1.    Introduction

Corrosion studies of aluminium and its alloys have received considerable attention because of their technological importance and industrial application. Aluminium and its alloys are mainly used as parts in automobiles, aerospace and building structures \citep*{pe2003} . A remarkable economic and industrial importance of aluminium and its alloy are due to their low cost, light weight, high thermal and electrical conductivity. Phosphoric acid medium is widely used for acid cleaning and electro-polishing of aluminium and its alloy \citep{2000} \citep{mk2015}\citep{c2004}. However, in the presence of aggressive environments the protective layer can be destroyed and corrosive attack takes place \citep{ma2004} \citep{as2008}. Therefore, it is necessary to use inhibitors for controlling the corrosion of aluminium in phosphoric acid solution during cleaning processes \citep{oa2009} \citep{m2012} \citep{p2013}.  The use of the natural products is more effective and highly environmentally benign and harmless compared to organic and inorganic inhibitors used in chemical or any industrial applications \citep{e2010} \citep{a2013}.
Several native Trigona sp. that produce honey in the Northern and North Eastern regions of Brazil which is known as indigenous bees. They are different from A. Mellifera at the subfamily level and it is known as Meliponinae which was recently renamed as Meliponini \citep{sd1992}. According to the Malaysian Agricultural Research and Development Institute (MARDI), Trigona sp. is more nutritious than ordinary honey. . Apis sp. in contrast with Trigona sp., is the member of the single genus Apis from the tribe of Apini where the most popular subspecies is Apis mellifera. According to \citep{g2009}, honey that comes from Apis mellifera are mainly produced and distributed in Asia and Europe. The corrosion inhibition properties of other types of honey have been investigated in various environment and different types of materials \citep{ay1998} \citep{m2000} \citep{j2008}.  Phenolic materials are commonly found in plants and human food such as phenolic acids and hydroxy-cinnamic. Phenolic acid accounted one-third of the overall phenolic materials in food. These materials can be divided into two classes, hydroxyl benzoic acid and hydroxyl cinnamic acid. It can be used as anticorrosion, acts as anticancer, antiseptic, and anti-inflammatory \citep{c2014}. P-coumaric acid is one of the major phenolic acids in honey. It is the highest antioxidant among other main phenolic acid in honey. It is a major component of pollen grains which is present in the natural diet of honey and also functions as a nutraceutical regulating immune and detoxification processes (Wenfu et al., 2013). There are several researches on p-coumaric acid as the compound in charge for antioxidant role (Simic et al., 2007; Teixeira et al., 2013). Antioxidant prevents or inhibits the natural phenomenon of oxidation which has many applications in various fields. 
It was demonstrates that honey was a mixture of various compounds containing carbon, oxygen (polyphenols), and nitrogen which was the reason of adsorption process occurring on the aluminium alloy (Berkovic, et al., 2008). Previous study stated that the adsorption of the honey molecules onto the surface of aluminium alloy may take place through all of this functional group (Rosliza and Nik, 2013) and showed good potential of corrosion inhibitor. The presence of p-coumaric acid compounds in Trigona sp. and Apis sp. will show a corrosion inhibitor property. The corrosion behavior of Trigona sp. comparing with Apis sp. on aluminium alloy was determined via weight loss measurement, potentiodynamic polarization measurement and electrochemical impedance spectroscopy (EIS). Furthermore, the adsorption mechanism and surface morphology analysis using scanning electron microscope (SEM) were also determined.

2.    Material and Methods

2.1`Aluminium Specimens
The study was performed with the specimen of Al-Mg alloy which was supplied directly from a local electronic company in Bayan Lepas, Pulau Pinang (2014). The outer diameter size of the alloy is 95 mm while the internal diameter is 24 mm with the thickness of about 2 mm. The composition analysis of the Al-Mg alloy was determined by using EDX spectroscopy (Oxford INCA 400). For electrochemical tests the aluminium alloy specimens with dimensions of 3 cm x 4 cm x 0.2 cm were polished with 400, 600, 800 grade of emery papers to ensure smoothness and cleanliness of the surface. It was then degreased with acetone and rinsed with distilled water just before each electrochemical test.
 2.2    Inhibitors
Two types of honey samples which represented the Trigona sp. ( stingless bee honey ) and Apis sp. ( honey bee ) were studied. The Trigona sp. was collected from a farm in Merbok, Kedah (2015) while the Apis sp. was from Kuala Berang, Terengganu (2015). 
2.3    Medium
Stock solution of 1 M H3POwas prepared by diluting analytical grade of phosphoric acid 85 %  with distilled water. AR grade H3PO4 and distilled water were used for preparing 1 M H3PO4 solution for experimental work. The concentration range of both honey employed was varied from 100 ppm to 2500 ppm. 
2.4    Weight loss measurement method
The corrosion inhibition by weight loss measurement followed the procedure by Yiase et al. (2014) with slight modifications. The aluminium alloy with dimensions 3.0 cm x 4.0 cm x 0.2 cm was polished with 400, 600, 800 and 1000 grits of emery paper. The specimen was rinsed with distilled water followed by isopropanol. The purpose was to remove the residue on the alloy. Aluminium specimens were immersed in the 100 mL of 1 M H3PO4 with the absence and presence of different concentrations of both honeys. The weight loss was determined after the immersion in 24 hours. The percentage of inhibition efficiency (IE %) was calculated based on the following equation: 
 
