Buchi DuMaster D-480, Operation Manual, Page: 54 / 354

Operating instructions DuMaster D-480 ©BÜCHI Labortechnik AG
4 - Understanding the instrument and planning its use 54
Use the following values for process standard deviation "Proc-SD" as guidelines:
Acceptable value:
Proc-SD = ±3 µg in the lower range
Proc-SD = ±15 µg in the upper range
Very good value: Proc-SD = ±1 µg
Use the following values for quality factor "r" as guidelines:
Good value for linear calibration function:
r = 0.99 (r should tend to 1)
Good value for polynomial as calibration function r = < 2-4 (r should tend to 0)
If you do not approximately achieve the listed values you can assume that you have chosen the
wrong calibration curve calculation method.
Calibration formulae
Calculation of absolute element content
Calibration is performed by analyzing samples with known concentrations (standard samples) where
the measured peak areas are correlated with the corresponding absolute element content.
For a standard sample, the following parameters are known:

element concentration

Weight
Absolute element content is computed according to this formula:
where:
a = absolute element content in [mg]
w = sample weight in [mg]
c = element concentration in [%]
Computing calibration coefficients
Once the absolute element contents have been computed and the peak areas have been measured
for the respective sample, the pairs of values (absolute content / peak area) are transferred into a
coordinate system where they form the calibration curve. By means of an algorithm, the calibration
coefficients a - e are computed which, entered in a polynomial, constitute the calibration curve.
Absolute element content is computed according to this formula:
where:
y = absolute element content in [mg]
a...e = calibration coefficients
x = peak area
Computing the quality factor for polynomials of the 1st degree
In order to assess how well the calibration curve described by the polynomial follows the actual
development of the pairs of values, the software computes a quality factor for each calibration
process. For a polynomial of the 1st degree (linear equation), this quality factor corresponds to the
correlation coefficient. The closer the correlation coefficient tends to 1, the better the fit.