1. business and industrial
2. energy
3. oil

CLIN. CHEM. 21/13,
1907-1917(1975)
Factors Influencing Evaporation from Sample Cups,
and Assessment of Their Effect on Analytical Error
C. A. Burtis, J. M. Begovich, and J. S. Watson
We studied sample evaporationanditseffect on analytical error. Several factors influencing evaporative loss
have been identified and measured: environmental, instrumental, and operational factors, and the chemical
and physical properties of the sample and its container.
Such losses from several different types of sample
cups have been measured, either chemically or grayimetrically, and compared with those calculated by
using a model that allows evaporative loss from a cup
of known geometry to be predicted under various environmental conditions. We discuss some steps that may
be taken to minimize evaporative loss and give an example to demonstrate that analytical error from this
source can be decreased to a routine 1-2% or less by
selecting a particular cup design.
ute
Addftlonal Keyphrases:
Instrumentation
error #{149}sample-cup
ing evaporative loss
chemistry
microliter samples
#{149} analytical
geometry
#{149}predicting and prevent#{149}
centrifugal analyzer #{149}pediatric
Various
newly developed
instruments
offer real
potential
advantages
for automated
clinical analyses. Because
increased
sensitivity
is typically
a feaand
ture of these instruments,
many factors that heretofore made relatively insignificant
contributions
to the
total analytical
error should now be examined and
evaluated.
One such factor is evaporative
ponents
poured
loss of volatile
com-
or liquid1 from a sample
after it has been
into its sample cup. This factor can contrib-
Oak Ridge National
Laboratory,
Oak Ridge, Tenn. 37830.
ORNL
is operated
for the Energy
Research
and Development
Administration
by the Union Carbide Corporation.
1 A volatile
solute can escape from the solvent, thereby
decreasing its concentration
in the sample, or the solvent (usually water)
can evaporate,
thereby
increasing
the concentration
of the nonvolatile solutes in the sample, or both. Although
this paper deals primarily with the second effect of evaporation,
most of the observations and discussion
pertain to both.
Received Aug. 22, 1975; accepted
Sept. 18, 1975.
a significant
analytical
error,
especially
when
small volumes are involved. Surprisingly,
the clinical
literature
hardly discusses this problem (1-3). We
have quantitatively
studied the process of evaporation and how it affects analytical error in the clinical
laboratory.
Here, we (a) present
the experimental
results of this study, (b) discuss several factors that influence evaporative
loss, (c) present a model that can
be used to estimate
the quantitative
and relative
evaporative
loss from a given sample cup under various environmental
conditions,
and (a) describe preventive or corrective measures that can be taken to
minimize sample evaporation during analysis.
Materials and Methods
The miniature Centrifugal
Fast Analyzer, the version of the centrifugal analyzer that we used in these
studies, has been previously described (4, 5), as has
its operation
(4).
Methodology
In the quantitative
samples were analyzed
okinase procedure (6).
Pack” kit; Calbiochem,
reconstituted
before
studies, control and aqueous
for glucose by a modified hexReagents for the assay (“StatSan Diego, Calif. 92112) were
analysis
by dissolving
the
con-
tents of one vial in 2 ml of water, which was then
added to and mixed with the contents of the second
vial. Lyophilized
control area (Monitrol
I and II;
Dade, Miami, Fla. 33152) were reconstituted
according to the manufacturer’s
recommendations.
Aqueous
standard
solutions
of NBS
terial glucose were prepared
ed distilled water.
Quantitative
measurement
We measured
Standard
Reference
Main benzoic acid-saturatof evaporative
the effects of evaporation
CUNICAL
CHEMISTRY,
losses.
both chemi-
Vol. 21, No. 13, 1975
1907
Table 1. Sample Cups Used in Evaporation Studies
Dimensions
Code
no,
Source
1
Laboratory
Area, cm2
cmb
0.357
0.1001
0.380
0.391
0.1134
0.1207
0.422
0.439
0.1399
0.1513
-300
EP-1500
0.439
0.91
0.1513
0.6503
2.18
1.75
1.35
0.97
0.60
0.27
0.39
UC-100
0.69
0.3739
0.98
.200
400
0.69
1.00
0.3739
0.7853
0.73
0.30
NI-bOO
1.21
1.1499
0.72
TC-500
0.70
0.3848
0.48
SM-bOO
1.25
1.25
1.25
1.25
1.25
0.45
1.227 1
1.2271
1.2271
1.2271
1.2271
0.1590
3.43
1.28
-100
-150
-200
Products
Ann Arbor, Mich 48106
-250
2
Eppendorf Tube
Brinkman Instruments, Inc.
Westbury, N.Y.
CentrifiChem Sample Cup (0.25 ml)
3
Union Carbide Corp.
Tarrytown, N.Y.
4
NilabSample
Cup
National Instrument Laboratory Co.
Silver Spring, Md.
0.5-mI AutoAnalyzer Cup
Technicon Corp.
Tarrytown, N.Y.
5-mI SMAC Tube
Technicon Corp.
5
6
7
-2000
-3000
4000
-5000
AB-50
ABS-i 00 Micro-Cup
Depth,
cm
MT-50
Micro-Tube
Hruden
Diameter,
Code and
vol testede
0.58
0.24
0.40
Abbott Diagnostic Division
South Pasadena,Calif.
acup code followed
bDjstance
from
by volume (&l) tested.
top of cup to surface of liquid.
and gravimetrically.
