J Am Oil Chem Soc DOI 10.1007/s11746-013-2300-6
ORIGINAL PAPER
Characterization of Carnauba Wax Inorganic Content Allan N. de S. Dantas • Ticiane A. Magalha˜es Wladiana O. Matos • Sandro T. Gouveia • Gisele S. Lopes
•
Received: 15 January 2013 / Revised: 12 May 2013 / Accepted: 4 July 2013 Ó AOCS 2013
Abstract In this work the analysis of inorganic elements (Al, Ca, Cu, Fe, K, Mg, Mn, Na and Zn) in different types of carnauba waxes (types 1, 3 and 4) was implemented. The Box-Behnken experimental design was used to optimize the digestion of the carnauba wax sample using a microwave-assisted approach. The following parameters were evaluated: microwave power applied (600–1,000 W), time of microwave power application (5–20 min) and nitric acid volume (1–4 mL). The residual carbon content (%RCC) was measured by ICP OES (inductively coupled plasma optical emission spectrometry) to evaluate the efficiency of the digestion. The %RCC values in all of the experiments were below 16 %. The best conditions for carnauba wax digestion were found: 800 W applied power for 15 min using 2.5 mL of HNO3. In these conditions the %RCC was lower than 4 %. The amounts of Al, Ca, Cu, Fe, K, Mg, Mn, Na and Zn in these samples were determined by ICP OES. The average contents of Al, Ca, Fe and K found in the carnauba wax type 1 were 28.6 ± 1.5, 33.8 ± 2.8, 18.5 ± 1.1 and 37.2 ± 2.5 mg kg-1, respectively. For carnauba wax types 3 and 4 larger amounts were found. The principal components analysis (PCA) showed
A. N. de S. Dantas T. A. Magalha˜es W. O. Matos S. T. Gouveia G. S. Lopes (&) Laborato´rio de Estudos em Quı´mica Aplicada (LEQA), Departamento de Quı´mica Analı´tica e Fı´sico-Quı´mica, Universidade Federal do Ceara´, Campus do Pici, Fortaleza, CE, Brazil e-mail:
[email protected] A. N. de S. Dantas Instituto Federal de Educac¸a˜o, Cieˆncia e Tecnologia do Rio Grande do Norte-IFRN, Caˆmpus Nova Cruz, Nova Cruz, RN, Brazil
three groups of carnauba wax with the first two principal components. Keywords Carnauba waxes Sample preparation Elemental analysis ICP OES
Introduction The carnauba palm tree (Copernicia prunifera) produces a powder that covers its leaves and aids in the conservation of moisture within the plant by preventing evaporation from the surface of the leaves. From this powder is extracted one of the most valuable waxes used in industry because it has the highest melting point among the commercial vegetable waxes and often varies to some extent depending not only on the environmental conditions but also on the growth stage of leaves from which the wax was collected [1]. According to the Official Brazilian Regulations entrusted with this subjected matter, there are three types of wax obtained from carnauba palm leaves. The first type (Type 1, T1: prime yellow) is obtained from the young leaves that have a lower maturation time. Other types (Type 3, T3: light fatty; and Type 4, T4: fatty grey) of wax are obtained from the adult leaves that present superior maturation time. The leaves are cut, left to dry and beaten by hand or machine and the powder they contain is produced. The powder is cleaned and dissolved in adequate aliphatic solvent. Processes such as distillation, filtration and chemical bleaching with hydrogen peroxide complete the cycle of carnauba wax production. One of the difficulties faced by the industry is that the carnauba waxes are commonly classified based only on the shade of color. The difficulties mentioned consist of finding a suitable way to distinguish between the products presented by wax
123
J Am Oil Chem Soc
refineries. Therefore, this fact shows a lack of a standard product to supply to an increasingly demanding market. Thereby a classification according to the chemical characteristics would be very helpful. Another problem in the industry is the lack of studies in the literature related to the inorganic chemical composition of the different types of carnauba wax. To ensure the quality of carnauba waxes produced in Brazil, standard methods of analysis for these samples must be developed. The ABNT (Associac¸a˜o Brasileira de Normas Te´cnicas) [2] is the Brazilian organization that is responsible for technical standardization. The ABNT establishes national legislation to determine the acidity index, saponification, ester index and melting range of carnauba waxes. Currently, the ABNT uses the color distinction to classify carnauba waxes by type. There is no legislation regarding