosilicate TMR Tetramethylrhodamine UV Ultraviolet Z N Ziegler Nichols NAC N Acetylcysteine

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degre e of Doctor of Philosophy IMPLEMENTATION OF FLOW MANUFACTURING AND PROCESS CONTROL IN NANOPARTICLE SYNTHESIS BY THE WET CHEMISTRY METHOD Jiaqing Zhou August 2012 Chair: Kevin William Powers Major: Materials Science and Engineering Nano particle manufac turing is a promising industry in the near future. Several methods are used for nano particle production. One method, called the wet chemistry technique, is widely used, but lacks reproducibility and scalability when batch processed. Possible solutions tha t avoid these problems are the flow synthesis system (FSS) and process control. However, despite their benefits, these methods are relatively new in the nano particle field. The combination of these two methods and their benefits shows potential in novel i ndustrial scale manufacturing of nano particles. In order to establish a system that monitors and controls the product quality, both online/inline measurements and sized map based/feedback process controls are introduced into the FSS. In order to study the efficacy of the process controls on particle properties such as size distribution, the Stober silica model was chosen to develop and test the FSS. Two types of process control were investigated in the Stober silica process. The size map based control was established by building an experimental database and using it to model the relationship between mean volume (M V concentration. The second method used feedback control with a PI controller. Its parameters were derived from the Cohen Coon method.

PAGE 14

14 Two high value colloidal products, dye doped silica and cadmium telluride quantum dots (CdTe QDs) were studied in the FSS as case studies. The synthesis of dye doped silica followed a modification of the Stober process to incorporate va rious fluorescent dyes into the product. Rubpy and Rhodamine 6G (R6G) dyes were physical adsorbed and fluorescein isothiocyanate (FITC) and 7 methoxycoumarin 3 carboxylic acid (MCA) were chemically bonded in doping the silica particles. CdTe QDs in the emi ssion range of 500 800nm were synthesized hydrothermally by controlling the reaction temperature and the residence time in the flow reactor. The effects of temperature, reagent concentration, and residence time on the emission spectrum were studied. The results indicated that higher concentrations of cadmium (Cd 2+ ) ions and lower concentrations of N acetylcysteine (NAC) produce QDs with a high quantum yield (QY) of 40 60% in a much reduced reaction time compared to batch synthesis. The process control of the CdTe QDs relies on a proportional integral ( PI ) controller. Both the Cohen Coon and the Ziegler Nichols tuning methods were used for the tuning parameters. The control algorithm was able to reach the desired emission wavelength in around 10 minutes with a precision of 2 nanometers (nm). Furthermore, a novel coating method for CdTe/CdS core/shell QDs was developed for the FSS using controlled degradation of sodium thiosulfate in an acidic environment. This resulted in a Type II quantum dot where the e mission spectrum of the QDs was red shifted up to 70nm.

PAGE 15

15 1 CHAPTER 1 INTRODUCTION Well P romising M aterial: Nano particle s N ano particle s or ultra fine particle s, are defined as materials with at least one dimension between 1 to 100 nm. Although this definition has been proposed only in recent times, nano particle s applications have been involved in human history ever since ancient time s As f a r back as the 4 th century, evidence showed that the Romans already mastered the technique to generate an optical dichroi c effect in glass vessels by using silver and gold nano particles 1 The pottery from the middle a ges and r enaissance were often covered by glaze layers that contain ed copper and silver nano particles 2 T he first scientific description of nanoparticle s was mentioned by Michael Faraday and described in his paper published in 1857 3 It was only in the early stages of the 20 th century that nano particles began to p lay a significant role in various technologies such as colloidal systems. Today these advanced technologies have greatly affect ed in areas that include energy, healthcare, computers, microelectronics, optical en gineering and many other areas using advanced materials The Emerging market for Nano particle s The Nanotechnology market is rapidly expanding in market value for a wide variety of applications. According to a report by Electronics.ca Publications the global market value for the nanotechnology was estimat ed to be $15.7 billion in 2010, and was expected to have a compound annual growth rate (CAGR) of 11.1% for the next 5 years. BY 2015, t he estimated global market in nanotechnology is expected to increase to over $27 billion. A s the largest segment in the market, the nanomaterial s market is expected

PAGE 16

16 to increase from nearly $10 billion in 2010 to $19.6 billion in 2015 with an annual growth rate of 14.7% 4 Despite the exploding market value, nanotechnology commercializatio n is still at a very early stage This immature market indication is illustrated by the large difference between the growth of patents and the number of products. According to the recent nanotechnology product inventory from t he project on emerging nanotechnologies at the Woodrow Wilson International Centre for Scholars ( www.nanotechproject.org ), 1317 nanotechnology related products or product lines were being produced globally in 2011, up substantially from 2006 when it liste d only 212 products. Synthesis of Nano particle s T here are four fundamental routes for nano materials synthesis including form in place processes mechanical processes gas ph ase synthesis, and wet chemistry processes 5 Each of these meth ods has its own advantages and limitations so that the resulting products have unique properties. F orm in place processes. Th ese include lithography, vacuum deposition and spray coating. These techniques directly generate nano materials as surface layers for other products. T he y are more suitable for nanostructured layers and coatings b ut they can still be used to manufacture nano particle s by separating deposits from collector s The limitations of th ese method s are the relatively low efficien cy when they are used for dry powders synthesis M echanical process es These are s that reduce particle size s by collision and attrition, i.e. grinding, milling and mechanical alloying techniques. A dvantages of these age old techniques are s imple, widely applicab le and low cost. However, it is hard to achieve fine particles by these methods due to the increasing

PAGE 17

17 surface energy and the tendency to agglomerat e Other difficulties include b road particle size distribution s and contamination from m illing media and equipment. G as phase synthesis This includes flame pyrolysis, electro explosion, laser ablation, high temperature evaporation and plasma synthesis techniques. These processes generate nanomaterials through chemical reaction s or physical evaporation at high temperature. The advantages of gas phase synthesis method are the clean and controllable environment and temperature. However, the high temperature feature also excludes the process ing of organic materials W et chemistr y pr ocesses Th ese are s that the formation of insoluble compounds start s from the mixture of ions or molecules. These processes include colloidal chemistr ies hydrothermal methods, sol gel, and other precipitation process T he Advantage s of W et C hemistry Processes W et chemistry processes currently provide better quality nano particle s which result from the following aspects. First ly agglomeration and aggregation of the products can be reduced or eliminated by designed inter particle forces. Second ly n ano particle s can be synthesized with narrow or mono disperse size distribution Finally it is capable of finely control ling nano chemical composition purity and morphology This is important for appliations that require high repeatability. C urrent B arriers for Commercialization of Nanotechnology T he main barriers for commercialization of nanotechnology deal with four domains p roject from European Commission 6 Those four domains are including manufacturing domain, technological domain, marketing & strategy domain, and investment &

PAGE 18

18 organization domain. The Manufacturing domain suffers from lack of maturity includ ing the lack of funding and equipment to achieve scale up of production. T he t echnological domain is related to the reproducibility and long term reliability of the system. The marketing and strategy domain involves the agreement between market oppo rtunities with technical development. The investment and organizational domain involves return on investment and the required dedicated manufacturing infrastructure. The lack of reproducibility and long term reliability are always technological problem s w hen the wet chemistry processes scales up to large quantities The major reason for this phenomenon is the difficulty in simultaneously controlling all parameters. The a gglomeration of nano particle s is aoften a problem due to the enhanced temperature and concentration gradient s in the pilot scale reactors 7 The P otential S olution: F low Chemistry Batch method s are generally used for preparing nano particle s by we t chemistry methods. T hey require the precise control of experiment conditions that determine properties of produced nano particle s. U npredictable deviations of exp eriment al conditions often result in disparit ies between different batches in terms of the size distribution, the zet a potential etc. Automated and miniaturized continuous flow synthesis methods, also known as flow chemistry are a well established technique for manufacturing large quantities of a given material and have been proposed as an improved alternative to overcome these limitations. In flow synthesis methods chemical reaction s run in continuous flow stream s rather than in batch container s i.e. reactions take place when reactive fluids are driven into tubes by pumps Compared to traditional batch mode reac tion s flow synthesis methods have several advantage s. Firstly, faster and uniform m ixing of reagents can be

PAGE 19

19 achieved because of the smaller cross section of tube Secondly, temperature in flow system is more controllable due to the decrease in thermal mas s. The system temperature can also be increased above the normal boiling point by applying pressure using a backpressure regulator Thirdly, t he reaction time can be determined precisely by calculating residence time in the tubing The introduction of addi tional reagents can be controlled precisely at desired time point 8 Fourthly, t he flow synthesis system can be automated with far less expense than batch system s It is possible to establish an automated system that can change reaction parameters to optimize products qualit ies with little intervention and loss 9 Finally, the flow synthesis system is able to scale up without losing control of reaction conditions by increasing the diameter of the sys tem or the number of tube reactors. Gap A nalysis and S tate of P roblem Flow synthesis methods are relatively new in the laboratory, especially in the a rea of nano particle synthesis 10 Research into flow synthesis methods most ly focus es on organic reactions that have quite different properties from nano particle s. The effect s of scale up on system performance s and process control are also absent Therefore, the first objective of this research is to establish a flow synthesis model system in a well known and representative nano particle synthesis process The S tober silica process was selected as the model process because t he relation between its particle size and reaction conditions is typical and well understood. The effects of Stober silica s properties on the performance of the flow synthesis system are evaluated The process control system for particle size is tested on this model system by tuning the concentration. Several particle size characterization instruments are modified as inline/online detectors

PAGE 20

20 The second goal is to extend the flow synthesis system into more functional and high value added nano products Two case studies were selected due to the high interest and high value added in these products : dye doped silica and quantum dot s D ye doped silica is one of the important post product s of the Stober silica process It has been extensively used in photonics materials 11 12 in nonlinear optical materials 13 and in the bioimaging and biochemical analysis applications 14 The synthesis of dye dope d silica in the flow synthesis system is studied by using four types of dye molecules. Quantum dot s are another good sample of high value added materials ($3000 $10000 per gram 15 ) They have been involved in various applications such as LEDs 16 solar cells 17 video displays, diode lasers 18 and bio imaging 19 In this study, t he hydrothermal synthesis route for CdTe quantum dots is applied in the flow synthesis system The system parameters including tempe rature, reagent concentration, and residence time are studied and optimized for the quantum yield The process control is based on the peak emission wavelength by tuning the residence time in the hot zone of the reactor As a bonus, a n in lin e CdS coating process was developed for the flow synthesis system to generate Type II CdTe/CdS core shell quantum dots in this project

PAGE 21

21 2 CHAPTER 2 BACKGROUND The Flow S y n thesis and the Process control The f low synth esis system has been found to be applicable in many new fields, including disciplines in chemistry and biology 20 In the field of the microfluidic system, more and more reports have been reported regarding in novative approaches, wh ere there is an integration of the advantages from the flow synthesis system and the economy due to the reduced volume The microfluidic system is ideal to process ing experiments with less costly materials. A whole flow synthesis system normal ly includes the following parts: Pumps The p recise control of transporting fluid is the foundation of a controllable FSS It can be ac hieved either with an integrated mechanical and electrical actuation or by temperature and pressure gradients. The s yringe pump is one of the mechanical pumps commonly used in the microfluidic system for non pulsating flow 21 The p iston pump is another source for the continuous and steady flow with a higher pressure and a larger reservoir. This is achieved by combining the two pistons which work interchangeably In addition to mechanical pumps, other researchers have developed several micro pumps based on pH gradients 22 pressure 23 laser induced cavitation 24 and temperature sensitive hydrogels 25 among others Mixers The behavior of fluids at the micro scale can be different from those at the macro scale. T he Reynolds number become s very low when the channel diameter ranges between 100 n ano m eters to several hundred micrometers 26 Therefore non turbulent flow in microfluidic channels makes mixing a challenging task because diffusion might be t oo slow so that the time or channel length becomes unacceptable 27

PAGE 22

22 Current mixing methods for microfluidic systems include static mixers 28 35 and external mechanical actuation methods such as acoustics 36 and electro osmotic based mixer s 37 39 Reactors Reactors for the flow synthesis system are typically tube like and fabricated by non reactive materials such as glass 40 silicon 41 stainless steel and polymers 42 43 Important propert ies of a suitable material include the cost of fabrication, machinabili ty, working temperature, thermal conductivity, inertness, surface charge, molecular adsorption, optical properties and other s 43 S urface modification may be required due to the specific desired surface properties from the flow system applications 44 46 The types of reactors include spinning disc reactors, multi cell flow reactors, oscillatory flow reactors, heat exchanger reactors and micro reactors among others Problems are generated when the the micro meter range T ube blockage be comes the biggest hurdle for an application involving particulates 47 48 Furthermore, any g as es that are generated from the reaction pressure decrease or temperature variance may affect the residence time of the reagents by pushing out fluid faster than expect ed Detectors Various detection systems have been rep orted for their reliab ility and repeatable online/inline measurements. Maimiroli.et al. reported a free jet micromixer that was combined with low angle X ray scattering for the study of fast chemical reactions 49 Amarie et al. introduced the surface plasmon resonance to study glucose oxidase binding activity in a microcavity 50 St aples and co workers have demonstrated mass spectroscopy/liquid chromatography detection method s for analyzing

PAGE 23

23 glycosaminoglycan on a chip 51 Carter et al. demonstrated online non intrusive measurement of particle size distribution through digital imaging 52 Electrostatic sensors are an additional technique for the particle size measurement 53 55 Traditional batch based instrument s also can provide pot ential online/inline measurements through the application of flow cell s Automatic control s : Three types of control systems are available: the o pen loop, the feed forward, and the feedback control system However, t he open loop system is a manual co ntrol, with no automatic response to the environmen tal disturbance. This system is commonly used in most of the lab experiments. The f eed forward control system has significant benefits when a predictable disturbance occurs upstream of the system i f the mathem atic al model is reliable and the control law is followed entirely ( i.e. the controller predicts the incoming disturbance and compensates for it ) The feed forward control relies on the accuracy of the disturbance measurement as well as the noise and the ac curacy of the feed forward gain and the timing An ideal feed forward system can overcome the oscillation and the delays of the output while maintain ing the system stability. The limitation of the fe ed forward control is obvious. The control system can only respond to the disturbance in a pre defined way, which usually means that the disturbance must be predictably stable with time. The introduction of any unknown disturbance or input will result in a n inaccuracy. Thus, the feed forward control system is best for a well understood process, or for those processes whose behaviors can be easily measured and replicated under known operating conditions.

