Embed Size (px)
Citation preview
From characterisation to validation, a journey through
MSc applied analytical chemistry.Dr Victoria Hilborne: [email protected]
Dr Anna RoffeyUniversity College LondonDepartment of Chemistry
1
• CareersAgriculture and Food.Environment.Toxicology.Oil and Gas.Pharmaceutical.Personal care.Forensics.
2
• Aim“This MSc seeks to train the next generation of analytical
scientists in state-of-the-art methods”
“is designed to provide comprehensive training in analytical chemistry and its implementation. A thorough understanding of error analysis, data processing and data presentation is at the foundation of this programme”.
3
• Exercise.From the following statements decide which is qualitative, quantitative, characterisation or fundamental analysis.
A. A waste incinerator is suspected of releasing more than the legal limit of dioxins into the atmosphere.
B. An art gallery suspects a masterpiece was switched to a forgery.
C. Airport security needs a more reliable method for detecting explosives in luggage.
D. The loss of metabolites of a new drug through excretion in urine need to be determined.
4
• Importance of measurement.Identification of impurities / contaminants.
- regulatory limits.- batch quality.
Characterisation of a new substance.- chemotherapy.
General assay – content or purity.- process improvement.- long term trends.
5
?
• Method performance characteristics.
Limit of detection LOD (MLD) - sensitivityLimit of detection LOQ (MLQ)Bias - accuracy.Precision (homogenous samples)
RepeatabiltyReproducability
Ruggedness – method control
6
• ChemometricsChemistry, chemical engineering, biochemistry, medicine. Factors, responses and variables.It was noted that an increase in foaming was accompanied by an increase in alcohol content.
7
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 𝑚𝑚𝑚𝑚𝑚𝑚𝑤𝑤𝑤𝑤𝑚𝑚𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑤𝑤𝑝𝑝𝑝𝑝 % alcoholfoamingsugar content
Foaming / CO2
• ChemometricsStatistical experiment design (Design of Experiment)1.
8
• Representative.• Variance.Sampling
• Extraction, clean up, pre concentration.• Data preprocessing.
Experiment design
• Variable reduction, pH, temperature…• Controlled, uncontrolled, masquerading.
Factor analysis
• SAMPLING
9
10
�̅�𝑥�̅�𝑥 − 𝑝𝑝 �̅�𝑥 + 𝑝𝑝�̅�𝑥 − 2𝑝𝑝�̅�𝑥 − 3𝑝𝑝 �̅�𝑥 + 3𝑝𝑝�̅�𝑥 + 2𝑝𝑝
𝑄𝑄𝑄 𝑄𝑄2 𝑄𝑄3
50%
68.3%
95.4%
99.7%
68.3% of area under the curve; �̅�𝑥 − 𝑝𝑝 < �̅�𝑥 < �̅�𝑥 + 𝑝𝑝
± 1 𝑝𝑝𝑤𝑤𝑚𝑚95.4% of area under the curve;
�̅�𝑥 − 2𝑝𝑝 < �̅�𝑥 < �̅�𝑥 + 2𝑝𝑝± 1.96 𝑝𝑝𝑤𝑤𝑚𝑚
99.7% of area under the curve;�̅�𝑥 − 3𝑝𝑝 < �̅�𝑥 < �̅�𝑥 + 3𝑝𝑝
±2.58 𝑝𝑝𝑤𝑤𝑚𝑚
• Data Analysis and Theory.
Descriptive statistics.
• Data Analysis and Theory: DESCRIPTIVE STATISTICS.
11
• Synthesis of cadmium sulphide crystal quantum dots by bubbling hydrogen sulphide (H2S) gas over a foam with cadmium ions.
• Diffractive imaging showed CdS quantum dot sizes ranging from 7 – 9 nm.
• Estimate 𝑑𝑑𝑑𝑑 (𝑁𝑁 − 𝑄) , �𝑥𝑥 , s2, s, cv and 𝑝𝑝�̅�𝑥 .