the presence of inorganic elements in carnauba waxes, even though this analysis has been necessary to export raw carnauba wax material to other countries. The standard rules for the quality control of the carnauba waxes are under study in Brazil [2]. Thus, the development of a method for inorganic trace element analysis is important to estimate the quality of the wax. Furthermore, special consideration about trace elements is important in order to know in which concentration range can be found in the different types of waxes (T1, T3 and T4) that are used as raw material in the manufacture of different types of materials. Carnauba waxes are complex mixtures that consist mainly of free fatty acids, alcohols and long-chain esters [3, 4]. Long chains of alcohols with even numbers of carbon atoms (C28–C34) and esters with even numbers of carbon atoms (C44–C62) have been observed [3–8]. Therefore, the organic composition of carnauba waxes has been reported in the literature several times. Few studies related to the inorganic composition of the vegetable wax and the preparation of carnauba wax samples for inorganic analysis were found in a search of the literature. Krivan et al. [9] investigated the use of wax from spruce needles as a bioindicator of the presence of heavy metals in the environment. The sample preparation of the spruce needle wax used chloroform to extract the wax from the tree needles. The chloroform was evaporated, and the wax was decomposed using an acid mixture (HNO3/HF/ HClO4) to determine Cd, Cu, Fe and Pb by atomic absorption spectrometry and neutron activation analysis. The authors do not comment on the volumes of the reactants or the heating procedures used during the sample preparation. Rademeyer et al. [10] studied a method using a V-groove hot nebulizer and spray chamber for introducing molten wax directly into the ICP OES. The sample introduction system heats and maintains the molten wax at the
123
appropriate temperature. However, the system used requires a complex apparatus to introduce the sample into the plasma, which may cause clogging problems due to an inadequate temperature control by the hot nebulization system (215 °C). Another factor to be noted is the impossibility of obtaining an analytical blank and to evaluate the background signal due to the high organic matter content of the injected sample. Microwave-assisted digestion procedures using closed systems have commonly been used because there is a fast and an efficient technique for sample preparation [11]. High pressures and temperatures are reached when decompositions are performed in a cavity microwave oven. Elevated temperature and pressure associated with oxidants as reactants are efficient for the degradation of organic matter. Several studies report the use of cavity microwave oven methods for sample preparation for food samples [12, 13], soils and sediments [14–16], biological materials [17] and other organic materials. This procedure is commonly associated with experimental designs [18] in order to reduce the number of experiments needed and to determine the conditions under which the residual carbon content can be used as reference to evaluate the quality of the digestion procedure. Principal Component Analysis (PCA) is a chemometric tool that has been extensively used for classification, pattern recognition and multivariate calibration. Principal component loadings identify responses that vary as a function of the experimental factors being manipulated by the investigator and they are highly correlated. Principal component scores allow a significant reduction in the number of response surfaces to be analyzed [19]. The aim of this work was to optimize an analytical procedure for the determination of elements in carnauba waxes samples by ICP OES with precision and accuracy.
Materials and Methods Samples, Reagents and Standards The samples were provided by Sindicato das Indu´strias Refinadoras de Cera de Carnau´ba do Ceara´ (SINDICAR´ BA, Ceara´, Brazil). Three samples of each type of NAU carnauba wax (1, 3 and 4) were collected from different producers of Ceara´, Brazil. All of the solutions were prepared using ultrapure water (resistivity 18.2 MX cm) obtained from a Milli-QÒ water purification system (Millipore, Bedford, MA, USA). All of the glassware and volumetric flasks were immersed in 10 % v/v nitric acid (Vetec, Rio de Janeiro, Brazil) for 24 h and rinsed with ultrapure water prior to use.