PAGE 24

24 The feedback control monitors system output through d etector s and check s the difference between the target value and the output, which is defined as the error. The control system a djusts the input to minimize this error. A familiar and fundamental example of feedback control is the on and off control, such as those found in ovens which utilize a common temperature control system to supply or not supply heat to the oven The fact that the feedback control obtains data at the process output brings both pros and cons. Although a full understanding is not required of the system or the mathematical model for the control system, the feedback control method requires time to correct the output after the disturbance occurs Extreme conditions such as large magnitude disturbance s or large time delay s may cause th e control system to work inefficiently. Silica S ynthesis and A pplication The Stober silica is a n example of a well characterized process for producing mono dispersed silica particle s Since Stober first described the growth of mono dispersed silica particle s in alcohol in 1968 56 hundre ds of papers have been published about the various applications of silica particles in bio imaging 57 nano carrier s 58 59 pigments, and stabilizer s 60 among others Mechanism The Stober process involves the complex reactions between water, Si(OR) 4 and ammonia. The overall reaction can be shown as: ( 2 1 )

PAGE 25

25 The reaction indicates that two moles of water are required to stoichiometrically react with one mole of TEOS The w ater, ammonia and Si(OR) 4 concentration as well as the type of alcohol solvent are co nsidered to be critical parameter s for affecting particle size. Unlike the multiple intermediate products during the acid catalyzed gel synthesis, the hydrolysis reaction produces only the single hydrolyzed monomer 61 63 which is accumulated at the beginning or induction time. Nucleation occurs when single hydrolyzed monomer s become saturated. Mono dispersed particle s retain the growth afterward s until all the reactants are exhaust ed The reaction mechanism is explained by the following two models: the LaMer mode l 64 66 which indicates that nucleation happened only once during the whole process followed by continuous particle growth, and the controlled aggregation model which suggests that the growth of the particle s result s from the aggregation of the small particles 67 69 Lee et al. 63 supported th e controlled aggregation model by examining the profile of the intermediate concentration using 29 Si NMR. However, Harris 70 and van Blaaderen et al. 71 suggested that both models contribute to the particle growth: the controlled aggregation model controls the reaction speed while the LaMer model makes the surface s moother Reproducibility The Stober silica process is a good example of a sensitive reaction in that the resulting particle size distribution can vary due to the influence of the conditions in the reaction. The results from Stober et al. 56 show an error in the range of hundreds of nanometers in experiment replications The precise control of the particle size is difficult due to the poor reproducibility. It is hard to validate the potential reasons. One explanation is that t he precise control of the nucleation depends heavily on the saturatio n

PAGE 26

26 of the hydrolyzed Si(OR) 4 yet any subtle changes of concentration can alter the induction time For instance, both the alcohol solvent and ammonia are volatile materials Application of S ilica in F low/ M icro S ystem Several group s reported the application of the Stober silica process in FSS. Ferguson et al. 72 tested a modified Stober process using a continuously stirred tank reactor yet the resulting particle size distribution was quite broad. Her et al. 73 reported the application of a static mixer tubular reactor with a 0.8 cm PTFE tube. They suggest ed that the reaction time in the continuous tubular reactor was narrow compared to the batch method. Herbert Giesche 74 established a FSS by peristaltic pump s, a mix er and 3mm/6mm diameter silane tube s His result s showed a broad deviation in repetition Furthermore, there was also a phase separation inside the tube where a particle deposit existed at the bottom of the tube. Ogihara et al. 75 introduced the Couette Taylor vortex FSS for the silica particle synthesis, which can continuously work for five hours giving comparable products to those obtained using the batch system. The process control and the online m easurement s w as absen t in the previous research studies Dye D oped S ilica D ye doped silica has wide application s in the biomedical field. It was first developed in 1992 by V an b laaderen et al. using the Stober method via f luorescein isothiocyanate (FITC) dye molecule conjugated with 3 a minopropyltriethoxysilane (APTS) 76 F ollowing this numerous studies were done in applying different dyes into silica matrix. Santra et al. reported the FITC doped silica particle s using the reverse microemulsion method 77 The Rubpy dye molecule is doped by the reverse microemulsion method 78 The fluorescence spectra, particle size and size distribution of

PAGE 27

27 these particles have been tested for optimization 79 Xiaojun Zhao et al. reported tetramethylrhodamine ( TMR ) doped silica particles by the reverse microemulsion method and also tested the leakage of the dyes 80 Core shell structures have been developed for further protection of dye molecule s from photo bleaching and leaking. S antra et al. reported a core shell silica particle with FITC dye doped in the core by both the Stober 81 and reverse microe mulsion methods 82 Hooisweng Ow et al. synthesized a TRITC doped core shell silica particle in 2005 83 Xichun Zhou et al. reported a hybrid core shell particle containing an Au core with Cyanine 3 (Cy3)/Cyanine 5 (Cy5) that was chemisorbed and a silica coating bearing thiol functional groups for microarray based DNA bioanalysis 8 4 Multi dye doped silica particle s have also been developed. Lin Wang et al. created silica particles entrapped with two fluorophores, OsBpy and RuBpy, simultaneously by reverse microemulsion 85 Flow S ynthesis plus F eedback C ontrol for CdTe Nano particle Studies in quantum dot s increased after Murray et al. 86 developed the conventional synthesis route. The excellent optical properties such as the quantum yield and the resistance to photo bleaching ma de quantum dots highly promising for applications in various fields like solar cells 87 and biological labels 88 The characteristics of the QDs c a me from the quantum confinement effect. When the size of the QDs is smaller than the critical characteristic length (Exciton Bohr radius), the original energy levels start to split into smaller ones with gaps between each successive level. The electronic and optical properties of the particles change with small enough particle size (typically less than 10nm) wit h the band gap increasing as particle size decreases QDs are direct band gap materials. The fluorescence is a result of the

PAGE 28

28 excited valance electron returning to the ground state combining with the hole. The fluorescent wavelength is determined by the siz e of the quantum dot since the energy of the emitted photon can been seen as a sum of band gap energy, the confinement energy of the hole and the excited electron, and the bound energy of the exciton. The o rganometallic method and the hydrothermal method are the two main methods for synthesis of QD s Alt hough the organometallic synthesis is the most widely used technique there is more and more interest in the aqueous synthesis method since it was introduced by Gaponik et al 89 Compared with the organometallic synthesis, the hydrothermal synthesis is less toxic, less cost ly, and more productive with a high stability and biological compatibility 89 However, the traditional disadvantage s of QD s made by the hydrothermal method include a broad emission peak, a longer process, and a relatively low quantum yield. 90 These disadvantages are mitgted using the Flow method described here. Various attempts have been made to explore the condition of hydrothermal QD synthesis and to improve its luminescent propert ies : different thiol compound s were tested as a stabilizing agent 91 ; the ratio of ligand and monomers were fine tuned by Guo et al 92 ; the relationship between h eating temperature and particle growth speed was reported by Zhang et al 93 ; Juandria et al 94 developed the rapid hot injection method by which the reaction time was reduced down to 1 10min at a high temperature of 200 240 C Core S hell QD Core shell QDs are one of the active fields in the QD research because of their novel properties. By coating higher band gap inorganic materials, the core shell QDs have a red shifted emission wave length and a longer decay lifetime due to the formation

PAGE 29

29 of the indirect electron excitation. The photoluminescence (PL) quantum yield and the photo stability are also improved due to the reduction of surface defects 95 D espite the core shell QDs being prepared through organometallic methods 96 98 preparation through the hydrotherma l method is more attractive 99 due to the advantages that this provides. QD in FSS There has been increasing use of such microfluidic devices in the production of various QDs 100 For example, CdSe 101 103 CdS 104 105 and InP 106 have been s ynthesized using the organometallic method with microfluidic techniques. Yang et al. reported core shell structure QDs using a microfluidic device by a two step organometallic method 107 However, no publication has related the hydrothermal method synthesis of QD in the FSS.

PAGE 30

30 3 CHAPTER 3 STOBER SILICA SYNTHESIS BY FLOW MANUFACTURING WITH PROCESS CONTROL Stober Silica Particles Made by Batch Synthesis Silica nano particles were first synthesized by th e batch method as a pre study and to provide a control for the implementation of the flow synthesis. These batches were carefully characterized through particle size analysis and by SEM imaging. Reagents included, ammonia (37%wt, Acros Organics), DI water (Barnstead Nanopure Infinity, 18M/cm 1) and Tetraethoxy Silane (TEOS 98%Acros Organics). The ammonia and water were carefully measured and mixed with half amount of required pure ethanol (200 proof) in a sealed glass flask with magnetic stirring for 2 minu tes. The tetraethyl orthosilicate (TEOS, 98%, Acros Organics) was diluted with the other half of the ethanol and poured slowly into the ammonium solution. The solution becomes opaque as the particle s nucleate and grow large enough to scatter light. Inducti on and growth can take several minutes to hours depending on the target size. The solution is kept at room temperature and stirred rapidly until the reaction is completed. The completion of the reaction is assessed by the cessation of particle growth as de termined by laser diffraction size analysis Quenching the reaction is possible by two methods: (1) Trimethylmethoxysilane can be added to react with active silanol sites on the hydrolyzed TEOS, interrupting the condensation and growth of the particles. (2) P ouring water into the system (best if reaction already passed induction time for several minutes and became turbidity) which dilutes the concentration of all reagents and inhibits further growth.

PAGE 31

31 The reaction time typically varies from 30 minutes to several hours depending on formula used. An hour is sufficient to end the reaction of those sizes larger than 150nm, while more time is required for smaller particles. The resulting suspensions are washed with ethanol and water by centrifuging (Beckman JA 21 Centrifuge) at 5000 rpm. Table 3 1 lists the common formulas for different sizes silica nano particles. T he A ssembly of the F low S ynthesis S ystem The initial model of FSS was based on the FRX100 from Syrris Co. which included three piston pumps, two tube reactors, a pre ssurization module and a sample collector. The wetted materials in the FSS include sapphire, PTFE, ruby and PEEK, which are chemically inert. The piston pumps ( Figure 3 1 ) are able to provide a non impulse continuous stream by the reverse stroke of two pistons. The flow rate provided by each pump ranges from 0.01 to 9.99 mL /min with a precision of 0.3% (measured at 1 mL /min) and an accuracy of 1% (measur ed at 1 mL /min). The two reactors are constructed of 0.8mm PTFE tubing with a volume of 4 mL and 16 mL respectively. The pressurization module is designed to control the backpressure of the FSS in the range 0 10bar. The FSS can be controlled manually or by co mputer using Labview software. Later on, other add ons were installed to enhance the system performance. Aldrich) were purchased to provide flexibility in controlling the linear flow rate (re sidence time) and to facilitate limited scale up studies. A hot plate was used to control the reaction temperature. An ultrasonicator was modified by connecting its control panel with a USB relay controller to achieve PC controlled periodic sonication. A d ilution system was constructed and connected at the end of tube reactor to adjust the particle concentration to that required for the online detector. Two dynamic light scattering (DLS) instruments,

PAGE 32

32 the Nanotrac (Micromeritics Corp) and the DelsaNano (Beck man Coulter Inc.), were adapted as the online detector for the particle size distribution measurement. Figure 3 2 shows a sketch of the whole system. Materials The Stober reaction is well suited as a model for designing and testing the flow system. The reagents are easily separated into three parts, amm onia, water and TEOS, and introduced independently by three computer controlled pumps. Since ethanol is required as a diluent, a 1:5 volume ratio of X (X=ammonia or water or TEOS) to ethanol is used in each stream. This enhances mixing and prevents the pre mature reaction of the precursors. The total flow rate is adjusted to achieve the desired residence time and initially was set to 0.7 mL /min, providing a 30min reaction time in the 20 mL FRX tube reactor and a 90min overall residence time in the system (63 mL PTFE tube loop). Characterization The LS13320 (Beckman Coulter, Inc.) was chosen to make the external particle size distribution measurements (as a reference) due to the excellent accuracy and precision of the laser diffraction (LD) technique. In the si ze ranges produced here, Stober silica can be considered an ideal particulate system for virtually all sizing techniques due to its spherical and monodisperse qualities. This can be seen by applying several common size measurement techniques as shown in Figure 3 3 Additional characterization was applied by scanning electron microscopy (SEM, JEOL 6335F FEG SEM) on filtered and air dried samples. Althoug h all are very close, laser diffraction had the closest mean value and distribution details (shoulder and tail in the left side of the distribution) to image analysis carried out by SEM. Consequently, LS13320x

PAGE 33

33 was also utilized to calibrate the results fro m DelsaNano (Beckman Coulter, Inc.) and its flow cell used as an in line size measurement. Online detector s DelsaNano The DelsaNanox is based on dynamic light scattering (DLS), which determines the particle size by detecting fluctuation rates of reflected or scattered laser intensity from dynamic information of the particles by autocorrelation function (ACF) which is used to derive the Diffusion coefficient of the p articles. Using the Stokes Einstein relation (Equation 3 1 ) the hydrodynamic diameter (size) of the particle is calculated. This technique is able to measure particle size from 1nm to several microns. ( 3 1 ) Where d h = hydrodynamic diameter, k B = Boltzman Constant, solvent viscosity and D T = Translational d iffusion c onstant. The precision and accuracy of the DelsaNano are lower than LS13320. In order to minimize these errors, 15 silica samples with gradually increased M.V. particle sizes from 80nm to 514nm were prepared and measured by both instruments for 5 replicates as show n in Figure 3 4 The results illustrate that the DelsaNano has a nearly linear deviation from the Coulter LS13320. A linear compensation factor can be calculated by the following equation: y = 1.0806x + 4.1316 ( 3 2 ) Although the Delsa Nano is designed as benchtop instrument it was adapted to online measurement through the use of a flow cell (Internation al Crystal Laboratories,

PAGE 34

34 UV VIS Cells/Type 42 Flow Through Cell with 10mm light path) instead of normal collects the reflection light from inner surface of flow cell instead of transmitted Still, a dilution system was required to improve the precision when the instrument was used for more concentrated samples. Figure 3 5 s hows the sketch of DelsaNano on line detector system and its operation sequence. The keyboard and mouse control software, Quick software. The measurement sequence starts with a 2.5 minute flush followed by the introduction of the new sample into the flow cell. Then a 2 min equilibration took place to measurement took 3.5 min to finish. Finally, the m easurement result was saved as a text file and loaded into the database by control software (Labview 6, National Instruments Corporation) while the flush for next measurement conducted. The design of dilution system required a very smooth flow of the silic a suspension to prevent clogging of the system which tends to happened at the T type tube connector and the pressure regulator. The silica suspension was continuously driven into the dilution system by slight pressure difference between two tubes. The pres sure difference resulted from the adjusted for a proper shunt ratio. Nanotrac (Microtrac Inc.) The Nanotrac is another instrument that relies on DLS technology. Different from DelsaNano, it has a laser backscatter probe that transmits reflected light from the sample through a sapphire window. A measurement chamber was designed and constructed to pass the flowing product over the Nanotrac probe. This arrangement served as an

PAGE 35

35 in working concentration. Thus it did not require auto dilution. The measurement sequence for Nanotrac was similar to that for DelsaNano, except for the absence of this dil ution system. Comparability between batch and flow synthesized Stober silica particle s Stober silica is well known for its mono dispersed particle size distribution, as shown in Figure 3 6 a. The SEM result ( Figure 3 8 ) confirmed its spherical shape and uniform size distribut ion. The breadth of the size distribution increases slowly as size increased, as shown in Figure 3 7 from a geometric standard deviation of 10nm (at a mean M.V. size = 50nm) to about 80nm ( M.V. size = 600nm). As particle size increases above 600nm a shoulder begins to appear indicating a bimodal distribution. This is a common characteristic of the Stober process and is caused by the high concentration of TEOS required and a second nucleation event ( Figure 3 3 ). The Stober silica made by the FSS showed similar properties as batch made silica. As show n in Figure 3 6 b, a typical flow synthesized silica particle was well mono dispersed and had a standard deviation equivalent or slightly smaller than b atch made silica particles. Optimization of the Flow System Sufficient Reaction time The reaction time of Stober silica process varies with the formula and the target particle size. In general, the higher the relative concentration of ammonia and TEOS, the higher the reaction rate, thus the shorter the reaction time to completion. A more succinct relationship is found between the M.V. particle size and reaction time due to the straightforward relation between reagent concentration and particle size. Gies che 108 quantified the growth of particle size as a function of time by the light scattering method

PAGE 36

36 with four formulas. Here, 20min is required for the synthesis of 417nm particles, (consistent with our experience). As expected, increasing temperature also reduces the reaction time significantly. Expand and support this statement During the design of the FSS, the residence time must be co nsidered at the beginning because it governs the overall flow rate by the equation: ( 3 3 ) Sufficient residence time maximizes the yield of reaction, diminishes the residu al reactants and reduces the dead time for quality control. For larger particles, a 30min residence time is generally sufficient but longer times are required as the target particle decreases. At a given flow rate, this requires the addition of a longer le ngth of tubing. Heating effect Heating is another option for controlling particle size distribution and reaction rate in the Stober silica reaction. The increase of reaction temperature leads to the decrease of particle size. According to Giesche 108 the particle size dropped from 665nm to 309nm when the temperature increased from 293k to 313K with the same reactant concentrations. The pa rticle size further decreased to 186nm at 333K (60 C ). The drop in particle size indicates an enhanced nucleation event which is the primary determinant of the final particle size. The increase of temperature also speeds up the reaction, obtaining these smaller particles in shorter time. While heating has several benefits for the system, the system was not yet configured for controlling temperature during these early studies. Therefore the flow rate was chosen as the only parameter for particle size contr ol in the Stober silica study.

PAGE 37

37 Stability and accuracy The accuracy of the FSS was a key to the design of the process control algorithm since it determined the precision with which the product size could be controlled. A series of experiments were made with 30min residence time to monitor the consistency of th e M.V. particle size and size distribution (S.D.). Samples were collected every 3min and measured off line by laser diffraction (LS13320) for the best accuracy and resolution. As shown in Figure 3 10 six experiments with different size ranges indicated that FSS was able to produce silica particles with steady particle size distribution in the 30min residence time reactor. Repeat experiments under the sa me reaction conditions ( Figure 3 11 ) confirmed that FSS was able to duplicate the same size silica particle (P value = 0.072 for 188nm particle and P v t test) at the same reaction conditions. Sedimentation in tube Although the FSS worked well with 30min residence time in the reactor, the long term test with 90min PTFE tube loop was initially interrupted with s erious problems. As shown in Figure 3 12 M.V. particle size continuously decreased during the operation of the FSS and lost about 400nm size in 18 ho urs. Further investigation revealed that silica particles were depositing on the walls of the tubing forming a thick layer which reduced the residence time ( Figure 3 13 ). There was also settling and stratification of the larger particles due to the low flow rates and lack of sufficient mixing The stratification of the suspension results from t sedimentation, and is one of the causes of tube blockage. This phenomenon is rarely observed in turbulent flow, but becomes distinct in laminar flow where the reaction is time consuming and the particle

PAGE 38

38 sedimentation in suspension as sh own in E quation 3 4 ( 3 4 ) where v s p is the mass density of particle (kg/m 3 3 2 ), g is the gravitational acceleration (m/s 2 ) and R is the radius of the particle (m). Figure 3 9 particle could settle in 90min in the ethanol solution. The 400nm silica par ticle, for Considering the gradient of velocity in laminar flow, the silica particles can precipitate even more at the flow conditions near the tubing wall, where the fl ow rate is much slower than that in the center of tube. As a result, particles begin to accumulate at the bottom of the tube with a much slower movement that are likely to clog the system when tube diameter changed at connector. Thus the size decrease of silica particle with time resulted from two conditions: the First, the product consisted of primarily the smaller particles exiting the reactor tubing with the larger particles settling out in the tubing. Thus the results only represented the tail of th e true distribution. Second, more particles were adsorbed on part of the tube wall which related to the induction section during the operation. The additional particles may work as the extra nucleus due to their large surface area, thus resulted less silic a precursor per particle, i.e., smaller particle size. The introduction of sonication to the flow loop solved this problem. The sonication helped mix the solution as well as preventing particles from adsorbing on the tube wall.

