Environmental monitoring using a conventional photographic digital camera for multianalyte disposable optical sensors.

Abstract

The primary interest of this study concerns the use of an inexpensive photographic digital camera as the detection system, using its own flash as the source of light to present a new analytical procedure to measure disposable multianalyte optical sensors for potassium, magnesium, hardness and conventional pH test strips. The camera arrangement was designed in a fixed position over an optical board with controllable ambient conditions. After acquiring the digital image, the analytical information contained in each test zone is analyzed using theRGB colour space. Reflectance measurements were developed to study the colourimetric and spectral characteristics of the test zones. We obtained the following application ranges and precision in terms of relative standard deviation (RSD %): for potassium from 3.2×10(-7) to 0.1 M with a precision between 3.3 and 4.0%, for magnesium from 2.7×10(-6) to 1.5 M showing a precision between 4.7 and 7.8% and finally for hardness from 4.3×10(-2) to 200,000 mg L(-1) CaCO(3) and between 5.1 and 7.0%. Moreover, the analytical characteristics of several optical procedures were compared with the results presented here. The proposed method was statistically validated against a reference procedure using samples of water from different sources and beverages, indicating that there are no significant statistical differences at a 95% confidence level.

Full text

Page 1

Analytica
Chimica
Acta
706 (2011) 328–
337
Contents
lists
available
at
SciVerse
ScienceDirect
Analytica
Chimica
Acta
j our
na l ho
me
p age:
www.elsevier.com/locate/aca
Environmental
multianalyte
monitoring
using
a
conventional
photographic
digital
camera
for
disposable
optical
sensors
A.
Lapresta-Fernández, L.F.
Capitán-Vallvey∗
ECsens,
Department
of
Analytical
Chemistry,
Campus
Fuentenueva,
University
of
Granada,
18071
Granada,
Spain
a
r
t
i
c
l
e

i
n
f
o
Article
Received
Received
Accepted
Available online 1 September 2011
history:
20
May
2011
in revised
form
11 August
2011
25
August
2011
Keywords:
Photographic
Image
Potassium,
determination
Disposable
Environmental
digital
camera
colour
analysis
magnesium
and
hardness
multianalyte
sensor
monitoring
a
b
s
t
r
a
c
t
The
the
measure
pH
controllable
each
the
ranges
0.1
between
and
results
using
statistical
primary
interest
of
this
study
concerns
the
use
of
an
inexpensive
photographic
digital
camera
as
detection
system,
using
its
own
flash
as
the
source
of
light
to
present
a
new
analytical
procedure
to
disposable
multianalyte
optical
sensors
for
potassium,
magnesium,
hardness
and
conventional
test
strips.
The
camera
arrangement
was
designed
in
a
fixed
position
over
an
optical
board
with
ambient
conditions.
After
acquiring
the
digital
image,
the
analytical
information
contained
in
test
zone
is
analyzed
using
theRGB
colour
space.
Reflectance
measurements
were
developed
to
study
colourimetric
and
spectral
characteristics
of
the
test
zones.
We
obtained
the
following
application
and
precision
in
terms
of relative
standard
deviation
(RSD
%):
for
potassium
from
3.2
×
10−7to
M
with
a
precision
between
3.3
and
4.0%,
for
magnesium
from
2.7
×
10−6to
1.5
M
showing
a
precision
4.7
and
7.8%
and
finally
for
hardness
from
4.3
×
10−2to
200,000
mg
L−1CaCO3and
between
5.1
7.0%.
Moreover,
the
analytical
characteristics
of
several
optical
procedures
were
compared
with
the
presented
here.
The
proposed
method
was
statistically
validated
against
a
reference
procedure
samples
of
water
from
different
sources
and
beverages,
indicating
that
there
are
no
significant
differences
at
a
95%
confidence
level.
© 2011 Elsevier B.V. All rights reserved.
1.
Introduction
The
current
trend
for
rapid
and
low-cost
methods
to assess
ana-
lytical
to
that
lowest
single
Visual
where
answer
achieved
Quantitative
the
with
out
simultaneous
by
Lemke
based
tions,
information
– composition,
quality,
discrimination
– has
led
the
development
of
devices
which
provide
relevant
information
is easy
to read.
One
suggestion
to obtain
methods
with
the
possible
cost
is the
use
of
disposable
sensor
devices,
mainly
analyte
but
also
multianalyte.
multianalyte
sensors
can
produce
qualitative
information
the
change
or
disappearance
of
colour
provides
the
yes/no
[1]
or a semiquantitative
estimation
of
concentration
is
by
comparing
the
colour
intensity
with
a colour
chart
[2].
multianalyte
disposable
sensors
can
be
based
on
spatial
resolution
of
different
analytes
on
the
same
substrate
or
without
separation
by lateral
fluid
flow.
One
example
with-
separation
is
the
multi-spot
test
zone
strip
developed
for
the
determination
of
several
pesticides
in water
samples
chemiluminescence
[3].
The
same
concept
was
developed
by
et
al.
[4]
as
a potentiometric
array
for
pH,
K(I),
Na(I)
and
Ca(II)
on
coated-film
electrodes.
In other
cases,
there
are
no separa-
but
the
sample
is delivered
using
a microfluidic
device
such
as
∗Corresponding
E-mail
author.
Tel.:
+34
958
248436;
fax:
+34
958
243328.
address:
lcapitan@ugr.es
(L.F.
Capitán-Vallvey).
the
microfluidic
When
usually
tianalyte
fluoroimmunosensor
simultaneous
In
analytical
by
the
[10]
have
desktop
webcam
colour
ble
analytical
ardising
conventional
its
alternative
colour
The
using
paper-made
multianalite
optical
device
containing
hydrophilic
patterns
and
sensing
areas
[5–7].
separation
is required
in disposable
devices,
it is
performed
by
lateral
fluid
flow.
One
example
of
a
mul-
disposable
optical
device
is the
four-band
capillary
presented
by
Mastichiadis
et
al.
[8]
for
the
determination
of
pesticides.
the
case
of
optical
sensing,
the
usual
way
to acquire
the
information
from
disposable
multianalyte
sensors
is
imaging
techniques.
The
most
common
are
CCD
cameras
for
measurement
of
fluorescence
[9],
electrochemiluminescence
or
effective
absorbance
[11]; although
other
imaging
devices
been
used
such
as
scanner,
including
hand-held
[12–15]
and
scanners
[16,17], camera
phone
[15], video
cameras
[18],
[19], digital
photographic
cameras
[20,21]
and
digital
analyzers
[22–24].
Most
of
these
systems
make
it possi-
to acquire
images
of
an analytical
element
containing
potential
information,
but
the
ease
of
use
is
quite
different,
jeop-
portability
in many
cases.
In this
respect,
the
use
of
a
digital
photographic
camera
as
an
imaging
device
and
built-in
flash
as
the
light
source
is an
interesting
and
practical
for
the
easy
readout
of
bulk
optode
films
where
the
indicates
the
uptake
of
the
analyte.
approach
presented
in this
paper
relies
on
the
idea
of
an inexpensive
photographic
digital
camera,
placed
in a fixed
0003-2670/$
doi:10.1016/j.aca.2011.08.042
– see
front
matter ©