J Am Oil Chem Soc
For sample digestions, concentrated nitric acid (65 % v/v) (Vetec, Rio de Janeiro, Brazil) and hydrogen peroxide (30 % w/w) (Vetec, Rio de Janeiro, Brazil) were employed. Reference solutions were prepared after successive dilutions from 1000 mg L-1 Al, Ca, Cu, Fe, K, Mg, Mn, Na and Zn stock solutions (Acros Organics, Geel, Belgium). Stock solutions containing carbon (5.0 % w/v) were prepared using urea (CH4N2O, Vetec, Rio de Janeiro, Brazil). Standard solutions containing 0.05, 0.1, 0.2, 0.5, 1.0 and 2.5 % carbon were prepared in 1.4 mol L-1 HNO3. The Milli-QÒ water used to prepare the carbon analytical curve was previously heated to eliminate dissolved CO2 to perform the measurements for %RCC. The accuracy of the method was evaluated by the analysis of certified reference material Apple Leaves 1515a (National Institute of Standards and Technology, Gaithersburg, MD, USA). Instrumentation For total digestion of samples, a cavity microwave oven (MultiwaveÒ, Anton Paar, Graz, Austria) equipped with six quartz closed vessels and temperature/pressure sensors was used. An ICP OES dual view Optima 4300 DV (PerkinElmer, Waltham, USA) was used for Al, Ca, Cu, Fe, K, Mg, Mn, Na and Zn determinations. The ICP sample introduction system was a cross-flow nebulizer with a double pass spray chamber. The ICP operating parameters were as follows: 40 MHz generator frequency; 1.1 kW radio frequency power; 15 L min-1 plasma argon-flow rate; 0.5 L min-1 auxiliary argon-flow rate; 0.8 L min-1 nebulizer argonflow rate; and 1.4 mL min-1 sample flow rate. A central tube torch with a 2.4 mm internal diameter was used. The wavelengths of the elements and the viewing position of the torch in the ICP OES are presented in Table 1.
Table 1 Wavelengths of the monitored elements and the viewing position of the ICP OES instrument Element
Wavelength (nm)
View position
Al
396.153
Axial
C
193.030
Axial
C
193.030
Radial
Ca
317.933
Axial
Cu
327.393
Axial
Fe K
238.204 766.490
Axial Radial
Mg
285.213
Mn
259.372
Na
588.995
Radial
Zn
213.857
Axial
Sample Preparation in a Microwave Oven For the cavity microwave oven decomposition, a sample mass of 250 mg was digested using 2.5 mL of 65 % v/v HNO3 plus 1 mL of 30 % w/w H2O2. After digestion, the solutions were diluted to 20 mL with ultrapure water. The microwave oven heating program is shown in Table 2. The carnauba wax samples were digested using the proposed optimized methodology, and the amounts of the elements were determined by ICP OES. Experimental Design and Statistical Analysis Experimental design is widely used to optimize the parameters in several processes, and the use of experimental design decreases the number of experiments required and minimizes the errors. In this study, the BoxBehnken Design [20] was applied to optimize sample preparation conditions for carnauba wax samples. The variables evaluated were heating power, time of heating and nitric acid volume. Table 3 shows the relationship between these factors and the levels established. The response variable adopted was residual carbon content, and all measurements were performed randomly. This experimental design, a rotatable second-order design based on a three-level incomplete factorial design, allows the number of design points to increase at the same rate as the number of polynomial coefficients. For three factors, the design can be constructed as three blocks of four experiments consisting of a full two-factor factorial design with the level of the third factor set at zero. For the Box-Behnken Design, the number of experiments (N) required can be defined by the equation N = 2 K(K-1) ? Co where K is related to the number of factors to be tested and Co is related to the number of center point replicates. The total experiments generated for this design were 15. In the present system involving three independent variables, the mathematical relationship to the response can be obtained by a quadratic (second degree) polynomial equation (Eq. 1): Y ¼ b0 þ b1 X1 þ b2 X2 þ B3 X3 þ b12 X1 X2 þ b13 X1 X3 þ b23 X2 X3 þ b11 X12 þ b22 X22 þ b33 X32
ð1Þ
Table 2 Heating program used for digestion in the microwave oven Step
Power (W)
Time (min)
Axial
1
100–500
5.0
Axial
2 3
X1 0
X2 15.0
X1: 600–1,000 W, X2: 5–20 min
123
J Am Oil Chem Soc Table 3 Experimental maximum and minimum levels of independent variables Variables
High level (?)