PAGE 39

39 The ultrasonicator was set to operate on aperiodic schedule, 1min every 3min and a cooling system was attached to keep the bath at room temperature. As shown in Figure 3 14 the M.V. particle size in a 22 hour running was stabilized at 285nm. The standard deviation of the mean volume size distribution throughout the experiment is 10.5nm. This minor fluctuation in the result is due to the low accuracy of DelsaNano at this size rang e. Process control The FSS, online detector and control software provide the tools for designing either a size map based or feedback control system. The size map based control relies on the precise database or library of initial conditions which can be r apidly generated by the flexibility of the system. The more conventional feedback control is dependent on designing a suitable algorithm that capitalizes on the real time on line/in line measurement of the product. These will be introduced in the coming p aragraph respectively. Size map based C ontrol Map the S ize R ange Mapping the size range of the FSS product was important to the size map based control system since it provides the information between particle size and flow rate which is necessary to the c ontrol algorithm. An initial design was made for the 20 mL tube reactor to explore the size range as shown in Table 3 2 In this design, the total flow rate was fixed at 0.7 mL /min to fix the reaction time. With the fixed flow rate, the FSS has two degrees of freedom left, any two of the concentrations of the three reactants (ammonia, water, TEOS) can be adjusted but the third must bring the sum to the total flow rate of 0.7 mL /min. The tri axial diagram ( Figure 3 15 ) depicts the relationship between three flow rates and the resulting particle size. T he characterization was done

PAGE 40

40 by both the LS13320 and Nanotrac to cover the whole range of particle size. The results show that particle size can be well controlled in the range from 29nm to 401nm Interestingly, the tri axial diagram indicates that in cer tain cases there are multiple conditions that can produce the same particle size. Samples with less than 100nm M.V. particle size were far from 100% yield, since the reaction time for such particle size usually takes hours. A more detailed experiment was done to gather information about the relationship between the particle size and the flow rate. Among the three reagents, the TEOS concentration is thought to have the smallest effect on the particle size 108 Thus, the ammonia and water flow rate were chosen as the main control parameters. The 63 mL PTFE tube loop was used in this design, ensuring enough residence time (90min) for the reaction. The DelsaNano and LS13320 were used for the size characterization, the DelsaNano for in line and the LS for off line post reaction sizing. The experimental de sign and results are shown in Table 3 3 The test flow rate range was located in the middle of the tri axial diagram from 0.2 to 0.3 mL /min for water a nd 0.1 to 0.2 mL /min for ammonia for the sake of balancing the consumption of reagents. This design provided a full coverage of particle size from 90nm to 495nm where both the reaction time and mono dispersity can be ensured by FSS. Furthermore, the partic le sizes corresponded to every 0.01 mL /min step change of flow rate ( Table 3 4 ). Conditions were selected to determine what the effect of TEOS concentra tion on the trend illustrated in the overview 3 D chart in Figure 3 16 Control A lgorithm Figure 3 17 shows the fundamental regulation of the size map based control system. The process starts by inputting the target M.V. particle size in the software after

PAGE 41

41 the FSS was in the common running state. O nce the target size was set, the software initiated a search for the optimum flow rate combination in the database which was indexed and calibrated from detailed mapping and calculation step. The flow rate combination was then delivered to pumps for the tu be reaction (dead time: 90min). After the reaction, the product was delivered by the dilution system to the online detector where the particle size distributions were measured and sent back to the control software. A reliability test was applied to the M.V size from online detector during the monitoring. Specifically, the standard deviation (SD) of the three latest M.V. size were calculated and compared with the resolution of DelsaNano at that particular size to confirm proper operation. The reliability of the size map based control depends on the validity of the mathematic model, the. database in this case. An additional control loop was added alongside with the main route to minimize the error from unpredictable disturbances, for example, the changes of r eagent concentration. After the reliable testing, the average of last three M.V. size were calculated and further compared with target size. If the S.D. is within the precision expected of the DelsaNano, it would be added to the database and reset the flow rate. Two samples for the size map based control and their final flow rate set are shown in Figure 3 18 The target sizes were set as 260nm and 430nm respectively and the resulting products have 1.9% and 2.6% deviations from the target. Feedback C ontrol The feedback control model is an importa nt fraction of process analytical technique that provide s the compensation mechanism to an unknown disturbance with relatively low workload and understanding of the system. The ultimate objective is to create a

PAGE 42

42 system that will provide feedback to the cont rol system to obtain a desired particle size during non steady state conditions by adjusting the flow rate of inputs. This system is integrated with a Proportional Integral (PI) controller. Methods First order plus time delay (FOPTD) FOPTD process is one that shows an exponential response to an input step change with a delayed response. The output response of FOPTD to a change in input can be mathematically presented as the following equation 109 : ( 3 5 ) where K is the input. The FOPTD Exxon Three Point Method is used to determine the above three unknown parameters This method involves finding the ti me s when process reach ed 25% and 75% of o utput. ( 3 6 ) ( 3 7 ) ( 3 8 ) where Y is the output and U is the input. Bode Plot. A bode plot can be constructed to determine frequency response information of a given transfer function. After calculating G u (the transfer function of FOPTD model, E quation 3 9 ) from the step up and step down experimental data and G c (the transfer function of feedback control ler, Equation 3 10 ) from the Cohen Coon

PAGE 43

43 method, the G OL (the product of all transfer function in the loop, Equation 3 11 ) can be calculated for the Bode plot. ( 3 9 ) ( 3 10 ) ( 3 11 ) From the Bode plot we can conclude whether the transfer function meets the Bode Stability Criterion. The criterion says that a closed loop system is stable only if the bode plot of G OL has: co ) < 1 ( 3 12 ) Or co )) < 0 ( 3 13 ) A measure of stability can be determined by calculating the gain margin (GM) and co )) co 180 A rule of thumb for safety says that the GM should be at least 1.7 and the PM should be at least 30 Tuning Methods The Cohen Coon (C C) tuning method is one method to tune the PI controller. Tuning parameters for control gain (Kc) and integral time constant the PI controller are determined by the following equatio ns : ( 3 14 ) ( 3 15 )

PAGE 44

44 Another tuning method is the Ziegler Nichols (Z N) tuning method. Its control parameter derived from the ultimate gain (Ku) and ultimate period (Pu) that brings the closed loop system to the verge of instability. Ku and Pu can also be derived from the ( 3 16 ) ( 3 17 ) The formulas used to find K c i are: ( 3 18 ) ( 3 19 ) Results and D iscussion First Order Plus Time Delay: To find the FOPTD model, the FSS was set on manual mode and waited till the process was at the steady state, with inputs at 0.15, 0.27 and 0.28 mL /min for ammonia, water and TEOS respectively. A step change was then introduced to the ammonia flow rate, increasing the flow rate from 0.15 mL /min to 0.17 mL /min. After 1.5 hours, when another steady state was reached, the FSS was set back to nominal steady state. The step change down was done in a same procedure by decreasing ammonia flow rate from 0.15 mL /min to 0.13 mL /min for a period of 1.5 hour. The Exxon method was applied to calculate the process gain, time constant and time delay 109 Figure 3 19 shows the step up increase in flow rate of ammonia, the flow rate was increased from 0.15 mL /min to 0.17 mL /min. K, and D were obtained by the following calculations:

PAGE 45

45 ( 3 20 ) ( 3 21 ) ( 3 22 ) Similarly, Figure 3 20 to Figure 3 22 changes made in ammonia and water and the results are shown in Table 3 5 G u (s) was derived by using the average values of the step change data from ammonia step changes, specifically: ( 3 23 ) ( 3 24 ) Tuning Methods: The average value of both ammonia and water flow rate were applied to calculate the parameters for the PI controller. In C C method, K c and I were calculated as followed: Ammonia: ( 3 25 ) ( 3 26 ) Water: ( 3 27 ) ( 3 28 ) In Ziegler Nichols (Z N) method the calculation for ammonia is show n below : ( 3 29 ) Since G u (0)= 2375 (positive) G u

PAGE 46

46 From Bode Plot of G u 180) Log AR= 3.353 Log W= 0.2849 ( 3 30 ) ( 3 31 ) ( 3 32 ) ( 3 33 ) ( 3 34 ) ( 3 35 ) ( 3 36 ) The calculation for water using averages is shown below: ( 3 37 ) Since G u (0) = 1250 (positive) G u u (o))* G u 180) l og AR=3.089 Log W=0.2822 ( 3 38 ) ( 3 39 ) ( 3 40 ) ( 3 41 ) ( 3 42 ) ( 3 43 ) ( 3 44 ) Bode Stability and safety margins : Using the G u calculated form step change and the G c G OL was acquired to input into the Bode plotting software by equation: ( 3 45 )

PAGE 47

47 Figure 3 23 shows the Bode Plot drawn from E quation 3 45 which can be used to determine the GM an d PM mathematically and graphically. The transfer function meets co )) is less than zero. ( 3 46 ) ( 3 47 ) The GM of 1.62 is close to the recommended valve of 1.7, indicating the small possibility of instability, while the PM of 33.8 is already at the safe range (above 30 degree). Simulation a nd experiment data : There are two differences between the ideal feedback model and the model used in the FSS. The main difference comes from the discreteness. The output flow rate from the ideal model is continuous, while the flow rate settings are actuall y discrete because of the minimum flow rate change of the pumps (0.01 mL /min). The input signal ( M.V. particle size) in the ideal model is also continuous, yet the real online detector takes 8min for each measurement. The other difference comes from the noise of the size measurement output, which might confuse the feedback control software in the wrong di rection. The simulation of the above differences gives a prediction of system behavior before actually applying model into real system. In the simulation, the change of M.V. particle size was calculated by E quation 3 5 i ) (0 < t i
PAGE 48

48 ( 3 48 ) The discrete of input M.V. particle size was also taken in account in the simulation system. Figure 3 25 show s how the system would react to a simple flow rate change with the same condition of former step change experiment The simulated result performed the same as experiment data with the same amount of size change, indicating that the simulation operates properly. Figure 3 26 shows an ideal feedback control (continuous flow rate) using Cohen Coon method by adjusting ammonia flow rate. 5 hours was required to reach the target size with no oscillation or steady state offset. However, the final flow rate ( 0.175 mL /min ) was not available with the real pump s The discrete flow rate was then added to the simulation for further prediction as shown in Figure 3 27 The flow rate calculated by the control algorithm (blue line) was rounded to two decimal (red line) to mimic the real pump. The resulting response of simulated FSS performed as cumulative step changes. A periodic oscillation appeared from 5hour due to the steady state offset that cannot be avoided by the current setting and instrument in FSS B ut its effect on broaden particle size distribution can be tempered if the offset is decreased. The noise was finally introduced to simulate the accuracy of online detector as shown in Figure 3 28 For simplification, noised data was rando ml y chosen in the range of 90%~110% of original data point instead of use Gaussian dispersion. The noise reduced the sensitivity of control algorithm and released the error accumula tion speed so that the oscillation from the discrete flow rate almost disappeared. The simulation of feedback model reveals two facts: First, the feedback control should be temporarily shut down to better achieve stabilization when particle size is

PAGE 49

49 close enough to the set point so that oscillation can be avoided. Second, a smaller offset would help eliminate the final oscillation. The Ammonia flow rate was set to be the prime parameter because its larger step change saves time, while the water flow rate wa s used for finer control of particle size. The final control algorithm is shown in Figure 3 30 Following these rules, the control algorithm was upda ted and the experiment data is given in Figure 3 29 The start point was set the same as step change experiment, with the M.V. particle size at 240nm, and the target size was set at 150nm. The ammonia flow rate dominates the control algorithm when the M.V. particle size is far from target size. The water flow rate is introduced at about 2.5hour when the difference between M.V. particle size and target si ze was smaller than the minimum adjustment from ammonia flow rate step change, although it switches between ammonia and water several time from 2.5hour to 3hour due to the variation of particle size. The fine adjustment from water flow rate took several ho urs since the error built up very slowly, but it reached set point at 7hour finally. Dye D oped S ilica The synthesis of Dye doped silica (DDS) is one of the extended applications for the FSS. Among all kinds of DDS synthesis methods, the modified Stober silica methods with chemical bonding and physical adsorption of dyes were chosen for the similarity and o perability with the earlier work. Four dyes were tested: FITC and 7 methoxycoumarin 3 carboxylic acid (MCA) were chemically bonding to silica particle by the pre reaction with APTS while Rubpy and Rhodamine 6G (R6G) were physically adsorbed. For FITC and M reaction, 1.5 times of APTS were mixed with 0.03mM sample in 2 mL solvent (ethanol for FITC and DMF for MCA). The solution was placed in darkness for 1 hour to enhance the fully reaction. The dye solution was then

PAGE 50

50 mixed with 100 mL of TEOS reagent ( 1:5 TEOS/Ethanol) and ready for FSS. The FSS system was shielded by aluminum foil to prevent any light induced oxidation. Since it is the modified Stober silica reaction with tiny amount of additives, the effect of dyes on particle size is limited and the feedback control algorithm is able to manipulate size distribution. Figure 3 31 shows a sample for each dye with difference sizes, which proved the abi lity of producing selected particle size distribution DDS by FSS.

PAGE 51

51 Table 3 1 Several formula for different size of Stober silica Target size (nm) Ammonia ( mL ) H2O ( mL ) Ethanol ( mL ) TEOS ( mL ) 203.5 2 0 50 4 12710 2 2 48 4 27112 2 3 47 4 3558.5 2 4 46 4 3745 2 6 44 4 52012 3 6 43 4 Table 3 2 Experiment set for mapping size range No. Ammonia ( mL /min) Water ( mL /min) TEOS ( mL /min) Mean volume Size (nm) 1 0.05 0.05 0.60 No particle observed 2 0.05 0.23 0.42 29 3 0.05 0.42 0.23 54 4 0.05 0.60 0.05 No particle observed 5 0.23 0.05 0.42 60.2 6 0.23 0.23 0.24 347 7 0.23 0.42 0.05 180 8 0.47 0.05 0.23 193 9 0.47 0.23 0.05 401 10 0.60 0.05 0.05 tube clog Table 3 3 Detailed experiment for covering silica size range No. Ammonia ( mL /min) Water ( mL /min) TEOS ( mL /min) Mean volume Size (nm) 1 0.1 0.2 0.4 90 2 0.1 0.25 0.35 96 3 0.1 0.3 0.3 153 4 0.15 0.2 0.35 143 5 0.15 0.25 0.3 198 6 0.15 0.3 0.25 365 7 0.2 0.2 0.3 223 8 0.2 0.25 0.25 405 9 0.2 0.3 0.2 495

PAGE 52

52 Table 3 4 Step change of flow rate and resulting particle size Water flow rate, mL /min 0.2 0.21 0.22 0.23 0.24 0.25 0.26 0.27 0.28 0.29 0.3 Ammonia flow rate, mL /min 0.1 90 87.12 86.3 87.5 90.7 96 103.3 112.7 124.1 137.5 153 0.11 98.4 93.6 92.1 94.1 99.3 108 120 135.5 154.2 176.4 202 0.12 108 102.6 101.6 104.8 112.4 124.2 140.3 160.7 185.4 214.4 247.6 0.13 118.6 114.2 114.7 119.9 129.8 144.6 164.1 188.5 217.5 251.4 290 0.14 130.2 128.4 131.3 139.1 151. 8 169.2 191.5 218.7 250.7 287.5 329.2 0.15 143 145 151.6 162.6 178 198 222.4 251.4 284.8 322.6 365 0.16 156.8 164.3 175.4 190.3 208.8 231 256.9 286.5 319.9 356.9 397.6 0.17 171.8 186.1 202.9 222.2 243.9 268.2 295 324.2 355.9 390.1 426.8 0.18 187.8 210.4 233.9 258.3 283.5 309.6 336.5 364.4 393 422.5 452.8 0.19 204.8 237.3 268.6 298.7 327.5 355.2 381.7 406.9 431 453.9 475.7 0.2 223 266.8 306.8 343.2 3 76 405 430.4 452 470 484.4 495

PAGE 53

53 Table 3 5 K, Changing Made K (nm/ mL /min) (Hours) D (Hours) NH 3 Step Up 2150 0.209 1.45 NH 3 Step Down 2600 0.132 1.49 NH 3 Step Averages 2375 0.171 1.47 H 2 O Step Up 1100 0.105 1.51 H 2 O Step Down 1400 0.0909 1.58 H 2 O Step Averages 1250 0.0977 1.55

PAGE 54

54 Figure 3 1 Piston pump from Syrris Co. which has two pistons works at reverse stroke. The self flush offered cleaning function for the piston.