In the chemical studies, the
Centrifugal Fast Analyzer
was used to analyze either control sera or an aqueous standard solution for glucose (6) as a function of time under several different experimental
conditions. In the gravimetnc studies, various volumes of water or serum were
dispensed into several different
types of tared sample
cups, and the weight loss of each was determined
as a
function
of time by weighing the cup and its contents
at 1-h intervals for as long as 8 h. We used an analytical balance that had a repeatability
of 0.1 mg and
that had been
calibrated
against
NBS
certified
weights. Table 1 lists the identification,
source, and
critical dimensions
of the various cups we used.
cally
miniature
Estimation
of Evaporative
over a specific
period
of time
from
a cup
of known geometry (Figure 1). The theoretical
basis
for the model was diffusion through a stagnant gas
film (7). The following assumptions
are made: (a)
water diffuses upward through a stagnant
film of air;
(b) the mole fraction of water at the gas/liquid
inter1908
CLINICALCFEMSTRY,
N
HZ-ZjI
Vol. 21, No. 13, 1975
R PT. (DH2o)
(Z2
Z1)
-
.
-
(1)
p1’
-
where
NH2OIz=zi
Losses
Results of the evaporative studies led to the development of a mathematical
model that allows one to
estimate the quantity of a particular
liquid lost by
evaporation
face can be calculated
from the vapor pressure of
water at room temperature;
(c) the mole fraction of
water in air at the top of the cup is the same as in the
laboratory
atmosphere;
(d) the air/water
mixture is
an ideal gas; (e) the solubility of air in water is negligible; (f) the entire system is at constant temperature
and pressure; and (g) that part of the cup that is
above the liquid level can be approximated
by a cylinder with the same diameter as the liquid interface.
The applicable equation is, then:2
=
rate of water evaporation
face, mol/cm2
=
gas/liquid
2Variables
at the inter-
.
interface,
and constants
cm;
included
in the
model
are given
metric units, because most current handbooka list them
In SI units, the following units would be substituted:
pressure,
gas constant
N/rn2 (1atm
=
=
101325N/rn2)
8.31433X 106
K
All other units are as listed in text.
cm3
-
m2
N
.
.
mol
this
in
way.
(Z2
Z1)
-
(Z2
The term in brackets is a constant (C) for a particular set of conditions, and can be readily calculated.
Integrating
Equation 5 between the initial height,
h1, and the final height after evaporation,
h1, gives
Zt)
-
ZI
(hf)2-
(h1)2
C.t.
(6)
From the geometry of the cup, one may calculate the
volume loss corresponding
to the change in height of
the stagnant
TIME
TIME
#{149}
Fig. 1. Cross-sectional
view of a typical
#{149}
t
sample
which evaporationhas occurred during the time
cup from
air mass
(h1
-
h1) for a given time,
tions It should be noted, however,
that losses
to convection
are not included
in the model,
tf
t.
This model can then be used to predict evaporative
loss from a given cup containing a specific volume of
liquid under a selected set of environmental
condiowing
Results and Discussion
=
p
Analytical
top of cup, cm;
The primary effect of evaporation
is on analytical
accuracy, because the concentration
of dissolved solutes in a sample will increase as solvent evaporation
continues,
as is demonstrated
in Figure 2, which
shows the glucose concentrations
of duplicate
ali-
atmospheric
pressure,
atm;
DH2Oair = diffusivity
of water in air, cm2/s;
R = gas constant
= 82.05
atm
cm3/mol.
K;
T = temperature,
K;
=
.
Pain
p
=
Pi-i2o(pH2o
-
=
vapor
pressure
of water);
p
(RH)
(pH2O)/100
(RH, relative humidity, %).
With the incorporation
of the appropriate
terms,
Equation 1 can be expressed in terms of the volume
lost at the interface, as follows:
Pair2
dv
=
-
cm3
at
.
/
s
mol
(NH2ZZ1,
=
-,
quots
of two control
__-)
The expression
=
(_J__,
-
.
(Ar, cm2).
g
PH2O
R
.
T
A1 .ln-siu,
.
h
(3)
C
rate of water evaporation
-#{231},______---
density
weight of water (g/mol);
of water
=
A1
area of sample
(g/cm3);
cup at the liquid/air
#{149}
-4.0
#{149}
-6.0
#{149}
-8.0
#{149}
0
1I
WITH
2I
3I
4
I
I
2
I
3
I
4
I
5
OIL
30
25
Z1 (cm), height of stagnant
air mass (i.e.,
the vertical distance
from the surface of the air/
MONITROL
-
20
45
liquid interface to the top of the sample cup). In
addition,
the relationship
between the rate of
evaporation,
dv/dt, and the rate of change in the
height of the stagnant
as follows:
air mass
E
I
dh
3 into Equation
dh = [p
dt
I
for dv/dt
from Equation
MH2O’ (DH2Oair)
PH2O
-
-20
0
-
I
I
TIME
4, we have:
.
-5
-40
-45 -
(4)
the expression
-O------0
Ui
z
dv
5/
10
C
can be expressed
U
Substituting
-2.0
interface,
cm2;
=
OIL
-0
:.:
at the interface,
cm3/s;
PH2O
WITHOUT
Pain
0.0
molecular
I
SO
where
=
To
40.0
MONITROL
(DH2Oair)
.
PH2O
MH2O =
of time.
(2)
then becomes:
dt
=
as a function
2
mol
dv/dt
sera
minimize evaporation,
75 zl of silicone oil (Sera-Seal;
Abbott
Diagnostics
Div., South Pasadena,
Calif.