Center level (0)
Low level (-)
X1: Power (W)
1,000
800
600
X2: Time (min)
20
12.5
5
X3: HNO3 volume (mL)
4.0
2.0
1.0
Figure of Merit for ICP OES Analysis Linearity was studied by means of an aqueous standard solution curve (n = 3) constructed with the elements cited previously in concentrations ranging from 0.5 to 12 mg L-1. The limits of detection and quantification (LOD and LOQ) were obtained employing robust conditions and following IUPAC [21] recommendations according to Eqs. (2) and (3), below. These equations use the background equivalent concentration (BEC) and signal-to-background ratio (SBR): BEC = Crs/SBR; SBR = Irs-Iblank/Iblank, where Crs is the concentration of the multi-elemental reference solution (5.0 mg L-1); Irs and Iblank are the emission intensities from the multi-elemental reference and blank solutions, respectively; and RSDblank is characterized as a relative standard deviation of the blank (n = 10). LOD ¼ ð3 BEC RSDblank Þ=100
ð2Þ
LOQ ¼ ð10 BEC RSDblank Þ=100
ð3Þ
Multivariate Analysis
Table 4 Box-Behnken design matrix with three independent variables expressed in coded units and response presented in function of residual carbon content (%RCC) Experiment
X1
X2
X3
%RCC
1
-1
-1
0
15.3
2
1
-1
0
3.8
3 4
-1 1
1 1
0 0
10.1 1.2
5
-1
0
-1
15
6
1
0
-1
7
-1
0
1
11.6
2.3
8
1
0
1
1.2
9
0
-1
-1
9.9
10
0
1
-1
7.6
11
0
-1
1
6.8
12
0
1
1
4.7
C*
0
0
0
5.8
C*
0
0
0
5.1
C*
0
0
0
4.8
* Central point
Multivariate analysis was applied for data association. The principal components analyses (PCA) were applied to the concentrations values obtained for the mineral composition of the carnauba waxes and these PCA were performed on auto-scaled data. The data matrix was constructed on columns for the samples and on rows for the elements and the software Unscrambler 7.5 (CAMO) was used.
Results and Discussion Sample Preparation Using Microwave Wet Digestion The total digestion of carnauba wax is necessary to perform elemental analysis by ICP OES since it has a high content of organic compounds mainly esters (86 %), accompanied by small amounts of free fatty acids and alcohols, hydrocarbons and resins [4]. Carnauba wax therefore requires special care in sample preparation. The heating program that was used consisted of several steps in which the microwave power was gradually increased. In an analytical sequence, the sample preparation stage often requires multivariate techniques because the variables may influence each other, impairing the final result. Table 4 shows
123
the results for the Box-Behnken Design and its response (%RCC) for each experiment tested. The %RCC values were \16 % for all experiments. Temperatures of approximately 200 °C were reached, and the only distinction observed among the experiments was a small difference in the color of the solution (slightly yellowish) when the minimum conditions were applied. All the other digestions produced clear solutions. This color difference is probably due to the greater organic matter content in the solutions that correspond to experiments 1 and 5. The lowest %RCC values were obtained in experiments 4, 6 and 8, in which the maximum power was applied, and in the center point experiments, the %RCC values were lower than 10 %. By applying multiple regression analysis to the variables and the respective responses, the final equation obtained can be observed below (Eq. 4). RCC ¼ 5:23 0:027 X1 0:036 X12 0:20 X2 0:60 X22 0:88 X3 þ 0:56 X32 þ 0:37 X1 X2 þ 0:33 X1 X3 þ 0:029 X2 X3
ð4Þ
In order to evaluate the accuracy of the developed model, an analysis of variance (ANOVA) was performed.