PAGE 55

55 Figure 3 2 Sketch of flow system. The regents are pumped into PTFE tube continuously in a designed flow rate which is controlled by software. The PTFE tube is used to reduce particle attachment on the inner wall of the tube. All tubes are immersed in the Ultrasonic bath for further prevention particle attachment. Temperature control is a combination of heater and cooling water system in the ultrasonic bath. Local network Collection Driven by height difference Ethanol

PAGE 56

56 Figure 3 3 Detection of bi dispersed Stober silica particle by multiple techniques. (a) LS13320; (b) Nanotrac; (c) DelsaNano; (d)CPS disc centrifuge; (e) image analysis. The mono dispersity of Stober silica particle is disturbed when M.V. particle size is above 550nm. Multiple test from different techniques shows that DLS method is not o ptimum for polydispersed samples. Compared with other instrument, the LS method has the best agreement with image analysis data. Mean value: 534nm SD: 82.20nm Mean value: 553nm SD: 114nm A B

PAGE 57

57 Figure 3 3 continue d Mean value: 532.1nm SD: 27.3nm Mean value: 463.5nm C D

PAGE 58

58 Figure 3 3 C ontinue d Mean value: 555.4nm E

PAGE 59

59 Figure 3 4 Calibrated offset of DelsaNano with LS13320

PAGE 60

60 Figure 3 5 (a)Sketch of online Delsa N ano and its dilution system; (b) flow chart for the DelsaNano online detector system Collection Ethanol Raw Stober silica suspension Peristaltic pump Waste bottle Received Start command from control software Peristaltic pump running (2.5min) Equilibration in flow cell (2min) Size measurement (3.5min) Save data and deliver to control software

PAGE 61

61 Figure 3 6 Particle size distribution of Stober silica measured by the Coulter LS13320 (a) made by batch synthesis, (b) made by flow synthesis.

PAGE 62

62 Figure 3 7 T he relationship between and mean size of batch made Stober silica (trend line added) 0 10 20 30 40 50 60 70 80 90 100 0 100 200 300 400 500 600 700 Standard deviation, nm MV particle size, nm The Standard deveiation of Stober silica particles Poly. (The Standard deveiation of Stober silica particles)

PAGE 63

63 Figure 3 8 SEM picture of batch made Stober silica.

PAGE 64

64 Figure 3 9 T he relationship of settling distance in 90min with M.V. particle size for the Stober silica suspension. 0 0.5 1 1.5 2 2.5 3 3.5 0 200 400 600 800 1000 1200 Sdimentation distance, mm MV Particle size, nm

PAGE 65

65 Figure 3 10 Stability of FSS The residence time was controlled at 30 minute s Measurements were made every 2.5 minutes.

PAGE 66

66 Figure 3 11 Repeat experiment s about the flow rate changed from 0.15 mL /min, 0.3 mL /min, 0.25 mL /min (Ammonia, Water, TEOS) to 0.22 mL /min, 0.26 mL /min, 0.22 mL /min in 30min tube reactor The reproducibility of flow system

PAGE 67

67 Figure 3 12 Gradually decrease of particle size during the long term operation of FSS without ultrasonication. 0 100 200 300 400 500 600 0 5 10 15 20 MV particle size, nm Time, hour

PAGE 68

68 Figure 3 13 Cross section of PTFE tubing showing the sedimentation of silica particle on the tube wall. The red line is the inner surface of the tube.

PAGE 69

69 Figure 3 14 Stability test on FSS with ultrasonicator.

PAGE 70

70 Figure 3 15 T ri axial diagram of particle size map. 347 401 60. 54 No particle 29 No particle 180 193 ml/min ml/min ml/min Agglomeration

PAGE 71

71 Figure 3 16 T hree dimensional graph of particle size map 0.1 0.2 0 50 100 150 200 250 300 350 400 450 500 0.2 0.3 Ammonia flow rate, ml./min MV particle size, nm Water flow rate, ml/min 450-500 400-450 350-400 300-350 250-300 200-250 150-200 100-150 50-100 0-50

PAGE 72

72 Figure 3 17 Flow chart of the size map based control Stop flow system Smaller Set target size First or not Database Calibrate target size Detector Resolutio n SD of last three results Set flow rate Tube Online detector Display MV size Large Reach target? No Yes No Start flow system Collect sample

PAGE 73

73 Figure 3 18 Size map based on size map control method. Target: 260nm Target: 430nm

PAGE 74

74 Figure 3 19 Ammonia step up data, with a 2 period moving average, ammonia flow rate increase from 0.15 mL /min to 0.17 mL /min. 230 235 240 245 250 255 260 265 270 275 280 285 290 295 300 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 Particle Size (nm) Time (h) Step up NH3 Series1 0.75y

PAGE 75

75 Figure 3 20 Ammonia step down data, with a 2 period moving average, ammonia flow rate decreased from 0.15 mL /min to 0.13 mL /min. 180 185 190 195 200 205 210 215 220 225 230 235 240 245 250 255 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5 5.1 5.2 5.3 5.4 5.5 Particle Size (nm) Time (h) Step down NH3 Series1 .75dy .25dy

PAGE 76

76 Figure 3 21 Water step up data, with a 2 period moving average, water flow rate increased from 0.15 mL /min to 0.17 mL /min. 230.0 235.0 240.0 245.0 250.0 255.0 260.0 265.0 270.0 6.5 7.0 7.5 8.0 8.5 Particle Size (nm) Time (h) Step up H2O Data .75y .25y 4 per. Mov. Avg. (Data)

PAGE 77

77 Figure 3 22 Water step down data, with a 4 period moving average, water flow rate decreased from 0.15 mL /min to 0.13 mL /min. 200 205 210 215 220 225 230 235 240 245 9.5 10.0 10.5 11.0 11.5 Particle Size (nm) Time (h) Step down H2O .75dy .25dy Data 4 per. Mov. Avg. (Data)

PAGE 78

78 Figure 3 23 Bode Plot created using AAS_ECH4323NP. The p lot showing Bode Stability lines for our transfer function.

PAGE 79

79 Figure 3 24 GM and PM from Bode plot:

PAGE 80

80 Figure 3 25 Simulation of simple step change of ammonia flow rate. 230 240 250 260 270 280 290 0 1 2 3 4 5 6 7 MV particle size, nm Time, hour 0.145 0.15 0.155 0.16 0.165 0.17 0.175 0 1 2 3 4 5 6 7 Ammonia flow rate, ml/min Time, hour

PAGE 81

81 Figure 3 26 Simulation of feedback control with set point changed from 240 to 300 0.145 0.15 0.155 0.16 0.165 0.17 0.175 0.18 0 2 4 6 8 10 12 MV particle szie, nm Time, hour 0 50 100 150 200 250 300 350 0 2 4 6 8 10 12 Ammonia flow rate, ml/min Time, hour

PAGE 82

82 Figure 3 27 Simulation of feedback control with discrete (stepped) flow rate 200 220 240 260 280 300 320 0 2 4 6 8 10 12 MV particle size, nm Time, hour 0.145 0.15 0.155 0.16 0.165 0.17 0.175 0.18 0.185 0 2 4 6 8 10 12 Ammonia flow rate, ml/min Time, hour Calculated Real

PAGE 83

83 Figure 3 28 Simulation of feedback control with discrete flow rate and noise of data 200 220 240 260 280 300 320 340 0 2 4 6 8 10 12 14 MV particle size, nm Time, hour Simulated output w/o noise Simulated data with noise 0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0 2 4 6 8 10 12 14 Ammonia flow rate, ml/min Time, hour Calculated flow rate Real flow rate

PAGE 84

84 Figure 3 29 Feedback control in the flow synthesis system including a set point change at time zero. 100 120 140 160 180 200 220 240 260 280 300 0 1 2 3 4 5 6 7 8 9 MV particle size, nm Time, hour Result from FSS 0.125 0.13 0.135 0.14 0.145 0.15 0.155 0 1 2 3 4 5 6 7 8 9 Ammonia flow rate, ml/min Time, hour Real flow rate Calculated flow rate Coarse tuning by ammonia flow rate 0.258 0.26 0.262 0.264 0.266 0.268 0.27 0.272 0 1 2 3 4 5 6 7 8 9 Water flow rate, ml/min Time, hour Real flow rate Calculated flow rate Fine tuning by water flow rate

PAGE 85

85 Figure 3 30 Flow chart of feedback control algorithm Online measurement Start FSS Set target size No Yes Is error larger than threshold? Ammonia flow rate adjust Water flow rate adjust Calculate flow rate output Set new flow rate Stop flow system

PAGE 86

86 Figure 3 31 Dye doped silica samples prepared by FSS. A: 112nm silica particle doped by Rubpy; B: 171nm silica particle doped by R6G; C: 370nm silica particle doped by MCA; D: 338nm silica particle doped by FITC

PAGE 87

87 4 CHAPTER 4 HYDROTHERMAL QUANTUM DOT SYNTHESIS IN FSS AND PROCESS CONTROL Convert from Batch to FSS The routine route for the hydrothermal CdTe QD synthesis method was reported by Guo et al 110 First, there is generat ion of Te 2 ions in the water by reduc ing the t e llurium powder with sodium borohydride (NaBH4, 98%) under nitrogen protection followed by storage in a refrigerator overnight Second, the Cd 2+ Te 2 and the li gand are mixed in a ratio of 2:1:4 in a Parr acid digestion bomb with an adjusted pH and then heat ed to the target temperature for the required time The resulting QD properties var y with different temperature s and reaction time s To achieve the desired emission wavelength, the reaction time can var y from 60 min to several hour s 111 112 After repeating the batch method hydrothermal CdTe QD synthesis, several problems were revealed. First, there w ere unreacted particles visible at the bottom of the solution after reduction, which may be contaminant s that resulted from the tellurium powder (99%, Fish erSci ) S econd, the Te 2 ions are highly sensitive to oxygen, which require s nitrogen protection during the entire s ynthesis process. Any leak can cause the formation of tellurium nano particle s that will turn the solu tion to pink or black. Th is poses a problem in determining the true concentration of Te 2 ions actually involved in the reaction because the tellurium metal can not react. S everal improvements were made on the formulas to avoid the above problems Tellurium metal particles were replaced by 100% soluble tellurium chloride salt ( TeCl 4 99%, Acros Organics ) in case any particles were brought into the FSS. However, t he reduction of TeCl 4 requires four times the amount of NaBH 4 compared with tellurium powder yet the reaction is faster and is not limited to the surface reaction. The

PAGE 88

88 concentration of the tellurium ions was reduced to 0.5mM compared to the 7.5mM in the original batch method 110 The low concentration of QD s not only reduces the risk of tube blockage caused by oxidized tellurium powder, but also may suggest a potential condition to obtain high photoluminescence of the QDs 113 because the excess ive tellurium ion damage the thermo dynamically favorable structur e. Briefly the TeCl 4 was first weigh e d and dissolved in DI water with droplets of 1M NaOH solution Then the solution wa s sealed in a reagent bottle and bubbled with nitrogen un til the whole process was finished An hour later NaBH 4 was weighed and dis solved with w ater (pH 9.3) and quickly injected into the reagent bottle. The whole bottle was then warmed by a hot plate set at 80 C with a stir rer to accelerate the red uction Any generated hydrogen gases were removed by nitrogen and a fume hood so that the risk of explosion was eliminated. After the reaction finished, the solution was cooled down to room temperature and then was ready to use. On the other hand, CdCl 2 was dissolved in DI water with N a cetylcysteine (NAC, 99%) in an other reagent bottle wit h pH adjusted to 9 The formation of the coordination compound at an alkaline pH is required to avoid Cd(OH) 2 precipitate s 114 A 1:1 molar ratio was found to give the lowest ratio at which NAC and CdCl 2 are completely dissolved at a pH of 9 The Cd reagent bottle was also bubbled with nitrogen for one hour to prevent any oxygen penetration. I nstrument and design Instruments The FSS is composed of two piston pumps (Syrris Co.), a PTFE tub e ( 0.75mm ID ), s tainless steel (SS) tube s ( ), and a backpressure regulator (IDEX Co.) A Hitachi F 2000 fluorescence spectrophotometer was modified as an inline detector for t he emission spectra by the application of a flow quartz cuvette of 1 0mm path length (NSG Precision Cells). All optical measurements were carried out at room

PAGE 89

89 temperature under ambient conditions. The pH measurement s were made by the AR60 pH meter (Fisher S ci ). Transmission electron microscopy (TEM, JEOL 2010F) was used to characterize the CdTe QDs. Labview 8.5 software was used to connect the pumps and the fluorescence spectrophotometer for the purpose of online measurement and flow rate control. Quantum yiel d was measured by a fluorometer (Horiba NanoLog) Micro reactor design and set up The capillary micro reactor synthesis system is shown in Figure 4 1 Two piston pumps were used to feed the precursor solution s prepared earlier and also the SS tubing with a designed flow rate. The nucleation and reaction take place in the heat zone of the tubes for PTFE t ubing ; 250 SS tube ; 2.4 mL ) which is coiled and set in an oil bath with a constant temperature. Next, the solution was immediately cooled though another coil ed bath to quench the reac tion and also to avoid any potential damage inside the back pressure regulator (9bar 1 3bar) due to high temperature s The fluorescence spectrophotometer was connected in an inline manner through the use of a flow cuvette (440uL) so that it could provide real time data for further analysis. Result s and discussion The reactor presented in Figure 4 1 enables an adjustable isothermal reaction condition, which minimize s the emission wavelength fluctuation due to temperature difference s Different residence time s and precursor ratio s can be achieved by changing the flow rate of the precursors By carefully tuning the se parameters, the FSS is able to produce the high quantum yield QDs with the max ranging from 500nm to 800nm. In the following section the results from the hydrothermal synthes is of CdTe QD at different

PAGE 90

90 reaction conditions under a steady state operation of the capillary micro reactor system are discussed Effect of reagent concentration on QDs The traditional precursor ratio of [Cd]:[ligand]:[Te] was chosen as 1:2.4:0.5 for the consumption of the tellurium using excessive cadmium 115 However, the concentration of the reagents is its PL propert y, especially in water base d QD s 116 T he effects of reagent concentration s on precise residence tim e and the ir PL property w as explored by comparing QDs that are emitted at 557nm made at different reagent concentrations, which is difficult to achieve with batch synthesis. Our experimental results indicated that the amounts of [Cd 2+ ] as well as [NAC] c a n strongly influence the PL properties of hydrothermally prepared CdTe QDs. As shown in Figure 4 2 the QY of CdTe QD gradually increased from 20% and stabilized at 45% as [Cd 2+ ] increased from 2.5mM to 12.5mM at 170C Meanwhile, the increase of [NAC] has a n opposite effect on the QY, red ucing the QY from 46% down to 20% as the [NAC] increased f rom 12.5mM to 30mM. The effect of reagent concentrations on reaction times required for 577nm QDs were contrary to that of QY s T he residence time was reduced from 7.6s (2.5mM [Cd 2+ ]) and 11s (30mM [NAC]) down to 3s for both 12.5mM [Cd 2+ ] and ~ 15mM [NAC] The variation of the QY results from the surface structure of QDs. Surface defect s which are controlled by the dynamic growth process of the QDs, are believed to be one of the main reasons for low QY With the equilibrium of dissolution and growth at the QD surface, defect s can be repair ed by the Ostwald ripening phenomenon 117 Bao et al revealed that QY can be gradually enhanced as time goes by without any treatment 118

PAGE 91

91 An alternative way to reduce the surface defect is by forming a thin tellurium poor layer t hat covers the origi nal defects, such as with a proper layer of organic ligands The ligand molecule s can interact with the s and thereby supply sulfur atom s into the crystal structure 119 Borchert et al showed that the highly luminescent CdTe QDs possess fewer tellurium atoms at the surface than QDs with low lumi nescence. 120 In our case, as the ratio of [Cd 2+ ]: [Te 2 ] increases the QD surface may be enrich ed in cadmium atoms thus providing more sites for ligand attachment. When the QD surface bec o me s full of cadmium atoms, further increasing the [Cd 2+ ]:[Te 2 ] ratio cannot drive ligand attachment. T herefore the QY growth trend slowed down as the ratio increased from 10mM to 12.5mM. On the other hand, the reduc tion of the residence time that is required for the same emission wavelength QD should be considered through the ki netic process. Generally, the nucleation and the growth speed are controlled by the conce ntration of the free precursor, such as any free c admium ions More free cadmium ions lead to a faster reaction. It has been demonstrated by Farideh et al. 114 that cadmium can form coordination compounds with NAC at near neutral pH as well as a high pH value which is also the key to prepare cadmium precursor s under an alkaline condition They illustrate that the free, reactive cadmium ions are only available at a very low concentration from the reversible reaction of Cd(I I) complexes. By either increasing the [Cd 2+ ] or decreasing [NAC], more free cadmium ions will be released in to the solution This provides a potential site for nucleation and accelerat ion of crystal growth. Similar results were observed in previous publications 113 121

PAGE 92

92 Effect of reaction temperature on QDs We explored the effect of the reaction temperature within a range of 115 C to 185 C The residence time was constant ly set at 5.8s to provid e enough time for QD growth at low temperatures. Figure 4 3 (a) shows the normalized emission wavelength spectrum, indicat ing that the emission wavelength and therefore the size of QD s accelerated with the rising temperature. max is clearer in Figure 4 3 max can be calculated with a given temperature using the following equation: ( 4 1 ) Every degree rise in the max change of 1.6nm o n average, which is small enough for the precise control of the emission wavelength. The polynomial trend line illustrates that the reaction temperature is raised which is consist ent with other studies 122 P icking a suitable r eaction temperature is a strategy depending on the target of the process. A higher temperature reduces the reaction time and thus increases the yield especially for near inf rared QDs. However, a low temperature is preferred sin ce a slow growth speed is believed to help reduc e the surface defect s th r ough Os t wald ripening 117 Although lower reaction temperature s w ere reported 118 the QDs synthesized at 115C in the FSS approach max ever reported 123 by the hydrothermal method On the other han d, the maximum temperature in our appar a tus is limited because of the limitation of the back pressure regulator to 10 bar. In practice, 170 C to 180 C temperatures were chosen by considering both reaction speed and PL properties. We have observed 40 60% qu antum yields from the QDs with a max between 510 nm and 730nm synthesized in this temperature range.

