Professional Documents
Culture Documents
PDF Quantum Dots Applications in Biology 3Rd Edition Adriana Fontes Ebook Full Chapter
PDF Quantum Dots Applications in Biology 3Rd Edition Adriana Fontes Ebook Full Chapter
PDF Quantum Dots Applications in Biology 3Rd Edition Adriana Fontes Ebook Full Chapter
https://textbookfull.com/product/quantum-systems-in-physics-
chemistry-and-biology-advances-in-concepts-and-applications-1st-
edition-alia-tadjer/
https://textbookfull.com/product/a-guide-to-experiments-in-
quantum-optics-3rd-edition-hans-a-bachor/
https://textbookfull.com/product/modern-quantum-mechanics-3rd-
edition-j-j-sakurai/
Spectroscopic Methods in Food Analysis 1st Edition
Adriana S. Franca
https://textbookfull.com/product/spectroscopic-methods-in-food-
analysis-1st-edition-adriana-s-franca/
https://textbookfull.com/product/quantum-mechanics-for-beginners-
with-applications-to-quantum-communication-and-quantum-
computing-1st-edition-m-suhail-zubairy/
https://textbookfull.com/product/emerging-trends-in-applications-
and-infrastructures-for-computational-biology-bioinformatics-and-
systems-biology-systems-and-applications-1st-edition-arabnia/
https://textbookfull.com/product/ideas-of-quantum-chemistry-
volume-1-from-quantum-physics-to-chemistry-3rd-edition-lucjan-
piela/
https://textbookfull.com/product/fundamentals-of-quantum-
mechanics-3rd-edition-james-e-house/
Volume 2135
Series Editor
John M. Walker
School of Life and Medical Sciences, University of Hertfordshire, Hatfield,
Hertfordshire, UK
Quantum Dots
Applications in Biology
3rd ed. 2020
Editors
Adriana Fontes
Biomedical Nanotechnology Group, Federal University of Pernambuco,
Recife, Pernambuco, Brazil
Beate S. Santos
Biomedical Nanotechnology Group, Federal University of Pernambuco,
Recife, Pernambuco, Brazil
The publisher, the authors, and the editors are safe to assume that the
advice and information in this book are believed to be true and accurate
at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed or implied, with respect to the
material contained herein or for any errors or omissions that may have
been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
Diogo B. Almeida
Quantum Electronics Department, Institute of Physics “Gleb Wataghin”,
University of Campinas – UNICAMP, Campinas, Sã o Paulo, Brazil
Frauke Alves
Translational Molecular Imaging, Max-Planck-Institute of Experimental
Medicine, Gö ttingen, Germany
Clinic of Haematology and Medical Oncology, University Medical Center
Gö ttingen, Gö ttingen, Germany
Nizar Ayadi
Group of Mechanism and Regulation of DNA Repair and IMPACT
Platform, UFIP UMR CNRS 6286/University of Nantes, Nantes, France
Houda Benhelli-Mokrani
Group of Mechanism and Regulation of DNA Repair and IMPACT
Platform, UFIP UMR CNRS 6286/University of Nantes, Nantes, France
Cathy Charlier
Group of Mechanism and Regulation of DNA Repair and IMPACT
Platform, UFIP UMR CNRS 6286/University of Nantes, Nantes, France
Hong-Yuan Chen
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Juan Chen
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, P. R. China
Margaret Chern
Division of Materials Science and Engineering, Boston University,
Boston, MA, USA
Renato E. de Araujo
Laboratory of Biomedical Optics and Imaging, Federal University of
Pernambuco, Recife, Brazil
André A. de Thomaz
Quantum Electronics Department, Institute of Physics “Gleb Wataghin”,
University of Campinas – UNICAMP, Campinas, Sã o Paulo, Brazil
Allison M. Dennis
Division of Materials Science and Engineering, Boston University,
Boston, MA, USA
Department of Biomedical Engineering, Boston University, Boston, MA,
USA
Emmanuel Derivery
MRC Laboratory of Molecular Biology, Cambridge, UK
Christian T. Dominguez
Physics Department, Federal University of Paraíba, Joã o Pessoa, Brazil
Andrew Edet Ekpenyong
Department of Physics, Creighton University, Omaha, NE, USA
Ali A. Ensafi
Department of Chemistry, Isfahan University of Technology, Isfahan,
Iran
Fabrice Fleury
Group of Mechanism and Regulation of DNA Repair and IMPACT
Platform, UFIP UMR CNRS 6286/University of Nantes, Nantes, France
Adriana Fontes
Biomedical Nanotechnology Group, Federal University of Pernambuco,
Recife, Pernambuco, Brazil
Chloe Grazon
Department of Chemistry, Boston University, Boston, MA, USA
CNRS, Bordeaux INP, LCPO, UMR 5629, University of Bordeaux, Pessac,
France
Travis Josephs
Neuroscience Program, Vanderbilt University, Nashville, TN, USA
Alexander Karaulov