J Am Oil Chem Soc
The ANOVA presents the values of variation due to the treatment (change in the combinations of variables levels) with the variation due to random errors inherent to the measurements of the generated responses. From this comparison it is possible to appraise the significance of the regression used to predict responses considering the sources of experimental variance. The significance of regression model can be estimated by the ratio between the media of the square of regression (MSSreg) and the media of the squares of residuals (MSSres) and by comparing these variation source using the Fisher distribution (F test). In Table 5, the ANOVA for the quadratic regression model was highly significant, as shown by the F-test (Fregression = 58.32), which was accompanied by a low probability value (Pregression = 0.00007). The calculated F value (58.32) was greater than the critical F value (4.77) at the 5 % level. The obtained model constructed after fitting to the data can sometimes not perfectly describe the experimental domain studied. In this purpose, the model obtained in this work was evaluated using the Lack of Fit and Pure Error statistical parameters [22]. As of central point replicates was possible to estimate the pure error associated with repetitions. If the obtained model is well fitted to the experimental data, MSSlof should reflect only the random errors inherent to the system. Additionally, MSSpe is also estimated of these random errors, and it is assumed that these two values are not statistically different [23]. The values obtained and showed in Table 5 for MSSlof at 0.879 and MSSpe at 0.00233. Thus the F-ratio (1.35 \ 9.28) for these statistical parameters demonstrated that the proposed model has no a lack of fit. From another view point it is possible to observe that the minimum region for the residual carbon content values is outside the experimental region. However, it is not possible to apply conditions higher than those used for the power (1,000 W is the maximum power achieved by the microwave device) variable compared to that the made in experimental design used in this work. The parameter that had the most influence on the response was the time of microwave irradiation. The interaction between the factors investigated had a positive influence on the response. The significance of the effects was evaluated by ANOVA, and the p value (significance level) was determined. The p value represents the probability of errors and can be explained by random factors. Using Eq. (4) and its solution and analyzing the response surface plot (Fig. 1), the optimum values for the selected test variables were able be obtained. The critical point was obtained by the model equation that provided 817 W of applied power for 27 min using 2.5 mL of nitric acid. This calculated time (27 min) would be very long for a small percentage reduction in %RCC. A value of 800 W of applied power for 15 min with 2.5 mL
Table 5 Analysis of variance (ANOVA) of the response surface model for %RCC prediction after sample digestion Factors
Statistics SS
DF
MSS
Fvalue
Pvalue
Intercept
82.16
1
82.16
104.15
0.0002
Power (X1)
236.53
1
236.53
299.84
0.0000
Time (X2) Volume (X3)
18.61 13.78
1 1
18.61 13.78
23.58 17.47
0.0046 0.0087
X21
6.44
1
6.44
8.16
0.036
X22 X33
4.04
1
4.04
5.12
0.073
3.48
1
3.48
4.41
0.090
X1X2
1.69
1
1.69
2.14
0.20
X1X3
1.32
1
1.32
1.68
0.25
X2X3
0.0010
1
0.001
0.0026
0.91
Regression
284.09
9
33.13
58.32
0.00070
Lack of fit
2.63
3
0.88
1.35
0.45
Pure error
1.31
2
0.65
–
–
Error
3.94
5
0.57
–
–
Total
288.03
14
–
–
–
Fig. 1 Response surface to the Box-Behnken design
of HNO3 was therefore adopted. Under these conditions, the %RCC is nearly 4.0 %, as noted on the response surface and verified by Eq. (4). This value of %RCC provides security from spectral interferences. Using the optimized conditions, the digested sample showed a clear final solution. Figures of Merits for the Analytical Method The characteristic parameters of the analytical method, such as the correlation coefficient (R2), the average relative
123
J Am Oil Chem Soc Table 6 Parameters of the analytical calibration curves. Average (RSD) for repeatability of calibration solutions measurements and limits of quantification (LOQ) and detection (LOD) for the elements studied Elements
Parameter Ibranco
Isr
SBR
RSD
BEC
LOD
LOQ
Al
607.5
40218.8
65.2
4.20
0.015
0.0010
0.063
Ca
1216.5
15009.1
11.3
2.12
0.088
0.0055
0.018
Cu Fe
15543.8 159.6
97226.5 7565.9
5.3 46.4
0.28 6.90
0.19 0.022
0.0016 0.0046
0.0053 0.015
K
36141.9
369810.5
9.2
3.90
0.108
0.012
0.042
Mg
1222.8
68001.1
54.6
9.80
0.018
0.0052
0.017
Mn
155.6
63793.9
408.9
8.20
0.0024
0.00059
0.002
Na
7632.8
113781.5
13.9
0.80
0.35
0.0084
0.028
Zn
300.2
4183.1
12.9
5.22
0.77
0.012
0.040