2011 Elsevier B.V. All rights reserved.

Page 2

A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
329
position
system
in
previously
and
of
paper
In
analyte
design
conventional
Although
fixed
nation,
portability,
posable
for
personnel.
Thus,
the
ysis.
and
for
over
an
optical
board,
with
a built-in
flash
as
a detection
to simultaneously
measure
the
colour
changes
that
appear
low-cost
transparent
disposable
optical
multisensors,
containing
reported
selective
test
zones
for
K (I)
[25], Mg
(II)
[26]
hardness
[27,28]. Moreover,
we
demonstrate
the
applicability
the
method
for
other
test
method
as
is commercial
pH
indicator
for
pH determination.
this
proof-of-concept
study,
image
analysis
from
the
multi-
optical
sensor
using
a digital
camera
in a simple
optical
was
compared
to other
imaging
and
spectrophotometric
setups.
the
digital
camera
measurements
were
made
over
a
position
on
an
optical
board
in the
absence
of
ambient
illumi-
the
shirt-pocket-size
camera
provides
easier
handling
and
and
along
with
the
highly
selective
multianalyte
dis-
sensor,
is
a sensitive,
low
cost,
stable,
and
robust
method
environmental
analysis
that
can
be
performed
by
non-skilled
the
only
steps
needed
are:
take
the
sensor,
dip
it,
develop
colour,
take
the
picture
and
make
the
subsequent
colour
anal-
Therefore,
the
two
principal
ideas
in this
paper
are
low
cost
simplicity
by using
a commercial
off-the-shelf
(COTS)
device
analytical
purposes.
2. Experimental
methods
2.1.
Apparatus
and
software
Sony
Cyber-shot
DSC-T9
digital
colour
photographic
camera
with
a maximum
New
Adobe
nia,
value
mounted
Boston,
Packard
(model
performed
version
matica
WA,
Reagents:
(2
reac,
and
pH
plied
To
molecular
furan
(4-chlorophenyl)borate
(DB18C6)
Madrid,
weresynthesised
(octadecanoylimino)-5H-benzo(a)phenoxazine-9-amine
5294)
tetraoxa-4,13,-diazacyclooctadecane
N,N-diheptyl-N,N-dimethylaspartamide
Mylar-type
to
were
grade
a
a CCD
sensor
super
HAD
CCD
1/2.5??with
6.0
megapixels
and
resolution
of
2816
× 2112
pixels
(Sony
Corporation,
York,
USA)
equipped
with
a 512
MB
PRO
Duo
memory
card.
Photoshop
CS ver.
8.0.1
(Adobe
System
Inc.
San
Jose,
Califor-
USA)
was
employed
as
software
to obtain
the
luminosity
colour
of
the
spot.
To maintain
a constant
geometry,
the
camera
was
on
a Vibraplane
optical
breadboard
(Kinetic
Systems,
Inc.,
MA).
Absorbance
measurements
were
done
by
a Hewlett
diode
array
spectrophotometer
(DAD
spectrophotometer)
8453;
Nortwalk,
CT,
US).
Later
statistical
calculations
were
with
the
Statgraphics
software
package
for
windows
4.1
Plus.
(Statistical
Graphics
Corporation,
USA,),
Graph-
for
Win
32 ver.
1:60d,
Excel
software
(Microsoft,
Redmond,
USA)
and
Origin
7.0
(Northampton,
MA,
USA).
calcium
(1
M),
magnesium
(1 M)
and
potassium
M)
stock
solutions
were
prepared
in water
from
CaCO3(Pan-
Barcelona,
Spain),
MgCO3 (Merck,
Darmstadt,
Germany)
KCl
(Aldrich,
Steinheim,
Germany),
respectively.
0.2
M
9.0
buffer
solution
was
prepared
from
Tris
and
HCl
sup-
by
Sigma
(Sigma–Aldrich
Química
S.A.,
Madrid,
Spain).
prepare
the
optode
films,
poly(vinylchloride)
(PVC;
high
weight),
tributylphosphate
(TBP)
and
tetrahydro-
(THF)
were
acquired
from
(TCPB),
Sigma,
dibenzo18-crown-6-ether
potassium
tetrakis

and
2-nitrophenyloctyl
ether
(NPOE)
from
Fluka
(Fluka,
Spain).
The
following
according
ionophore
to
and
chromoionophores
references:
N,N-diethyl-5-


(ETH
[29]; 4,13-(bis(N-adamantylcarbamoyl)acetyl)-1,7,10,16-

(K22B5)

[27,30];

and
(ETH
2220)
[31]. Sheets
of
polyester
(Goodfellow,
Cambridge,
UK)
were
used
as
support
the
sensor
membranes.
The
pH indicator
papers
used
from
Merck.
The
chemicals
used
were
of
analytical-reagent
and
the
water
used
for
preparing
solutions
was
purified
with
Milli-RO
12
plus
Milli-Q
water
system
(Millipore,
Bedford,
MA).
Fig.
sensors.
1.
Experimental
arrangement
for
image
acquisition
of
the
disposable
optical
2.2.
Test
zone
preparation
of
the
multianalyte
sensor
On
a Mylar
support
(14
mm
×40
mm
×0.5
mm
thick)
we
placed
an array
around
using
test
63.00
of
26.00
of
sensitive
PVC,
and
dissolved
of
three
different
test
zones
(12
mm
in diameter
and
5 ?m in thickness
each)
per
analyte,
prepared
from
20
?L
a
spin-coating
technique.
Mixtures
for the
hardness-sensitive
zones
were
made
from
a mixture
containing
25.00
mg
of
PVC,
mg
of
TBP,
0.50
mg
of
K22B5,
0.85
mg
of
ETH
5294
and
0.71
mg
TCPB.
For
potassium
test
zones,
the
mixture
composition
was:
mg
of
PVC,
63.50
mg
of
NPOE,
0.80
mg
of
DB18C6,
1.30
mg
ETH
5294
and
1.10
mg
of
TCPB.
In the
case
of
the
magnesium-
test
zones,
the
composition
used
contained:
25.00
mg
of
63.00
mg
of
TBP,
0.95
mg
of
ETH
2220,
0.78
mg
of
ETH
5294,
1.33
mg
of
TCPB.
For
each
test
zone,
the
components
were
in 1 mL
of freshly
distilled
THF.
2.3.
Imaging
set-up
All
measurements
were
made
without
any
temperature
con-
trol,
in
reflections.
For
9 cm with
cover
ble
zone
In
sure
flash
in
posable
uniform
working
around
20–25◦C and
in the
absence
of ambient
light
a
dark
room
in order
to prevent
external
light
interference
and
the
image
acquisition,
the
camera
was
placed
at
a height
of
an
angle
of
8◦from
the
normal
to the
detecting
area,
to
a detection
field
of
11
× 8 cm (see
Fig.
1),
where
it
is possi-
to position
the
multianalyte
sensor
with
except
in the
central
(2 × 8 cm)
because
the
flash
light
produces
light
saturation.
this
light
saturated
field
(see
Fig.
3),
it is
not
possible
to mea-
any
colour
change
in the
sensor
since
the
intensity
of
the
overlaps
any
colour
change
because
all
pixel
luminosities
are,
consequence,
in their
maximum
values.
The
multianalyte
dis-
sensors
were
placed
on
a
white
matte
sheet
to provide
a
background
and
reduce
glare
and
specular
reflections
[32].

