Thus, the analytical curve is presented as −log (I/I0) vs. the cop- per concentration.

For this, a
proper R G or B channel was used. Other mathematical manipulation could be performed such
as the R, G, or B values directly, or its combination as R × G × B generating exponential,
polynomial or linear behavior versus concentration. However, −log (I/ I0) is commonly used
once it is linearly correlated with the concentra- tion being very employed for colorimetry and
spectrophotometry [30, 37,38]. By using this calculation, it is possible to use a reference value
(Blank) and to avoid performing the subtraction from the analytical sig- nal, being similar to the
Lambert-Beer law for colorimetry. The signal captured from a digital camera is due to
reflective, however, this signal is correlated with the absorbed signal, and so, the logarithm
signal can be used without loss of performance.

1.1. Optimization of Operating Parameters of Imaging System

To evaluate the effect of light intensity on the analytical response it was employed a
luximeter and the light intensity incident varied at five levels: 20, 35, 90, 240 and 980 LUX.
To evaluate the effect of sample volume on the analytical response, it was varied the volume of
the solu- tion of the dye Brilliant Blue 1.0 × 10−5 mol L−1 of 100 to 800 μL with 100 μL of
increment on the spot. To verify the RGB channels linear re- sponse due to the analyte
concentration, analytical curves were con- structed employing the dyes Brilliant Blue,
Tartrazine and Direct Red 80, whose concentration ranges were tested from
1.00
× 10−6 mol L−1 to 1.00 × 10−4 mol L−1, using the light intensity and
the volume determined in earlier studies. The curves were constructed in triplicate, an image for
each curve was captured and the coefficient of determination was evaluated.
The effect of exchanging cameras on the analytical response was assessed by comparison
of two standard curves constructed using Bril- liant Blue dye, and the images were captured by
a camera Sony Cyber- shot DSC-W610 with resolution 14.1 Megapixels (MP) and also through
camera Sony Cyber-shot DSC-S730 with 7.2 MP. Assays were performed in replicate (n = 7)
with only one image being captured for each test. The results were compared using the paired
t-test with 95% confidence.
1.2. Copper(II) Determination in Sugar Cane Spirit

For the copper determination in sugar cane spirit, it was employed the cuprizone, a
bidentate organic reagent, which forms with copper a blue chelate in a slightly alkaline
medium [39], as shown in Fig. 2.
Initially, studies were conducted to select the concentration of cuprizone that would
generate a range of suitable colors for screening analysis type. The concentration of
cuprizone was evaluated in four levels using a fixed concentration of Cu(II) (1.25 mg
L−1; 0.0019 × 10−3 mol L−1): 44 (0.16 × 10−3 mol L−1), 88 (0.32
× 10−3 mol L−1), 132 (0.48 × 10−3 mol L−1) and 176 mg L−1 (0.64 × 10−3 mol L−1) all of
them with a fixed volume of 100 μL. Furthermore, to verify the effect of the kinetic of reaction
on the analytical response, the reaction was monitored at 1 min intervals for 10 min employing
a solution of Cu(II) 1.25 mg L−1 (0.0019 × 10−3 mol L−1) at a volume of 400 μL. The analytical
blank was obtained by replacing the Cu(II) so- lution for distilled water. After reaction
optimization, the volumes were fixed in: 100 μL phosphate buffer pH 9, 400 μL solution of
Cu(II)1.25 mg L−1 or sample and 100 μL of cuprizone. The tests were per- formed in
triplicate, and an image was captured for each test.
Effects of possible interferents were evaluated for determining Cu(II) in different proportions
of analyte/interferent. The study of the matrix effect was done by comparing the parameters of
the analytical curve obtained for standard aqueous solutions of Cu(II) with the param- eters of
analytical curve using a sugar cane spirit free of copper(II). The mean responses were compared
using statistical test F (Snedecor) of homogeneity of variances and through the paired t-test with
95% confidence.
Recovery studies were tested in unaged and aged sugar cane spirit samples employing
standard addition method using three concentra- tion levels: 0.75; 2.88 and 5.00 mg L−1. The
accuracy of the method was evaluated by comparing the results obtained by digital analysis
method with the reference method (Flame Atomic Absorption Spectro- photometry) [40]. The
results obtained by both methods were com- pared using the paired t-test at 95% confidence.
The DIB method was applied for determination of Cu(II) in samples of aged and unaged sugar
cane spirit collected in supermarkets in the cities of Viçosa and Bom Jesus do Amparo - Minas
Gerais/Brazil. No treatment of samples was need before the analysis.