PAGE 93

93 Effect of residen ce time The effect of the residence time is demonstrated in Figure 4 4 a with a reaction temperature of 180 C Lo nger residence times result ed in longer wavelength emission s The max of CdTe QD range s from 509 nm to 641nm attributable to 25 40 particles 124 Figure 4 4 b shows that the residen ce time has a logarithmic relationship with the emission wavelength as defined in the following equation: ( 4 2 ) Moreover, by converting the emission wavelength calculated in Equation 4 2 using the below equation, the band gap of the QD s can be calculated: ( 4 3 ) Sotirios et al. 125 calculated that the effective band gap energies for CdTe QD were a function of the dot radius. Th erefore the residence time can be directly related to the average radius of the QDs ( Figure 4 5 a ). The linear plot of the cube of the average QD radius ( Figure 4 5 b) is consistent w ith the Ostwald ripening growth mechanism, which support s the hypothesis that the growth mechanism of QD s is mainly controlled by the Ostwald ripening 117 Compared to conventional batch methods, the reaction time required for the same emi tted QD at the same reaction temperature is dra matically lower using the FSS. According to previous reports, several studies required at least 30 minutes to reach the minimum emission wavelength 126 The h ot injection method reduc ed the reaction time to the 2 minute scale 94 which is still 10 times longer and is difficult to scale up. The c hemical aerosol flow method provides an alter native approach giving a comparable

PAGE 94

94 reaction time at 200C 270C, al though further treatm ent s such as coating and/or bio conjunction are limited. 122 The reason for the reaction time variation may be due to th e different thermal conductivit ies between each system. For example, the traditional batch methods involve the heating of both stain less steel jacket s and PTFE vessel s via an oven so that the heating ratio is mainly controlled by the thermal conductivity of the air. While applying the hot injection method, the container and part of the solution are already heated T hus the reaction time can be reduced substantially. The FSS takes the advantage of the relatively high thermal conductivity of both the PTFE and SS tub es and also the thin cross section of the tubes, where the liquid can approach its target temperature almost instantaneously so that no time is wasted on heating. Since the flow rate is adjustable with a 0.02 mL /min minimum step change, the max is reduced to 0.5nm, which is the best resolution achieved A straightforward example is given in Figure 4 6 to demonstrate the ability of tun ing the emission wavelength of the aqueous CdTe QDs. XRD characterization of CdTe QDs XRD patterns of CdTe QD s obtained by flow synthesis at different reside nce time s are shown in Figure 4 7 The green QD pattern is consiste nt with the bulk CdTe materials, which belongs to cubic (zinc blende) structure How ever, the other pattern s from the yellow and red QD s reflect that the crystal structure of QD s shifted from the cub ic CdTe towards the cubic CdS as the residence time increased. Similar results in the XRD pattern have been reported in the synthesis of CdTe using thiol group ligand. 89 This phenomenon is consistent with the theory that a sulfur shell is generated from the thiol group of a partially hydrolyzed ligand. It can be pr evented by u sing DMF as the solvent 127

PAGE 95

95 or performing synthesis at comparatively low pH (5.6 5.9) in the presence of 2 mercaptoethylamine as the stabilize r 89 This limit s the in corporation of sulfur into the growing CdTe QD s TEM characterization of CdTe QDs synthesized at 180C The CdTe QDs synthesized at 180C were characterized by TEM as shown in Figure 4 8 The distinguishable lattice planes reveal the high crystalline of QD s It also indicate s that CdTe QDs produced by FSS ha d a narrow particle size distribution and were we ll dispersed in the solution. The average siz e was around 2nm from the TEM pic ture and was also consistent with the estimated mean particle size by using the effective mass approximation. 128 Thermal Control As the understanding of the QD reaction goes deeper the stability of the temperature is wavelength because of the temperature sensitivity of the reaction. As discussed in the previous chapter, a simple oil bath made by using a glass c ontainer and a hot plate with 1 2 C variation can not suppor t the requirement of manufacturing a 1nm resolution QD, A 1 C temperature change contribute s to a 1.6 nm EW change. Figure 4 14 gives an max is affected by the oscillation of the reaction temperature. The new heating system was designed and composed with a heating mantel, PTFE and P I board, external stir rer, a peristaltic pump, and a thermal couple for the precise control of the reaction temperature A s shown in Figure 4 9 the temperature inside the new heating system is determined by the interaction of the heating mantel and the cooling water while the external stir rer ensures the adequate heat exchange. The heating mantel is able to heat the oil bath at 4.5 C /min at its full power. The cooling water

PAGE 96

96 is pumped by a digital peristaltic pump whose flow rate can be adjusted by Labview software. Moreover, due to the excellent the rmal conductivity of copper tubing the cooling water could remove heat by evaporation. In the following test, the flow rate of the cooling water is limited so that it could reach the same cooling speed as heating ( 4.5 C /min). The SS reaction tube sits in the innermost space, which is adja c en t to a thermal couple in order to get the precise detection of the reaction temperature. The new device was tested by three control algorithm s for the purpose of a fast and accura te control algorithm : the P controller, PI controller and the on off controller. The Good Gain method was applied to find out the suitable Kp for the P controller without the need for specific knowledge about the new heating system This method involves a series of adjustments in altering the set point with the Kp value increasing from 0 or 1 until the system response is acceptable. In a brief test, Kp was set as 20, 40, and 80 respectively as shown from Figure 4 10 to Figure 4 12 Both the phenomenon of overshoot ing and oscillation were observed in a ll three cases, which is not desirable A detailed mathematical model was built for the PI controller in order for a smooth er temperature alter ation. Figure 4 13 indicates the step change data for the manipulated variable ( MV ) decrease from 40 down to 35, where K, and D were calculated from the tangent line: In Z N method, the calculation of the control parameter s is shown below: ( 4 4 )

PAGE 97

97 From t he Bode plot of G u (s) ( obtained from program AAS_ECH4323Noline.exe) at 180 the l og AR = 0.9 and the l og W co = 0.4. ( 4 5 ) ( 4 6 ) ( 4 7 ) ( 4 8 ) In the control algorithm, the following rules w ere set up in case the temperature was out side a reasonable control range : 1. If error>10, MV = 100. 2. If 10
PAGE 98

98 set point and continues heating the oil bath even when powered off for a long time On the other hand, the oil bath c an be rapidly co oled and thus can confuse the control algorithm by unnecessarily increasing MV. Reduc ing the oscillation is possible by further calibrating both K c and but it is very likely to depart from the goal of fast tuning temperature. Lastly, t he on off controller was tested since it is a simple yet effective method for these condition s The fully open (MV=100) and fully close d (MV= 100) control algorithm ensured that the fastest response to the set point change happened around the set point des pite a dead band of 0.5 C This was done to reduce any potential overshoot ing due to the large heat capacity of the heating system. The MV was set to 20% inside the dead band so that the temperature could be fine tuned and oscillation constrained to a sma ll range In general, the following rules were applied for the on off controller. 1. If error>0.5, MV=100 2. If 0.5>error>0, MV=20 3. If 0>error> 0.5, MV= 20 4. If 0.5>error, MV= 100 Figure 4 18 (a) shows the performance of the on off controller when increasing the set point from 160 C to 170 C With the heating mantel set at full power until the temperature reached 169.5 C an unavoidable overshoot of 1 C was observed for around 60 seconds This is similar to the P controller but comparatively better than the PI controller. The following oscillation w as expected due to the nature of the on off controller, yet in a much smaller temperature range ( 0.2 C ). D espite the smaller temperature deviation, the drawback of the on off system is reflected by its high frequency oscillation of MV (changed working state rando ml y from 1 to 9 seconds) which would accelerate the wear and the tear of the system. Nevertheless, the on off

PAGE 99

99 controller did improve the precision of the QD system as shown in Figure 4 18 (b). At constant temperature (170 C ) and flow rate (1.5 mL /min per pump), the standard deviation of the emission wavelength produced by the FSS was reduced to 0.69nm at the mean value of 581nm. Such precision i s already above the resolution of the fluorometer and is considered to be sufficient for further application Process control The size of the QDs and their PL properties can be adjusted by varying the temperature as well as changing the r esidence time. It is apparent that temperature cannot be rapidly and precisely controlled compared with tuning the pump flow rate. Meanwhile, keeping temperature inside a suitable range is important: a high temperature is preferred because of the relevant fast reaction speed yet the temperature cannot be too high to overload the pressure limitation of the system. Thus the establishment of the process control system was based on setting an appropriate temperature and tuning the flow rate ( i.e. residence time ) The following section introduce s a first order plus time delay ( FOPTD ) model structure for analyzing the emission wavelength of QD s produced by the FSS that is based on step response data. Two different tuning methods w ere used to calculate the cont rol parameters for the close d loop system with the feedback controller Those two tuning methods are the Cohen Coon and the Ziegler Nichols method. Although the reaction conditions were completely different from the silica model, as discussed in Chapter 3 the same procedure was applied when build ing the feedback control model Graphical process identification from step responses The FOPTD process is a proper assumption for the QD FSS because the typical FOPTD feature c an be observed from its step change r esponse curve: a sigmoidal

PAGE 100

10 0 response with no oscillations or inverse response s A g raphical identification procedure w as made t o identify the three parameters ( 4 9 ) Figure 4 20 shows the step responses of the FSS in different flow rate ranges. A simple step change experiment i s not enough to cover the whole working flow range because the curve of the flow rate (MV) to emission wavelength (PV) might be asymptot ical as shown in Figure 4 19 This is based on the assumption that the wavelength (QD particle size) is pr oportional to the reaction time. Therefore four step responses were made at 0.5 mL /min, 1.5 mL /min, 2.5 mL /min and 3.5 mL /min This provides coverage of most of the flow rate range as shown in Figure 4 20 A gradual increase of the step change of 0.1 mL /mi n at 0.5 mL /min and 1.5 mL /min, 0. 2 mL /min at 2.5 mL /min and 0.3 mL /min at 3.5 mL /min were done to ensure the detectable differences The Ks for e ach step change were calculated from the equation and are shown below: ( 4 10 ) ( 4 11 ) ( 4 12 ) ( 4 13 ) By simulating the power trend line, where y represents K and x is the flow rate ( Figure 4 21 a), we get ( 4 14 ) Although K can be directly calculated from the harder to determine because of the limitation of the discrete wavelength measurement s

PAGE 101

101 Table 4 1 that are determined graphically from the step is distributed around 0. 4 min and D become s negative at a high er flow rate. To explain the discrepancy in these parameters, a possible reason is that this could have result ed from the higher interval of each wavelength measurement compared to the time cost for each step change so that the real step response curves are elongat ed A lternative me thods were applied to simulat e and estimat e using the residence time. According to the definition of dead time, D should be equal to the residence time that measures from the pump, where step change occurs, to the fluorometer, where the waveleng th change is observed Therefore, the following equation holds: ( 4 15 ) where V is the absolute volume of the FSS and v is the flow rate. Considering the limitation of the measurement interval, D is restricted above 0.5min so that the tuning program wo uld n o t be improperly affected by the delayed responses. Figure 4 21 c shows the calculated D which is discontinuous at the flow rate of 1.22 mL /min and gives a 0.5min dead time. First, by denoting t with the FOPTD function becomes : ( 4 16 ) ( 4 17 ) Next, assuming that the gradient of the response curve is proportional to the residence time, we have :

PAGE 102

102 ( 4 18 ) where c is an unidentified coefficient. Since Equation 4 17 can be rewrit t e n as : ( 4 19 ) Three values of 1, 0.5 and 0.25, were chosen to optimize the coefficient c ( Figure 4 21 b), which is described in detail in the following chapter. Cohen Coon tuning method The Cohen Coon tuning method was applied with an control and programmatically convenient mathematic al relationship ( 3 14 and 3 15 ). Figure 4 22 indicates that the was the same, where the absolute value of Kc and increased with an While the increase of Kc would tun e the system faster, a larger could reduce the st ability of the system. Figure 4 23 shows the results of the feedback control on the QD FSS by the PI controller and the Cohen Coon tuning parameter as displayed in Figure 4 22 at the 0.5min interval. All experiments started with a 1.5 mL /min flow rate which produced the 650nm emission wavelength QD with a target ed 570nm emission wavelength for comparison. O vershoot s were observed in all three experiments and were antiparallel to the effect, which is contrary to observing an expected parallel and stronger overshoot for a higher Kc. The maximum flow rate gradually decreased from c=0.25 (absolute value of Kc is lowest ) t o c=1 (absolute value of Kc is highest ) Further exploration of the control algorithm provided more details about the overshoot s The discretized PI controller uses the following equation:

PAGE 103

103 ( 4 20 ) w here represents the proportion of Kc and represents the proportion of Figure 4 24 shows the variation of both parts in the tuning process which indicates that the tuning process was mainly controlled by as the Kc proportion increases the oscillation amplitude. Since the ov ershoots were contributed by the increase of from c=0.25 to c=1 resulted in a decrease of the maximum for the overshoot s Alt hough the overshoot was restrained at c=1, the oscillation became unacceptable due to the increasing K c proportion and its interactivity with a delayed effect. The best performance belongs to c=0.25, which has the best stability once the set point is reached. It takes 7min to reach the set point with three attenuated peaks. However, further decreasing the coefficient c will not speed up the tuning process because the overshoot will be further amplified by the increasing and thus the oscillation will be enlarged. Ziegler Nichols tuning method An alternative, yet robust and popular method, the Ziegler Nichols tuning method was also test ed in case it could provide a smoother tuning. The open loop transfer function was determined by plug ging values gathered at the 0.1 mL /min flow rate mark from Figure 4 21 (c=0.5) into the below equation: ( 4 21 ) The Bode Plot s of E quation 4 21 were taken by the program named FrespAsthaASS_ECH4323NoLine .exe and the log AR and l were recorded at

PAGE 104

104 phase lag 180. By applying E quation s 3 16 3 17 3 18 and 3 19 the complete series of Kc and values covering the whole flow rate were calculated as shown in Figure 4 25 Both parameters can be broken into two parts at 1.22 mL /min flow rate due to the turning point of D. Kc can be represented by the following equations: ( 4 22 ) can be represented by the following equation: ( 4 23 ) w here x is the flow rate. The Kc and were then plugged into Equation 4 20 for the tuning calculation. Figure 4 26 shows the tuning result by the Ziegler Nichol s method at three different set points. The tuning curve s were smooth but slower compared to Cohen Coon method, taking about 10 min to 14min to reach the set point The parameter generated by the Ziegler Nichols method was much higher than th at derived from the Cohen Coon me thod (1.5 3 times for and 3.8 7.5 times for ), which indicates th at the tuning process was less a ffected by However, the analysis of the tuning equation ( Figure 4 27 ) illustrate s that the tuning mainly followed the trend of Core shell QD in FSS The coating technique for CdTe /CdS QD s in the hydrothermal method relies on the controlled reaction of S 2 with Cd 2+ wh ere the S 2 ion could come from either the NAC ligand 118 129 or Na 2 S 130 131 W hile the ligand s provide the S 2 by self degradation and surface reaction s that link themselves with the QD s t hey cannot shift the wavelengths

PAGE 105

105 more towards the region of red emission due to the functionality of the ligands The ligands maintain a level of stability that is insufficient to supply an adequate source of S 2 to coat the QDs. On the other hand, Na 2 S re mains a suf ficient and direct source of S 2 ions. The only disadvantage of Na 2 S come s from the CdS reaction where t he reaction speed is so fast that a single crystal of CdS instead of CdS shell can form in the solution unless the speed at which Na 2 S is added is high ly limited A third and alternative method was developed during the course of this research to supply the S 2 ion s in order to bypass the complication of the flow system when Na 2 S is applied. This method combines the ability of precise ly control ling the residence time ( or reaction time) and the nature of sodium thiosulfate which will slowly degrade in the acidic environment. The sodium thiosulfate decomposes at p H <7, as shown below: ( 4 24 ) Wh en mixed with the Cd 2+ ion, three thiosulfate compounds can be formed according to the concentration of : and All three compounds degrade slowly under UV or acidic environment s at room temperature but the two coordination compounds have a lower photostability 132 The overall reaction can be written as follow s : ( 4 25 ) While it takes hours to form the CdS precipitate s at room temperature the reaction is accelerated dramatically w hen the temperature is increased. Combining this with the reaction time control from the FSS, the controlled CdS coating process becomes possible.