Page 3

330
A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
The
digital
camera
lens
with
an aperture
of
f/3.5,
a
shutter
speed
of
maintaining
the
ture
studied
1/400
s and
the
white
balance
in the
HAD
CCD
chip
setting
were
constant.
The
disposable
sensors
were
focuses
by
using
autofocus
function
and
no
additional
filters
were
used.
The
cap-
conditions
in terms
of
digital
camera
sensitivity
(ISO)
were
and
optimized,
selecting
a sensitivity
of
ISO
400.
2.3.1.
Region
The
of
interest
(ROI)
image
was
stored
in JPEG
format
and
transferred
to the
PC
by
fixed
ity
analytical
nosity
the
vertical
study
of
up
The
symmetry
240
a generic
card
reader
via
USB.
The
ROI
– a homogeneous
circle
of
diameter
– was
selected,
and
the
average
ROI
pixel
luminos-
was
taken
directly
using
Adobe
Photoshop
in order
to build
the
parameter.
To select
the
ROI
size
we
analyzed
the
lumi-
[17,33], using
the
red
channel
for
hardness
and
Mg(II)
and
green
channel
for
K(I).
The
circular
image
was
divided
by
both
and
horizontal
diameters
into
squares
of
10
×10
pixels
to
the
colour
symmetry,
and
additionally,
a
study
was
made
the
circular
integration
of
ROI
using
circles
of
increasing
size,
to 300
× 300
pixels,
in order
to study
the
ROI
homogeneity.
highest
size
of
ROI
that
shows
good
signal
homogeneity
and
which
avoid
any
sensing
film
imperfections
was
that
of
×240
pixels,
which
means
an
average
of
45,239
pixels.
2.4.
Measurements
The
sensor
was
dipped
for
5 min
into
a polyethylene
tube
(10
1 mL
vated
Tris
10−2M NaOH
forms
tively.
Calibration
dently
application
different
pH
from
the
× 1.5
cm)
containing
9 mL
of
the
test
solution
together
with
of
0.2
M pH 9.0
Tris
buffer.
Initially,
all
sensors
had
to be acti-
in HCl
10−2M and
after
that
were
equilibrated
in 2 × 10−2M
buffer
(IXbuffer), followed
by
the
test
solution
(IX) and
finally
by
(IXNaOH).
The
protonated
and
the
fully
deprotonated
of
the
chromoionophore
were
(IXbuffer) and
IXNaOH, respec-
(see
Eq.
(4)).
studies
for
each
sensor
were
performed
indepen-
using
sensors
containing
three
equal
test
zones,
and
for
to real
samples
we
also
used
sensors
containing
all
three
test
zones
for
K(I),
Mg(II)
and
hardness,
respectively.
For
measurements
using
indicator
papers,
the
pH values
ranging
6.5
to 10.0
were
measured
using
pH
papers
equilibrated
with
above
Tris
buffer
solution
for
3 min
using
the
same
procedure.
3.
Theoretical
basis
3.1.
General
models
The
sensing
mechanisms
of
this
multianalyte
ionophore-based
sensor
theoretical
by
[11,36]:For
?
For
?
where
scripts
?
follow
the
ion-exchange
equilibrium
[34,35], where
the
response
function
for
each
analyte
can
be
described
the
following
linearised
equations
considering
the
references
potassium:
?1
log
˛eff
1 − ˛eff
2log
KK+
exch
aH+
+1
2log
aK+
(1)
hardness:
log
˛eff
1 − ˛eff
?
+1
2log
˛eff
8 − ˛eff
?
1
2log
KM2+
exch
a2
H+
√C
+1
2log
aM2+
(2)
aM2+ is
expressed
as
CaCO3and
C = CR= CC= 1/2CL(see
sub-
below)And
for
magnesium:
?
log
˛eff
1 − ˛eff
?
+1
2log
2˛eff−
?
1
5 − 2˛eff
?
=1
2log
KMg2+
exch
a2
H+
√C
+1
2log
aMg2+
(3)
where
ion
hydrogen
analytical
lipophilic
analyte
The
the
moionophore
pixel
nels)
C = CL= CR= 1/2CC (see
subscripts
below)In
all
cases
the
activity
aI?+ is related
to the
equilibrium
constant
Kexch, the
ion
activity
aH+, the
experimental
parameter
˛
and
the
concentrations
of
ionophore,
CLchromoionophore
CCand
anion
CR, p is the
stoichiometric
factor
for
the
ionophore-
complex
and
? is the
analyte
charge.
extension
of
the
ion-exchange
process
can
be measured
by
so-called
effective
degree
of
protonation
1 −
˛effof
the
chro-
(Eq.
(3)), which
was
calculated
using
the
averaged
value
of
the
selected
channel
(IX) (X:
blue,
red
or
green
chan-
from
the
ROI:
1
− ˛eff=
log
IX− log
IXNaOH
log
IXbuffer− log
IXNaOH
(4)
Each
and
els
(Y(˛)
D(˛),
ered
(B
test
zone
was
also
equilibrated
in NaOH
solution
IXNaOH
in buffer
IXbuffer.In a simplified
way,
the
linearised
mod-
can
be
described
as
the
common
straight
line
equation
+ D(˛)
= A + BX)
with
the
introduction
of
a disturbance
term
where
X is the
decimal
logarithm
of
the
activity
of
the
consid-
cationic
species,
A is an
independent
term
and
B is
the
slope
= 1/?).
4.
Results
and
discussion
4.1.
Colourimetric
and
spectral
characteristics
of
the
test
zones
The
spectral
characteristics
and
colour
description
of
the
sensing
zone
branes
they
ing
each
troradiometric
to the
when
band
point
best
and
colour
chromoionophore
compositions,
different
colour
nesium
from
the
membranes
were
checked.
As
the
different
mem-
include
the
same
chromoionophore
in
their
composition,
will
have
approximately
the
same
reflectance
spectra,
depend-
on
the
extension
of
the
recognition
reaction
for
the
analytes
in
case.
The
reflectance
spectra
of
the
membranes
from
the
spec-
measurements
show
a
maximum
at
482
nm due
protonated
form
of
the
chromoionophore,
which
decreases
the
analyte
concentration
increases
at the
same
time
as
a
at
660
nm
due
to the
deprotonated
form
through
an isosbestic
at 565
nm [37]
(Fig.
2).
In this
respect,
we
expected
to find
the
correlation
with
the
blue
channel
due
to the
peak
at 482
nm
with
the
red
channel
due
to the
lower
peak
at
660
nm in the
RGB
space.
Although
all
the
membranes
are
based
on the
same
in all
cases,
they
have
different
components
and
making
the
observed
visual
colour
change
slightly
in the
three
membranes
used.
Thus,
a
predominant
red
can
be
observed
in the
cases
of
both
hardness
and
mag-
membranes,
and
a
blue
colour
for
potassium
membranes,
700600500400
0.00
0.04
0.08
0.12
1 M
1·10- 7M
Basic form
Acid form
reflectance
Basic form
Acid form
λ (nm)
[K+]
1 M
1·10-7M
Fig.
ferent
2. Spectral
reflectance
of
potassium
sensing
membranes
equilibrated
with
dif-
standard
potassium
solutions.