Department of Clinical Immunology and Allergology, Sechenov First
Moscow State Medical University, Moscow, Russian Federation
Joshua C. Kays
Department of Biomedical Engineering, Boston University, Boston, MA,
USA
Nafiseh Kazemifard
Department of Chemistry, Isfahan University of Technology, Isfahan,
Iran
Karel Klepárník
Institute of Analytical Chemistry, Czech Academy of Sciences, Brno,
Czech Republic
Oleg Kovtun
Department of Chemistry, Vanderbilt University, Nashville, TN, USA
Florian Lafont
Group of Mechanism and Regulation of DNA Repair and IMPACT
Platform, UFIP UMR CNRS 6286/University of Nantes, Nantes, France
Jianjun Li
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, China
Zhaohui Li
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, China
Juanzu Liu
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, China
Melissa Massey
Department of Chemistry, University of British Columbia, Vancouver,
BC, Canada
Hong-Min Meng
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, P. R. China
Pavlína Modlitbová
Central European Institute of Technology, Brno University of
Technology, Brno, Czech Republic
Camila Aparecida P. Monteiro
Biomedical Nanotechnology Group, Federal University of Pernambuco,
Recife, Pernambuco, Brazil
Igor Nabiev
Laboratory of Nano-bioengineering, National Research Nuclear
University MEPhI (Moscow Engineering Physics Institute), Moscow,
Russian Federation
Laboratoire de Recherche en Nanosciences, LRN-EA4682, Université de
Reims Champagne-Ardenne, Reims, France
Galina Nifontova
Laboratory of Nano-bioengineering, National Research Nuclear
University MEPhI (Moscow Engineering Physics Institute), Moscow,
Russian Federation
Lingbo Qu
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, P. R. China
Fernanda Ramos-Gomes
Translational Molecular Imaging, Max-Planck-Institute of Experimental
Medicine, Gö ttingen, Germany
Kelly Rees
Department of Chemistry, University of British Columbia, Vancouver,
BC, Canada
Sandra J. Rosenthal
Department of Chemistry, Vanderbilt University, Nashville, TN, USA
Vanderbilt Institute of Chemical Biology, Vanderbilt University,
Nashville, TN, USA
Department of Pharmacology, Vanderbilt University, Nashville, TN, USA
Department of Physics and Astronomy, Vanderbilt University, Nashville,
TN, USA
Department of Chemical and Biomolecular Engineering, Vanderbilt
University, Nashville, TN, USA
Yi-Fan Ruan
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Zeinab Saberi
Department of Chemistry, Isfahan University of Technology, Isfahan,
Iran
Alexander M. Saeboe
Division of Materials Science and Engineering, Boston University,
Boston, MA, USA
Beate S. Santos
Biomedical Nanotechnology Group, Federal University of Pernambuco,
Recife, Pernambuco, Brazil
Xiao-Mei Shi
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Pavel Sokolov
Laboratory of Nano-Bioengineering, National Research Nuclear
University MEPhI (Moscow Engineering Physics Institute), Moscow,
Russian Federation
Alessandra Stangherlin
MRC Laboratory of Molecular Biology, Cambridge, UK
Alyona Sukhanova
Laboratoire de Recherche en Nanosciences, LRN-EA4682, Université de
Reims Champagne-Ardenne, Reims, France
Sindhuja Suresh
Department of Physics, Creighton University, Omaha, NE, USA
Lucas B. Thal
Department of Chemistry, Vanderbilt University, Nashville, TN, USA
Vanderbilt Institute of Chemical Biology, Vanderbilt University,
Nashville, TN, USA
Michael V. Tran
Department of Chemistry, University of British Columbia, Vancouver,
BC, Canada
Tatiana Tsoy
Laboratory of Nano-Bioengineering, National Research Nuclear
University MEPhI (Moscow Engineering Physics Institute), Moscow,
Russian Federation
Hai-Yan Wang
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Joseph L. Watson
MRC Laboratory of Molecular Biology, Cambridge, UK
Jing-Juan Xu
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Lin Zhang
College of Chemistry, Henan Joint International Research Laboratory of
Green Construction of Functional Molecules and Their Bioanalytical
Applications, Zhengzhou University, Zhengzhou, China
Wei-Wei Zhao
State Key Laboratory of Analytical Chemistry for Life Science, School of
Chemistry and Chemical Engineering, Nanjing University, Nanjing,
China
Part I
Tutorials
© Springer Science+Business Media, LLC, part of Springer Nature 2020
A. Fontes, B. S. Santos (eds.), Quantum Dots, Methods in Molecular Biology 2135
https://doi.org/10.1007/978-1-0716-0463-2_1
Abstract
In this chapter, we present some chemometric tools used in the area of experimental design and
