Lecture 1
Dimensional Concentration
Analysis and Solutions
• Mass units
• Volume units
• Concentration units
Dimensional Analysis:
Multiply by 1
Dimensional Analysis:
Examples
You have 34.5 kg of sodium sulfate, how many grams do you have?
You have 1.0 quart of milk, how many milliliters do you have?
Hi Students: So in the original I had
grams in the problem on the bottom, and
of course it should have been kilograms. Dimensional Analysis:
Thanks for the catch and its fine now!
More Examples
Dr. Colvin
You have 34.5 kg of sodium sulfate, how many moles do you have?
You have .100 moles of aluminum nitrate, how much is that in kilograms?
About
Dimensional
Instrumental
Analysis
Analysis (1)
Lecture
Lecture 22: Dimensional Analysis
Dimensional Concentration
Analysis and Solutions
Moles Moles
(molecule) (atom) Molarity
Weight
(grams) = ppb (ug/L)
Volume
ppm (mg/L)
(liters)
Examples of simple
concentration calculations
You have .345 mg of sodium sulfate in 1L of water, what is the molarity of the solution?
You have .100 moles of aluminum nitrate in .500 L of water, what is the ppm of Al?
Hi Students: When I did the first problem, I typed in sodium phosphate into
wikipedia to get the molecular weight. A common short-cut and confusing
in this case! It gave me the molecular weight for monosodium phosphate
(120), not trisodium phosphate which is what I show below. Always best to
More
calculate from the formula which is what I used to do, and will now do
again! It did not affect the numeric answer in this example, but best idea is
to calculate MW from formula weights.
concentration
Had a transcription error – right answer is .03069 as shown not .0369 as
originally written. This is corrected.
Dr. C
examples
You have 1.023 molar tri-sodium phosphate, how many moles sodium are in 10.0 ml?
164 gm/mole
.03069
You diluted 2.00 ml of an unknown lead solution into .500 L and found a 10.3 ppb
concentration in the diluted sample. What was the molarity of the unknown?
Units of Concentration in Analysis
Lecture 3A
But is that
Every instrument has an instrument tolerance – the accuracy one can expect
Expected error:
Instrument Tolerances
Graduated cylinder ± 2 ml
http://en.wikipedia.org/wiki/Significant_figures
Analysis Error:
Significant Figures
Lecture 4B- Examples
When rounding – look at all digits beyond last place desired. If exactly halfway, round
to nearest even digit (why do we do this?)
Independent Exercise:
Significant Figures
Ex. 1: Round
(A) 1.2367 to 4 SF:
(B) 0.1352 to 3 SF:
(C ) 2.051 to 2 SF:
(D) 2.0050 to 3 SF:
(E) 3.0150 to 3 SF:
Ex 3: Find the molecular weight of KrF2. We know Fluorine’s atomic weight to very high
precision (18.9984032 ± .0000005 g/mol). Krypton is less well defined (83.90 ± .01
g/mol).
Ex. 4: Rewrite the number 3.12356 ± 0.16789% in the forms (a) the absolute uncertainty and (b)
the relative uncertainty. Use the correct number of digits.
Analysis Error: Significant Figures
Lecture 4
(x i x) 2
s i 1
N 1
Example: Average and
Standard Deviation
Hi Students: Note, when I did this the first time I used Excel’s stdev function and somehow got
a stdev which is a divide by N standard deviation – excel now has a whole variety of stdev –
stdev.p, stdev.s for example. So, I redid it with the stdev.s function which is the sample
standard deviation and the answer has been updated –note it hardly changed the answer
(before it was s=1.85 with N=6) but I wanted to be as clear as possible.
Thanks! Dr. C
You are measuring the lead in teething toys, and find in 6 samples lead measurements
of 102.3, 100.1, 98.56, 104.3, 99.89 and 101.0 ppm. What is the average and
uncertainty (both absolute and relative) of this dataset? How do you report the data?
You would report either 101.0 ± 2.0 ppm Or you would report 101.0 ± 2%
You should only seldom keep more than one significant figure in the standard deviation
because it has error too – possibly two significant figures are OK but never more.