91030) was layered over one of the aliquots. The results very clearly indicated that evaporation
had occurred in the uncovered samples, because their glu-
cmOs
\
(MHZO
h
Effects of Evaporation
R T
-
.
ln P21
Pain]
.
I
h
(5)
(HOURSI
FIg. 2. Analytical effect of sample evaporation (change In solute concn) as a function of time, with and without oil coverIng
sample surface
CLINICAL CHEMISTRY,
Vol. 21 No. 13, 1975
1909
Table
-
MICR010BE CUP
2. Factors Affecting Evaporative Loss
30
I. Time
SAMPLE
VOLUME . 200 pA
SAMPLE
DEPTH . 5
em
SAMPLE
AREA.
0.1399
en,2
BAROMETRIC
PRESSURE - 740
TIME.
6 HOURS
II. Environmental
factors
A. Ambient temperature
50
mm
7
Hg
.
.-
B. Relativehumidity
C. Air
..-
20
flow
Ill. Sample cup
properties
-
thus
time
is a very
Loss
critical
factor
in deter-
mining the magnitude
of evaporative
loss. From the
moment the sample is withdrawn
from the patient
until
the
final
quantitative
measurement
is made,
evaporation
can occur throughout
the analytical process and cause an appreciable
analytical
error. For
example, the data that will be subsequently
presented and discussed will demonstrate
that this error can
be as high as 40 to 50% over an 8-h period.
Environmental
factors.
As would
be expected,
en-
vironmental
factors have a pronounced
influence on
the evaporative
loss of a liquid from a sample cup.
The mathematical
model discussed earlier (Equation
3) can be used to demonstrate
the effect of environmental factors on evaporative
loss. Figure 3 shows
evaporative
loss as a function of ambient temperature and relative humidity; as may be seen, it is directly proportional
to ambient temperature
and inversely proportional
to relative humidity. The strong
influence that these two environmental
factors exert
on evaporation
may be partially
responsible
for the
seasonal
fluctuations
some laboratories
experience
in
their quality-control
data.
A third environmental
factor that potentially
can
cause the greatest
evaporative
loss is the magnitude
of air flow within
1910
the laboratory.
-
.-,-
.-
-.----.
50
_.__.
IS
-- -
20
22
24
I
70
26
28
30
80
32
90
Fig. 3. Effect of ambient temperature and relative humidity as
a function of time on evaporative
loss from the Microtube
sample cup
technique
Many factors influence the magnitude of evaporative loss; several of the more obvious ones are listed
in Table 2. By using either actual measurements
of
the evaporative
loss or the experimental
model to estimate it, we have attempted
to quantitate
the effects
of the more important factors.
Time. Evaporation
is a time-dependent
phenomeand
6
60
cose concentrations
increased as a function of time
while the glucose concentrations
of the covered samples remained nearly constant throughout
the 5-h experiment.
non,
#{149}.-
I
C
B. Sample cup environment
C. Sample protection
Evaporative
-.--
.-.
V. Instrumental
factors
A. Mode of operation
Factors Affecting
OO
.-
A. Design
B. Volume capacity
IV. Sample properties
A. Nature of solvent
B. Solutes
C. Surface effects
VI. Operator
.-‘
As the air flow in-
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
0.4
ee
0.9 cm
1.2cm
1.2cm
uu
-0.2
,nI
2-1.5 eel
3-025m1
1.25en,
4-I
085
U
5-0.5 ml
Ol
cm
7-0050
ml
6-10-5,0 eel
Fig. 4. Cross-sectional
views and diameters
cups used in evaporation studies
Numbers correspond
to those in Table 1
for seven sample
creases
across the surface
of a liquid/air
interface,
vapor is swept with it, providing
a driving force for
evaporation.
For this reason, sample covers are provided with most instruments.
However,
as will be discussed later, appreciable
evaporative
loss can occur
even when such covers are used.
Sample
cup properties.
Although
the evaporation
of liquid from a sample
cup is basically
a function
of
the interaction
of several environmental
factors,
its
magnitude
is influenced
by the geometry,
design, and
physical properties
of the cup.
The effect of sample-cup geometry was determined
by placing various volumes of distilled
water into various types of tared sample
cups and then weighing
them
at specified
times
to determine
evaporative
losses. Between
weighings,
unless otherwise
specified,
the sample
cups were covered.
The cups that were
3.
Table
Cup
Environmental
Relative
humidity
Ambient
temperature
(#{176}C)
a
Conditions
and.
Estted.
Ba rometric
pressure
(mmlig)
()
vs
Measured
Evaporation
Vapor
Cups
Relative
loss
(%/h)
3.7
5.5
11.0
16.0
16.2
1.1
0.5
2.3
14.0
0.5
0.8
1.1
1.5
2.2
3.2
1.0
0.8
0.7
0.8
0.9
1.1
17.2
22.7
2.3
2.9
3.7
5.3
8.2
0.3
0.2
0.2
0.2
rate
Predicted
(mug)
Sample
Measured
Evaporation
pressurec
Diffusivityb
(2/s)
for Several
Rates
(l/h)
AB-50
25.14
53
736.6
0,2660
214.2914
EP-1500
22.3
38
752.9
0.25147
20.11s9
t.-5O
-ioo
-150
-200
-250
-300
25.1
56
736.6
0.26142
23.977
NI-bOO
22.3
38
752.9
0.25147
20.1149
SM-bOO
-2000
-3000
23.9
63
7146.6
0.2593
22.2143
TC-500
22.3
38
752.9
0.25147
20.1149
12.9
12.9
2.6
UC-l00
-200
22.3
36
71114.5
o.z6oi
20.1149
14.14
5.8
8.0
11.5
14.5
5.9
8.2
11.1
14.9
3,0
2.8
3.1
22.3
38
752.9
0.25147
20,1149
114.6
149
0.6
0.9
1.14
2.7
3.7
5.14
9,7
-18ooo
-300
..1iyj
uc_3ood
aSee
1 for
Table
bOiffusivity
identification
calculated
the
D
where
T
=
ambient
Reference
cValues
McGraw
dsample
Hill,
cups
from
K;
and
Perry,
R.