1. Results and Discussion

1.1. Imaging System Optimization

After preliminary tests, the light intensity was set at 20 LUX due to the greater magnitude
of the analytical signal allowing the measure- ment of lower concentrations of the analyte.
To verify the effect on the analytical response, volumes of solutions from 400 to 800 μL
were used on the spot, and the intensity of the analytical signal remained nearly constant,
with a slight increase until 600 μL. Using 600 μL, an ex- cess of light “brightness regions”
appeared only on the edges of the spot, and thus in the center more homogeneous regions
were obtained. Ac- cording to Benedetti et al. [29], the analytical response can be affected
by sample volume similar to optical path in spectrophotometry. In fact, the depth of the
vessel is correlated with the absorbance and con- sequently the reflected RGB values. Few
microliters of the colorimetric reaction generate poor sensitivity. However, high volumes
generate a great amount of residue. Thus, 600 μL was selected for further studies.
Calibration curves were built to verify the linearity of concentration range to some solution
based on dyes. The linear ranges obtained for the concentration of Brilliant Blue, Direct Red
80 and Tartrazine were: 1.75
× 10−5 to −41.00 × 10−1
−4
mol L−1, 1.00 × 10−6 to 8.35 × 10−5 mol L−1 and 21.00 × 10−6 to
1.00 × 10 mol L , respectively. The coefficients of de- termination (R ) were obtained
exceeding 0.99, what indicated that the channels R, G and B linearly respond to these
studied concentration range. According to Choodum [34], R, G and B channels absorb
wave- lengths of light in the range 580–700 nm, 500–580 nm and 400– 500 nm,
respectively. Therefore, the spectrums obtained for the dyes were compared with the
response obtained by the RGB model showing agreement results and that the Blue
Coomassie Brilliant presents re- sponse in the R channel, Direct Red 80 in the G channel
and Tartrazine on channel B.
The effect of the camera resolution was evaluated by comparing standard curves
constructed using Brilliant blue dye. The test was done by comparing all analytical
responses for each concentration level. The 0.50 value t calculated was lower than the
critical t (3.18), in- dicating that the camera resolution did not influence the analytical
 methodology. Thus, low cost photographic cameras and smartphones can be used, being
very attractive and encouraging for small producers to use them for quality control
purposes.

1.3. Copper(II) Determination in Sugar Cane Spirit

The colorimetric reaction between Cu(II) and cuprizone is a mixture of colors and therefore,
more than one channel can present a satisfacto- ry response to variations in the concentrations
of the chelate. Thus, R, G and B channels were monitored in order to determine which of them
has greater sensitivity as a function of the copper(II) concentration. The channel R showed a
better response, so it was employed for the de- termination of copper in the carried out studies.
In fact, the red color is exactly the complementary color from the reaction blue color. Green
and blue channel presented lower sensitivity than red, but the blue channel presented lower
sensitivity than expected once the blue radia- tion is reflected by blue solution.
For analyzes of screening type (fast response) involving colorimetric reactions of spot tests,
it is essential to find the most stable and color gradient - changing in color levels. In this
regard, different concentra- tions of cuprizone were tested from 44 to 176 mg L −1. The
evaluated cuprizone concentrations were always higher than the range of concen- tration of the
tested copper solutions to ensure its maximum complexation.
For all evaluated levels, a fixed volume of 100 μL of cuprizone solu- tion was used. For
lower concentrations of cuprizone such as 44 and 88 mg L−1, the color remained almost the
same in all spots. In fact, low concentrations of chelant reduce the formation of the product,
and so, lower sensitivity. Therefore, the experiments were continued employing the cuprizone
solution of 132 mg L−1. For more concentrat- ed cuprizone solution, the analytical signal did
not increase significantly.
It was evaluated the reaction kinetic of cuprizone and copper ions. The analytical signal
increased up to 7 min of reaction as shown in Fig. After this time, there was a decrease of the
analytical signal, probably due to the low stability of the generated chelate.
After checking the optimal reagent concentrations, volumes used for each solution were set
as follows: 400 μL Cu(II), 100 μL of phosphate buffer pH 9 and 100 μL of cuprizone solution.
Fig. 4 shows an image of
spot test containing copper concentrations in the range from 0 to 5 mg L−1 after optimized
reaction conditions.

