Measurement 40 (2007) 338–342

www.elsevier.com/locate/measurement

Calibration of micropipettes: Test methods and

uncertainty analysis
a,*
E. Batista , L. Pinto a, E. Filipe a, A.M.H. van der Veen b

a
Instituto Português da Qualidade (IPQ), Rua António Gião, 2, Caparica, Portugal
b
Nederlands Meetinstituut (NMi), Schoemakerstraat 97, 2628 VK Delft, The Netherlands

Received 9 September 2005; accepted 23 May 2006

Available online 2 June 2006

Abstract

Micropipettes or piston pipettes are used to make the most of volume measurements in fields like health, chemistry,
biology, pharmacy and genetics. Laboratories must ensure that obtained results using these instruments are reliable, for
that is necessary to calibrate micropipettes using the correct method and estimate uncertainties. The objective of this work
is to present the best method to calibrate micropipettes according to the describe measurement conditions. A model for the
calculation of the measurement uncertainty is also described.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Micropipette; Uncertainty; Measurement; Calibration

1. Introduction in the final result. In order to reduce and identify

Pipetting in the microliter range is now a current
and necessary task for almost every field of chemis-
try. New dispensing systems allow experiments to be
simpler and more automated, but at the same time
new fields like genetics put heavier demands on
the reliability of the results. It is, therefore, impor-
tant to focus on calibration and uncertainty related
with this kind of equipment.
Volume measurement is critical in any analytic
laboratory, but in particular in highly sensitive tests
(that use a very small amount of liquid) where a
small mistake in pipetting may cause a large error
*
Corresponding author.
E-mail address: ebatista@mail.ipq.pt (E. Batista).
possible errors in liquid handling it is necessary to
calibrate micropipettes under similar conditions as
the ones used in the laboratory.
This paper has the main purpose to test and dis-
cuss several calibration procedures and find the best
method for calibrating micropipettes.
The methodology applied to evaluate all uncer-
tainty components and a model for evaluating the
measurement uncertainty is described.
2. Experimental methods
The principle of the operation of dispensing is
simple: a disposable tip, usually of polypropylene,
is attached to the piston pipette and, with the piston
at the aspiration lower limit, the tip is dipped in the
liquid. When moved to the aspiration upper limit the

0263-2241/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.measurement.2006.05.012
 E. Batista et al. / Measurement 40 (2007) 338–342
piston aspirates the liquid. To expel the liquid the
piston is slid between the volume defining marks.
The volume delivered by the micropipette is deter-
mined using a gravimetric method and water as the
calibration liquid at a temperature of (20 ± 0.5) °C.
The calibration of micropipettes was carried out
using four different gravimetric methods based on
different ISO standards.
In all four methods a suitable mass comparator is
needed for registering the mass delivered for a given
pipetted volume. The water and room temperature
must be carefully controlled during the measure-
ments.
The first method, the ‘‘normal’’ method, is based
on ISO 4787 [1] and ISO 8655 [2]. The micropi-
pette is wetted five times to reach equilibrium with
respect to humidity. Then the tip is changed and
wetted again. The weight of the dry beaker is deter-
mined, and so is the volume pipetted. The beaker is
weighted again and the mass obtained by difference
is a measure for the volume of the micropipette. The
beaker is cleaned after each volume additions and
this procedure is repeated ten times.
The second method is the ‘‘evaporation’’ method
based only in ISO 8655. The micropipette is wetted
five times and then the tip is changed and wetted
one time. The weight of the beaker with a 3 mm
water film is determined (m0). A stopwatch is started
and the testing volume is pipetted. The beaker is
weighted again (m1), then the test volume is pipetted
again and added to the beaker (m2). The mass
obtained by difference (m1  m0, m2  m1, . . . ,
m10  m9) is a measure for the volume of the micro-
pipette. The stopwatch is stopped after the tenth
measurement and the weighing vessel is left on the
balance for the time used for the ten measurements
and its mass is recorded (m11). The value determined
due to evaporation loss is added to each measure-
ment. To calculate the evaporation loss the follow-
ing formula is used: (m10  m11)/10.
The third method is the ‘‘without evaporation’’
method, is also based on the ISO 8655 but the evap-
oration loss is not taken in count. The micropipette is
wetted five times, and then the tip is changed and wet-
ted once more. The weight of the beaker with a 3 mm
water film is determined and the testing volume is
pipetted. The beaker is weighted again and the mass
obtained by difference is a measure for the volume of
the micropipette. This cycle is repeated ten times.
The last method is the ‘‘without evaporation
using oil’’ method. This method is very similar to
the third and is also based in ISO 8655. First of
339
all, the beaker must be filled with 5 mm film of
oil. Due to the fact that the oil does not evaporate
because it has a very high vapour pressure the bea-
ker can be weighted any time before the start of the
measurements. As in the others methods, the micro-
pipette is wetted five times and the tips are replaced
as described before. A new tip is wetted and filled
with the testing volume and delivered into the bea-
ker, with caution in order to ensure that the water
goes beneath the oil. It is important not to deliver
near the beaker walls. After delivering, the beaker
is weighted again and the mass obtained by differ-
ence is a measure for the volume of the micropi-
pette. The procedure is repeated ten times. Within
each measurement the mass of the beaker must be
registered.
Five tests were performed to each micropipette of
5 ll, 100 ll, 250 ll and 1000 ll in all four methods.
3. Evaluation of measurement uncertainty
The evaluation of the uncertainty of all four
methods follows what is commonly referred as
‘‘mainstream GUM’’ [3] and is similar to the one
described in the work of Batista et al. [4].
To determine the volume of a micropipette the
following formula (1) is used:
 