PAGE 106

106 Materials and method The raw QD solution was prepared using the same method mentioned earlier in the section describing the conversion of the batch method to the FSS method Briefly, the Te precursor and the Cd precursor were prepared as follow s : Te precursor: 125mg of TeCl 4 was dissolved in 500 mL of DI water by drop addition of 1M sodium hydroxide until the solution became clear. The solution was then sealed and bubbled by N 2 After 30 min, 250mg of NaBH 4 was added and the solution was heated to 80 C until it became colorles s again. Cd precursor: 2.292g of CdCl 2 and 2.448g of NAC were dissolved with 500 mL of DI water and the pH was adjusted to 9. The solution was ready after 30min of N 2 bubbling. Several QD raw solutions with different emission wavelength s were collected in advance. In the preliminary test by the batch method 10 mL of QD raw solution was mixed with 0.1g of sodium thiosulfate Afterwards, the pH of the solution was adjusted to 5 using 3.3wt% HCl. The resulting solution was sealed and then h eated at 90 C to observe the wavelength shift. During the heating, the samples were quickly taken from the glass vial, quenched and then measured for their fluorescence by the Hitachi F2000 spectrophotometer. In the FSS, t he reagents for the core shell QD reaction were separate d into three parts and pumped respectively : the raw QD solution which contains the unreacted Cd 2+ ion and the core QD; the diluted HCl solution; and the sodium thiosulfate solution. Following the reaction and the online fluorescence detection, the products were collected with excess NaOH to quench the reaction. Furthermore, the products were centrifuged and washed by DI water to remove the unreacted sodium thiosulfate. Precise emission

PAGE 107

107 spectrums were measured offli ne by a Horiba Nanolog UV/NIR spectrophotometer, which gives better accuracy at wavelengths above 650nm. Results and D iscussion A preliminary test was done prior to tuning the reaction in the FSS. By recycling the unreacted Cd 2+ in the raw QD solution, th e addition of any extra chemicals w as limited to diluted HCl solution and sodium thiosulfate. While the unreacted Cd 2+ ion was approximately 12mM, 0.1g of sodium thiosulfate supplied a 5:1 molar ratio to the Cd 2+ ion so that all the remaining Cd 2+ would be in the form of coordination compounds. The presence of excess sodium thiosulfate could help accelerate the reaction in order to fit the time requirement for the FSS. The control group was also induced by adjustin g the pH of the raw QD solution to 5 without adding any excess sodium thiosulfate. As shown in Table 4 2 the control group has a 2nm absolute shift from the raw solu tion during the heating, which may be caused by the pH changes 133 On the other hand, the sample with the sodium thiosulfate experience d a red shift of the emission wavelength at the speed around 2.4nm/min and finally become agglomerated due to the overreaction. The stationary emission wavelength of the c ontrol group proved that the growth of the QD s stopped after it was produced by the FSS because all the free Te 2 ions were blocked by the dissolved O 2 in the solution Therefore, the red shift of the emission wavelength only results from the degradation of sodium thiosulfate Further explorations were carried out on the FSS in order to stop the reaction at the target wavelength. Compared to the batch test, a more accelerated reaction was preferred to fit the residence time range of the FSS. Table 4 3 shows that the red shift of the emission wavelength is affected by reaction time, temperature, molar ratio and the pH. The red shift of the emission wav elength enlarged as the residence time increase d

PAGE 108

108 However, the tuning of the residence time shows the limitation of the maximum red shift at about 70nm ( Figure 4 28 ) before precipitation This is consistent with other literature involving hydrothermal synthesized core shell CdTe/Cds QD s 134 A f urther shift was limited by the small difference between the conduction bands of the CdTe core and the CdS shell (about 0.1eV) 131 On the other hand, the organometallic method g ives a maximum red shift of 120 nm for the CdTe/CdS system 135 The difference in the maximum red shift between the se two methods may be attribu ted to the increase of the CdS concentration gradient towards the surface in the hydrothermal method 136 138 The results also indicate d that the reaction temperature and the pH are two critical parameters The increase in temperature from 90 C to 170 C reduce d the reaction time from tens of minutes (batch) to second s Although high temperature s such as 160 C dramatically accelerate the reaction, it is difficult to approach the maximum red shift because the boundary residence time between the coating and the precipitation are quite blur red Conversely, the lower temperature with a much mild er rea ction provides a sufficiently large time zone for tuning without the risk of tube blockage The pH also has the same function as temperature in that it catalyz es the reaction However, it is more difficult to monitor during the process since degradation of sodium thiosulfate produces H + ion s throughout the reaction

PAGE 109

109 Table 4 1 Calculated Step change data for K D Flow rate, mL /min K min D min 0.5 90 0.447761 1.302238806 1.5 40 0.402985 0.342014925 2.5 30 0.447761 0.322238806 3.5 20 0.223881 11.5238806 Table 4 2 P reliminary batch test of coating effect by sodium thiosulfate Time, min Emission wavelength, nm (Control group) Emission wavelength, nm (Core shell group) 0 624 624 10 626 641 18 626 668 26 626 Agglomerate Table 4 3 Residence time and temperature effect on CdS coating Residence time, s Flow rate ratio Cd2+: Temperature, C Emission wavelength, nm 72 2:1:1 1:5 120 682 96 2:1:1 1:5 120 693 144 2:1:1 1:5 120 697 144 2:1:1 1:5 130 precipitation 7.5 2:1:1 1:5 150 670 7.5 2:1:1 1:5 160 681 9.375 2:1:1 1:5 160 686 10 2:1:1 1:5 160 688 15 2:1:1 1:5 160 698 30 2:1:1 1:5 160 precipitation 15 2:1:1 1:5 170 715 73 2:0.5:0.5 1:2.5 130 663 72 1:0.25:0.75 1:2.5 130 722 96 1:0.25:0.75 1:2.5 130 precipitation

PAGE 110

110 Figure 4 1 Flow system for QD synthesis. 1. piston pump; 2. oil/heating bath; 3. condenser/cooling bath; 4. back pressure regulator; 5. fluorometer; 6. sample collector; 7. data acquisition (and proposed feedback control).

PAGE 111

111 Figure 4 2 Concentration effect of the QD were adjusted to the same emission wavelength for comparison. The basic condition is [Cd2+] = 12.5mM, [Te2 ] = 0.5mM, [NAC] = 15mM. Reaction temperature was set to be 170C. 0% 10% 20% 30% 40% 50% 60% 0 1 2 3 4 5 6 7 8 9 10 12.5 10 7.5 5 2.5 Quantum yield Recation time, second [Cd 2+ ], mM Residence time QY% 0% 10% 20% 30% 40% 50% 60% 0 2 4 6 8 10 12 14 12.5 16 19.5 23 26.5 30 Qtantum yield Reaction time, second [NAC], mM RT QY%

PAGE 112

112 Figure 4 3 (a) Normalized emission spectra for QDs synthesized at different temperatures, with a constant residence time of 5.8s showing the tunability of emission wavelength (excitation 350 nm). Temperature were 115C, 125C, 135C, 145C, 155C, 165C, 175C, 185C from left to right respectively. (b) max and temperature for QDs synthesized at different temperatures 0 0.2 0.4 0.6 0.8 1 450 475 500 525 550 575 600 625 650 675 A.U. Wavelength, nm 115 125 135 145 155 165 175 185 y = 0.014x 2 2.5857x + 629.91 500 525 550 575 600 625 650 100 150 200 Temperature, C

PAGE 113

113 Figure 4 4 (a) Normalized emission spectra for QDs synthesized with different residence time at the constant temperature of 1 70 C (excitation 350 nm). (b) max and residence time for aqueous QD s synthesized with different residence times. 0 0.2 0.4 0.6 0.8 1 450 475 500 525 550 575 600 625 650 675 700 A.U. Wavelength, nm 5.88s 3.92s 2.35s 2.35s 1.96s 1.63s 1.37s 1.18s 0.98s y = 75.283ln(x) + 510.6 500 520 540 560 580 600 620 640 0 1 2 3 4 5 6 7 max, nm Residence time, second

PAGE 114

114 Figure 4 5 (a) The calculated QD average radius as the function of residence time. (b) the plot of cube of average QD radius as a function of residence time. 0 0.5 1 1.5 2 2.5 3 0 1 2 3 4 5 6 7 QD average radius, nm Residence time, second 0 5 10 15 20 25 0 1 2 3 4 5 6 7 3 Residence time, second

PAGE 115

115 Figure 4 6 Images of QDs prepared via continuous flow under room light (left) and under UV excitation (right). QDs were synthesized at 180 C by decreasing the residence time at constant interva ls to obtain emission wavelengths ranging from 530 to 730 nm.

PAGE 116

116 Figure 4 7 XRD patterns of the CdTe QD by flow synthesis at different residence time. CdS CdTe Red Yellow Green

PAGE 117

117 Figure 4 8 TEM image of QD produced under 180C with a residence time of 3.5 seconds. 5nm

PAGE 118

118 Figure 4 9 Sketch of heating system Thermal couple External stir SS tube Cooling water PTFE stand

PAGE 119

119 Figure 4 10 The temperature response of new heating system with increase set point. ( Kp= 2 0, P controller) 0 20 40 60 80 100 120 140 0 500 1000 1500 2000 2500 3000 3500 Temperature, C Time, second 125 127 129 131 133 135 0 500 1000 1500 2000 2500 3000 3500 Temperature C Time, second -80 -60 -40 -20 0 20 40 60 80 100 120 0 500 1000 1500 2000 2500 3000 3500 Manipulated Variable(MV) Time, second

PAGE 120

120 Figure 4 11 The temperature response of new heating system with increase set point. ( Kp=40, P controller) 125 130 135 140 145 150 0 200 400 600 800 1000 1200 1400 1600 1800 Temperature, C Time, second 142 143 144 145 146 147 148 0 200 400 600 800 1000 1200 1400 1600 1800 Temperature, C Time, second -100 -50 0 50 100 150 0 200 400 600 800 1000 1200 1400 1600 1800 Manipulated Variable(MV) Time, second

PAGE 121

121 Figure 4 12 The temperature response of new heating system with decrease set point. ( Kp=80, P controller) 170 175 180 185 190 195 200 205 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 Temperature, C Time(second) 199 199.5 200 200.5 201 600 800 1000 1200 1400 1600 1800 2000 2200 2400 Temperature, C Time(second) -110 -60 -10 40 90 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 Manipulated Variable(MV) Time(second)

PAGE 122

122 Figure 4 13 Step change data for heating system from MV 40 to 35 140 145 150 155 160 165 170 175 0 50 100 150 200 250 300 350 Temperature C Time, minute

PAGE 123

123 Figure 4 14 QD emission wavelength disturbed by temperature deviation 611 612 613 614 615 616 617 618 0 200 400 600 800 1000 1200 1400 1600 Emission wavelength, nm Time, second 169 169.5 170 170.5 171 0 200 400 600 800 1000 1200 1400 1600 Temperature, C Time, second

PAGE 124

124 Figure 4 15 Performance of heating system with PI controller (Kc= 3.6, =13.15, set point at 120 / 135 / 14 5) 115 117 119 121 123 125 0 10 20 30 40 50 60 70 Temperature, C Time,second -150 -100 -50 0 50 100 150 0 10 20 30 40 50 60 70 MV Time,second

PAGE 125

125 Figure 4 15 C ontinue d 115 120 125 130 135 140 145 0 20 40 60 80 100 120 140 Temperature, C Time,second 0 20 40 60 80 100 120 0 20 40 60 80 100 120 140 MV Time,second

PAGE 126

126 Figure 4 15 C ontinue d 140 142 144 146 148 150 0 10 20 30 40 50 60 70 Temperature, C Time,second -120 -100 -80 -60 -40 -20 0 0 10 20 30 40 50 60 70 MV Time,second

PAGE 127

127 Figure 4 16 Performance of heating system with PI controller (Kc= 1.8, =13.15, set point at 120) 115 117 119 121 123 125 0 20 40 60 80 100 120 140 Temperature, C Time,min -150 -100 -50 0 50 100 150 0 20 40 60 80 100 120 140 MV Time,min

PAGE 128

128 Figure 4 17 Performance of heating system with PI controller (Kc= 0.9, =13.15, set point at 130) 115 120 125 130 135 0 50 100 150 200 250 Temperature, C Time,min -150 -100 -50 0 50 100 150 0 50 100 150 200 250 MV Time,min

PAGE 129

129 Figure 4 18 (a)The performance of heating system with on off controller and (b)its effect on stabilizing the QD emission wavelength. 169 169.2 169.4 169.6 169.8 170 170.2 170.4 170.6 170.8 171 0 200 400 600 800 1000 1200 1400 Temperature, C Time, second -150 -100 -50 0 50 100 150 0 200 400 600 800 1000 1200 1400 MV Time, second

PAGE 130

130 Figure 4 18 C ontinue d. 169.7 169.8 169.9 170 170.1 170.2 0 500 1000 1500 2000 2500 3000 Temperature, C Time,s -40 -20 0 20 40 60 80 100 120 0 500 1000 1500 2000 2500 3000 MV Time,s 578 579 580 581 582 583 584 500 1000 1500 2000 2500 3000 3500 Emission peak, nm Time,second

PAGE 131

131 Figure 4 19 The potential relationship between flow rate, reaction time and emission wavelength Reaction time, second Flow rate, ml/min Emission wavelength, nm

PAGE 132

132 Figure 4 20 Step change from 0.5 to 0.6 mL /min, 1.5 to 1.6 mL /min, 2.5 to 2.7 mL /min, 3.5 to 3.8 mL /min 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 0 1 2 3 4 5 6 7 8 Emission wavelength, nm Time, minute 576 577 578 579 580 581 582 583 0 1 2 3 4 5 6 7 8 Wavelength, nm Time, minute

PAGE 133

133 Figure 4 20 C ontinue d 542 543 544 545 546 547 548 549 550 551 552 553 554 0 1 2 3 4 5 6 7 8 Wavelength, nm Time, minute 518 519 520 521 522 523 524 525 526 0 1 2 3 4 5 6 Wavelength, nm Time, minute

PAGE 134

134 Figure 4 21 -140 -120 -100 -80 -60 -40 -20 0 0 1 2 3 4 5 6 7 K flow rate,ml/min y = 54.404x 0.744 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7 Tau Flow rate, ml/min Tau1 tau0.5 Tau0.25 0 0.5 1 1.5 2 2.5 0 1 2 3 4 5 6 7 D Flow rate, ml/min

PAGE 135

135 Figure 4 22 C C method Kc and -0.015 -0.01 -0.005 0 0 1 2 3 4 5 6 7 Kc Flow rate,ml/min COHEN COON method Kc tau1 Tau0.5 tau0.25 0 0.5 1 1.5 2 0 1 2 3 4 5 6 7 Ti Flow rate,ml/min COHEN COON method Ti tau1 tau0.5 tau0.25

PAGE 136

136 Figure 4 23 C C method tuning (Feedback control for Stainless steel tubing with c=(a)1, (b) 0.5, (c) 0.25) 540 550 560 570 580 590 600 610 620 630 640 650 660 670 0 2 4 6 8 10 12 wavelength, nm Time, min 0 1 2 3 4 5 6 0 2 4 6 8 10 12 Total flow rate, ml/min Time, min

PAGE 137

137 Figure 4 23 C ontinue d. 540 560 580 600 620 640 660 680 0 2 4 6 8 10 12 14 16 wavelength, nm Time, min 0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 16 Total flow rate, ml/min Time, min

PAGE 138

138 Figure 4 23 C ontinue d. 540 560 580 600 620 640 660 680 0 2 4 6 8 10 12 14 wavelength, nm Time, min 0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 Total flow rate, ml/min Time, min

PAGE 139

139 Figure 4 24 T he weight of Kc and in tuning program for c=0.25(a), 0.5(b) and 1 .0 (c) -300 -250 -200 -150 -100 -50 0 50 100 150 0 2 4 6 8 10 12 14 Time, min (e-et-1) dt/taui*et e(t)-e(t-1)+dt/Taui*e(t) -200 -150 -100 -50 0 50 100 0 2 4 6 8 10 12 14 Time, min e(t)-e(t-1) dt/Taui*e(t) e(t)-e(t-1)+dt/Taui*e(t)