Page 4

A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
331
Fig.
background
analyte
3.
Sequence
of
images
for
six
multianalyte
disposable
sensors
after
equilibration
with
different
standards
of
Mg(II),
K(I)
and
hardness
and
placed
over
the
white
sheet
in the
11
×8 cm detection
area.
Concentrations
are
in M.
This
image
capture
goes
from
B: buffer
to OH−: basic
form.
They
show
two
disposable
sensors
per
with
three
test
zones
each.
as
hardness
channel,
tion
may
to
should
cover
located,
the
can
be
seen
in Fig.
3.
For
that
reason,
the
relationship
with
the
and
magnesium
concentration
was
found
using
the
red
and
for
potassium
with
the
green
channel.
The
lack
of
rela-
found
with
the
potassium
concentration
in the
blue
channel
be
due
to the
nature
of
the
Bayer
matrix
used
in this
camera
register
the
different
colour.
The
filters
used
in the
Bayer
matrix
have
an
overlap
between
the
blue
and
the
green
filters
that
the
spectral
region
where
the
maximum
peak
at 482
nm is
with
the
corresponding
green
channel
showing
better
than
blue
one.
4.2.
Optimisation
procedure
The
selected
working
pH for
this
multianalyte
sensor
was
9.0
because
alkaline
In
noise
ables
and
From
used
since
level
In
the
was
tolerance
a concentration
culated
the
librated
˛
signal
at this
pH
the
sensor
selectivity
is sufficient
for
common
and
alkaline
earth
concentrations
in natural
waters
[37].
terms
of
optimising
the
procedure
to improve
the
signal
to
ratio
and
the
dynamic
range,
the
authors
checked
several
vari-
such
as
background
colour,
the
illumination
field
of the
flash,
the
gain
of
the
detector
as
ISO
parameter.
different
colours,
a white
matte
background
colour
was
to place
the
multianalyte
sensors
as
previously
optimised
[11],
it showed
the
highest
(near
1.0)
and
most
stable
reflectance
from
430
nm to 750
nm.
order
to check
the
effect
on
the
response
of
the
sensors
due
to
non-homogenous
illumination
by the
flash,
the
detection
field
divided
into
seven
different
zones
(see
Fig.
2),
establishing
a
interval,
in each
zone,
around
to the
value
of
1 − ˛efffor
of
10−3M for
each
analyte.
This
interval
was
cal-
as
follows:
predicted
response
±
(tsR/√n), with
SRbeing
standard
deviation
for
4 different
test
zone
replicates
(n)
equi-
in 10−3M,
t is
the
t-student
for
n − 2 freedom
degrees
and
= 0.05.
We
used
this
interval
as
the
reference
and
the
maximum
value
for
the
sensor.
All
studied
zones
gave
an
experimental
value
the
interval,
We also
towards
a digital
be
larger
ing
saturation-based
dynamic
four
400
this
better
imental
theoretically)
Apart
ground
that
the
light
The
given
aperture
ate
was
figuration
values
and
constant
Accordingly,
for
assessed
in this
interval
with
the
exception
of
the
central
zone
where
experimental
value
was
beyond
the
limit
values
of the
tolerance
showing
a light
saturated
field.
(see
Fig.
2).
studied
the
CCD
sensor
sensitivity
of
the
camera
the
light
in terms
of
the
ISO
system.
The
ISO
number
in
camera
corresponds
to the
gain
of
the
detector
and
should
kept
at the
minimum
in order
to reduce
noise.
Nevertheless,
CCD
gains
could
correspond
to better
sensitivities
regard-
the
reflected
colour
that
reaches
the
CCD
sensor.
Thus,
as
the
indices
of
exposure
can
be
varied
by
adjusting
the
range
of the
camera,
the
digital
camera
was
configured
at
different
ISOs,
80,
200,
400
and
640.
From
all
of
them,
ISO
was
selected
for
the
multianalyte
sensor
due
to the
fact
that
option
showed
the
best
precision
(between
3 and
7%),
yielded
correspondence
with
the
theoretical
models
(as
the
exper-
slope
values
are
in
close
relation
with
the
predicted
ones,
and
the
widest
application
range
in overall
terms.
from
that,
selecting
the
ISO
400
we
obtain
an
uniform
back-
without
any
vignetting
effect
[38]. This
last
one
is
an
artefact
it corresponds
to a light
fall
from
the
centre
to the
border
of
image
due
to the
lens
aperture,
creating
uneven
responses
to
across
the
image.
rest
of
the
settings
correspond
to the
exposure,
and
for
a
illumination
the
only
two
parameters
to optimise
are
the
and
the
exposure
time.
The
shutter
speed
and
appropri-
aperture
were
modified
automatically
by the
camera
since
it
not
possible
to make
a systematic
optimisation
due
to the
con-
limitations
of the
camera.
In our
work
conditions,
the
were
1/400
s and
f/3.5,
corresponding
to the
shutter
speed
aperture
respectively.
These
values
remained
automatically
in all
measurements.
the
suitability
of
this
photographic
digital
camera
the
quantitative
ratiometric
measurements
has
been
carefully
in this
sense.
Nevertheless,
it is highly
advisable
to work

Page 5

332
A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
0
0.2
0.4
0.6
0.8
1
0-2-4-6-8
log a(K+)
log a(K+)
1-αeff
y = 0.9623x + 3.3461
R2 = 0.9664
-4
-2
0
2
4
0-2-4 -6-8
log(αeff/(1-αeff))+D(αeff)
BA
Fig.
concentration.
4.
Calibration
functions
for
potassium:
Theoretical
sigmoidal
function
(A)
and
linearised
model
(B).
Error
bars
indicate
standard
deviation
of
measurements
in each
Experimental
points
in M:
1 × 10−7, 5 × 10−7, 1 × 10−6, 5 × 10−6, 1 × 10−5, 5 × 10−5, 5 × 10−4, 1 × 10−3, 5 × 10−3, 1 × 10−2, 5 × 10−2and
0.1.
with
surrounding
in
use
1
a reference
signal
[39]
which
may
decrease
the
influence
of
the
media
or
the
light
source
intensity.
In this
sense
and
an attempt
to avoid
external
fluctuations,
it could
be
possible
to
the
following
expression
to obtain
the
normalized
parameter
− ˛eff
1 − ˛eff=
(−
log
IX/I0) − (− log
I(NaOH)X/I0)
(− log
I(buffer)X/I0) −
(− log
I(NaOH)X/I0)
(5)
where
ratio
nel
its
background.
Accordingly,
from
metric
On
during
the
luminosity
of
the
sensing
zone
is defined
as
the
between
the
average
colour
intensity
of
the
correct
chan-
obtained
from
histogram
(IX) (being
X:
red,
green
or blue)
and
maximum
value
(I0), obtained
from
the
reflected
light
on
the
considering
I0as
the
reference
signal
and
changes
one
measurement
to another,
it could
be
possible
to use
ratio-
measurements
to decrease
external
variations.
the
other
hand,
by
assuming
that
the
value
I0is constants
all
the
measurements
this
expression
can
be
simplified
to:
1 − ˛eff=
log(IX) − log(I(NaOH))
log(I(buffer)X) − log(I(NaOH)X)
(6)
In
our
case,
the
value
of
I0it was
usually
near
to its
theoretical
maximum
value
of
255
[17]. In consequence,
we
used
the
Eq.
(6).
4.3.
Linearised
model
and
performance
characteristics
Under
optimal
conditions,
the
linearity
of
the
analytical
parame-
ter
between
(two
analysis
taken
The
and
eral
(Figs.
The
and
with
tration,
sensors
The
spondence
of
try.
log(˛eff/1 − ˛eff) + D(˛eff) instead
latter
the
a black
mode
was
established
from
12
calibration
levels
for
the
three
analytes
1 × 10−7and
1.0
M,
at pH
9.0,
using
6 replicates
each
time
test
strips
for
each
analyte
with
3 test
zones
each).
The
global
sequence
was
joined
in the
Fig.
3 where
each
image
was
and
analyzed
individually.
linearisation
of
the
whole
sigmoidal
function
(Figs.
4B,
5B
6B)
makes
it possible
to increase
the
measuring
range
by
sev-
orders
of
magnitude,
comparing
with
the
sigmoidal
function
3A,
4A and
5A)
[36].
statistic
parameters
calculated
from
least-square
regression
the
performance
characteristics
are
presented
in Table
1 along
the
obtained
precision,
expressed
as
logarithms
of
concen-
at
three
concentration
levels
using
10
different
disposable
each
time.
experimental
data
for
potassium
suggest
a
better
corre-
with
the
theoretical
functions
when
a stoichiometry
p = 0.5
is used
rather
than
the
usual
1:1
(p = 1)
stoichiome-
In line
with
this,
the
analytical
linear
parameter
for
K(I)
is
of
log(˛eff/1
− ˛eff) because
the
does
not
generate
a straight
line
with
p = 0.5.
The
effect
of
disturbance
term
D(˛eff), already
observed
by us
working
with
and
white
CCD
camera
[11]
and
a scanner
in reflectance
[17], involves
higher
slopes
for
potassium
in both
sigmoidal
0.0
0.2
0.4
0.6
0.8
1.0
1-1-3-5-7
1-αeff
y = 0.4947x + 0.9207
R2 = 0.9624
-3.0
-2.0
-1.0
0.0
1.0
-1-3 -5-7
log(αeff/(1-αeff))
B
A
log a(Mg2+)log a(Mg2+)
Fig.
concentration.
5.
Calibration
functions
for
magnesium:
Theoretical
sigmoidal
function
(A)
and
linearised
model
(B).
Error
bars
indicate
standard
deviation
of measurements
in each
Experimental
points
in M:
5 ×10−7, 1 × 10−6, 5 × 10−6, 1 × 10−5, 1 × 10−4, 5 × 10−4, 1 × 10−3, 5 × 10−3, 1 × 10−2, 5 × 10−2, 0.1,
0.5
and
1.