uncovered
1973,
T
(0.220)(-)
=
W. P.,
New York,
were
of
the
T0
=
cups.
and
p
(-),
273 K; P0=
Braltain,
H.,
1.75
W. H.,
Chilton,
760
nHg;
Interdiffusion
C.
H.,
Chemical
P
pressure,
barometric
of
Gases
Engineers
and
mm}fg (1 nisHg
=
133 Pa).
Vapors.
Handbook,
5th
ed.,
p 3_145.
during
the
entire
experiment.
studied
and their critical dimensions
and volumes are
listed in Table 1. Cross-sectional
views of the cups
are illustrated
in Figure 4.
The evaporation
sample
individual
relationship
temperature,
Baynton,
obtained
sources
and
from
--
rates obtained
for these cups and
the environmental
conditions
under which they were
measured
are summarized
in Table 3. In general, the
rate of evaporation
was proportional
to the surface
area of the liquid and inversely
proportional
to the
vertical
height of the stagnant
air mass which is located between
the liquid surface and the top of the
sample cup.
The experiments
summarized
in Table 3 indicated
that evaporative
loss depends
on the type of sample
cup used. Because
a gravimetric
study is tiresome
and time-consuming
and there are literally
hundreds
of different
types of sample cups, it was desirable
to
develop a model that could be used to predict evaporative loss for any cup of known geometry
under specific environmental
conditions.
Consequently,
we developed the model described
in the section on Materials and Methods.
The predictive
capabilities
of the model were eval-
uated
by comparing
gravimetrically
calculated
agreement
the evaporative
losses
for the seven cups studied
measured
with those
by the model (Table
3). In general,
the
was good, indicating
that the model is ap-
proximately
correct. Note that the discrepancies
encountered always occurred when the liquid level was
near the top of the cup. Under these conditions, differences between the predicted and measured values
can be attributed
to convective air currents and measurement errors. When the liquid level is near the top
of the cup, convective air currents would be expected
to have a more important
effect. Since the model is
based
on diffusion
of a vapor
through
a stagnant
air
mass, air convection would be expected to decrease
the predictive power of the model, which is consistent
with
the
data
obtained
in our
study.
In addition,
when the liquid level is near the top of the cup, it is
more difficult
to measure
the depth of the stagnant
air mass; hence the predictive
value of the model is
more subject to error and uncertainties
under these
conditions.
The results
of these experiments
and those preCUNICAL CHEMISTRY, Vol. 21, No. 13, 1975
1911
dicted by the model indicated
a direct relationship
between the volume of liquid placed in a sample cup
of a particular
design and the absolute quantity of
liquid lost by evaporation.
For example, when the
evaporation
data obtained from the UC cup are plotted as a function
of time
5), the largest quantitative
400-gil sample,
with
and sample volume (Figure
loss was observed
with the
lesser
losses
observed
for the
300-, 200-, and l00-jl
samples.
Similar
results were
also obtained
with the various volumes
placed in the
SMAC and MT sample cups (Table 3).
When relating
evaporative
loss in terms of its effect on analytical error, it can be demonstrated
that
it is proportional
to the relative
loss of liquid from
the sample. For example,
the initial concentration
of
a given solute can be expressed
as:
on
0
U-
(7)
C0=3-,
V0
where Co = initial concentration
of solute; q = quantity of substance;
Vo = initial volume of sample.
After a finite time (t) in which an evaporative
volume
loss of Ve has occurred, the concentration
of the so-
lute then becomes:
TIME
(8)
Cf=VOVe
Relating the initial
be shown that:
to the final concentration,
Co
Cf
V0
-
V0
-
()
Ve
greatest
percentage
and 300-id samples.
loss, followed
by the 400-, 200-,
Thus one would expect the analytical error caused by evaporation
to be greater
in
the 100-al sample than in the 400-, 200-, or
samples, an expectation
confirmed
experimentally
by
analyzing
various volumes
of an aqueous
glucose solution for glucose content
as a function
of time. The
data obtained in this experiment
(Table 4) are con-
sistent with the above observation
and indicate a
greater analytical error owing to evaporation
for the
lO0-il sample; the smallest error was observed for the
3OO-il sample. Therefore,
when the problem of evaporation is considered, one should relate the quantitative evaporative
loss to the starting volume, because
the analytical error is proportional
to the relative loss
instead of the quantitative
loss. Consequently,
the
relative evaporative
losses determined
for the various
sample cups are also listed in Table 3.
The experiments
discussed
above demonstrated
that evaporative
loss should be expressed
on a percentage or fractional
basis in order to relate it to analytical
error. This fractional
loss can also be exCLINICAL
CHEMISTRY,Vol. 21, No. 13, 1975
of cup volume and time
it can
Thus the change in concentration
of the substance is
proportional
to the relative loss of liquid from the example.