multivariate optimization. To make the subject more understandable, a didactic example employing
colloidal aqueous synthesis of quantum dots is employed. We start with the factorial design that is
very useful in screening which factors are important to the response of interest. All statistical
calculations and interpretations of individual and interaction effects are detailed. Then, we describe
how to build and evaluate empirical models by analysis of variance (ANOVA) to explain the
behavior of the data set. Finally, the response surface methodology (RSM) is described. We expect
this chapter to be useful as a guide for those who seek to solve synthetic problems in a quicker and
more objective way, providing particularly a wider perception of the experimental factors that
dominate the responses of interest of a system under study.
Key words Chemometrics – Experimental design – Factorial design – Empirical models – ANOVA –
Response surface methodology – Multivariate optimization
1 Introduction
A critical step to obtain an experimental procedure or method is the search of how the factors or
independentvariables of the experiment may affect the desired response or dependent variable, that
is, the property under study, in order to find the best conditions for the target goal [1]. With this
aim, the univariate optimization or one-factor-at-a-time method is the most commonly used
strategy, which consists in varying a single factor per step while keeping the others constant. Once
the best condition for the factor under study is found, this “optimal value” is set and another factor
will be varied in the following step. This process is repeated until optimal conditions are obtained
for all the factors.
Let us assume that a researcher needs to evaluate the fluorescence quantum yield (QY) of
quantum dots (QDs) of a hypothetical semiconductor XY, produced in aqueous medium. Aiming to
maximize the QY, the researcher wants to vary the pH and the synthesistemperature, which will be
the factors studied. By univariate optimization of the QY, represented by the black dots in the
scheme of Fig. 1, initially it would be set a pH value pointed out in the literature as the best one (e.g.,
pH 10), while the response would be evaluated by varying the temperature by 50, 55, 60, 65, and
70 °C. After finding the temperature that leads to the highest QY, which is 60 °C in this example, this
value would be fixed to then vary the pH by 5, 7, 9, 11, and 13. In this way, pH 9 and temperature
60 °C would be set as “optimal conditions” to maximize the QY of the XY QDs under study.
Fig. 1 Comparison of the distribution of experimental points in the factor space for univariate optimization (black
dots) and multivariate optimization (cyan dots)
However, there are some drawbacks associated with the univariate optimization strategy. First,
since a single factor is studied at each stage, it will be very difficult to check for interaction between
two or more factors. In addition, a large number of runs are required in one-factor-at-a-time testing
to reach an optimal response. As can be seen, the response considered as “optimal” after ten runs in
the given example does not correspond to the maximum QY that can be found in the contour map of
Fig. 1.
In this context, the multivariate optimization has been shown as a very interesting alternative,
since it allows for varying simultaneously all the factors under study and thus assessing their
individual and interactions effects, with a certain level of statistical reliability. In multivariate
optimization, we employ experimental design techniques, which outline the experimental points in
the factor space in such a way as to provide a more systematized and representative scan of the
experimental domain [2]. For the previous example, from a 22factorial design with only the four
cyan dots shown in Fig. 1, it would be possible to define the direction of increase in the response,
represented by the yellow arrow. Therefore, multivariate optimization allows for finding the
optimal response with fewer runs compared to univariate optimization and in a systematic way.
2 23 Factorial Design
Factorial designs are the simplest to interpret , very useful in screening investigations. The
minimum number of runs is given by nk, where k is the number of factors and n is the number of
levels for each factor, which is the number of different values that we want to test for a given factor.