1.4. Analytical Features

An analytical curve was constructed containing copper concentra- tions in the range from
0.75 to 5.00 mg L−1 using the R channel as shown in Fig. 4. The linearity of the developed
method was verified by calibration curve analysis using the method named least squares
methods (LSM) following the regression equation: −log (I/I0) = −0.0113 + 0.3290 ×
[Copper(II)], where [Copper(II)], is mg L−1 and r = 0.9999. The limits of detection (LOD)
and quantification (LOQ) were calculated using the relations 3 × σ/m and 10 × σ/m,
respectively, where σ is the standard deviation of ten times blank analytical and m the slope
of the analytical curve. Thus, the LOD and LOQ were 0.078 mg L−1 and 0.26 mg L−1.
Some metallic ions that could be present in sugar cane spirit were tested, as described
by [41] which consequently may form chelates with cuprizone. Previously, potential
interferences showed in Table 1 were analyzed by Flame Atomic Absorption Spectrometry
(FAAS) in all sugar cane spirit samples used in this study. The aluminum and zinc were
found in some samples, however in concentrations below the quantification limit. The iron
ions were found in only one sample at a concentration below the quantification limit. Other
metals were not detected in sugar cane spirit samples. Therefore, the metals ana- lyzed
were not considered important in determining interfering Cu(II) employing cuprizone as a
complexing agent.
Recovery studies were established in aged and unaged sugar cane spirits in three
concentration levels: 0.75; 2.88 and 5.00 mg L−1. Recov- ery percentages ranged from 98.3
to 104% were obtained. The results were compared with the values described by [40] that
recommends a recovery interval between 80 and 110% to concentration levels of
0.1 mg L−1 to mg L−1. Therefore, all the values obtained from the recov- ery study were
satisfactory. The analytical curves obtained from aque- ous solutions and sugar cane spirit in
the absence of Cu(II) were compared using F test (Snedecor), and it was found that there are
no differences among the variances for the obtained curves.
For comparison of means pairs in each concentration level it was used the paired t-test
with 95% confidence with n = 3. The 1.19 t value was lower than the t critical (2.77),
showing no statistically signif- icant differences in the results by both generated curves. It
can be con- cluded that there were no significant effects of matrices in samples of sugar
cane spirit. In addition to the recovery tests, the accuracy was evaluated by comparing the
results obtained using the method devel- oped by the results provided by the reference
method (Flame Atomic Absorption Spectrophotometry) [41].
Some samples of sugar cane spirit were analyzed using the proposed method and the reference
method. The results are shown in Table 2. The 1.29 t value calculated was lower than the
critical t (2.30) for n = 3, therefore, no statistically significant differences in the results were
gen- erated by both methods at a confidence level of 95% [42]. It can be con- cluded that
the six samples of sugar cane spirit analyzed presented copper(II) concentrations within
the limits required by the Ministry of Agriculture, Livestock and Supply, except AE11 sample,
whose concen- tration of Cu(II) was approximately eight times higher than allowed by law.
In Table 3 some analytical features acquired from DIB were com- pared to those
obtained by other methods reported in the literature for determination of Cu(II) in sugar
cane to evaluate its performance. Molecular spectrophotometry, atomic absorption
spectrophotometry and voltammetric methods are initially presented, followed by the dig-
ital image based method.

3. Conclusions

The method developed for the determination of copper(II) in sugar cane spirit presented
advantages such as practicality, quickness, low cost, reduced reagent consumption and
waste generation. The device employed for image capture is easy to build and it uses low cost
compo- nents such as LEDs, battery and accessible apparatus as smartphones or digital
camera. In addition, it is versatile, robust and portable, very inter- esting for in situ analysis.
The digital image method presented good repeatability and robust- ness to determine copper,
mainly because of the apparatus developed to avoid lack of homogenous with excess of lighting
or shadow. Besides, with this method, only one photography is enough to complete the an-
alytical analysis- analytical curve and sample- generating only 600 μL/ spot (miniaturization)
at 7 min of analysis. Thus, each analysis could be performed at 35 s. The DIB method presented
a good linear range, precision and LOD compared to some spectrophotometric methods as
expected, and compatible performance as the reported voltammetric method. Thus, the spot test
with digital analysis is ideal to green chem- istry purposes. Moreover, this method is a useful
analytical tool for lab- oratory with limited financial resource and for quality control in industries
and small producers.
Besides, comparing with a work in the literature employing an in- ductively coupled
plasma atomic emission spectroscopy (ICP OES), it showed that all metals tested were
found in concentrations lower than the copper concentration in sugar cane spirit samples
ratifying the obtained results. Finally, the method can be used for chemical, clin- ical, medical,
forensic, environmental analysis, and as an analytical tool with high potential.