When the data shown in Figure 5 are replotted as a
relative evaporative
loss vs. time and sample volume
(Figure 6), the l0O-tl sample is found to have the
1912
(HOURS)
Fig. 5. Evaporative loss as a function
for the Union Carbide sample cup
pressed
sample
in terms of the change in depth of the liquid
as evaporation
occurs. For example,
for a cup
of constant diameter (Figure 1) the fractional loss (F)
can be expressed
as:
(10)
where F
liquid;3
ration;
shown
model,
fraction
of liquid lost; h0
= initial
depth of
of stagnant
air mass after evapoh1 = initial height of stagnant
air mass, As
earlier
in Equation
6 of the evaporation
the final height of the stagnant
air mass is
=
hf = height
given, after rearrangement,
by the equation:
(11)
hf=v/h12+2.C.t=hl\/1+2hC2t
The square root term for this expression can be expanded into a power series, and for small evaporation
losses (e.g., when 2 C. ti/u2 << 1) the final height of
the stagnant air mass is approximately:
.
(12)
hf=hl(1+-+...)
The initial depth of the liquid
calculated
from the relationship:
(h0)
can be
measured
directly
or
4. V0
h0
=
-.,
where
Vo = initial
volume of sample, and
d = diameter
of the cylindrical
section of the cup.
It should also be noted that this expression
can be used
an effective liquid depth should the cup contain a conical
to obtain
bottom.
36
/
32
00
H1
9 300
#{149}
400
#{149}
200
H1
H’
o
28
24
S
on
I-
-
/,
0
-
air mass
spheric
parameters
over it as well as the atmo-
located
included
in the constant
C.
The
height of the stagnant
air mass can also be related
the depth of liquid by the expression:
to
h=h0+h1,
(15)
where h = the
rate of fractional
losses) is:
-
20
16
stagnant
R
-
dF
dt
-
(h
height of the sample
cup. Then the
liquid loss, R, for short times (small
C
ho)
-
p
.
-
MH2O.
-ho-pH2o.R.T
(h-h0)
h
(DH2O_air)
ln2.
(16)
Pain
#{182}2
-
In addition
to predicting
the evaporation
rate for a
given cup, Equation
16 also suggests
an optimum
depth for filling the cup with liquid to reduce the effects of evaporation.
The rate of fractional
evaporation loss, R, is a function
of h0 and is a minimum
when the denominator
of Equation
16 is a maximum.
This occurs when:
o,’II
0
I
2
3
4
5
6
7
d[(h
TIME (HOURS)
FIg. 6. Relative evaporative loss as a function
and time for the Union Carbide sample cup
-h0).ho]
0h
-2h0;
(17)
dh0
of cup volume
e.g., when
h
Table
14.
Effect
Error
of
Sample
in Glucose
Cup Volume
on the
Measureisents5
Sample
100 H1
due
4goitude
to
Evaporation
(%)
Thus, to minimize
the fractional
evaporative
loss
from a sample cup, the cup initially
should be only
half full. This somewhat
surprising
relationship
has
been confirmed
experimentally,
as shown by the data
in Table 3. For example, when the fractional
evapo-
0.0
rative losses obtained
cuge volume
200 Hl
4
300
Error
0
1911
0.0
0.9
1970
3.1
1936
1.5
1928
1.1
19140
1.9
1.7
20110
6.8
1982
3.9
1972
3.14
1990
14.5
2023
6.1
2055
6.0
2187
assay
2176
with
blank
esmple
-
30.O’C;
Substituting
the fractional
2 ci;
0.0
114.1
measured
volume
-
1907
7.9
2058
25.9
Error
0.0
1907
concentration
temperature
(%)
(seg/I)
lIe.14
21105
aGle
sample
(%)
Cone.
(mg/I)
the
reaction
reaction
time
(%)
2080
under
6 sin;
-
of
7.9
conditions:
wavelength
assay
=
3140 Os;
equilibrium,
correction.
this expression
for hf into
loss of liquid becomes:
h1 (i+---)
C.t
Equation
10,
-h1
(13)
which
reduces
from the MT sample cups were
plotted
as a function
of cup capacity
(Figure 7), the
minimum
loss occurred
when the cups were approximately half full. Similar curves were also obtained
for
the SMAC and UC cups. The data also suggest that
for a given sample cup a useful volumetric
range to
minimize
evaporation
would be from 25 to 75% of the
10.14
2102
analytical
122 4;
type
Error
19014
9.1
following
volume
-
Cone.
(mg/I)
4
Cone.
(mg/i)
3.5
Error
1400
Time
(h)
Cone.
(18)
=
of Analytical
to
cup capacity.
Sample
volumes
below or above
this
range would be more severely
affected
by evaporation.
Sample
properties.
The chemical and physical
properties
of a sample
will affect evaporative
loss.
For example,
the properties
of the solvent
in which
the solutes of the sample are dissolved
or suspended
are important
in determining
the magnitude
of the
loss. In biological samples, this solvent is water;
therefore, evaporation
from such samples is a function of the physical properties
of water, including
vapor
pressure,
diffusivity,
density,
molecular
weight,
etc. When a liquid other than water is used as the solF
Note that
liquid
given
depth
Equation
(14)
C.t
h1.h0
14 predicts
vent
(e.g., an organic
cess), its physical
that the fraction
of
loss will increase
linearly with time, and for a
cup will depend
on the product
of the initial
of the liquid in the cup and the height of the
solvent
properties
from
an extractive
will be different
pro-
and the
rate of evaporation
will be affected
accordingly.