The simplest factorial design is the 22 factorial design, but we will start with an example of the 23
factorial design. Let us suppose that the objective of the optimization is to maximize the QY of QDs
of the hypothetical semiconductor XY and, for this, we want to evaluate the influence of pH,
temperature, and stabilizing agent on the synthesis procedure.
The stabilizing agent is a qualitative or categorical factor, as its levels are classes, while the pH
and the temperature are quantitative factors because they support numerical values as levels [1]. In
the factorial design, we assign −1 or simply the minus sign to represent the coded lower level of a
factor, which is the lower value under investigation. Likewise, the coded upper level is represented
by +1 or the plus sign, corresponding to the higher value studied for that factor. The lower levels
(−1) will be 5, A, and 50 °C for pH, stabilizing agent, and temperature, respectively, and the upper
levels (+1) will be 13, B, and 70 °C.
When it comes to qualitative factors, the assignment of the lower and upper levels to the classes
is arbitrary, as long as it is the same during the data treatment and the interpretation of the
information extracted from the design. The levels should be chosen based on a careful review of the
literature and taking into account prior knowledge about the problem addressed. It is
recommended to explore ranges of levels slightly further than the known effective range. For
example, if it is known that stabilizing agents A and B can be successfully used in the pH range from
6 to 12, then in this fictitious optimization we will investigate the pH in the range of 5 to 13, a bit
wider range than that supposedly effective found in literature.
To investigate 2 levels for each of the 3 factors, we will have a 23 factorial design, which results
in at least eight different experiments. The design matrix can be built in the standard order as in
Table 1. Columns 2, 3, and 4 are obtained starting with the lower level (−1). The signal alternates
one to one for the first factor (column 2: −1, +1, −1, +1, …), two to two for the second factor (column
3: −1, −1, +1, +1, …), and four to four for the third factor (column 4: −1, −1, −1, −1, +1, +1, +1, +1). If
it had a fourth factor, the signal would alternate from eight to eight, so that for k factors, the last
factor would alternate the signals every 2k − 1 [1]. In this way, it is possible to easily build the design
matrix for any 2k factorial design.
Table 1 Results of a 23factorial design to evaluate the effect of pH, stabilizing agent, and temperature on the quantum
yield (%) of quantum dots
Table 1 presents the QY results for all experiments, performed in a random sequence. Some
experiments are performed in triplicate (combinations 2, 3, 4, 5, and 8), others in duplicate
(combinations 1 and 6), and the combination 7 is performed only once. All other synthetic variables
(stirring time, synthesis volume, initial concentration of X and Y precursors, etc.) that are not under
study in the optimization must be kept constant, and the instrumental conditions for acquisition of
QY must be set to the same value for all experiments. Thus, we avoid that the response is influenced
by sources of variation besides the factors under study.
According to the results of Table 1, when using stabilizing agent A with temperature 50 °C, and
increasing the pH from 5 to 13, the mean QY increases by 9%, since the response varies from 33%
to 42%. Using the stabilizing agent B and keeping the temperature at 50 °C, the same pH variation
from 5 to 13 causes the mean QY to decrease from 72% to 26%, resulting in a decrease of 46%. This
demonstrates that the effect of pH on the mean QY of QDs depends on the type of stabilizing agent
we are using, that is, there is an interaction effect between these two factors. This is an observation
that could hardly be found from the results of a univariate optimization.
The main effect of pH, which will be represented here by H, is the difference between the mean
QY of the QDs synthesized at the upper level for pH and the mean QY obtained at the lower level for
this factor. Using to represent the mean response of QYs obtained in the combination i, the main
effect of pH can be calculated as follows [1, 3]:
(1)
The interpretation that can be extracted from this main effect is that when we increase the pH of
the synthesis from its lower level to its upper level, that is, from pH 5 to 13, the QY on average
decreases by 19%. However, this analysis is not complete, since the type of stabilizing agent may
interact with other factors. Therefore, we also need to calculate the effects of stabilizing agent (S)
and temperature (T) and the interaction between these factors to make the joint interpretation and
draw reliable conclusions about the results of the experimental design.
Similar to pH, the calculation of the main effect of temperature is given by Eq. 2. It is noteworthy
that experiments in which the temperature is at its upper level will not always be the same
experiments in which the pH (or the stabilizing agent) is at its upper level.