                    
                                                                                     (3.1)
Where  and  are the weight losses of aluminium (g) without and with the inhibitors respectively.
2.5    Potentiodynamic polarization
Potentiodynamic polarization studies were performed after half an hour immersion at open circuit potential (OCP). The polarization were recorded by scanning the potential for range ± 250 mV with respect to open circuit potential (OCP) at room temperature with the scan rate of 1 mV s-1. Inhibition efficiencies (IE %) were calculated using the following equation:
                                                                                    (3.2)
where and ’ are the corrosion current densities (mA cm-2) obtained in the presence and absence of inhibitor, respectively. These procedures were used for both stingless bee honey and honey bee solution.
2.6    Electrochemical impedance spectroscopy (EIS)
The EIS measurement was conducted after 30 minutes of immersion at open circuit potential. The EIS were carried out with the frequency range of 0.01Hz to 1 MHz root means square (rms) amplitude signal of 10 mV. Then, the IE % were calculated by the following equation:
 
                                                                                   (3.3)
Where  and  are defined as the charge transfer resistance (Ω cm2) in absence and presence of inhibitors, respectively.
2.7    Surface analysis by using SEM-EDX
The scanning electron microscope images were recorded by using SEM analysis (Leo Supra 50 VP) to study the surface morphological with 140x magnification of aluminium in the presence and absence of inhibitors. The images were taken after 24 hours immersion of the samples at room temperature in 1 M H3PO4 for the optimum inhibition concentration of stingless bee honey and honey bee when compared with inhibited sample.
3.    Result and Discussion
3.1    Weight loss measurement method 
The weight loss of aluminium in 1 M H3PO4 with and without various concentrations of Trigona sp. honey and Apis sp. honey were determined after 24 hours of immersion at room temperature as shown in Tables 1.1 and 1.2. The inhibition efficiency increases as the concentration of both honeys increases. The highest percentage of inhibition efficiency for Trigona sp. honey was found to be 85.1 % while for  Apis sp. honey was 75.8 % at 1500 ppm. The results indicated that the inhibitors were “concentration-dependent” because as concentration increased, inhibition efficiency increased. The value of the highest percentage of inhibition efficiency was observed at 1500 ppm from both honeys which were considered as optimum concentration. As shown in Tables 4.3 and 4.4, the percentages of inhibition efficiency decreased beyond the optimum concentration for  Trigona sp. honey and  Apis sp. honey. 
Table 1.1: Weight loss measurement of aluminium in 1 M H3PO4 with and without various concentrations of Trigona sp. honey at 303 K for 24 hours.
 
Concentration (ppm)
Weight loss (g)
CR (µg cm-2 h-1)
IE (%)
0
100
500
1000
1500
2000
2500
0.2710  
0.0463
0.0435
0.0416
0.0405
0.0675
0.0695
0.9410
0.1608
0.1510
0.1444
0.1406
0.2344
0.2413
-
82.9
84.0
84.7
85.1
75.1
74.4
Table 1.2: Weight loss measurement of aluminium in 1 M H3PO4 with and without various concentrations of  Apis sp. honey at 303 K for 24 hours.
 
Concentration (ppm)
 
Weight loss (g)
 
CR (µg cm-2 h-1)
 
IE (%)
0
100
500
1000
1500
2000
2500
0.2710
0.0730
0.0672
0.0661
0.0655
0.0757
0.0783
0.9410
0.2535
0.2333
0.2295
0.2274
0.2628
0.2719
-
73.1
75.2
75.6
75.8
72.1
71.1
 