Error Bars from Replicate
Measurements Lecture 5
x i
x i 1
N
Answer x (factor ) s
• Type 1: you have a set of experimental data and can find ‘s’ from
calculating the standard deviation. You wish to report the error bars
and specify a confidence limit:
t s
Answer x
Specify degrees of freedom (N-1), confidence limit
Then use t-table to look up the “t” parameter
N
• Type 2: You know the s for a method (past history/experience/ error
validations) that it is the REAL . You collect experimental data with a
given N and wish to report the error bars and confidence limit:
t s t
Some books call this t the
Answer x Answer x
parameter ‘z’ and pulls out the
N=infinity t-table in a separate
N N table
Specify an infinite degree of freedom and desired confidence to look up ‘t’
Remember for ‘N’ to use the number relevant in specific experimental case
Hi Students: Note, when I did this the first time I used Excel’s stdev function which is a divide
by N standard deviation – even though it claims it isn’t! So, I redid it with the stdev.s function
which is the sample standard deviation and the answer has been updated –note it hardly
changed the answer (before it was s=.47) but I wanted to be as clear as possible.
Note that a similar issue was in lecture 5 as well. Check my posts to see more information.
Thanks! Dr. C
ts ts
Answer x Answer 14.9 mg Zn
N N
2.45 .52
Answer 14.91 mg Zn 14.9 mg Zn 0.5 mg Zn
7
Example: You measure the zinc level in vitamin tablets
using a NIST method that reports the standard error to
be ± .30 mg Zinc. How many replicate measurements
would you need to make to get the 95% confidence limits
to under ± .42 mg Zinc?
t 1.96 .30
2 2
N 1.96 round to 2
error .42
Error Bars with Confidence
Lecture 6
Tolerance is _______
Tolerance +/- .1 grams
credit: PRHaney
Answer = a + b - c
Answer = a·b/c
Example: Error Propagation
You measure a metal rod of unknown composition and find using a graduated cylinder that it has a volume of
50.3 ml and you weigh it on an analytical balance and find a weight of 361 grams. (A) What are the standard
errors for volume and weight? (B) What is the density with error? (C) What is the rod made of?
Remember, we use instrument tolerance if we did not do replicate measurements
(A) For graduated cylinders it is ± .2 ml and for analytical balances it is ± .0005 gm (see first page)
Density is defined as mass over volume, so this means we need relative errors
.2 .0005
.00397 .0040 1.39 6
50.3 361
Now you just divide mass by volume, and propagate the error
361
7.18 /
50.3
So the answer would be 7.18 gm/ml ± .4% or if you convert back to absolute 7.18 ± .03 gm/ml
Example: A 218.44 ± 0.01 gram cylinder has a diameter of 2.50
± 0.01 cm and is 5.00 ± 0.01 cm long. What is the density of
the metal?
Note: There
was a really
interesting
thread on
whether you
could lower the
error by writing
the volume in
terms of
diameter! An
interesting
conversation,
with no exact
resolution. My
take is that I
had envisioned
the error as
due to calipers
– so the error
on 1.25 would
be the same
as on 2.50
.2
247
1.9% or 2%
Combining Error from
Multiple Sources
Using Error Bars to Make Decisions
Lecture 8
1: Systematic error check. You are comparing an experimental result against a known
precise and accurate value. If the precise and accurate value is within your error bounds
then there is no systematic error.
x known
t N if t t X , N then with X% confidence they are the same
sx
Now you simply plug in the values from the problem to get an actual t value
20.06 20.00
t 6 5.9 2.571 t95,5
0.025
The computed t-value is 5.9 which is greater than the t(95,5) or 2.571
We thus conclude that the known value and experimental values are not the same and
that there is a source of systematic error
The Q-Test: Tossing out a Datapoint
• Q: Divide the ‘gap’ between outlier and nearest neighbor by
range of all measurements – call this Qexp
• If calculated Qexp is larger than Qrej from the table below,
you can neglect the datapoint in question
• This is a subjective decision about how confident you want or
need your measurement to be
Example: You have six measurements: 102, 97, 90, 110, 108, 65. Can you throw one out?
First put in descending order – outliers have to be biggest or smallest: 110, 108,102,97,90,60
The top is easy then, just the outlier – nearest value: 90-60 = 30 (note take the absolute values, sign won’t matter)
You can throw out 60 with 90% but not 95% confidence!
Number 3 4 5 6 7 8 9 10
of values:
Q90%: 0.941 0.765 0.642 0.560 0.507 0.468 0.437 0.412