Page 6

A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
333
0.0
0.2
0.4
0.6
0.8
1.0
0 -2 -4-6-8
log a(Mg2+,Ca2+

)
log a(Mg2+,Ca2+

)
1-αeff
y = 0.4289x + 1.1498
R2 = 0.9446
-2.5
-1.5
-0.5
0.5
1.5
0-2-4 -6-8
log(αeff/(1-αeff))
B
A
Fig.
concentration.
6.
Calibration
functions
for
hardness:
Theoretical
sigmoidal
function
(A)
and
linearised
model
(B).
Error
bars
indicate
standard
deviation
of
measurements
in each
Experimental
points
in M:
1 × 10−7, 5 × 10−7, 1 × 10−6, 5 × 10−6, 5 × 10−5, 1 × 10−4, 5 × 10−4, 1 × 10−3, 5 × 10−3, 1 × 10−2, 5 × 10−2and
0.1.
and
(p
when
Mg(II)
D(˛eff) the
ing
influence
the
value
concentration
zero,
dition
˛ values
resulting
bration
(7)):
linear
models;
with
the
theoretical
linear
slope
being
equal
to 1
= 0.5),
obtaining
0.9623
experimentally,
instead
of
a slope
of
0.5
we
work
with
p = 1.
This
behaviour
has
small
influence
for
test
zones
where
due
to the
effect
of
this
disturbance
term
theoretical
value
decreases
from
0.5
to 0.4765
obtain-
0.4947
experimentally.
In the
case
of
hardness,
the
effect
of
the
of D(˛eff) is
strong,
decreasing
the
theoretical
slope
of
linear
function
from
0.5
to 0.3907,
resulting
experimentally
a
of
0.4289.
Finally
for
the
LOD
is
necessary
to calculate
a cation
which
yields
an
a value
significantly
different
from
establishing
a minimum
value
for
a
such
that
fulfil
the
con-
a ≥ 3s0, where
s0is the
standard
deviation
of
experimental
when
the
analyte
concentration
is
near
to zero
[36]. The
LOD
is obtained
by
introducing
this
value
of
a in the
cali-
linearised
function,
obtaining
the
following
expression
(Eq.
LD3 ·s0
b
− a
(7)
where
the
the
s0is
the
standard
deviation
of experimental
˛
values
when
analyte
concentration
is
near
to zero,
b and
a are
the
slope
and
origin
ordinate,
respectively.
4.4.
different
Performance
comparison
of multianalyte
sensors
using
instrumental
systems
We
carried
out
this
study
with
the
aim
of
comparing
and
evaluating
described
spectrophotometry
a
nal
Table
procedures.
and
In
ology,
using
more,
spectrophotometer.
theoretical
log
era
digital
the
analytical
characteristics
of
the
proposed
method
here
with
other
possible
measurement
systems
such
as
and
imaging,
with
a DAD
spectrophotometer,
black
and
white
CCD
camera
and
a digital
camera
using
an
exter-
light
source
(halogen
lamp)
[40]
with
both
imaging
techniques.
2 presents
the
performance
comparison
between
different
This
comparison
study
was
developed
in terms
of
LODs
precision
as
relative
standard
deviation
in %.
the
case
of
potassium,
and
employing
the
sigmoidal
method-
the
best
LOD
corresponds
to the
photographic
digital
camera
a halogen
lamp
as
the
light
source
(5.4
× 10−6M).
Further-
the
use
of
cameras
produces
a
better
LOD
than
using
a
This
might
be
due
to the
displacement
of
the
curve
towards
lower
activities
as
shown
by
the
values
of
Kexchcomparing
the
spectrophotometer
(−5.45),
the
CCD
cam-
(−4.88),
the
digital
camera
with
halogen
lamp
(−4.94),
and
the
camera
with
built-in
flash
studied
here
(−5.19).
With
regard
to the
linearised
model,
although
the
obtained
value
the
noticed
slightly
flash),
In terms
camera,
cision
acceptable
than
(2.4–7.6%)
For magnesium,
involves
concentration,
spectrophotometer
function
camera
gen
(2.7
Even
that
are
procedure
ical
the
tometric
Finally,
tion,
(1.2
ter
(1.7
this
the
by
applied,
ter.
present
els
9.8–13.1%
in both
camera
Accordingly,
tographic
response
of
the
LOD
using
a spectrophotometer
is 1.6
×
10−7M and
corresponding
one
in a B&W
CCD
camera
is 9.9
× 10−8M,
we
that
using
a photographic
digital
camera,
the
LOD
are
high,
2.7
× 10−7(halogen
lamp)
and
3.2
× 10−7M
(built-in
although
acceptable
for
analytical
purposes.
of
precision,
it is clear
that
the
photographic
digital
not
designed
for
analysis
purposes,
produces
worse
pre-
than
the
spectrophotometer
(1.4–3.8%)
although
it shows
and
highly
promising
values
(3.3–4.2%),
even
better
when
using
the
photographic
camera
with
a
halogen
lamp
[41].
the
working
pH
used
for
this
multisensor
a
displacement
in the
sigmoidal
model
to a higher
obtaining
a higher
LOD
compared
with
the
DAD
procedure
(9.0
× 10−6M).
When
the
sigmoidal
is
linearised,
the
best
LOD
is
achieved
with
the
B&W
CCD
(1.9
× 10−6M)
compared
to digital
camera
with
a
halo-
lamp
(2.0
× 10−6M)
and
digital
camera
with
built-in
flash
× 10−6M).
though
the
precision
of
the
current
study
(4.7–7.8%)
and
of
the
photographic
camera
using
a halogen
lamp
(6.8–7.8%)
smaller
than
with
the
developed
black
and
white
CCD
camera
(5.01–6.83%),
they
are
considered
adequate
for
analyt-
purposes.
Furthermore,
considering
the
sigmoidal
calibration,
imaging
techniques
show
better
precision
than
the
spectropho-
procedure
(8.0–39.8%).
in the
case
of
the
hardness
sensor
and
sigmoidal
calibra-
the
best
LOD
was
obtained
by
the
black
and
white
CCD
camera
mg
L−1of
CaCO3), but
methodologies
using
a spectrophotome-
(1.9
mg
L−1of
CaCO3) and
digital
camera
with
halogen
lamp
mg
L−1of
CaCO3) gave
a similar
LOD,
with
the
LOD
obtained
in
paper
being
the
highest
(4.2
mg
L−1of
CaCO3).
An
evaluation
of
imaging
techniques
reveals
that
in all
cases
the
LOD
increased
two
orders
of magnitude
when
the
linearising
procedure
was
similar
to the
value
obtained
using
the
spectrophotome-
Moreover,
considering
the
precision,
all
the
imaging
techniques
better
precision
in both
the
sigmoidal
and
linear
mod-
than
the
spectrophotometer
(5.4–7.6%
sigmoidal
model
and
with
the
linearised
model),
with
the
best
values
being
the
sigmoidal
and
linearised
models
when
using
the
digital
with
built-in
flash
(4.8–7.0%
and
3.5–5.0%,
respectively).
one
of
the
intrinsical
benefits
of
using
the
pho-
digital
camera
with
its
own
built-in
flash
is that
a
fast
can
be
obtained
at
times,
without
the
need
to wait

Page 7

334
A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
Table
Analytical
1
performance
of
the
multianalyte
disposable
sensors
studied.a
Parameter