1 q
V 20 ¼ ðI L  I E Þ   1 A
qW  q A qB
 ½1  cðt  20Þ ð1Þ
where
V20 volume, at temperature of 20 °C, in ll;
IL weighting result of the recipient after the
pipetting in mg;
IE weighting result of the recipient before the
pipetting, in mg;
qW water density, at calibration temperature t,
in mg/ll;
qA air density, in mg/ll (0.0012 mg/ll);
qB mass pieces density (8.0 mg/ll);
c material cubic thermal expansion coeffi-
cient of the micropipette, in 1/°C;
t water temperature, in °C.
Defining 1/(qW  qA) as A, 1  qA/qB as B,
1  c(t  20) as C, and m as IL  IE, Eq. (1) can
be rewritten in a simpler way as
V 20 ¼ m  A  B  C
Some of the variables in (1) depend on others. A
complete overview is given in Fig. 1.
340 E. Batista et al. / Measurement 40 (2007) 338–342

t The uncertainty of the temperature is essentially

A evaluated in the same way: the calibration of the
P
thermometer and its repeatability are the main
H uncertainty components.
t twater ¼ t þ dtcal ð4Þ
ρ The evaluation of the uncertainty of the density
B
B P of water is more problematic. The ISO 4787 pro-
H
vides a table with water densities. For the calibra-
V(20) tion uncertainty of the thermometer (0.1 °C), the
t uncertainty of the water density must be obtained.
γ This is performed by interpolating the tabled values.
C
t
For the air density, the standard uncertainty is
estimated to be smaller than 0.0005 g/ml, which is
obtained from a review from Spieweck and Bettin
[5]. The standard uncertainty of the density of the
rpt.
mass pieces is obtained from the certificate of the
m res. comparator. The standard uncertainty of the expan-
sion coefficient is obtained from the last significant
cal.
number in the table described in the ASTM E542 [6].
Fig. 1. Structure of model.
For all variables the assumed distributions are
given. For the volume V20, the normal distribution
can be assumed. The results are summarized in
The expression for the standard uncertainty asso-
Table 1 for a micropipette of 5 ll.
ciated with V20 is developed from (1), using the law
of propagation of uncertainty. The resulting expres-
sion reads as: 4. Discussion of results
" 2  2
oV 20 oV 20 The described work had the main purpose of
uðV 20 Þ ¼ u2 ðmÞ þ u2 ðqW Þ
om oqW finding the suitable method to calibrate micropi-
 2  2 pettes according to laboratory requirements and
oV 20 oV 20 based on existing ISO standards.
þ 2
u ðqA Þ þ uðqB Þ2
oqA oqB Four air displacement micropipettes were tested.
 2  2 #12 Fixed volume micropipettes of 5 ll, 100 ll and
oV 20 2 oV 20 2
þ uðcÞ þ uðtÞ ð2Þ 250 ll and a variable micropipette of 1000 ll. The
oc ot results are shown in Figs. 2–5.
As it can be seen from the figures, the best repeat-
One of the most important uncertainty compo-
able method with smallest uncertainty for all micro-
nents is the uncertainty of the operator that is
pipettes in almost every test is the method ‘‘without
included in the standard deviation of the mean.
evaporation’’. This method is faster and more reli-
For variable micropipettes instead of doing all the
able because it does not have any awaiting time dur-
10 measurements with the same adjustment of the
ing the procedure. Furthermore, there is no volume
scale, the experiments were made with scale adjust-
correction necessary after the measurements, neither
ing within each measurement.
a cleaning step for the beaker with solvents to
remove the oil afterwards.
3.1. Evaluation of standard uncertainty of input
The ‘‘normal’’ method is the one with the lower
variables
volume results due to the evaporation loss caused
by the major evaporation surface when the liquid
The uncertainty of the mass is a combination of
touches the glass. It is also the slower method,
the uncertainty of the calibration of the comparator,
because it is necessary to clean the beaker after each
the repeatability of weighting and the resolution of
volume addition.
reading.
The volume correction in the ‘‘evaporation’’
m ¼ M þ dmres þ dmcal ð3Þ method can lead to results that are not correct, as
 E. Batista et al. / Measurement 40 (2007) 338–342 341