standard deviation of all measurements for each element in the calibration curves, and the limits of detection and quantification, are presented in Table 6. The average RSD for repeatability of the method was in the range of 0.28–9.80 %. Considering the sample mass of 250 mg and the final volume of 25 mL, the limits of detection were in the range of 0.00059–0.012 mg L-1 for Al, Cu, Fe, Mn and Zn and in the range of 0.0052–0.012 mg L-1 for Ca, Mg, Na and K. Determination of Inorganic Elements in the Carnauba Waxes The optimized sample preparation method was implemented to digest carnauba wax type 1, 3 and 4 samples from three different manufacturers. Figure 2 shows the results for the elements determined by ICP OES. The concentrations of Ca, K, Mg and Na are higher than the concentrations of other elements in the samples. The concentrations of Ca, K, Mg and Na content in T1 carnauba wax samples were 33.83 ± 2.80 mg kg-1, 37.18 ± 2.48 mg kg-1, 6.76 ± 0.69 and 4.53 ± 0.46, respectively (Fig. 2). For wax T3, values of 92.61 ± 4.25, 131.66 ± 5.22 mg kg-1, 105.62 ± 2.24 and 77.83 ± 4.25 were found for Ca, K, Mg and Na, respectively. The amounts of the same preceding elements were 87.76 ± 5.55, 180.56 ± 2.49 mg kg-1, 116.27 ± 1.90 and 65.63 ± 2.31, respectively, in the wax T4 sample. The RSDs for these elements are lower than 10 % for all samples. Al, Fe and Mn are reported to be 28.63 ± 1.46, 18.54 ± 1.10, 1.77 ± 0.60 mg kg-1, respectively, in T1 carnauba wax. These results were plotted and are shown in Fig. 3. Zn was not detected in T1 carnauba wax. T3 carnauba wax presents Al, Fe, Mn and Zn concentrations of 108.52 ± 8,05, 104.47 ± 3.23, 18.35 ± 1.64, 9.35 ± 1.29 mg kg-1, respectively (Fig. 3). The results for T4 carnauba
123
Fig. 2 Concentration of Ca, Mg, Na and K in T1, T3 and T4
wax are 166.03 ± 3.75, 105.58 ± 4.15, 19.98 ± 1.32, 12.34 ± 0.81 mg kg-1 for Al, Fe, Mn and Zn, respectively. Mg is an element found in chlorophyll, the green pigment present in plants (used in the production of plant food by photosynthesis). The presence of chlorophyll in carnauba wax is another possible explanation for the difference in the color observed in the types of carnauba waxes. In general, carnauba wax T1 shows lower levels of inorganic elements than the waxes T3 and T4, this can be explained by the fact that the waxes obtained from the opened leaves experienced greater water loss due to the exposure to solar radiation, and also because of maturation of the leaves, which results in preconcentration of elements in the powder that is extracted from the leaves to manufacture these waxes (T3 and T4). T4 carnauba waxes generally present higher content of elements measured. It is supposed that iron and manganese are responsible for the darkening of the wax, because the darkness of the waxes increases from T1 \ T3 \ T4. The Fe content in T1 was
J Am Oil Chem Soc Table 7 Results of the certified reference material NIST SRM 1515a Apple leaves, N = 3 Elements (lg g-1)
Obtained valuea
Certified valueb
Recovery (%)
Cac
1.571 ± 0.104
1.526 ± 0.015
103,0
Kc
1,63 ± 0,005
1,61 ± 0,02
101,0
Mgc Al
0.268 ± 0.006 288,05 ± 3,59
0.271 ± 0.008 286 ± 9
99,11 100,7
Cu
5.25 ± 0.38
5.64 ± 0.24
93,09
Fe
67,21 ± 0,59
83 ± 5
81,57
Mn
53,78 ± 0,61
54 ± 3
99,60
Na
24.10 ± 0,12
24,04 ± 1.2
100,2
Zn
12,51 ± 0,2
12.5 ± 0.3
100,1
a,b
Fig. 3 Concentration of Al, Fe, Mn and Zn in T1, T3 and T4
82 % lower than in T3 and 85 % lower than T4. This result may be an indication that Fe contributes to the darkening of the color of the waxes. Iron is a component found in the cellulosic material that is present in the leaves [7], so it could naturally be transferred to the wax powder. The manufacturing processes could also be source of Fe due to the used devices. The quality of the carnauba wax produced and its chemical composition can be affected by atmospheric conditions, such as solar intensity or rainfall, also by local moisture and soil characteristics. Climatic conditions are natural agents that modify the biochemical cycle of the trees and provide changes in the physical and chemical powder characteristics. A greater number of carnauba wax samples must be evaluated to understand the complete chemical composition and to correlate the results obtained in this work with the climatic conditions and soil characteristics. Establishing consistent standards for each type of carnauba wax has been found to be difficult because the manufacturing industries purchase the powder from several different locations in the northwest of Brazil. Accuracy No certified reference material for carnauba waxes is commercially available, thus, the accuracy of the method was evaluated by the analysis of botanical certified reference material Apple Leaves NIST 1515a, purchase from the National Institute of Standards and Technology (Gaithersburg, MD, USA). The results obtained for the certified reference material showed good agreement for most elements, according to the t test at a 95 % confidence level. The results are shown in Table 7. The percentage of recovery varied from 81 to 103 % for the elements availed. These values reflect the efficiency of the sample decomposition method, as well as the precision of the results.