In
most cases, the organic
solvent
will have a greater
vapor pressure
than water and a greater
evaporative
loss.
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
1913
2.0
#{149}
CALCULATED
#{149}
ExPERNTAI.
DATA
DATA
1.8
1.6
1.4
1.2
2
on
a
1.0
0
-J
U428
0.6
422
20
40
60
80
100
% OP CUP CAPACITY
3
Fig. 7. Relative evaporative loss as a function of sample cup
capacity, for the Microtube sample cup
The inter- and intramolecular
forces exerted
between the solutes and solvent of a liquid sample will
also affect evaporative
loss. For example,
it has been
reported
that evaporation
of water was more rapid
from buffered
solutions
of ovalbumin
or hemoglobin
than from solutions
of various surface-active
agents
or from buffer alone (8). In a gravimetric
study of the
comparative
evaporative
losses of water versus serum
samples
as a function
of volume size, we found the
evaporative
loss from serum
samples
to be consistently higher than those from water samples
(Figure
8).
There
is also an interaction
between
the surface
properties
of the sample and the material
used to fabricate
the sample cup. If the surface of the cup is
wet by the sample, the deeper meniscus
will provide
additional
surface
for evaporation.
If the surface
of
the cup is nonwettable,
the sample will have a shallow meniscus
that will result in relatively
less evaporative loss than from a similar cup fabricated
from a
wettable
material.
Thus, one means by which evaporative loss can be decreased
is to use cups made of
nonwettable
material.
Instrumental
factors.
Several
instrumental
factors, for example,
the type of instrument
used and
the method
of operation,
influence
the analytical
error caused by evaporation.
In a continuous-flow
or
discrete-type
analyzer
that is electronically
calibrated against the chemical
responses
obtained
from reference standards,
evaporative
loss may not be a problem, especially
if the reference
and patients’
samples
1914
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
4
TIME (HOURS)
Fig. 8. Evaporative loss as a function of cup volume and time,
for both serum and water samples
are treated
exactly the same. Under these conditions,
as evaporation
occurs, a similar volume of liquid is
lost from each; thus the analytical
error caused by
evaporation
would be masked because of the ratioing
mode in which the machine
is calibrated.
However,
one should be aware that, because
evaporation
will
occur at different
rates for serum and aqueous
samples (see Figure 8), evaporation
may cause analytical
problems
when aqueous
solutions
of standards
are
used for calibration
and sera obtained
from patients
are the samples
being analyzed.
In addition,
to minimize the effects of evaporation,
similar
volumes
of
standards
and patient
samples
should be placed into
the sample cups.
The instruments
in which evaporation
can potentially cause the greatest
analytical
error are the discrete-type
analyzers
that are operated
in the batch
mode, offer microvolume
capabilities,
and use stoichiometric
factors
instead
of chemical
calibration
factors.
For example,
with these instruments,
small
volumes of samples (usually 500 jzl or less) are poured
into cups that are subsequently
placed
into a carousel. During the course of the day, aliquots
of sample are repetitively
sampled
from these cups as the
technologist
performs
several
different
chemical
analyses on them, sequentially
and batchwise.
Hence,
the samples
may remain
in the cups for as long as
several
hours.
Under
these conditions,
progressive
evaporation
of solvent
from the small volumes
involved will significantly
increase the concentration
of
the dissolved
analytes,
especially
if preventive
measures are not taken.
Previously
we have discussed
the role that environmental
factors have on the magnitude
of evaporative
loss. The manner
in which an instrument
maintains
temperature
control
can also influence
evaporative
loss if it changes the immediate
or proximal
environment of the samples.
For example,
evaporative
loss
will be increased
in a machine
that maintains
the
samples
at an elevated
temperature
such as 30 or 37
#{176}C.
In addition,
if an open water bath is used to
maintain
temperature,
it may increase
the relative
humidity
and thus decrease evaporative
loss.
A fourth
factor is governing
the evaporative
loss
that might be expected
from a particular
instrument
is the technique
or means used to prevent
or minimize evaporation.
In most instruments,
this consists
of a cover that fits over the sample carousel and
thereby
protects
the surface of the sample. However,
one should
be aware that appreciable
evaporative
loss can occur even if the sample cup is covered and
convective
air currents
are eliminated.
This loss is
predicted
by the evaporation
model and was confirmed experimentally
by measuring
the evaporative
loss from both covered and uncovered
UC cups that
contained
3OO-tl aliquots
of water. In this study, the
covered and uncovered
sample cups had evaporative
losses of 2.8% and 4.9% per hour, respectively
(Table
3). Thus evaporative
loss was retarded
by covering
the sample,
but the loss even from the covered
cup
was appreciable.
Technique.
Another
factor
that
influences
the
magnitude
of evaporative
loss is operator
technique.
For the most part, technique
involves the use of common sense and good laboratory
practice.
For example, if the system requires
sample covers, the operator should make a conscientious
effort to use them,
even though it may delay his work schedule slightly.
In addition,
he should not place a loaded sample car-
ousel in an area where there is a draft or in a sunny
location near a window. In general, evaporative
loss
attributable
to technique
can be minimized
if the operator is aware of the problem
and exercises
care in
his work.
Minimizing Evaporative
Loss
These studies show
ous analytical
problem
that evaporative
loss is a seriand must be minimized.
Several solutions to achieve this goal are possible:
1. Control and maintenance
of the laboratory
environment.
