Homework 8: Analyzing Rate Data & Regression

CHE 3184

1. The gas phase reaction A→ B + C was carried out in an isothermal PFR (negligible
pressure drop) where the following rate measurements were recorded at very low
conversions:

-rA (mol/(dm3 sec)) 0.26 0.11 0.1 0.067 0.03 0.024 0.01 0.0055
3
CA0 (mol/dm ) 1.7 1.4 1.2 0.95 0.75 0.65 0.53 0.4

Use linear regression to determine the reaction order n and the reaction rate constant k

To use linear regression, we must first get a linear equation. We start with the general
reaction rate equation:

−𝑟𝐴 = 𝑟 = 𝑘𝐶𝐴𝑛 = 𝑘𝐶𝐴0

𝑛

We can linearize this by taking the natural log

ln(−𝑟𝐴 ) = ln(𝑘) + 𝑛 ∗ ln(𝐶𝐴0 )

Compare this to the equation for a line and you have that:

𝑦 = ln(−𝑟𝐴 ) 𝑏 = ln(𝑘) 𝑚=𝑛 𝑥 = ln(𝐶𝐴0 )

This gives a slope of approximately 2.6 and an intercept of approximately -2.8.
Our slope is exactly our order, however to get the k value, we need to convert it back using
the inverse of our transformation equation. i.e.
𝑏 = ln(𝑘) → 𝑘 = exp(𝑏) = exp(−2.8) = 0.061
The units here are not given as they do not make much sense as non-whole units will be
included.
The equation is then:

−𝑟𝐴 = 0.061 ∗ 𝐶𝐴2.6

Where the concentration is given in moles per decimeter cubed and the rate is given in mole
per decimeter cubed per second.
 2. The liquid-phase irreversible reaction
A→B+C
is carried out in an isothermal PFR (negligible pressure drop). To determine the rate law,
the volumetric flow rate, υ0, (hence τ = V/υ0) is varied and the effluent concentrations of
species A are recorded as a function of the space time τ. Pure A enters the reactor at a
concentration of 2 mol.dm-3. The measurements were conducted under steady-state
conditions.
τ (min) 15 38 75 250 700 1400 1700
CA (mol.dm-3) 1.8 1.5 1.3 0.7 0.3 0.13 0.09

Determine the reaction order (1, 2 or 3) and rate constant using two integral methods

First let’s get the system function for this reactor and these two variables. Starting from the
design equation:

𝑋 𝑋
𝑑𝑋 𝑑𝑋
𝑉 = 𝐹𝐴0 ∫ = 𝐶𝐴0 𝑣0 ∫ 𝑛 𝑛
0 −𝑟𝐴 0 𝑘𝐶𝐴0 (1 − 𝑋)

1−𝑛 𝑋
𝐶𝐴0 𝑑𝑋
𝜏= ∫
𝑘 0 (1 − 𝑋)𝑛

Solving this generically we get:

1−𝑛 (1
𝐶𝐴0 − 𝑋)1−𝑛 − 1
𝜏=
𝑘 𝑛−1

This has a discontinuity at n=1, luckily, if we resolve the integral at x=1. We get the simpler
function:

−1 1 1
𝜏= ln(1 − 𝑋) = ln ( )
𝑘 𝑘 1−𝑋

Now, we have concentrations instead of conversions, we can adjust for this by remembering
the definition of conversion.

𝐶𝐴
𝐶𝐴 = 𝐶𝐴0 (1 − 𝑋) → 𝑋 = 1 −
𝐶𝐴0
Now we have:

𝐶𝐴 1−𝑛
1−𝑛
𝐶𝐴0 (1 − 1 + ) −1 −1 𝐶𝐴
𝐶𝐴0
𝜏= 𝜏= ln (1 − 1 + )
𝑘 𝑛−1 𝑘 𝐶𝐴0

1−𝑛
𝐶𝐴0 𝐶𝐴 1−𝑛 1 𝐶𝐴0
𝜏= (( ) − 1) 𝜏= ln ( )
𝑘(𝑛 − 1) 𝐶𝐴0 𝑘 𝐶𝐴

This should look familiar. These are the Essen and van’t Hoff Equations to fit. To get the
half-life form, we just substitute the definition of half-life into it:

𝐶𝐴0 𝐶𝐴0
𝐶𝐴 = →2=
2 𝐶𝐴

1−𝑛
𝐶𝐴0 1
𝜏= ((2)𝑛−1 − 1) 𝜏= ln(2)
𝑘(𝑛 − 1) 𝑘

Plotting these:

It is apparent that the 2nd order one is the most linear. The 1st order one bows up and the 3rd
order bows down.
The van’t Hoff method is very similar, but with the y variable switched such that you are now
plotting k:

1−𝑛
𝐶𝐴0 𝐶 1−𝑛 1 𝐶𝐴0
𝑘= ((𝐶 𝐴 ) − 1) 𝑘 = ln ( )
𝜏(𝑛−1) 𝐴0 𝜏 𝐶𝐴

Here we can see that the 2nd order is the most flat, implying that the reaction is closest to
second order. Perhaps it is a little lower than 2nd order, however it is often best to round to
whole numbers if more accurate values must be guessed (in this method at least).

Considering there is not a point near the half-life, I would not recommend that method for
this problem.
This part was not in the HW but here’s the best way to extract the order. This is done by
extracting the order and rate constant simultaneously using non-linear regression (i.e. do not
assume order to get k)

If we go back to the equation

1−𝑛
𝐶𝐴0 𝐶𝐴 1−𝑛
𝜏= (( ) − 1)
𝑘(𝑛 − 1) 𝐶𝐴0

We can easily start with this and then minimize the error in the equation. Note that it is
preferred to match the dependent variable, which in this case is the concentration. So if we
solve for the concentration:

1
𝜏𝑘(𝑛 − 1) 1−𝑛
𝐶𝐴 = ( 1−𝑛 + 1) 𝐶𝐴0
𝐶𝐴0

tau (min) C_A (mol/dm^3) C_A calc Error

15 1.8 1.800418711 0.0000001753188622
38 1.5 1.558519472 0.003424528627
75 1.3 1.277015197 0.0005283011844
250 0.7 0.6724406411 0.0007595182636
700 0.3 0.2888156754 0.0001250891168
1400 0.13 0.146950517 0.0002873200264
1700 0.09 0.1204086691 0.0009246871569
0.006049619694
fitted params
k 0.003973875904
n 1.886297856

other params
C_A0 2 mol/dm^3

Manually fitting these using Google Sheet’s goal seek gives the following. Where the order is
1.9 and the k value is 0.0040. Or

𝑟 = 0.0040 ∗ 𝐶𝐴1.9
Where the units on the rate are moles per minute per decimeter cubed and the units on
concentration are in moles per decimeter cubed.

tau (min) C_A (mol/dm^3) C_A calc Error

15 1.8 1.831884025 0.0003137626644
38 1.5 1.616475237 0.006029547019
75 1.3 1.348772439 0.001407544833
250 0.7 0.7107792785 0.0002371282536
700 0.3 0.2761917988 0.006298116058
1400 0.13 0.1231162614 0.002803896864
1700 0.09 0.09646816736 0.005165085064
0.02225508076
fitted params
k 0.003744234731
n 1.688447124

other params
C_A0 2 mol/dm^3

Normalizing the error gives the slightly difference answer:

𝑟 = 0.0038 ∗ 𝐶𝐴1.7
 3. The following temperature dependent measurements were performed at low conversions
for the oxidation of CO to CO2
CO +0.5O2 → CO2

T(K) 312.5 318.5 322.6 327.9 333.3

-1
k (s ) 0.00031 0.0011 0.00197 0.0033 0.007

Determine the activation energy (Ea) and the pre-exponential factor (A) using linear
regression

Starting with the general form:

𝐸𝐴
𝑘 = 𝐴 ∗ exp (− )
𝑅𝑇

Linearizing that equation gives:

𝐸𝐴
ln(𝑘) = ln(𝐴) −
𝑅𝑇

Where,

1 𝐸𝐴
𝑦 = ln(𝑘) 𝑏 = ln(𝐴) 𝑥= 𝑚=−
𝑇 𝑅

And as such:

𝐴 = exp(𝑏) 𝐸𝐴 = −𝑚 ∗ 𝑅

Linearizing and graphing give us the following values.

Which corresponds to

𝑘𝐽 𝑘𝐽
𝐸𝐴 = −𝑚 ∗ 𝑅 = −15000 𝐾 ∗ 0.00831 = 125
𝑚𝑜𝑙 ∗ 𝐾 𝑚𝑜𝑙
1
𝐴 = exp(𝑏) = exp(40. ) = 3.1 ∗ 1017
𝑠
This gives the equation:
𝑘𝐽
1 125
𝑘 = 3.1 ∗ 1017 exp (− 𝑚𝑜𝑙)
𝑠 𝑅𝑇