Table 1
Evaluation of uncertainty of a micropipette volume
Variable Xi Estimate xi Distribution Standard uncertainty u(xi) Sensitivity coefficient ci Contribution ui(y) (ll)
M (lg) 5.033 Normal 0.0088 1.00 0.0088
dmres (lg) 0 Rectangular 0.001 1.00 0.001
dmcal (lg) 0 Normal (k = 2) 0.005 1.00 0.005
t (°C) 19.5 Normal 0 0.00121 0
dtcal (°C) 0 Normal (k = 2) 0.05 0.00121 0.0000606
qW (lg/ll) 0.9983 Rectangular 0.0000116 5.06 0.0000588
qA (lg/ll) 0.0012 Rectangular 0.000000289 4.43 0.00000128
qB (lg/ll) 7.96 Rectangular 0.0346 0.000096 0.00000332
c (1/°C) 0.000240 Rectangular 0.00000289 2.52 0.00000729
V20 (ll) 5.05 U(V20), k = 2 0.02

5.5 255.0
5 μl micropipette 250 μl micropipette
5.4 254.0

5.3 253.0
5.2 252.0
Volume (μl)
Volume (μl)

5.1
251.0
5.0
250.0
4.9
249.0
4.8
248.0 With evaporation Without evaporation
4.7 Normal With Oil
With evaporation Without evaporation 247.0
4.6 Normal With oil 1 2 3 4 5
4.5 Test
1 2 3 4 5
Test
Fig. 4. Calibration of a 250 ll micropipette.
Fig. 2. Calibration of a 5 ll micropipette.

102.0 100 μl micropipette 1011.0 1000 μl micropipette

1008.0
101.5
1005.0
101.0
1002.0
Volume (μ l)
Volume (μ l)

100.5 999.0
100.0 996.0

99.5 993.0
990.0
99.0
987.0 With evaporation Without evaporation
98.5 With evaporation Without evaportion Normal With oil
Normal With Oil
984.0
98.0 1 2 3 4 5
1 2 3 4 5 Test
Test
Fig. 5. Calibration of a 1000 ll micropipette.
Fig. 3. Calibration of a 100 ll micropipette.
surements. This can decrease the reliability of the
the major part of the evaporation occurs between method.
additions, and not between the additions and the The use of oil was thought to prevent evapora-
balance stability, so when a correction is made the tion of water while the calibration is undergoing,
measured volume is increased in a incorrect way in order to have better results but after evaluating
when compared with the real value. The correction the results, it is clear that this method does not have
value is a mean value added to each of the ten mea- any benefit over the ‘‘without evaporation’’ method.
342 E. Batista et al. / Measurement 40 (2007) 338–342
In the case of the 5 ll micropipette, the results are
even worse. There is a great difficulty in delivering
a small quantity of liquid inside the oil without
touching the oil or the wall of the beaker, which
leads to evaporation loss and a reduction in the
volume.
5. Concluding remarks
Several tests were preformed in order to deter-
mine the best method for calibration of micropi-
pettes based on the existing ISO standards. After
the evaluation of results and uncertainty it can be
concluded that the ‘‘without evaporation’’ method
is the most efficient.
The described model for the calculation of the
measurement uncertainty was described and it
proved to be completely suitable for this type of
gravimetric method.
The presented uncertainty analysis can also be
used for other piston operating volumetric equip-
ment and laboratory volumetric glassware if the
same gravimetric method is applied.
References
[1] ISO 4787, 1984, Laboratory glassware – Volumetric glass-
ware – Methods for use and testing of capacity.
[2] ISO 8655-1/2/6, 2002, Piston-operated volumetric apparatus.
[3] BIPM, IEC, IFCC, ISO, IUPAC, IUPAP, OIML, Guide to
the expression of uncertainty in measurement (GUM),
Geneva, 1995.
[4] E. Batista, A. Sabrosa, M. Ferreira, I. Castanheira, A.M.H.
Van der Veen, ‘‘Uncertainty calculation in the calibration of
volumetric laboratory glassware’’, in Metrologia 2000, Brasil,
2000.
[5] F. Spieweck, H. Bettin, Review: Solid Liquid density deter-
mination, Tm – Technisches Messen 59 (1992) 237–292.
[6] ASTM E542, Standard Practice for Calibration of Laboratory
Volumetric Apparatus, 2000.