(2)
In addition to the main effects, we have three interactions between two factors (HS, HT, and ST)
and one interaction between the three factors (HST). The signs for the interactions are obtained by
multiplying the signs of their respective factors for each experiment. The signs for the interaction
ST will be the product of signs for stabilizing agent and temperature: for the combination 3, for
example, ST will be the product (+1) × (−1). The signs for the interaction HST will be obtained by
multiplying the signs for the three factors: for the combination 1, it will be (−1) × (−1) × (−1). The
calculations of the effects of the interactionsST and HST are represented in Eqs. 3 and 4,
respectively, and follow the same reasoning as in Eqs. 1 and 2. Note that the results of all the
experiments are used to calculate each of the effects. Therefore, 8 is the minimum number of factor-
level combinations required to calculate the effects from a 2-level factorial design with 3 factors.
(3)
(4)
In order to evaluate the statistical significance of the effects, it is necessary to compare them
with the experimental error calculated from authentic repetitions (or genuine run replicates) of the
experiments, that is, replicates in which the whole process from the synthesis up to the
measurement of the QY is performed again [4]. Therefore, repeating only the measurement of QY
does not count as an authentic repetition, for instance. It is important that the repetitions are
representative of the variability of the whole process or experiment [1]. If it is not possible to
repeat all the experiments, some tests representative of the experimental domain should be
selected to be repeated, for example the extreme factor-level combinations ((−1, −1, −1) and (+1,
+1, +1), for three factors). In addition, it is essential to run the experiments in random order to
prevent occasional variables from systematically and significantly influencing the results and to
contaminate the effects of the factors under screening and optimization.
The pooled estimate of the experimental run variance is calculated from the authentic
repetitions by Eq. 5, where ν represents the degrees of freedom and s2 is the sample variance of
each experiment with repetition [3]. The sample variances are calculated and entered in Table 1,
and the degrees of freedom are obtained by subtracting 1 from the number of replicates for each
experiment. An experiment with three replicates, for example, will have 2 degrees of freedom. In
this example, the numbers of replicates are different to illustrate the calculation of the pooled
variance because in real optimizations it is not always possible to perform the same number of
repetitions for all the experiments.
(5)
The pooled variance of the QY of QDs is 4.32%, and the standard error can be obtained by taking
the square root of that value, which is 2.08%. This is the standard deviation associated with each
experiment [4]. The standard error for the effects is calculated from the pooled estimate of variance
by Eq. 6, where n is the total number of syntheses performed.
(6)
The standard error for the mean is calculated by Eq. 7, using the same pooled estimate of the
variance obtained by Eq. 5.
(7)
The confidence interval for each effect is calculated by Eq. 8 using the standard error for the
effects and Student’s t critical value with the sum of degrees of freedom used to calculate the pooled
variance [1]. The 95% confidence interval is the most used in statistical treatments, but the
Student’s t critical values at 90% and 99% confidence levels are also provided in Table 2.
(8)
Table 3 Effects estimates of pH, stabilizing agent, temperature, and their interactions, for the results of the quantum
yield (%) of quantum dots obtained by a 23factorial design
Factor Effect seffect × t12 −95% Confidence interval +95% Confidence interval
Mean 43.40 1.013 42.39 44.41
H −19.21 2.025 −21.24 −17.19
S 12.37 2.025 10.35 14.40
T 0.13 2.025 −1.90 2.16
HS −21.96 2.025 −23.99 −19.94
HT −0.54 2.025 −2.57 1.49
ST 0.38 2.025 −1.65 2.41
HST 5.38 2.025 3.36 7.41
Based on the confidence intervals, it can be stated that the main effect of temperature, the
interaction between pH and temperature, and the interaction between the stabilizing agent and
temperature are not statistically significant at 95% confidence level. A more didactic way of
representing these data is by the Pareto chart, commonly presented in articles to replace the effect
estimates table. The Pareto chart is a bar chart representing the effects estimates divided by the
standard error for the effects. Thus, each standardized effect estimate should be compared simply
to Student’s t critical value, which corresponds to the level of significance 0.05 represented by the
red line in Fig. 2. Using the Pareto chart, one can easily identify the significant effects, which should
be consistent with the conclusion drawn from Table 3.