Potassium

Magnesium

Hardness
Sigmoidal
model

Linear
model

Sigmoidal
model

Linear
model

Sigmoidal
model

Linear
model
Intercept
Intercept
Slope
Slope
Determination
Linearity
RSD(%)

−0.456
0.020
−0.256
0.006
0.986
97.8

3.346
0.108
0.962
0.024
0.966
97.6
4.0%
3.3%
4.2%
3.2
3.2

0.075
0.016
−0.219
0.007
0.957
96.6

0.920
0.041
0.495
0.012
0.962
97.6
7.8%
4.7%
6.7%
2.7
2.7

−0.023
0.019
−0.215
0.007
0.954
96.6

1.149
0.054
0.429
0.013
0.945
96.9
7.0%
5.1%
4.8%
4.3
4.3
standard
deviation


standard
deviation
coefficient
(R2)
on-line
(%)

log
cb
5 × 10−5M
1
1
6.4
6.4
4.1%
3.1%
5.0%

5 × 10−4M
5 × 10−3M
0.1
1.4
1.4
5.5%
4.6%
6.9%

5 mg
100
10,000
4.2
4.2–200,000
L−1
3.5%
4.4%
5.0%

× 10−4M mg
L−1

× 10−3M M mg
L−1

Limit
Linear
of
detection
(LOD)
× 10−6
× 10−7
× 10−4
× 10−6
× 10−2
rangec
×10−6–1.0
× 10−2
× 10−7–0.1
× 10−4–1.0
× 10−6–1.5
× 10−2–200,000
aThe
bPrecision
cFor
highest
concentration
units
for
the
detection
limit
and
range
are
M for
potassium
and
magnesium
and
in mg
L−1of CaCO3for
hardness.
is expressed
as RSD(%)
of
the
logarithm
of
the
concentration
at three
different
levels.
the
sigmoidal
function,
the
LOD
was
obtained
from
the
intersection
between
the
fitted
central
zone
of
the
sigmoidal
function
and
the
linear
function
adjusted
at
low
analyte
concentration,
and
the
UL
was
defined
as
the
value
tested
that
fit
the
linear
range.
Table
Application
2
range
comparison.
Study
for
LOD
and
precision
between
different
methods
used
with
these
multianalyte
disposable
sensors.
Precision
is given
as
RSD(%)
of
the
logarithm
of
the
concentration
at three
different
levels.
Method

Model
Function

Potassiuma
Magnesiumb
Hardnessc
Linear
range
(M)

Precision

Linear
range
(M)

Precision

Linear
range
(mg
L−1)
Precision
Low

Medium

High

Low

Medium

High

Low

Medium

High
DAD
Spectrophotometer
Sigmoidal
Linear

1.3
1.6
× 10−5–0.1

3.3
3.8
3.4
1.9
7.7
1.4
9.0
× 10−6–0.16

8.0
14.1

39.8
1.9–15,000

7.0
10.1
7.6
9.8

5.4
13.1
× 10−7–1.5
8.2
9.9

1.7
× 10−2–300,000
1.2–20,000
1.

Black
and
white
CCD
camera
Sigmoidal
Linear
× 10−6–1 × 10−2
2.49
3.63

3.75
3.83

2.27
2.35

6.9
1.9
× 10−5–1.0

5.35
6.83

6.23
5.01

6.81
5.20

3.88
5.88

7.71
5.74

4.49
5.41
× 10−8–0.1
5.4
2.7
× 10−6–1.0
1.4
2.0
× 10−2–20,000
1.7–200,000
2 × 10−2–200,000
4.2–200,000
4.3

Photographic
lamp)
digital
camera
(Halogen
Sigmoidal
Linear
× 10−6–1 × 10−2
7.2
7.6
5.4
4.8
2.2
2.4
× 10−4–1.5

4.7
6.8
6.8
7.7

8.7
7.8

6.1
6.6
4.4
4.3

6.7
7.8
× 10−7–0.1
6.4
3.2
× 10−6–1.5
1.4
2.7

Photographic
digital
camera
(Flash)
Sigmoidal
Linear
× 10−6–1.0
× 10−24.1
3.1
3.3
5.0
4.2
× 10−4–1.0

5.5
7.8
4.6
4.7

6.9
6.7

3.5
7.0
4.4
5.1

5.0
4.8
× 10−7–0.1

4.0 × 10−6–1.5
× 10−2–200,000

aK(I):
low,
medium
and
high
concentrations
used
for
DAD
spectrophotometer
(sigmoidal
model):
5 × 10−4, 5 × 10−3, 5 × 10−2(M);
(linear
model)
2.5
× 10−5, 2.5
× 10−4,
2.5
× 10−3(M);
the
other
methods
were
5 × 10−5, 1 × 10−4,
1
×10−3(M).
bMg(II):
cHardness:
concentrations
used
for
DAD
spectrophotometer
(sigmoidal
model):
1 × 10−3, 1 × 10−2, 5 × 10−1(M);
the
other
methods
were
5 × 10−4, 5 × 10−3, 1 × 10−1(M).
concentrations
used
for
DAD
spectrophotometer
(sigmoidal
model)
4,
17,
7980
(mg
L−1); the
other
methods
were
5,
100,
10,000
(mg
L−1).

Page 8

A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
335
Table
Determination
3
of
potassium,
magnesium
and
hardness
in different
types
of
samples
using
AAS
as reference
method.
Sample

Disposable
sensor

AAS

Pval(%)
K (mM)

s(1)
K (mM)

s(1)
Mineral
Tap
Tap
Tap
Stream
Snow
Rain
Well
Fruit
Beer
Red
water
(San
Vicente)

1.1 × 10−2
1.0 × 10−1
9.5
3.1
3.4
8.9 × 10−3
1.7 × 10−3
2.5 × 10−2
54
9.2
32
2.0 × 10−3
2.0 × 10−2
2.3 ×10−3
9.0 × 10−3
1.1 × 10−1
2.9 × 10−3
4.4 × 10−4
1.0 × 10−2
2.2
3.1
10
1.2 × 10−2
1.2 × 10−1
9.0 ×10−3
3.6 × 10−2
2.5 × 10−1
1.1 × 10−2
1.5 × 10−3
2.0 × 10−1
51
10
33
4.0 × 10−4
1.9 × 10−4
2.7
3.3 × 10−4
4.1 × 10−3
1.8 × 10−4
5.0 × 10−5
3.5 × 10−3
1.1
9.9
9.9 × 10−2
62.2
28.7
65.9
45.6
33.2
41.1
40.0
9.0
9.7
44.2
64.4
water
(Almu˜ necar,
Granada)

water
(Viznar,
Granada)
×10−3
×10−4
water
(Ogijares,
Granada)
× 10−2
water
(Lachar,
Granada)
× 10−1
water
(Sierra
Nevada,
Granada)

water
(Granada)

water
(Otura,
Granada)

juice
×10−1

×10−2
wine

SampleMg
(mM)

s(1)
Mg
(mM)

s(1)
Pval(%)
Mineral
Mineral
Mineral water
Tap
Tank
Snow
water
(Solán
de
Cabras)

1.2
3.0
2.7 × 10−2
1.2
1.9
2.0 × 10−3
4.5 × 10−1
1.2 × 10−1
1.1 × 10−3
6.1 × 10−2
4.2 × 10−2
1.0 × 10−3
8.9 × 10−1
3.0 × 10−1
2.7 × 10−2
9.8 × 10−1
1.5
2.2 × 10−3
8.8 × 10−3
1.8 × 10−3
1.9 × 10−4
1.4 × 10−2
3.8 × 10−2
6.3 × 10−5
22.6
18.0
98.7
27.7
14.8
83.7
water
(Lanjarón)
×10−1
(Bezoya)

water
(Ogijares,
Granada)

water
(Granada)

water
(Sierra
Nevada,
Granada)