Huang W.J, Zuo J.M…Nat. Phys. Letters., 2009, 5, 129
7.5 nm 9 nm 8.5 nm 8 nm 7 nm
7 nm 9 nm 7.5 nm 8 nm 8.5 nm
�𝒙𝒙 = s2 = s = cv = 𝒔𝒔�𝒙𝒙 =
• SIZE OF DATA SETS…..and DISTRIBUTION.Atmospheric CO2 at Mauna Loa in June – August in 1985: 334.6 to 348.7 ppm. Assuming a uniform distribution plot a;
• histogram of the running means of eight observations. • continuous normal distribution using the parameters.
12
02468
1012141618
freq
uenc
y
means of 8 [CO2]/ppm
Running mean of 8 [CO2] / ppm
0.00
0.02
0.04
0.06
0.08
0.10
332 334 336 338 340 342 344 346 348 350
prob
abili
ty
[CO2] / ppm
f([CO2])
• Error and Uncertainty
• Uncertainty estimate. • Uncertainty budget.
13
Contribution to uncertainty.a Intercept of calibration curve.b Slope of calibration curve.c Intermediate precision.d Concentration of calibration solutions.
Uncertainty in response 𝑦𝑦;
𝑢𝑢𝑦𝑦 = 𝑢𝑢𝐴𝐴2 + 𝑢𝑢𝐵𝐵2𝑢𝑢𝐴𝐴
𝑢𝑢𝐵𝐵
δz = z 𝑤𝑤𝛿𝛿𝑚𝑚𝑚𝑚
2
+ 𝑚𝑚𝛿𝛿𝛿𝛿𝛿𝛿
2
+ 𝑝𝑝𝛿𝛿𝑝𝑝𝑝𝑝
2
+ 𝑞𝑞𝛿𝛿𝑑𝑑𝑑𝑑
2
𝑢𝑢𝑦𝑦
Quality
ASSESSMENT – How? Planned set of activities, documentation. ISO / IEC.
CONTROL – What? Programme of activities, what is done. Control Charts.
14
2.5
3.5
4.5
5.5
6.5
7.5
1 2 3 4 5 6 7 8 9 10 11
µg c
m-3
Control Chart
• Inferential Statistics.
𝑭𝑭 variance ratio, 𝒁𝒁 tests.Student 𝒕𝒕 tests.
Levels of geraniol in a suspected fake perfume sample.ANOVA.
Ochratoxin A levels in samples from bulk wheat storage.GLM – ANOVA.
Significance of difference between levels of acetaminophen glucuronide in different sections of a river.
Chi Square (𝝌𝝌𝟐𝟐). Presence of uni and divalent cations in fermentation tank effect alcohol level.
15
How many sets of data do you want
to compare>
Find a relationship
between the two
sets?
TWO THREE OR MORE
PAIRED t-TEST ONE_WAY ANOVASTUDENT t-TEST
Find a relationship
between the sets?
NONO
YESAre the
data paired? YES
YES
NO
Regression and correlation
YES
Are the data
paired?
Matched samples ANOVA
TWO_WAY ANOVA
1 or 2 independ
ent variables
NO
ONE TWO
16
• Parametric tests
• Student t test: Comparing two methods.
Vitamin B9 recovered from green kiwi fruit using 2 extraction methods.
𝐹𝐹 test = 𝑠𝑠12
𝑠𝑠22
combined standard deviation, 𝑝𝑝𝑐𝑐 = 𝑠𝑠12 𝑛𝑛1−1 +𝑠𝑠22 𝑛𝑛2−1𝑛𝑛1+𝑛𝑛2−2
𝑡𝑡 = �̅�𝑥1−�̅�𝑥2
𝑠𝑠𝑐𝑐1𝑛𝑛1+ 1𝑛𝑛2
𝑑𝑑𝑑𝑑 = 𝑤𝑤1 + 𝑤𝑤2 − 2 𝑡𝑡𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 < 𝑡𝑡𝑒𝑒𝑠𝑠𝑐𝑐𝑐𝑐𝑒𝑒𝑐𝑐𝑐𝑐𝑒𝑒
17
• Practical: Quantification of D-Limonene in perfume using GCMS.