Fig. 2 Pareto chart of the standardized effect estimates, for the results of the quantum yield of quantum dots obtained
by the 23 factorial design
The effects of factors and interactions can also be interpreted as geometric contrasts. The three-
factor representation corresponds to a cube whose vertices are the mean response of each of the
eight experiments, as shown in the diagram of Fig. 3. The main effects can be understood as
contrasts between opposing faces. The effect of pH, for example, is the difference between the mean
of the four positive sign experiments (back side of the cube) and the mean of the four negative sign
experiments (front side), which results in −19.21. This means that increasing the pH from 5 to 13
causes a mean decrease of 19.21% in the QY of the QDs. The effect of interactions between two
factors is the contrast between two diagonal planes, and the effect of interaction between three
factors is the contrast between the two twisted shapes represented in Fig. 3.
Fig. 3 Diagram for interpretation of effects of the 23factorial design
As the interaction effect between pH, stabilizing agent, and temperature is significant, it is
necessary to make the joint interpretation of these three factors. The interpretation of the
interaction HST predominates over the interpretation of the interaction HS between pH and
stabilizing agent (which is also significant), since these two factors also interact significantly with
temperature.
(equal to the average of the maximum and minimum experimental levels), ΔXi is the step of
variation (calculated by (Xmax − Xmin)/2), and α is the maximum coded level [5].
(9)
We will keep on using the example of 23 factorial design to study the QY of QDs by varying
temperature, stabilizing agent, and pH during the syntheses. The coded values associated with the
minimum and maximum experimental levels of each factor have been shown in Table 1.
Nevertheless, let us suppose that one wants to calculate the coded level for the temperature of
66 °C. Because it is a factorial design, the maximum coded level (α) is 1. is 60 °C, which is the
average of 50 °C (minimum experimental level) and 70 °C (maximum experimental level), and ΔXi is
10 °C, resulting from (70 − 50)/2. Introducing all these values in Eq. 9, one can find the value 0.6 for
Ci, which corresponds to the coded level for the experimental temperature of 66 °C.
What about the coded level for pH 7? Once again α is 1, and given 5 as the minimum
experimental level and 13 as the maximum experimental level, ΔXi and can be easily calculated,
resulting in 4 and 9, respectively. Thus, Ci is −0.5 for pH 7. This means that, considering the pH
experimental range from 5 to 13 coded from −1 to +1, pH 7 at its coded level is represented by −0.5.
The stabilizing agent, as a qualitative factor, can only be represented by −1 or +1 in the factorial
design, and those levels are assigned to the stabilizing agents A and B, respectively.
Equation 10 gives the general equation for calculating the response y by a factorial design with
three factors (represented by x1, x2 and x3) [1]. The model for a factorial design with two factors is
even simpler, obtained by Eq. 10 without the terms dependent of the third factor (β3x3, β13x1x3,
β23x2x3, β123x1x2x3).
(10)
If we consider the three factors pH (H), stabilizing agent (S), and temperature (T) given in coded
levels, the intercept (β0) of Eq. 10 is the average of the 8 combinations of 23factorial design, which is
43.4 for the results of Table 1. If the main effect of pH is −19.2, this means that a change from level
−1 (pH 5) to level +1 (pH 13) causes an average decrease of 19.2% in the QY. As the response varies
by −19.2% with the variation of 2 units at coded levels of pH, then the change of 1 unit causes the
variation of −9.6% in the QY. Therefore, the coefficient β1 of Eq. 10 for the main effect of pH is equal
to one-half of the effect estimate H (Table 3). The same reasoning applies to the other main effects
and interactions, giving rise to Eq. 11.
(11)
The terms whose effects are not statistically significant at the confidence level studied (95%, for
the example under discussion) should be removed from the equation. Hence, the terms referring to
T, HT, and ST are not present in Eq. 11. Below the terms of the equation, the corresponding
standard errors are written, which are also one-half of the standard error of their respective effects
(except for the mean).
With Eq. 11, it is possible to estimate QY values (given in %) from coded values of temperature,
pH and stabilizing agent. This equation is an empirical model that expresses the direct relationship
between the response and the factors within the ranges of levels studied, so it is a local model. For
that reason, we can only estimate the response or make conclusions about the factors within the
ranges studied (temperature from 50 to 70 °C, pH from 5 to 13, and stabilizing agents A and B), and
extrapolations are strictly not recommended.
So, what is the estimated QY for QDs of the hypothetical semiconductor XY synthesized with
stabilizing agent B, at 66 °C and pH 7? The first step is to convert these values to coded levels by Eq.