Sample Hardness
(mg
L−1CaCO3)s(1)
Hardness
(mg
L−1CaCO3)s(1)
Pval(%)
Mineral
Mineral
Mineral
Tap
Well
Sea
Snow
Rain
Tank
water
(Mondariz)

37
90
12
236
432
8804

3.7
23
4.3
36
49
815
5.0 × 10−1
3.6 × 10−1
49
37
88
13

4.8 × 10−1
2.5 × 10−1
4.5 × 10−2
8.4 × 10−3
11
143
6.1 × 10−2
1.5 × 10−2
6.8
99.6
14.6
64.4
68.1
71.2
77.7
25.1
22.1
78.5
water
(Lanjarón)

water
(Bezoya)

water
(Monachil,
Granada)

227
421
8659

water
(Otura,
Granada)

water
(Almería)

water
(Sierra
Nevada,
Granada)1.2
2.7

8.0 × 10−1
2.3
418
water
(Granada)

water
(Granada)

409

EPA
case.
hardness
classification
in mg
L−1: Soft
(0–17.1),
Slightly
hard
(17.1–60),
Moderately
hard
(60–120),
Hard
(120–180),
Very
hard
(180
and
over);
s(1): Three
replicates
each
until
camera
neous
that
here
same
spectrophotometric
eras,
analyte
device.
the
external
light
source
– used
in the
black
and
white
CCD
and
in the
photographic
digital
camera
– reaches
a homoge-
and
stable
illumination,
and
moreover,
these
results
suggest
the
two
dimensional
CCD
sensor
array
technology
studied
by means
of
a photographic
digital
camera
can
yield
the
analytical
performance
compared
with
those
achieved
by
procedures
and
black
and
white
CCD
cam-
thus
making
quantitative
and
fast
determination
of
different
concentrations
possible
with
an
easy-to-use
handheld
4.5.
pH measurements
We also
lose
indicator
ranged
according
7.1,
bration
Tris
acquired
different
lamp.
Experimental
a
studied
test
strips
prepared
as
usual
on
an opaque
cellu-
support,
namely
commercial
pH strips.
Here,
we
employed
pH
papers
commonly
used
to estimate
pH visually,
which
between
6.5
and
10.0,
with
the
pH scale
being
divided,
to the
manufacturer,
at the
following
pH levels:
6.5,
6.8,
7.4,
7.7,
7.9,
8.1,
8.3,
8.5,
8.7,
9.0,
9.5
and
10.0.
The
pH cali-
was
performed
after
equilibrating
the
strips
for
3 min
with
buffer
and
using
4 replicates
in each
case.
The
images
were
with
the
same
photographic
digital
camera
using
two
light
sources,
the
built-in
flash
and
an external
halogen
data
were
adjusted
to a sigmoidal
function
using
Boltzmann
type
equation
(Eq.
(8)):
Channel
luminosity
= a2+
a1−
a2
1 +
exp(([H+] − ao)/a3)
(8)
where
using
a0to a3are
adjusting
coefficients.
The
best
fit
was
obtained
red
channel
data
when
the
halogen
lamp
was
used
as
a
flash.
with
functions.
The
using
4.6%
halogen
ing
determination
large
light
source
and
the
blue
channel
when
using
the
built-in
Fig.
7 presents
a
sequence
of
imaged
pH strips
captured
the
two
light
sources
noted
above
and
the
calibration
precision
was
evaluated
at
three
pH
levels
7.1,
8.1
and
9.0
8 different
pH strips
in each
case,
obtaining
RSD
of
2.1,
3.8,
and
between
1.3,
1.1
and
1.2%
using
the
built-in
flash
and
a
lamp,
respectively.
Furthermore,
using
the
digital
imag-
to determine
the
pH,
we
avoided
the
subjectivity
of
naked
eye
and,
additionally,
it makes
it possible
to monitor
numbers
of
sensors
simultaneously.
4.6.
Validation
and
application
Sample
analysis
of
waters
coming
from
different
sources
such
as
ages
were
in
using
in Table
(AAS)
able
ing
the
level.
In
lyzed
(RMSRE)
the
reference
mineral,
snow,
rain
and
sea
water,
as
well
as
juices
and
bever-
whose
analyte
concentration
covers
the
entire
linearised
range
checked
in order
to study
how
suitable
this
methodology
is
order
to assess
its
usefulness.
The
representative
measurements
the
multianalyte
disposable
sensor
procedure
are
outlined
3 and
compared
to the
atomic
absorption
spectrometry
method
used
as
a
reference.
With
these
results,
we
were
to assess
the
usefulness
of
the
proposed
methodology,
show-
that
there
are
no significant
statistical
differences
between
proposed
and
the
reference
method
at a
95%
of
confidence
addition,
and
considering
the
wide
concentration
range
ana-
in the
real
samples,
the
root
mean
squared
relative
error
[36]
was
selected
as
an appropriate
tool
for
evaluating
proposed
calibration’s
capacity
for
prediction
in relation
to the
values.
The
obtained
value
from
the
data
set
presented