The three factors that may be controlled
to minimize
evaporative
loss are temperature,
rela-
tive humidity,
and air flow. As in most situations,
a
compromise
must be made, since the environmental
conditions
that would result in decreased
evaporative
loss (i.e., low ambient
temperature,
high relative
humidity, and decreased
air flow) would be very uncomfortable
for the operator
and might also result in op-
erating
vious,
problems
then,
that
with
the system.
a compromise
must
It becomes
ob-
be made
be-
tween
comfort
and
minimized
evaporative
loss,
which, under these conditions
might mean an airconditioned
laboratory
having an ambient
temperature of 21 #{176}C
(70 #{176}F)
and 50% relative humidity.
If fluctuations
in the laboratory
environment
are a
recognized
problem,
one possible solution4
is the use
of a chamber
in which the samples
are maintained
at
a reduced
temperature
and at an increased
relative
humidity
during the time they are not actively
in the
loading process.
However,
care should be taken that
condensate
collecting
on the walls and ceiling of the
chamber does not fall into the sample cups and dilute
their contents. A simple alternative
to the use of an
environmental
chamber
would be to cover and place
the sample
carousel
in a refrigerator
between
loadings.
2. Covering
the sample.
Theoretically,
evaporation can be minimized
by keeping the specimen
stoppered and the sample
covered
at all times and by
minimizing
the volume of the stagnant
air mass located between
the liquid level of the sample and the
cover. However,
since aliquots
of the specimen
must
be sampled
for analysis,
this is often not a practical
solution
unless
a means
of sampling
through
the
cover is available.
Two options are available
for this:
(a) the cover would be of a rigid, semirigid,
or permeable membrane
that is punctured
by the sampling
probe; or (b) the cover could be an immiscible
liquid
such as a silicone
fluid layered
over the top of the
sample and through
which the probe can easily pass.
Both
of these
options
have disadvantages;
the first
requires
a strong and inflexible
probe, the availability of which is limited,
especially
for 1- to 10-id volumes of sample, while the second risks contaminating
the sample
probe with the silicone
fluid. Also, as
shown in Figure 9, care must be taken to ensure that
an adequate
volume of silicone fluid is used; otherwiser part of the surface
of the sample
will be exposed and evaporative
loss will still occur.
3. Sample
cup selection.
As demonstrated
earlier
(Table 3), evaporative
loss occurs at different
rates
from cups having different
cross-sectional
designs. In
general,
evaporative
loss is less for tall cups (cups 1,
2, and 6) and greater for short cups (cups 3, 4, 5, and
7). Thus evaporative
loss can be minimized
by the
use of a particular
cup design
having
a relatively
large height-to-diameter
ratio. This condition
is fulfilled by the 0.2-mi Microtube
(cup 1, Figure 4). The
validity
of this selection
is predicted
by the mathematical
model and has also been confirmed
experimentally
(Figure 7, Table 3).
The minimal
evaporative
loss from the Microtube
led us to consider
using this type of cup in the sample/reagent
loader
(5) that has been developed
for
use with the miniature
Centrifugal
Fast Analyzer.
When
selecting
both theoretical
our experimental
Dr. James
co, California;
Penton,
personal
a sample
cup, one has to consider
and practical
aspects.
For example,
results, as well as the mathematical
Institute
for Health
communication.
Research,
San Francis-
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
1915
I-
on
0
-I
1.
TIME
(HOURS)
Fig. 9. Evaporativelossand percent error for the Abbott
ABA-i 00 sample cup as a function of surface treatment and
time
Fig. 10. Modif led turntable
of the sample/reagent
loader (min-
iature Centrifugal Fast Analyzer)
model, indicated that evaporative
mized by decreasing the diameter
loss can be miniof a cup relative to
its height. However,
there is a practical
limitation
to
the benefit
gained by this approach,
because
filling
the cup with sample becomes
an increasing
problem
as the diameter
of the cup is decreased.
In addition,
the sampling
probe that is inserted
into the cup may
also impose a practical
limitation
on the minimum
diameter of the cup. These problems were considered
specifically with regard to the dimensions of the Microtube
cup.
It was found
that
samples
could
be very
easily introduced
into this cup via a disposable
pipette or a similar type of pipetting device. In addition, we found that it was possible, with the assistance of a probe guide, to position
a sampling
probe
over the cup and then direct the probe into its con-
tents for sampling. Consequently,
the turntable
of
the sample/reagent
loader was modified to accept a
carousel that holds 17 of the Microtubes
(Figure 10).
To demonstrate
that the use of Microtubes
will
minimize
the analytical
conducted
an experiment
of an aqueous
standard
effects
of evaporation,
we
in which various volumes
solution
of glucose
were
and their glucose content was
placed into Microtubes
measured as a function of time. The volumes studied
were 50, 100, and 200 zl, with four replicate samples
1916
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
being assayed
for each volume.
For reference
purposes, four tubes were filled with 200 l each and
capped between
loadings.
The other tubes remained
open during the entire experiment.
The results (summarized in Table 5) indicate that very little evaporation occurred. For example, the 8-h loss was only
3.5% for the 50-il sample,
while the losses for the
100- and 200-gil samples
were hardly detectable.
The
samples
were also assayed
24 h after the start of the
experiment.
At this time, evaporation
had begun to
have a noticeable
effect on the glucose concentration
of the uncapped
samples;
however,
this is merely illustrative
and would not present a problem
in routine
analyses,
because samples
are not usually allowed to
remain in their cups that long.
This study with the Microtubes
pointed
out an additional
advantage
that was gained
by using these
cups, i.e., sample volumes as small as 50 tl could be
used and processed.
