SUPPLEMENTARY INFORMATION

Analytical Chemistry (2014) doi: 10.1021/ac5014749

Coordinate swapping in standard addition graphs for analytical chemistry:
a simplified path for uncertainty calculation in linear and nonlinear plots
Juris Meija*, Enea Pagliano, and Zoltn Mester
National Research Council Canada, 1200 Montreal Road, Ottawa, Ontario K1A 0R6, Canada
* Corresponding author: juris.meija@nrc.ca

Experimental part
Reagents and materials
Bromide Standard Solution (NIST SRM-3184, Gaithersburg MA, USA) was used as the primary
standard while groundwater certified reference material (IRMM BCR-611, No. 0049; Geel,
Belgium) was analyzed for the content of bromide. Aqueous ammonia (Tamapure AA-100;
w(NH3) = 0.2 g/g) and triethyloxonium tetrafluoroborate (Fluka; w > 0.97 g/g) were used for
sample preparation and derivatization respectively.
Safety considerations
Triethyloxonium tetrafluoroborate is a strong ethylating agent and attention should be paid
to avoid any exposure to this substance and to the fumes arising from its decomposition
products, diethyl ether and hydrofluoric acid.
Sample preparation
Determination of bromide in BCR-611 was performed using gravimetric standard addition
calibration. For this purpose five 2 mL aliquots of sample were transferred into 4 mL
borosilicate glass vials. The mass of each sample solution, mA(AA*), was measured. Four
aliquots were then spiked with 50, 100, 150, and 200 L of bromide standard solution
respectively (primary reference standard with w(Br) = wA* = 1925(2)k=1 ng g1), and the mass
of the added standard solution, mA*(AA*), was measured by weighing the vials before and after
the spiking. At this point all five solutions were diluted to equal volume (2.20 mL) with water,
alkalized with 50 L ammonia solution, w(NH3) = 0.2 g/g, and then spiked with 100 L of
freshly prepared aqueous Et3O+[BF4] solution, w(Et3O+[BF4]) = 0.25 g/g, in order to ethylate
bromide and iodide (iodide was present in the groundwater sample as was used as the
endogenous internal standard). All vials were immediately sealed with a screw-cap
containing an unpunctured poly(tetrafluoroethylene)/silicone septum and the mass of the
resulting solutions, mfinal, was measured. All samples were kept in the dark at room
temperature for at least 3 h before the analysis.
Instrumentation
A Hewlett-Packard 6890 gas chromatograph equipped with a Hewlett-Packard 5973 mass
selective detector was operated at constant flow of helium (0.9 mL/min) with a 60 m DB-624
capillary column. A narrow inlet liner designed for SPME (internal diameter of 0.75 mm) was

used and a headspace volume of 250 L was injected into the GCMS. The GC was operated in
pulsed split mode (5:1) with pulse pressure of 172 kPa for 1 min. Inlet temperature was set at
100 C; the oven was heated to 30 C for 7 min and then 20 C/min to 200 C; the transfer line
temperature was set at 230 C. Working conditions of MS were standard (EI at 70 eV,
temperature of ion source and quadrupole set at 250 C and 150 C, respectively). Detection
of compounds was performed in single-ion monitoring mode at m/z = 108 Da (C2H579Br+) and
156 Da (C2H5I+) with a 100 ms dwell time for each ion and a typical chromatogram is given in
Figure S1. Background-subtracted peak areas, denoted as AEtBr and AEtI, were used for the
instrumental response.

Figure S1. A typical GC/MS chromatogram of bromide and iodide sampled from the
headspace of the groundwater reference material BCR-611 with w(Br) = 93 4 ng/g
(certified quantity value) and w(I) = 9.0 ng/g (additional information). Bromide and
iodide were converted to EtBr, and EtI by reaction with Et 3O+[BF4] and detected
simultaneously by GCMS.

Theoretical part
Pad[1,1] model
The analytical result from a Pad[1,1] standard addition fit, y = (a0 + a1x)/(1 + a2x), is
x0 = a0/a1
(s1)
and the uncertainty of x0 is:
u2 (x0 ) u2 (a0 ) u2 (a1 )
cov(a0 , a1 )
(s2)

2 2
2
2
x0
a0
a1
a0a1
where
[x2 y][xy] [x][x2 y 2 ]
(s3)
cov(a0 , a1 ) u2 (a1 )
N[x2 y2 ] [xy][xy]

Here and hereafter the bracket notation [x] signifies the sum over all values of xi (i = 1N),
[xy] signifies the sum of products xiyi (i = 1N) and so on.
Pad[2,1] model
The analytical result from a Pad[2,1] standard addition fit, y = (a0 + a1x + a2x2)/(1 + a3x), is
x0

a1 a12 4a0a2

2a0
and the uncertainty calculation is prohibitively complex:
u2 (a0 ) q2u2 (a1 ) d2u2 (a2 )
u2 ( x0 )

z
4a22 z
4a24 z
q cov(a0 , a1 ) d cov(a0 , a2 ) qd cov(a1 , a2 )

a2 z
a22 z
2a23 z

(s4)

(s5)

where z a12 4a0a2 , q a1

z , d a1q 2a0a2 as in the quadratic model. The parameter
covariances, however, are given by the following expressions:
cov(a0 , a1 ) u2 (a0 )
[x3 y] [x][x 3 y] [x 3 ][xy] [x2 y 2 ] [x2 ][x 3 ] [x][x 4 ] [x 2 y] [x 4 ][xy] [x 3 y][x 2 ]

[x3 y] [x 3 ][x2 y] [x2 ][x 3 y] [x2 y 2 ] [x2 ][x 4 ] [x 3 ][x 3 ] [x2 y] [x 3 ][x 3 y] [x 2 y][x 4 ]

(s6)
cov(a0 , a2 ) u (a0 )
2

[x3 y] [x2 ][xy] [x][x2 y] [x2 y 2 ] [x][x 3 ] [x2 ][x2 ] [x2 y] [x2 ][x2 y] [xy][x 3 ]

[x3 y] [x3 ][x2 y] [x2 ][x 3 y] [x2 y 2 ] [x2 ][x 4 ] [x 3 ][x 3 ] [x2 y] [x 3 ][x 3 y] [x 2 y][x 4 ]

(s7)
cov(a1 , a2 ) u (a1 )
2

[x3 y] N[x2 y] [x][xy] [x2 y 2 ] [x][x2 ] N[x 3 ] [xy] [x 3 ][xy] [x 2 y][x 2 ]

[x3 y] [x2 ][xy] N[x 3 y] [x2 y 2 ] N[x 4 ] [x 2 ][x 2 ] [xy] [x 2 ][x 3 y] [xy][x 4 ]

(s8)