Page 9

336
A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
Fig.
6.5
(2)
7.
Images
of
pH
paper
strips
equilibrated
at
different
pH
values
ranging
from
to 10
and
calibration
functions
with
experimental
data
for
pH,
(1)
Halogen
lamp,
Built-in
flash.
Error
bars
indicate
standard
deviation
of
each
pH measurement.
in Table
hardness.
3 is 0.13
for
potassium,
0.20
for
magnesium
and
0.16
for
5.
Conclusion
In
summary,
we
have
developed
here
a novel
approach
based
on
sive
magnesium
ages
We have
a
moidal
source
itation
of
paring
notice
for
those
easy
untrained
spectrometers.
a specific
arrangement
the
more
foresee
was
up.
the
use
of
digital
cameras
as
a rapid,
sensitive
and
inexpen-
measurement
system
to simultaneously
determine
potassium,
and
hardness
in a wide
variety
of
waters
and
bever-
without
the
need
for
expensive
devices
or additional
optics.
also
applied
the
procedure
to pH determination
using
commercial
pH strip,
adjusting
the
experimental
data
to a sig-
calibration
function.
The
use
of
a built-in
flash
as
the
light
simplifies
the
analytical
procedure
with
only
a
small
lim-
in the
analytical
performance.
Accordingly,
the
sensitivity
the
proposed
method
is
suitable
for
analytical
purposes
com-
with
the
reference
methods
reported
in the
work.
We
must
that
we
are
using
a device
that
it has
not
been
designed
analytical
purposes,
reaching
similar
values
with
respect
to
well
thought
for
it.
These
devices
are
relatively
inexpensive,
to use
and
ubiquitous
showing
capacity
for
in situ
analysis
by
people,
being
suggested
as
an alternative
over
traditional
As
an inherently
portable
device,
further
design
of
system
containing
the
digital
camera
with
the
studied
would
ensure
the
portability
of
this
system
beyond
controllable
conditions
presented
here.
Ongoing
research
in
practical
conditions,
under
sunlight,
is being
developed.
We
similar
results
with
those
presented
here,
since
the
flash
able
to overlap
the
external
illumination
in the
current
set-
This
line
is
feasible
and
will
expand
the
scope
of
this
method.
Other
halogen
arrangement.
Nevertheless,
analytical
address
user-friendly
sources
of
light
have
been
checked,
such
as
fluorescent
and
lamp
light,
observing
similar
results
in the
current
imaging
the
aim
of
this
paper
is
to present
an
easy-to-use
system
along
with
disposable
multianalyte
sensors
to
the
need
for
analytical
results
for
multiple
analytes
with
tools.
Acknowledgements
We
acknowledge
financial
support
from
the
Ministerio
de Cien-
cia
Plan
CTQ2005-09060-CO2-02,
14428-C02-02);
P08-FQM-3535).
pean
Domíguez-Meister
e Innovación,
Dirección
General
de
Investigación
y Gestión
del
Nacional
de I+D+i
(Spain)
(Projects
CTQ2005-09060-CO2-01,
CTQ2009-14428-C02-01
and
CTQ2009-
and
the
Junta
de
Andalucía
(Proyecto
de Excelencia
These
projects
were
partially
supported
by Euro-
Regional
Development
Funds
(ERDF).
Thanks
also
to S.
for
his
valuables
discussions
about
cameras.
References
[1]
[2]
[3] M.Y.
10
[4] U.
7
[5] K.
[6]
Jain,
[7] A.W.
(2008)
[8] C.
Misiakos,
[9]
[10] B.P.
[11] A.
Acta
[12]
[13] C. Zhang,
[14] M.C.
(2006)
[15]
Anal. Chem.
[16] J.
Methods
[17]
694.
[18]
[19] D.
[20]
[21] M.I.J.
204.
[22] E.
465.
[23] K.
74
[24]
(2005) 343.
[25]
88
[26]
Talanta
[27]
Santoyo-Gonzalez,
[28]
Ramos,
1215.
[29] W.E.
[30] K.
Hisamoto,
[31]
2013.
[32]
M.C.U.
[33] K.L.
[34] E.
[35]
ical Applications,
N.A.
Rakow,
K.S.
Suslick,
Nature
406
(2000)
710.
D.
Kutter,
G.
Figueiredo,
L.
Klemmer,
J. Clin.
Chem.
Clin.
Biochem.
25
(1987)
91.
Rubtsova,
J.V.
Samsonova,
A.M.
Egorov,
R.D.
Schmid,
Food
Agric.
Immunol.
(1998)
223.
Lemke,
K.
Cammann,
C.
Koetter,
C.
Sundermeier,
M.
Knoll,
Sens.
Actuators
B
(1992)
488.
Abe,
K.
Suzuki,
D.
Citterio,
Anal.
Chem.
80
(2008)
6928.
A.K.
Ellerbee,
S.T.
Phillips,
A.C.
Siegel,
K.A.
Mirica,
A.W.
Martinez,
P.
Striehl,
N.
M. Prentiss,
G.M.
Whitesides,
Anal.
Chem.
81
(2009)
8447.
Martinez,
S.T.
Phillips,
G.M.
Whitesides,
Proc.
Natl.
Acad.
Sci.
U.S.A.
105
19606.
Mastichiadis,
S.E.
Kakabakos,
I.
Christofidis,
M.A.
Koupparis,
C.
Willetts,
K.
Anal.
Chem.
74 (2002)
6064.
C.
Malins,
M.
Niggemann,
B.D.
MacCraith,
Meas.
Sci.
Technol.
11 (2000)
1105.
Corgier,
C.A.
Marquette,
L.J.
Blum,
Anal.
Chim.
Acta
538
(2005)
1.
Lapresta-Fernandez,
R.
Huertas,
M.
Melgosa,
L.F.
Capitan-Vallvey,
Anal.
Chim.
636
(2009)
210.
M.
Kompany-Zareh,
M.
Mansourian,
F.
Ravaee,
Anal.
Chim.
Acta
471
(2002)
97.
K.S.
Suslick,
J. Am.
Chem.
Soc.
127
(2005)
11548.
Janzen,
J.B.
Ponder,
D.P.
Bailey,
C.K.
Ingison,
K.S.
Suslick,
Anal.
Chem.
78
3591.
A.W.
Martinez,
S.T.
Phillips,
E.
Carrilho,
S.W.
Thomas,
H.
Sindi,
G.M.
Whitesides,
80
(2008)
3699.
Gabrielson,
M.
Hart,
A.
Jarelov,
I.
Kuhn,
D.
McKenzie,
R.
Mollby,
J. Microbiol.
50
(2002)
63.
A.
Lapresta-Fernandez,
L.F.
Capitan-Vallvey,
Sens.
Actuators
B 134
(2008)
K.
Tohda,
M. Gratzl,
Anal.
Sci.
22
(2006)
383.
Filippini,
I. Lundstrom,
The
Analyst
131
(2006)
111.
V.V.
Apyari,
S.G.
Dmitrienko,
J. Anal.
Chem.
63
(2008)
530.
Stich,
S.M.
Borisov,
U.
Henne,
M.
Schaeferling,
Sens.
Actuators
B 139
(2009)
Hirayama,
T.
Sugiyama,
H.
Hisamoto,
K.
Suzuki,
Anal.
Chem.
72
(2000)
Suzuki,
E.
Hirayama,
T. Sugiyama,
K.
Yasuda,
H.
Okabe,
D.
Citterio,
Anal.
Chem.
(2002)
5766.
Y.
Suzuki,
K.
Suzuki,
Front.
Chem.
Sens.,
Springer
Ser.
Chem.
Sens.
Biosens.
3
L.F.
Capitan-Vallvey,
M.D.
Fernandez
Ramos,
M.
Al
Natsheh,
Sens.
Actuators
B
(2003)
217.
L.F.
Capitan-Vallvey,
M.D.
Fernandez-Ramos,
P. Alvarez
de Cienfuegos,
J. Gomez,
65 (2005)
239.
L.F.
Capitan-Vallvey,
M.D.
Fernandez-Ramos,
P.
Alvarez
de
Cienfuegos,
F.
Anal.
Chim.
Acta
481
(2003)
139.
A.J.
Palma,
A.
Lapresta-Fernandez,
J.M.
Ortigosa-Moreno,
M.D.
Fernandez-
M.A.
Carvajal,
L.F.
Capitan-Vallvey,
Anal.
Bioanal.
Chem.
386
(2006)
Morf,
K.
Seiler,
B.
Rusterholz,
W.
Simon,
Anal.
Chem.
62
(1990)
738.
Suzuki,
K.
Watanabe,
Y.
Matsumoto,
M.
Kobayashi,
S.
Sato,
D.
Siswanta,
H.
Anal.
Chem.
67
(1995)
324.
M.
Rouilly,
M. Badertscher,
E.
Pretsch,
G.
Suter,
W.
Simon,
Anal.
Chem.
60
(1988)
E.d.N.
Gaiao,
V.L.
Martins,
W.d.S.
Lyra,
L.
Farias
de Almeida,
E. Cirino
da
Silva,
Araujo,
Anal.
Chim.
Acta
570
(2006)
283.
Yam,
S.E.
Papadakis,
J. Food
Eng.
61
(2004)
137.
Bakker,
P.
Bühlmann,
E.
Pretsch,
Chem.
Rev.
97
(1997)
3083.
U.E.
Spichiger-Keller,
Chemical
Sensors
and
Biosensors
for
Medical
and
Biolog-
1st
ed.,
Wiley-VCH,
Weinheim,
1998.

Page 10

A.
Lapresta-Fernández,
L.F.
Capitán-Vallvey
/ Analytica
Chimica
Acta
706 (2011) 328–
337
337
[36]
L.F.
Cuadros-Rodriguez,
[37] A.
Bioanal.
[38] V.
(2008)
Capitan-Vallvey,
A.
Lapresta-Fernandez,
M.D.
Fernandez-Ramos,
L.
Sens.
Actuators
B 117
(2006)
27.
Lapresta-Fernandez,
R.
Huertas,
M.
Melgosa,
L.F.
Capitan-Vallvey,
Anal.
Chem.
393
(2009)
1361.
Lebourgeois,
A.
Bégué,
S.
Labbe,
B.
Mallavan,
L.
Prevot,
B.
Roux,
Sensors
8
7300.
[39]
X.D.
4907.
Wang,
R.J.
Meier,
M.
Link,
O.S.
Wolfbeis,
Angew.
Chem.
Int.
Ed.
49 (2010)
[40]
A.
c1an15204a.
[41] Lapresta-Fernandez,
Lapresta-Fernandez,
L.F.
Capitan-Vallvey,
Analyst
(2011),
doi:10.1039/
A.
PhD
Thesis
University
of
Granada,
2007.