PAGE 140

140 Figure 4 24 C ontinue d -120 -100 -80 -60 -40 -20 0 20 40 60 80 0 2 4 6 8 10 Time, minute e(t)-e(t-1) dt/Taui*e(t) e(t)-e(t-1)+dt/Taui*e(t)

PAGE 141

141 Figure 4 25 Z N method Kc y = 0.0016x 2 0.011x 0.0012 y = 8E 05x 2 0.0047x 0.0066 -0.04 -0.035 -0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0 0 1 2 3 4 5 6 7 Kc Flow rate,ml/min y = 1.2873x 1 y = 0.0081x 2 0.0894x + 1.1336 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5 6 7 I flow rate,ml/min

PAGE 142

142 Figure 4 26 Z N method tuning (530nm(a), 580nm(b), 637nm(c) 520 530 540 550 560 570 580 590 0 2 4 6 8 10 12 14 Wavelength, nm time, min 0.6 1.6 2.6 3.6 4.6 5.6 6.6 7.6 8.6 0 2 4 6 8 10 12 14 Total flow rate,ml/min time, min

PAGE 143

143 Figure 4 26 C ontinue d 570 580 590 600 610 620 630 640 650 0 2 4 6 8 10 12 14 16 18 Wavelength, nm time, min 0.6 1.1 1.6 2.1 2.6 3.1 0 2 4 6 8 10 12 14 16 18 Total flow rate,ml/min time, min

PAGE 144

144 Figure 4 26 C ontinue d 580 590 600 610 620 630 640 650 0 5 10 15 20 25 Wavelength, nm time, min 0.6 0.8 1 1.2 1.4 1.6 1.8 0 5 10 15 20 25 Total flow rate,ml/min time, min

PAGE 145

145 Figure 4 27 T he weight of Kc and in tuning program for set point=530 -30 -25 -20 -15 -10 -5 0 5 10 15 0 2 4 6 8 10 12 14 Time, minute (e-et-1) dt/Taui*e(t) e(t)-e(t-1)+dt/Taui*e(t)

PAGE 146

146 Figure 4 28 Red shift of emission wavelength from the coating of CdS shell at 120 C 0 0.2 0.4 0.6 0.8 1 1.2 500 550 600 650 700 750 800 A.U Wavelength, nm FR=0.75 FR=1.52 FR=2 Raw QD

PAGE 147

147 5 CHAPTER 5 CONCLUSION AND FU TURE WORK SUMMARY In this research we demonstrate d the application of the flow synthesis system and the process control in nano particle manufacturing by modeling the Stober silica particle synthesis process. The FSS was built with multiple improvements to solve the unique problems that are associated with nano particle production with the Stober process In additio n, b oth size map based and the feed back control s were investigated for the Stober process for its improvement Each of these controls demonstrated both advantages and disadvantages during the model establishment and the adjustment of process controls. The size map based control works well for the Stober model given a known range of tuning parameters while the feedback control requires less information from the process but gives better suitability because a similar output is achieved. C ase studies about dye doped silica particles and QDs indicate the potential of the FSS in the nano particle field. The addition of dye molecule s in the precursor molecule TEOS enables the synthesis of silica particles with adjustable size s using multiple dyes which span the entire visible light region. The CdTe QDs synthesized by the hydrothermal method were successfully converted into the FSS The QDs synthesized by the FSS emitted wavelengths ranging in the visible and NIR regions, from roughly 500 800nm with an offset of 2 nm and a 40% 6 0% quantum yield The advantage of the FSS allows for precise and quantitative studies on the effect s of reagent concentration, reaction temperature and residence time. The results indicate d that high Cd 2+ concentration and low ligand c oncentration were p referred for preparing high quality QDs. The FSS also compressed the reaction time from hours to

PAGE 148

148 second s with no adverse effect on the QD s and the PL properties. Two feedback tuning methods were applied on the residence time to control the emission wavelength of the QDs which are able to reach the set point at around 10min. A specialized CdS coating method for CdTe was developed for the FSS by controlling the degradation speed of Na 2 S This method was able to create a CdS shell that shifted the emission wavelength towards the red regions of the original QDs up to a maximum of 70nm. The concentration of Na 2 S temperature, and residence time can be used as control parameters. CONCLUSION This study confirms the fe asibility of applying the flow synthesis system in nanoparticle manufacturing. The following tip s would be useful to convert a batch synthesis into the FSS. 1. The reaction should be studied by batch before any attempts in FSS so that agglomerations and large particles can be avoided in the selected reaction condition range s 2. After the determination of reaction conditions, the reagents should be separated into groups where they can keep unreactive. 3. The FSS is a good tool to optimize the reaction conditio ns. The required reaction temperature and reaction time may vary a lot due to the high thermal transfer rate of the tube reactor. 4. The online/inline detector s may be transformed from benchtop instrument s by automatic sampling. F UTURE WORK In this research, the promising application of the FSS in the nano particle manufacturing field was reported. The research can be further extended in various ways. Possible directions that directly related to this study are reported as follows:

PAGE 149

149 Stober silica process The addition of a passive mix er or micro scal e mix er may improve the st ability of the products and also shorten the residence time by mixing the reagents faster and giving better uniformity. Also, t he temperature can be considered as a control parameter in order to accelerate the reaction time for the Stober silica process, which should reduce the tube blockage by intensif ying the Brownian motion The technique barrier is the lack of an instrument that can do sonication and precise temperature control sim ultaneously. CdTe QDs More studies is necessary for the synthesis of core/shell structure CdTe/CdS quantum dots. The effects from different control parameters can be compared and optimized for the yield and long term system reliability. A more complex system can be developed for the core/shell QDs. The quantum yield can be online detected by the combination of absorption and emission spectrums. The synthesis of core QDs and the coating process can be integrated into one flow system. The r elationship between quantum yield and shell thickness can be stud ied so that a control algorithm may be designed to produce the QDs with desired emission spectrum and high est quantum yield thickness Besides the future steps that directly related to the above two synthesis, the FSS is also promising in many other colloidal processes taking advantage of fast tuning parameters and hydrothermal ability

PAGE 150

150 LIST OF REFERENCES 1 Ruivo, A. et al. Gold nanoparticles in ancient and contemporary ruby glass. J Cult Herit 9 E134 E137 (2008). 2 Rawson, P. S. Ceramics (University of Pennsylvania Press, 1984). 3 Faraday, M. The b akerian l ecture: e xperimental r elations of g old (and o ther m etals) to l ight Phil Trans R Soc London 147 145 181 (1857). 4 Wellesley, M. N anotechnology : a r ealistic market assessment (BCC Research LLC, 2010). 5 Pitkethly, M. J. Nanomaterials the driving force. Materials Today 7 20 29 (2004). 6 F illon B. Mukish, P. Witomski A. Duret D. Lowering b arriers for n ano technology c ommercialisation. ( NanoCom, 2011). 7 Gomez de Pedro, S., Puyol, M. & Alonso Chamarro, J. Continuous flow synthesis of nanoparticles using ceramic microfluidic devices. Nanotechnology 21 (2010) 8 Wiles, C., Watts, P. & Haswell, S. J. The use of solid supported reagents for the multi step synthesis of analytically pure alpha,beta unsaturated compounds in miniaturized flow reactors. Lab Chip 7 322 330 (2007). 9 Jas, G. & Kirschning, A. Continuou s flow techniques in organic synthesis. Chem Eur J 9 5708 5723 (2003). 10 Kirschning, A. Chemistry in flow systems. Beilstein J Org Chem 5 (2009). 11 Wiersma, D. S. & Lagendijk, A. Light diffusion with gain and random lasers. Phys Rev E 54 4256 4265 (199 6). 12 Nagamura, T. & Shimizu, H. Organic dye doped silica films by the liquid phase deposition method as potential new hybrid materials for photonics. Mol Cryst Liq Crys A 267 169 174 (1995). 13 Bentivegna, F. et al. Reverse s aturable a bsorption in s olid x erogel m atrices. Appl Phys Lett 62 1721 1723 (1993). 14 Zhao, X. J., Tapec Dytioco, R., Wang, K. M. & Tan, W. H. Collection of trace amounts of DNA/mRNA molecules using genomagnetic nanocapturers. Anal Chem 75 3476 3483 (2003). 15 Sanderson, K. Qua ntum dots go large. Nature 459 760 761 (2009). 16 Kim, T. H. et al. Full colour quantum dot displays fabricated by transfer printing. Nat Photonics 5 176 182 (2011).

PAGE 151

151 17 Lee, Y. I., Youn, J. H., Ryu, M. S., Kim, J. & Jang, J. CdSe quantum dot cathode buff er for inverted organic bulk hetero junction solar cells. Org Electron 13 1302 1307 (2012). 18 Leyman, R., Nikitichev, D. I., Bazieva, N. & Rafailov, E. U. Multimodal spectral control of a quantum dot diode laser for THz difference frequency generation. A ppl Phys Lett 99 171107 (2011). 19 Jun, B. H. et al. Ultrasensitive, b iocompatible, q uantum d ot e mbedded s ilica n anoparticles for b ioimaging. Adv Funct Mater 22 1843 1849 (2012). 20 Arora, A., Simone, G., Salieb Beugelaar, G. B., Kim, J. T. & Manz, A. Latest d evelopments in m icro t otal a nalysis s ystems. Anal Chem 82 4830 4847 (2010). 21 Edel, J. B., Fortt, R., deMello, J. C. & deMello, A. J. Microfluidic routes to the controlled production of nanoparticles. Chem Commun 1136 1137 (2002). 22 Lagzi, I., Soh, S., Wesson, P. J., Browne, K. P. & Grzybowski, B. A. Maze s olving by c hemotactic d roplets. J Am Chem Soc 132 1198 1199 (2010). 23 Peng, X. Y. et al. A micro surface tension alveolus (MISTA) in a glass microchip. Lab Chip 9 3251 3254 (2009 ). 24 Dijkink, R. & Ohl, C. D. Laser induced cavitation based micropump. Lab Chip 8 1676 1681 (2008). 25 Richter, A., Klatt, S., Paschew, G. & Klenke, C. Micropumps operated by swelling and shrinking of temperature sensitive hydrogels. Lab Chip 9 613 618 (2009). 26 Tabeling, P. Introduction to Microfluidics (Oxford University Press, 2005). 27 Kirby, B. Micro and nanoscale fluid mechanics: transport in microfluidic devices (Cambridge University Press, 2010). 28 Mair, D. A., Schwei, T. R., Dinio, T. S., Svec, F. & Frechet, J. M. J. Use of photopatterned porous polymer monoliths as passive micromixers to enhance mixing efficiency for on chip labeling reactions. Lab Chip 9 877 883 (2009). 29 Abonnenc, M., Dayon, L., Perruche, B., Lion, N. & Girault, H. H. Electrospray micromixer chip for on line derivatization and kinetic studies. Anal Chem 80 3372 3378 (2008). 30 Asgar, A., Bhagat, S. & Papautsky, I. Enhancing particle dispersion in a passive planar micromixer using rectangular obstacles. J Micromech Micr oeng 18 (2008). 31 Egawa, T., Durand, J. L., Hayden, E. Y., Rousseau, D. L. & Yeh, S. R. Design and e valuation of a p assive a lcove Based m icrofluidic m ixer. Anal Chem 81 1622 1627 (2009).

PAGE 152

152 32 Mei, Q., Xia, Z., Xu, F., Soper, S. A. & Fan, Z. H. Fabrication of microfluidic reactors and mixing studies for luciferase detection. Anal Chem 80 6045 6050 (2008). 33 Kane, A. S. et al. Microfluidic Mixers for the i nvestigation of r apid p rotein f olding k inetics u sing s ynchrotron r adiation c irc ular d ichroism s pectroscopy. Anal Chem 80 9534 9541 (2008). 34 Park, S. G., Lee, S. K., Moon, J. H. & Yang, S. M. Holographic fabrication of three dimensional nanostructures for microfluidic passive mixing. Lab Chip 9 3144 3150 (2009). 35 Park, J. M., Se o, K. D. & Kwon, T. H. A chaotic micromixer using obstruction pairs. J Micromech Microeng 20 (2010). 36 Ahmed, D., Mao, X. L., Shi, J. J., Juluri, B. K. & Huang, T. J. A millisecond micromixer via single bubble based acoustic streaming. Lab Chip 9 2738 27 41 (2009). 37 Ng, W. Y., Goh, S., Lam, Y. C., Yang, C. & Rodriguez, I. DC biased AC electroosmotic and AC electrothermal flow mixing in microchannels. Lab Chip 9 802 809 (2009). 38 Wu, Z. M. & Li, D. Q. Micromixing using induced charge electrokinetic flow Electrochim Acta 53 5827 5835 (2008). 39 Harnett, C. K., Templeton, J., Dunphy Guzman, K. A., Senousy, Y. M. & Kanouff, M. P. Model based design of a microfluidic mixer driven by induced charge electroosmosis. Lab Chip 8 565 572 (2008). 40 Chan, E. M., Mathies, R. A. & Alivisatos, A. P. Size controlled growth of CdSe nanocrystals in microfluidic reactors. Nano Lett 3 199 201 (2003). 41 Bilitewski, U., Genrich, M., Kadow, S. & Mersal, G. Biochemical analysis with microfluidic systems. Anal Bioanal Chem 377 556 569 (2003). 42 Ho, Y. P., Chen, H. H., Leong, K. W. & Wang, T. H. The convergence of quantum dot mediated fluorescence resonance energy transfer and microfluidics for monitoring DNA polyplex self assembly in real time. Nanotechnology 20 (2009). 43 Becker, H. & Locascio, L. E. Polymer microfluidic devices. Talanta 56 267 287 (2002). 44 Abate, A. R., Lee, D., Do, T., Holtze, C. & Weitz, D. A. Glass coating for PDMS microfluidic channels by sol gel methods. Lab Chip 8 516 518 (2008).

PAGE 153

153 45 Sun, X. F., Liu, J. K. & Lee, M. L. Surface modification of polymer microfluidic devices using in channel atom transfer radical polymerization. Electrophoresis 29 2760 2767 (2008). 46 Radha, B. & Kulkarni, G. U. A m odified m icromolding m ethod for s ub 100 nm d irect p atterning of Pd n anowires. Small 5 2271 2275 (2009). 47 Wyss, H. M., Blair, D. L., Morris, J. F., Stone, H. A. & Weitz, D. A. Mechanism for clogging of microchannels. Phys Rev E 74 (2006). 48 Metz, S., Trautmann, C., Bertsch, A. & Renaud, P. Po lyimide microfluidic devices with integrated nanoporous filtration areas manufactured by micromachining and ion track technology. J Micromech Microeng 14 324 331 (2004). 49 Marmiroli, B. et al. Free jet micromixer to study fast chemical reactions by small angle X ray scattering. Lab Chip 9 2063 2069 (2009). 50 Amarie, D., Alileche, A., Dragnea, B. & Glazier, J. A. Microfluidic d evices i ntegrating m icrocavity s urface p lasmon r esonance s ensors: g lucose o xidase b inding a ctivity d etection. Anal Chem 82 343 352 (2010). 51 Staples, G. O. et al. Improved h ydrophilic i nteraction c hromatography LC/MS of h eparinoids u sing a c hip with p ostcolumn m akeup f low. Anal Chem 82 516 522 (2010). 52 Carter, R. M., Yan, Y. & Lee, P. On line nonintrusive measurement of pa rticle size distribution through digital imaging. Ieee T Instrum Meas 55 2034 2038 (2006). 53 Zhang, J. Q. & Yan, Y. On line continuous measurement of particle size using electrostatic sensors. Powder Technol 135 164 168 (2003). 54 Carter, R. M., Yan, Y. & Cameron, S. D. On line measurement of particle size distribution and mass flow rate of particles in a pneumatic suspension using combined imaging and electrostatic sensors. Flow Meas Instrum 16 309 314 (2005). 55 Krabicka, J. & Yan, Y. Finite e lement m odeling of e lectrostatic s ensors for the f low m easurement of p articles in p neumatic p ipelines. Ieee T Instrum Meas 58 2730 2736 (2009). 56 Stober, W., Fink, A. & Bohn, E. Controlled g rowth of m onodisperse s ilica s pheres in m icron s ize r ange. J Colloid Interf Sci 26 62 67 (1968). 57 Wu, C. L. et al. Hybrid silica nanocrystal organic dye superstructures as post encoding fluorescent probes. Angew Chem Int Edit 46 5393 5396 (2007). 58 Costa, C. A. R., Leite, C. A. P. & Galembeck, F. ESI TEM Imaging of sur factants and ions sorbed in Stober silica nanoparticles. Langmuir 22 7159 7166 (2006).