9. The coded level for the stabilizing agent B is +1. For pH 7 and temperature 66 °C, according to
calculations previously presented, the coded levels are −0.5 and 0.6, respectively. The coded level of
the interactionHS is −0.5, which is the product of the coded levels of pH and stabilizing agent.
Likewise, the coded level of the interaction HST is −0.3 (resulting from the product
(−0.5) × (+1) × (0.6)). Introducing these values in Eq. 11, the estimated QY is 59.1%.
We can evaluate the fitted model by performing the analysis of variance(ANOVA) with the
responses estimated by Eq. 11. The equations for ANOVA calculations can be found in Table 4,
where n is the total number of observations, m is the number of distinct levels in the experimental
design, and p is the number of parameters of model. To calculate the sum of squares for ANOVA, we
use the values of yij (experimental value of the replicate j performed in the level i), (overall mean
of the experimental responses), (mean experimental response for the level i), and (response
estimated by the model, for the level i) [1, 5].
Total n−1
The structure from Table 4 is how the calculations of ANOVA are commonly summarized in
academic works. However, the detailed calculations are presented in Table 5 for a better
understanding. The last three rows of Table 5 present the sum of squares (SS), degrees of freedom
(DF) and mean of squares (MS) for each source of variation, and corresponds to the overall mean
for the 20 runs. For that dataset, n is 20, m is 8 (corresponding to eight different factor-level
combinations performed), and p is 5, since Eq. 11 has five terms (counting with the intercept).
Table 5 Detailed calculations of ANOVA for the equation fitted to estimate the quantum yield (%) of quantum dots
obtained by a 23factorial design
Another random document with
no related content on Scribd:
Fig. 43
Fig. 44
Fig. 45
Fig. 46
Fig. 51
The bottom end should be slightly slotted, in order to receive the
end of a crowbar (see fig. 52).
It is now placed in position, and gently tightened up by the
leverage of a crowbar acting in the slot, and using the sole piece as
a fulcrum.
The advantage of the sole piece not being at right angles to the
shore can now be seen, as if it were so laid no tightening could be
gained by the leverage. This system is an improvement upon the
tightening up by wedges, as the structure is not jarred in any
manner. If the frame is to have more than one shore, they are
erected in the same manner, the bottom shore being the first put up,
the others succeeding in their turn. When in position the shores are
dogged to the sole piece and a cleat is nailed down on the outer side
of the system. The bottom ends are then bound together by hoop
iron just above the ground level. To prevent the shores sagging,
struts are fixed as shown on fig. 49.
Fig. 52
Besides preventing the sagging these struts serve the purpose of
keeping the shores in position. They may be fixed as nearly at right
angles to the shores as possible, or at right angles to the wall; in any
case they should reach to the wall plate at a point just below the
needle. The struts should be nailed to the shores and wall plate. If
the latter is wider than the shores, it should be cut to receive the
struts.
It sometimes occurs that the timbers are of insufficient length to
reach from the sole piece to wall plate. To overcome this difficulty, a
short timber is laid on the sole piece against and parallel to the next
middle raker, and on this short timber a rider shore stands reaching
to its position on the wall plate (see fig. 49).
When this is done the top middle raker should be stiffer to resist
the increased cross strain. Stiffness is gained by increasing the
depth. A rider shore is tightened by oak folding wedges driven
between the foot of the shore and the short timber which supports it.
Note must be taken that the outer raker is not carried too near the
top of the building, or else the upward thrust of the shores, which
always exists with raking shores, might force the bond or joints.
Fir is the best wood for shoring owing to the ease with which it can
be obtained in good length. Another advantage is its straightness of
fibre; although, as it is more easily crushed by pressure across the
grain, it does not answer so well as oak for wedges, sole pieces, &c.
In erecting flying or raking shores, notice should be taken of the
following points.
The systems should be placed from 12 to 15 feet apart if on a wall
without openings, otherwise on the piers between the openings.
In very defective walls it is an advantage to use lighter scantlings,
the systems being placed closer together. Heavy timbers handled
carelessly may precipitate the collapse which it is the intention to
avoid.
Wedge driving and tightening should be done as gently as
possible. It should be remembered that support only is to be given,
and not new thrusts set up, which may result in more harm than
good.
Fig. 54
In carrying out these operations note should be taken of the
following points:—
1. That the dead shores should not stand over cellars or such
places. It is better to continue the needle to such a length that solid
ground is reached, and the needle can then be strutted from the
dead shore.