18
0.00E+00
1.00E+08
2.00E+08
3.00E+08
4.00E+08
5.00E+08
6.00E+08
0 10 20 30 40 50
peak
are
a
D - limonene /µg cm-3
GCMS calibration of D-Limonene
-3.00E+08
-2.00E+08
-1.00E+08
0.00E+00
1.00E+08
2.00E+08
3.00E+08
0 10 20 30 40 50
y -y
estim
ate
D-Limonene µg cm-3
residuals
0.00E+00
5.00E+07
1.00E+08
1.50E+08
2.00E+08
2.50E+08
3.00E+08
0 10 20 30 40 50
peak
are
a
D-Limonine µg cm-3
GCMS calibration of D-Limonine
• Data: Estimate and Discuss.
Uncertainty from: slope of calibration, repeatability,concentration of calibration solutions.
𝐹𝐹 test: - variance ratio repeats of sample perfume versus real perfume.- student data comparison in pairs.
𝑡𝑡 test: two sample, two tail.
THEIR Method Limit of Quantification (MLQ), Bias.
19
• Steroids in Soya Milk – LCMS3
20
Extraction
• Liquid –liquid to
• liquid -solid.
Derivatisation
• Cholesterol oxidase.
• Girard P.• Overnight.
Extraction
• SPE• Washed
and eluted x 3.
Analysis
• ThermoAccela Ion Trap LCMS3
• HypersilGold C18
Campesterol
• Arduino – Analytical Device Construction.
21
SampleThe analyte in a
matrix: a mixture of potential
interferants and solvent.
TransducerConverts energyfrom one form to
another. Chemical, electromagnetic.
OutputDisplay of data- raw and processed for
interpretation.
• Analytical Strategies - ARDUINO projects https://www.arduino.cc/.
Programmable boards, electronics – sensors, mechanical parts and open source software (coding).
22
16 Steps to a spectrophotometerhttps://www.instructables.com/id/An-Arduino-powered-microplate-spectrophotometer/
• Arduino - Apples to Radon.
23
Integrated board
I2Cpower
RD200M Radon Sensor
LED Display
BME280CCS811
• Arduino- NIR Plastics Identification.
24
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Sign
al a
.u.
Mean signal per sample. Single Factor ANOVA.(𝜶𝜶 = 𝟎𝟎.𝟎𝟎𝟎𝟎)
F p F critical
10.18 0.0016 3.68
?
• Analytical Strategies II: Approach to Laboratory Diagnosis of Infection.
25
By --Ygonaar 23:09, 7 March 2006 (UTC) - It's a graph create by Ygonaar with Power point, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=607322)
Source of diseases.Metrology in microbial diagnosis.Antimicrobial resistance.
• Analytical Strategies II: Forensic Toxicology.
26
Drink and drug driving (abuse).Action in the body.OpioidsAnalysis – GCMSCase studies.
• Analytical Strategies II: Environmental Pollution monitoring.
27
Air, soil, water pollutionSampling.Complex sample matrices.Remote sensing.Sensor networks.Data processing.
• Masters level research projects:Contamination of waste and surface waters with phenylenediaminesfrom hair dyes.
Sampling and quantification of Iron particulates in underground train stations.
Nanodevices for detecting the destruction of cancer cells.
Photoelectron spectroscopy study of single molecules in the gas phase.
Measuring lung microbiome in the study of chronic respiratory disease.
28
Dr Vicki Barwick Dr Anna Roffey UK National Measurement Laboratory. Programme manager.
Professor Daren Caruana Dr Mike Parkes Physical Chemistry. Senior Research Associate.
Dr Kersti Karu Mass Spectrometry Lab Manager
29
• Acknowledgements
