- #1
andrewr
- 263
- 0
Hi!
I'm interested in computing empirical formulas with integer subscripts from mass percentages (or measured mass of substance.)
The standard technique is to compute a possible mass for each element (either measured, or mass percentage times a constant like 100g); and then convert this value to moles of substance for each element.
Once the moles of each element are known, all values are divided by the number of moles found in the element with least moles.
The result is the ratio of the number of atoms of one element to the minimal element; and still may contain decimal fractions. The remainder of the task is simply to find the best "fit" of an integer multiplication that will remove the decimal fractions from all elements.
But... There are the problems that I see by using the outlined method indiscriminately, and perhaps there are even others problems that I don't see; and I wonder what is available that overcomes these pitfalls.
Since the knee jerk method merely divides out the element with minimum mass or moles; it's clear that statistically, close to the largest possible error is being made.
Does anyone know of papers addressing any of the issues I am mentioning? or can anyone outline an improved method for determining empirical formulas from raw mass measurements?
Thanks!
I'm interested in computing empirical formulas with integer subscripts from mass percentages (or measured mass of substance.)
The standard technique is to compute a possible mass for each element (either measured, or mass percentage times a constant like 100g); and then convert this value to moles of substance for each element.
Once the moles of each element are known, all values are divided by the number of moles found in the element with least moles.
The result is the ratio of the number of atoms of one element to the minimal element; and still may contain decimal fractions. The remainder of the task is simply to find the best "fit" of an integer multiplication that will remove the decimal fractions from all elements.
But... There are the problems that I see by using the outlined method indiscriminately, and perhaps there are even others problems that I don't see; and I wonder what is available that overcomes these pitfalls.
- A measurement has an error of granularity and of scale; Smaller masses have high error.
- Often elements with least moles have nearly the largest random error.
- Sometimes parts of equations are not strictly related to other parts of equations (eg:Water of crystallization may vary).
Since the knee jerk method merely divides out the element with minimum mass or moles; it's clear that statistically, close to the largest possible error is being made.
Does anyone know of papers addressing any of the issues I am mentioning? or can anyone outline an improved method for determining empirical formulas from raw mass measurements?
Thanks!