PAGE 154

154 59 Yao, K. X. & Zeng, H. C. Simultaneous c hemical m odification and s tructural t ransformation of s tober s ilica s pheres for i ntegration of n anocatalysts. Chem Mater 24 1 40 148 (2012). 60 Haibach, K., Menner, A., Powell, R. & Bismarck, A. Tailoring mechanical properties of highly porous polymer foams: s ilica particle reinforced polymer foams via emulsion templating. Polymer 47 4513 4519 (2006). 61 Brinker, C. J., Keefer, K. D., Schaefer, D. W. & Ashley, C. S. Sol Gel t ransition in s imple s ilicates. J Non Cryst Solids 48 47 64 (1982). 62 Bailey, J. K. & Mecartney, M. L. Formation of c olloidal s ilica p articles from a lkoxides. Colloid Surface 63 151 161 (1992). 63 Lee, K., Look, J. L., Harris, M. T. & McCormick, A. V. Assessing extreme models of the s tober synthesis using transients under a range of initial composition. J Colloid Interf Sci 194 78 88 (1997). 64 Matsoukas, T. & Gulari, E. Dynamics of g rowth of s ilic a p articles from a mmonia c atalyzed h ydrolysis of t etra e thyl o rthosilicate. J Colloid Interf Sci 124 252 261 (1988). 65 Matsoukas, T. & Gulari, E. Monomer a ddition g rowth with a s low i nitiation s tep a g rowth m odel for s ilica p articles from a lkoxides. J Colloid Interf Sci 132 13 21 (1989). 66 Matsoukas, T. & Gulari, E. Self s harpening d istributions r evisited p olydispersity in g rowth by m onomer a ddition. J Colloid Interf Sci 145 557 562 (1991). 67 Bogush, G. H. & Zukoski, C. F. Studies of the k inetics of the p recipitation of u niform s ilica p articles through the h ydrolysis and c ondensation of s ilicon a lkoxides. J Colloid Interf Sci 142 1 18 (1991). 68 Bogush, G. H. & Zukoski, C. F. Uniform s ilica p article p recipitation an a ggregative g rowth Mo del. J Colloid Interf Sci 142 19 34 (1991). 69 Lee, K. T., Sathyagal, A. N. & McCormick, A. V. A closer look at an aggregation model of the Stober process. Colloid Surface A 144 115 125 (1998). 70 Harris, M. T., Brunson, R. R. & Byers, C. H. The b ase c atalyzed h ydrolysis and c ondensation r eactions of d ilute and c oncentrated T EOS s olutions. J Non Cryst Solids 121 397 403 (1990). 71 Vanblaaderen, A., Vangeest, J. & Vrij, A. Monodisperse c olloidal s ilica s pheres from t etraalkoxysilanes p article f ormation and g rowth m echanism. J Colloid Interf Sci 154 481 501 (1992).

PAGE 155

155 72 Ferguson, C. R. R., T.A. Synthesis of narrow size distribution SiO2 powders in a continuous stirred tank reactor. Ceramic Powder Procesing Science, Proceedings of the Second Inter national conference 93 99 (1989). 73 Her, Y. S., Lee, S. H. & Matijevic, E. Continuous precipitation of monodispersed colloidal particles .2. SiO2, Al(OH)(3), and BaTiO3. J Mater Res 11 156 161 (1996). 74 Giesche, H. Synthesis of m onodispersed s ilica p ow ders. 2 Controlled g rowth r eaction and c ontinuous p roduction p rocess. J Eur Ceram Soc 14 205 214 (1994). 75 Ogihara, T. et al. Continuous d ynthesis of m onodispersed s ilica p articles u sing c ouette t aylor v ortex f low. Nippon Seram Kyo Gak 103 151 154 (1995). 76 Vanblaaderen, A. & Vrij, A. Synthesis and c haracterization of c olloidal d ispersions of f luorescent, m onodisperse s ilica s pheres. Langmuir 8 2921 2931 (1992). 77 Santra, S., Wang, K. M., Tapec, R. & Tan, W. H. Development of novel dye doped sili ca nanoparticles for biomarker application. J Biomed Opt 6 160 166 (2001). 78 Santra, S., Zhang, P., Wang, K. M., Tapec, R. & Tan, W. H. Conjugation of biomolecules with luminophore doped silica nanoparticles for photostable biomarkers. Anal Chem 73 4988 4993 (2001). 79 Bagwe, R. P., Yang, C. Y., Hilliard, L. R. & Tan, W. H. Optimization of dye doped silica nanoparticles prepared using a reverse microemulsion method. Langmuir 20 8336 8342 (2004). 80 Zhao, X. J., Bagwe, R. P. & Tan, W. H. Development of o rganic dye doped silica nanoparticles in a reverse microemulsion. Adv Mater 16 173 174 (2004). 81 Santra, S. et al. Fluorescence lifetime measurements to determine the core shell nanostructure of FITC doped silica nanoparticles: An optical approach to eva luate nanoparticle photostability. J Lumin 117 75 82 (2006). 82 Santra, S., Dutta, D. & Moudgil, B. M. Functional dye doped silica nanoparticles for bioimaging, diagnostics and therapeutics. Food Bioprod Process 83 136 140 (2005). 83 Ow, H. et al. Bright and stable core shell fluorescent silica nanoparticles. Nano Lett 5 113 117 (2005). 84 Zhou, X. C. & Zhou, J. Z. Improving the signal sensitivity and photostability of DNA hybridizations on microarrays by using dye doped core shell silica nanopart icles. Anal Chem 76 5302 5312 (2004).

PAGE 156

156 85 Wang, L., Yang, C. Y. & Tan, W. H. Dual luminophore doped silica nanoparticles for multiplexed signaling. Nano Lett 5 37 43 (2005). 86 Murray, C. B., Norris, D. J. & Bawendi, M. G. Synthesis and c haracterization of n early m onodisperse Cd E (E = S, Se, Te) s emiconductor n anocrystallites. J Am Chem Soc 115 8706 8715 (1993). 87 Coe, S., Woo, W. K., Bawendi, M. & Bulovic, V. Electroluminescence from single monolayers of nanocrystals in molecular organic devices. Natur e 420 800 803 (2002). 88 Bruchez, M., Moronne, M., Gin, P., Weiss, S. & Alivisatos, A. P. Semiconductor nanocrystals as fluorescent biological labels. Science 281 2013 2016 (1998). 89 Gaponik, N. et al. Thiol capping of CdTe nanocrystals: An alternative to organometallic synthetic routes. J Phys Chem B 106 7177 7185 (2002). 90 Kurth, D. G., Lehmann, P. & Lesser, C. Engineering the surface chemical properties of semiconductor nanoparticles: surfactant encapsulated CdTe clusters. Chem Commun 949 950 (2000). 91 Rogach, A. L., Kornowski, A., Gao, M. Y., Eychmuller, A. & Weller, H. Synthesis and characterization of a size series of extremely small thiol stabilized CdSe nanocrystals. J Phys Chem B 103 3065 3069 (1999). 92 Guo, J., Yang, W. L. & Wang, C. C. Systematic study of the photoluminescence dependence of thiol capped CdTe nanocrystals on the reaction conditions. J Phys Chem B 109 17467 17473 (2005). 93 Zhang, H., Zhou, Z., Yang, B. & Gao, M. Y. The influence of carboxyl groups on the photoluminesc ence of mercaptocarboxylic acid stabilized CdTe nanoparticles. J Phys Chem B 107 8 13 (2003). 94 Williams, J. V., Kotov, N. A. & Savage, P. E. A Rapid h ot i njection m ethod for the i mproved h ydrothermal s ynthesis of CdSe n anoparticles. Ind Eng Chem Res 48 4316 4321 (2009). 95 Deng, Z. T. et al. Aqueous s ynthesis of z inc b lende CdTe/CdS m agic c ore/ t hick s hell t etrahedral s haped n anocrystals with e mission t unable to n ear i nfrared. J Am Chem Soc 132 5592 5593 (2010). 96 Liu, Y. F. & Yu, J. S. In situ synthesis of highly luminescent glutathione capped CdTe/ZnS quantum dots with biocompatibility. J Colloid Interf Sci 351 1 9 (2010). 97 Kim, Y. T., Il Kim, J., Lee, J. K. & Chae, H. K. Post synthesis e mission c ontrol of CdSe ZnS q uantum d ots by s urface o xidation through i n s itu g eneration of o xidant under m ild r eaction c onditions. Chem Lett 38 862 863 (2009).

PAGE 157

157 98 Seo, H. & Kim, S. W. In situ synthesis of CdTe/CdSe core shell quantum dots. Chem Mater 19 2715 2717 (2007). 99 Wang, J. & Ha n, H. Y. Hydrothermal synthesis of high quality type II CdTe/CdSe quantum dots with near infrared fluorescence. J Colloid Interf Sci 351 83 87 (2010). 100 Nightingale, A. M. & de Mello, J. C. Microscale synthesis of quantum dots. J Mater Chem 20 8454 846 3 (2010). 101 Nose, K. et al. Chemical role of amines in the colloidal synthesis of CdSe quantum dots and their luminescence properties. J Lumin 126 21 26 (2007). 102 Yang, H. W., Fan, N. N., Luan, W. L. & Tu, S. T. Synthesis of m onodisperse n anocrystals via m icroreaction: o pen to a ir s ynthesis with o leylamine as a c oligand. Nanoscale Res Lett 4 344 352 (2009). 103 Yen, B. K. H., Gunther, A., Schmidt, M. A., Jensen, K. F. & Bawendi, M. G. A microfabricated gas liquid segmented flow reactor for high temperature synthesis: The case of CdSe quantum dots. Angew Chem Int Edit 44 5447 5451 (2005). 104 Hung, L. H. et al. Alternating droplet generation and controlled dynamic drople t fusion in microfluidic device for CdS nanoparticle synthesis. Lab Chip 6 174 178 (2006). 105 Shestopalov, I., Tice, J. D. & Ismagilov, R. F. Multi step synthesis of nanoparticles performed on millisecond time scale in a microfluidic droplet based system Lab Chip 4 316 321 (2004). 106 Nightingale, A. M. & de Mello, J. C. Controlled s ynthesis of III V q uantum d ots in m icrofluidic r eactors. Chemphyschem 10 2612 2614 (2009). 107 Yang, H. W. et al. Continuous Synthesis of f ull c olor e mitting c ore/ s hell q ua ntum d ots via m icroreaction. Cryst Growth Des 9 4807 4813 (2009). 108 Giesche, H. Synthesis of m onodispersed s ilica p owders 1. p article p roperties and r eaction k inetics. J Eur Ceram Soc 14 189 204 (1994). 109 Silva, G. J., Datta, A. & Bhattacharyya, S. P. PI stabilization of first order systems with time delay. Automatica 37 2025 2031 (2001). 110 Guo, X., Wang, C. F., Fang, Y. A., Chen, L. & Chen, S. Fast synthesis of versatile nanocrystal embedded hydrogels toward the sensing of heav y metal ions and organoamines. J Mater Chem 21 1124 1129 (2011).

PAGE 158

158 111 Khatei, J. & Rao, K. S. R. K. Hydrothermal synthesis of CdTe QDs: t heir luminescence quenching in the presence of bio molecules and observation of bistable memory effect in CdTe QD/PED OT:PSS heterostructure. Mater Chem Phys 130 159 164 (2011). 112 Zhang, H. et al. L Cysteine capped CdTe CdS core shell quantum dots: preparation, characterization and immuno labeling of HeLa cells. Luminescence 26 86 92 (2011). 113 Wang, C. C., Guo, J. & Yang, W. L. Systematic study of the photoluminescence dependence of thiol capped CdTe nanocrystals on the reaction conditions. J Phys Chem B 109 17467 17473 (2005). 114 Jalilehvand, F., Amini, Z., Parmar, K. & Kang, E. Y. Cadmium(II) N acetylcysteine com plex formation in aqueous solution. Dalton T 40 12771 12778 (2011). 115 Zhao, D., He, Z. K., Chan, W. H. & Choi, M. M. F. Synthesis and c haracterization of h igh q uality w ater s oluble n ear i nfrared e mitting CdTe/CdS q uantum d ots c apped by N Acetyl L cysteine v ia h ydrothermal m ethod. J Phys Chem C 113 1293 1300 (2009). 116 Zou, L. et al. Ultrafast synthesis of highly luminescent green to near infrared emitting CdTe nanocrystals in aqueous phase. J Mater Chem 18 2807 2815 (2008). 117 Talapin, D. V. et al. Dynamic distribution of growth rates within the ensembles of colloidal II VI and III V semiconductor nanocrystals as a factor governing their photoluminescence efficiency. J Am Chem Soc 124 5782 5790 (2002). 118 Bao, H. B., Gong, Y. J., Li, Z. & Gao, M. Y. Enhancement effect of illumination on the photoluminescence of water soluble CdTe nanocrystals: Toward highly fluorescent CdTe/CdS core shell structure. Chem Mater 16 3853 3859 (2004). 119 Wang, P. N. et al. Time dependent phot oluminescence blue shift of the quantum dots in living cells: e ffect of oxidation by singlet oxygen. J Am Chem Soc 128 13396 13401 (2006). 120 Borchert, H. et al. Relations between the photoluminescence efficiency of CdTe nanocrystals and their surface properties revealed by synchrotron XPS. J Phys Chem B 107 9662 9668 (2003). 121 Peng, X. G. & Yu, W. W. Formation of high quality CdS and other II VI semiconductor nanocrystals in noncoordinating solvents: Tunable reactivity of monomers. Angew Chem Int Edit 41 2368 2371 (2002). 122 Li, H. B. et al. Ultrafast synthesis of water soluble nanocrystals by the chemical aerosol flow method. Crystengcomm 11 1231 1234 (2009 ).

PAGE 159

159 123 Li, M. Y. et al. Hydrothermal synthesis of highly luminescent CdTe quantum dots by adjusting precursors' concentration and their conjunction with BSA as biological fluorescent probes. Talanta 72 89 94 (2007). 124 Rajh, T., Micic, O. I. & Nozik, A. J. Synthesis and c haracterization of s urface m odified c olloidal C d T e q uantum d ots. J Phys Chem Us 97 11999 12003 (1993). 125 Baskoutas, S. & Terzis, A. F. Size dependent band gap of colloidal quantum dots. J Appl Phys 99 (2006). 126 Zhang, H. et al. Hydrothermal synthesis for high quality CdTe nanocrystals. Adv Mater 15 1712 1713 (2003). 127 Rogach, A. L. Nanocrystalline CdTe and CdTe(S) particles: wet chemical preparation, size dependent optical properties and perspectives of optoelectronic applica tions. Mat Sci Eng B Solid 69 435 440 (2000). 128 Brus, L. E. Electron Electron and e lectron h ole i nteractions in s mall s emi conductor c rystallites the s ize d ependence of the l owest e xcited e lectronic s tate. J Chem Phys 80 4403 4409 (1984). 129 Peng, H., Zhang, L. J., Soeller, C. & Travas Sejdic, J. Preparation of water soluble CdTe/CdS core/shell quantum dots with enhanced photostability. J Lumin 127 721 726 (2007). 130 He, Y. et al. Microwave synthesis of water dispersed CdTe/CdS/ZnS core shel l shell quantum dots with excellent photostability and biocompatibility. Adv Mater 20 3416 3421 (2008). 131 Zeng, Q. H. et al. Synthesis and optical properties of type II CdTe/CdS core/shell quantum dots in aqueous solution via successive ion layer adsor ption and reaction. J Phys Chem C 112 8587 8593 (2008). 132 Egorov, N. B. et al. Products of photolysis of cadmium thiosulfate aqueous solutions. High Energ Chem 42 119 122 (2008). 133 Wolcott, A. et al. Silica coated CdTe quantum dots functionalized wit h thiols for bioconjugation to IgG proteins. J Phys Chem B 110 5779 5789 (2006). 134 Fei, X. N., Jia, G. Z. & Wang, J. Synthesis and s tokes s hift of w ater s oluble Cd T e/Cds c ore s hell s tructrue q uantum d ots. Chalcogenide Lett 7 83 87 (2010). 135 Wang, J., Long, Y. T., Zhang, Y. L., Zhong, X. H. & Zhu, L. Y. Preparation of h ighly l uminescent CdTe/CdS c ore/ s hell q uantum d ots. Chemphyschem 10 680 685 (2009).

PAGE 160

160 136 He, Y. et al. Synthesis of CdTe nanocrystals through program process of microwave irradiation. J Phys Chem B 110 13352 13356 (2006). 137 Wang, C. L. et al. Sodium c itrate a ssisted s ynthesis of a queous CdTe n ano crystals: g iving n ew i nsight into the e ffect of l igand s hell. J Phys Chem C 113 827 833 (2009). 138 Gu, Z. Y., Zou, L., Fang, Z., Zhu, W. H. & Zhong, X. H. One pot synthesis of highly luminescent CdTe/CdS core/shell nanocrystals in aqueous phase. Nanotechnology 19 135604 (2008).

PAGE 161

161 BIOGRAPHICAL SKETCH Jiaqing Zhou was born in Shanghai, P.R.China in 1984. He obtained his bachelor degree in Materials Science and Engineering i n July 2006 in Tongji University. He then work ed for half a year in as an engineer in Shanghai SBS Zipper Science & Technology Co. Ltd. Later he continued his education at University of Florida beginning in 2007 an d joine He received his Ph.D. from the University of Florida in the summer of 2012.