In fact, by positioning
the sample probe to sample at the bottom of the tube, a 10-tl
aliquot can be obtained from a total sample
as small as 20 gil. This low-volume
capability
prove very useful
cal studies.
in pediatric
and small-animal
volume
should
clini-
Table
5.
Effect
of
of
Samples
Evaporation
Placed
in
on
the
Microtube
Glucose
Sample
Content
Cups
_____________________________________________________
We thank J. B. Overton for performing
some of the clinical analyses. The technical
advice and direction
provided
by C. D. Scott,
who initiated
the development
of the evaporation
model, and by S.
E. Shumate,
who suggested
descriptive
improvements
in it, are
greatly appreciated.
The technical
assistance
of W. F. Johnson,
W.
A. Walker, R. A. Mathis, and D. V. Maxey in designing
and modifying the sample-reagent
loader is also acknowledged.
In addition
to these Oak Ridge personnel,
we are indebted
to Dr. Donald
S.
Young, NIH, Bethesda,
Maryland,
for his helpful
recommendations and for providing
some of the sample cups used in this study;
to Dr. Harvey
Mohrenweiser,
National
Center
for Toxicological
_________
_________
_________
__________
cap
200 4
200 Il
100 4
50 cl
Sampl a volume
+
Time
(h)
ssa
RSD
(%)
0
2002
0.7
1
1996
0.5
2
l
o.1e
a
531)
a
531)
531)
1997
0.9
1999
0.6
2000
0.6
0.14
1999
0.5
1997
11
1996
0.7
1998
0.6
2002
0.6
a
3
1989
1.3
2008
0.6
1
0.9
1990
0.7
14
2039
0.7
2026
0.8
2025
0.7
2005
0.5
Research,
Jefferson,
Arkansas,
rotubes
n the sample-reagent
source of them;
and to Dr.
Health
Research,
San
Francisco,
Rodgerson,
UCLA, Los Angeles,
criticism
of this manuscript.
5
2030
0.1
2010
0.3
2037
0.5
2008
0.6
Work described
stitute of General
6
203].
0.9
2030
0.5
2033
0.5
1999
1.5
Space
7
2057
0.14
2007
0.6
2022
0.9
1995
0.14
8
search,
tion.
24270
0.2
2017
o.6
2027
0.6
1985
0.8
2311
0.6
2176
0.9
0.6
2019
214
aGlumee
concentration
In mg/liter,
2180
mean
of
four
measurements
Administration,
and
the
for suggesting
the use of the Mic.
loader and for directing
us to the
Widdowson,
Institute
for
Graham
California,
California,
and Dr. Denis
for their advice
in this study was supported
by the National
Medical Sciences,
the National
Aeronautics
the National
Center
for Toxicological
Energy
Research
and
Development
0.
and
Inand
Re-
Administra-
1.1
at
each
time.
References
In summary,
that
current
the loss of liquid
oration
studies
from
can be a serious
have demonstrated
a sample
problem,
caused
especially
by evap-
when mi-
croliter volumes are involved.
The magnitude
of this
loss is determined
by the interrelationship
of a complex set of variables
that include environmental,
instrumental,
and operational
factors
as well as the
chemical
and physical
properties
of the sample and
its container.
Because the error caused by evaporation can, in some instances,
be as high as 25 to 50%,
steps must be undertaken
to minimize
its effect.
Evaporative
loss can be minimized
by controlling
the
laboratory
environment,
selecting
a sample cup of a
favorable design, filling the cup only half full of sample, using effective sample covers, and following good
laboratory
practice.
even with microliter
to 1 to 2% or less.
In this way, the analytical
error,
volumes
of sample,
can be held
1. Loeb, H. G., Tramell,
P. R. Szepessy,
G., et al., Effect of a sample evaporation
retardant
on enzyme activity
results with the Abbott Bichromatic
Analyzer-100.
Clin. Chem. 18,698 (1972).
R. E., Automation
and on-line computers.
In Clinical
Principles
and Technics,
2nd. ed., R. J. Henry, D. C.
Cannon,
and J. W. Winkelman,
Eds., Harper and Row, Hagerstown, Md., 1974, p 214.
2. Thiers,
Chemistry,
3. Nishi, H. H., and Young, D. S., Concepts
and design of a mechanized microchemistry
system for discrete
samples.
In Microchemical Techniques,
M. Werner, Ed., John Wiley and Sons, in press.
4. Burtis,
C. A., Johnson,
W. F., Mailen, J. C., et al., Development
of an analytical
system based around
a miniature
Fast Analyzer.
Clin. Chem. 19,895 (1973).
5. Burtis,
C. A., Johnson,
W. F., and Overton,
J. B., Automated
loading of discrete,
microliter
volumes of liquids into a miniature
Fast Analyzer.
Anal. Chem. 46,786 (1974).
6. Tiffany,
T. 0., Jansen,
J. M., Burtis, C. A., et al., Enzymatic
kinetic rate and end-point
analyses
of substrate
by use of a GeMSAEC Fast Analyzer.
Clin. Chem. 18, 829 (1972).
7. Bird, R. B., Stewart,
Phenomena,
John Wiley
W. E., and Lightfoot,
E. N., Transport
Sons, New York, N.Y., 1960, p 522.
and
8. James, L. K., and Berry, D. J. 0., Evaporation
protein films. Science
140, 312 (1963).
enhancement
CLINICAL CHEMISTRY, Vol. 21, No. 13, 1975
by
1917