c
Mean ± standard deviation
Concentration units in % m m-1
Principal Component Analysis PCA is a mathematical manipulation of a data matrix with the goal of representing the variation presents in many variables using a small number of factors. In other words, PCA is a bilinear modeling method that gives an interpretable overview of the main information in a multidimensional data table. The information carried by the original variables is projected onto a smaller number of underlying variables called the principal components [24]. The exploratory analysis of data set through the PCA was performed using the element amounts determined by ICP OES obtained (Figs. 2, 3). The data were preprocessed by auto scaling prior to analysis, which allows the analysis to consider that every variable has the same importance in the discrimination of the samples. Sample patterns were detected using the first two principal components. A plot of PC2 versus PC1 scores (Fig. 4) shows the sample distribution and relationships between scores, yielding groups that indicate similarities. A small amount of variation is described by PC2 relative to PC1 (15.41 vs. 62.43 %). The first Principal Component (PC1) promotes the separation of the groups, so that the addition of the second component (PC2) improves the separation observed between the groups. Thus it was observed the formation of three groups for the studied elements in carnauba waxes. The loading plot (Fig. 5) shows the distribution of samples and the relationships among the variables, as well as the relative contribution of each one to the data variability. The samples T3 have higher Na content than T1 and T4 waxes. However, samples T4 present higher amounts of Al, Ca, Fe, Mn, Mg, K and Zn in the studied waxes. More studies should be conducted using a larger set of samples to investigate the variability of elements depending on the season and weather
123
J Am Oil Chem Soc
Fig. 4 Scores plot of T1, T3 and T4 carnauba wax samples
Fig. 5 Loadings plot of T1, T3 and T4 carnauba wax samples
conditions, which can influence the final characteristics of the manufactured waxes. In this study, a simple and fast microwave-assisted method for the preparation of carnauba wax samples and a robust and reliable analytical method for inorganic element analysis by ICP OES were developed. The use of the BoxBehnken experimental design to provide optimized
123
conditions for the microwave program was tested successfully. Using this statistical tool, only 15 experiments were necessary to determine the most favorable conditions for sample preparation. The advantages of using an experimental design were clear for this analytical step. This preparation and analysis procedure was implemented successfully for the determination of several elements by
J Am Oil Chem Soc
ICP OES. The results presented in this paper expand the knowledge of sample preparation and the elemental composition of carnauba wax and will be of interest for government agencies to approve new quality control assays of the manufactured material. This study may spread knowledge about this raw material and establish a new classification of the different types of wax that is based on both color and inorganic chemical composition. The characterization of carnauba wax in future studies is extremely important so that the quality of the raw material can be improved, which will add value to the product. Acknowledgments The authors are grateful to the Capes (Coordenac¸a˜o de Aperfeic¸oamento de Pessoal de Nı´vel Superior, Brazil) for financial support and to the A. N. S. Dantas fellowship and the ´ BA partnership for generously providing the carSINDICARNAU nauba waxes.
References 1. Cruz MA, Gomez VM, Fernandes KVS, Machado OLT (2002) Identification and Partial characterization of a chitinase and a b-1,3-glucanase from Copernicia cerifera wax. Plant Physiol Biochem 40:11–16 2. ABNT (Associac¸a˜o Brasileira de Normas Te´cnicas), Brazil (2012) http://www.abnt.org.br. Accessed on May 2013 3. Vanderburg LE, Wilder EA (1967) Aromatic acids of carnauba wax. J Am Oil Chem Soc 44:659–662 4. Vanderburg LE, Wilder EA (1970) The structural constituents of carnauba wax. J Am Oil Chem Soc 47:514–518 5. Asperger A, Engewald W, Fabian G (1999) Advances in the analysis of natural waxes provided by thermally assisted hydrolysis and methylation (THM) in combination with GC/MS. J Anal Appl Pyrolysis 52:51–63 6. Asperger A, Engewald W, Fabian G (1999) Analytical characterization of natural waxes employing pyrolysis—gas chromatography—mass spectrometry. J Anal Appl Pyrolysis 50:103–115 7. Wang L, Ando S, Ishida Y, Ohtani H, Tsuge S, Nakayama T (2001) Quantitative and discriminative analysis of carnauba waxes by reactive pyrolysis-GC in the presence of organic alkaline using a vertical microfurnace pyrolyzer. J Anal Appl Pyrolysis 58:525–537 8. Cysne JB, Braz-Filho R, Assunc¸a˜o MV, de Uchoa DEA, Silveira ER, Pessoa ODL (2006) Spectral assignments and reference data. Magn Reson Chem 44:641–643 9. Krivan V, Scha¨fer F (1989) Surface deposits on spruce needles as a possible indicator for the degree of heavy metal pollution of the atmosphere. Fresenius Z Anal Chem. 333:726 10. Rademeyer CJ, Fischer JL (1993) Determination of metals in waxes by Inductively Coupled Plasma Atomic Emission Spectrometry using high-temperature nebulization of the molten wax. J Anal At Spectrom 8:487–492
11. Kingston HM, Haswell SJ (1997) Chapter 2—Overview of microwave-assisted sample preparation. In: Microwave-enhanced chemistry: fundamentals, sample preparations and applications, 1st ed.; Kingston HM, Haswell SJ, Eds. Washington, DC American Chemical Society, p 55 12. Santos DM, Pedroso MM, Costa LM, Nogueira ARA, No´brega JA (2005) A new procedure for bovine milk digestion in a focused microwave oven: gradual sample addition to pre-heated acid. Talanta 65:505–510 13. Khajed M (2009) Optimization of microwave-assisted extraction procedure for zinc and copper determination in food samples by Box-Behnken Design. J Food Comp Anal 22:343–346 14. Bettinelli M, Beolne GM, Spezia S, Baffi C (2000) Determination of heavy metals in soils and sediments by microwave-assisted digestion and inductively coupled plasma optical emission spectrometry analysis. Anal Chim Acta 424:289–296 15. Navarro P, Raposo JR, Arana G, Etxebarria N (2006) Optimization of microwave assisted digestion of sediments and determination of Sn and Hg. Anal Chim Acta 566:37–44 16. Sakan S, Ðord¯evic´ D, Devic´ G, Relic´ D, And¯elkovic´ I, Ðuricˇic´ J (2011) A study of trace element contamination in river sediments in Serbia using microwave-assisted aqua regia digestion and multivariate statistical analysis. Microchem J 99:492–502 17. Mesko MF, de Moraes DP, Barin JS, Dressler VL, Knapp G, Flores EMM (2006) Digestion of biological materials using the microwave-assisted sample combustion technique. Microchem J 82:183–188 18. Dantas ANS, Costa RS, Gouveia ST, Lopes GS (2010) Development of a microwave-assisted digestion method using ICP OES to measure metals in gum deposits of internal combustion engines. Fuel Process Technol 91:1422–1427 19. Bezerra MA, Bruns RE, Ferreira SLC (2006) Statistical designprincipal component analysis optimization of a multiple response procedure using cloud point extraction and simultaneous determination of a metals by ICP OES. Anal Chim Acta 580:251–257 20. Ferreira SLC, Bruns RE, Ferreira HS, Matos GD, David JM, Branda˜o GC, da Silva EGP, Portugal LA, dos Reis PS, Souza AS, dos Santos WNL (2007) Box Behnken Design: an alternative for the optimization of analytical methods. Anal Chim Acta 597: 179–186 21. IUPAC (1978) Analytical chemistry division, commission on spectrochemical and optical procedures for analysis, nomenclature, symbols, units and their usage in spectrochemical analysis II. Spectrochim. Acta part B: At Spectrosc 33:242–245 22. Myers HR, Montgomery DC, Anderson-Cook CM (2009) Chapter 6—The analysis of second-order response surfaces. In: Response surface methodology: process and product optimization using designed experiments, 3rd edn. Wiley, Hoboken, p 219 23. Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA (2008) Response surface methodology (RSM) as a tool for optimization in analytical chemistry. Talanta 76:965–977 24. Correia PRM, Ferreira MMC (2007) Reconhecimento de padro˜es por me´todos na˜o supervisionados: explorando procedimentos quimiome´tricos para tratamento de dados analı´ticos. Quı´m Nova 30:481–487
123