3.2    Potentiodynamic polarization measurement
 
The inhibition efficiency (IE %) was calculated for the aluminium alloy in 1 M H3PO4 containing 100 ppm, 500 ppm, 1000 ppm, 1500 ppm, 2000 ppm and 2500 ppm of Trigona sp. honey and  Apis sp. honey by using Equation 3.2. The percentage inhibition efficiency value was obtained from icorr data for both honey. From the result, it was obvious that current density of corrosion decreases as the concentration of the inhibitors increases. Better results and performances were observed when 1500 ppm of Trigona sp. and Apis sp. was used. The anodic and cathodic reactions were affected by the addition of Trigona sp. honey and also Apis sp. honey as seen in Figures 4.8 and 4.9. This situation may lead to the reducing of metal dissolution at anode and retarding the hydrogen evolution reaction at cathodic site after the addition of inhibitors. This resulted from the adsorption of inhibitor molecules on the aluminium surface (Farag et al., 2015). 
The value of corrosion potential Ecorr indicates two different polarizations which are anodic polarization and cathodic polarization. As the concentrations of inhibitors increase, the anodic and cathodic current densities will decrease. It was proven that the additive affected both anodic and cathodic processes. The value Ecorr shifts to positive (noble) direction when anodic polarization occurs and it will shift to negative (active) direction for cathodic polarization (Fouda et al., 2014). As seen in Table 4.5, compared to the Ecorr of  Trigona sp. honey concentration with the 1 M H3PO4 solution (blank), the values of Ecorr was less negative (anodically) and slightly more negative (cathodically) from 1500 ppm to 2500 ppm. This was consistent for  Apis sp. honey inhibitor. Therefore, the presence of inhibitors resulted in the shifting of the corrosion potential (Ecorr) towards the positive direction. According to literature report, when the displacement in Ecorr was bigger than 85 mV with respect to the corrosion potential of the blank, the inhibitor can be classified as a cathodic or anodic type (Shylesha et al., 2011). Otherwise, if the displacement is less than 85 mV, the inhibitor can be considered as mixed type (Abdullahi et al., 2014). In this study, the maximum displacement of both honeys was lower than 85 mV which can prove that the  Trigona sp. honey and  Apis sp. honey act as mixed type inhibitors with predominant anodically.
By extrapolating the anodic and cathodic Tafel lines to corrosion potential of the polarization curves, corrosion current density, icorr was determined. The icorr will decrease with increase in the concentration of the inhibitor. It was suggested that the adsorption of inhibitor molecules at the surface of aluminium occurred to form protective film on the aluminium surface (Yadav et al., 2015). Thus, the inhibitors act via adsorption on aluminium surface and blocking the active corrosion sites (Yurt et al., 2014). It was also due to the phenyl ring which has high electron density in the p-coumaric acid in the inhibitors. As a result, it will block the active site of the aluminium surface (Chakravarthy and Mohana, 2014). From the Tables 4.5 and 4.6, the value of anodic Tafel slope (βa) was observed to change with the addition of the inhibitors. It was suggested that the inhibitors were first adsorbed onto the metal surface and blocked the reaction sites of the metal surface resulting in the effecting of the anodic mechanism. However, the values of cathodic Tafel slope constant (βc) almost remain unchanged in the presence of the inhibitors. This situation suggested that the inhibitors did not interupt the mechanism of cathodic reaction with hydrogen evolution reaction and the inhibitors inhibit the corrosion by blocking the active sites on the aluminium surface (Caliskan and Akbas, 2012)
Table 4.5: Data from polarization measurement of aluminium in 1 M H3PO4 solution in the absence and presence of various concentration of stingless bee honey.
Concentration
(ppm)
Ecorr
(mV)
icorr
(mA cm-2)
βa
(mV decades-1)
-βc
(mV decades-1)
     CR
   (mpy)
IE%
0
 
100
 
500
 
1000
 
1500
 
2000
 
2500
 
-1090
 
-1030
 
-1060
 
-1070
 
-1120
 
-1080
 
-1070
0.2520
 
0.1170
 
0.0775
 
0.0717
 
0.0540
 
0.0693
 
0.0736
523.7
 
566.4
 
517.7
 
498.4
 
509.4
 
511.6
 
549.6
271.6
 
275.1
 
225.2
 
276.8
 
288.1
 
245.7
 
294.1
109.07
 
50.64
 
33.55
 
31.04
 
23.37
 
29.99
 
31.85
-
 
53.6
 
69.2
 
71.5
 
78.6
 
72.5
 
70.8
Table 4.6: Data from polarization measurement of aluminium in 1 M H3PO4 solution in the absence and presence of various concentration of natural honey.
Concentration
(ppm)
Ecorr
(mV)
icorr
(mA cm-2)
βa
(mV decades-1)
-βc
     (mV decades-1)
CR
(mpy)
IE%
0
 
100
 
500
 
1000
 
1500
 
2000
 
2500
 
-1090
 
-1030
 
-1070
 
-1100
 
-1120
 
-1100
 
-1020
0.2520
 
0.1120
 
0.0809
 
0.0777
 
0.0658
 
0.0768
 
0.1130
523.7
 
533.8
 
590.2
 
588.6
 
579.1
 
563.5
 
557.1
271.6
 
265.7
 
303.3
 
287.7
 
306.2
 
292.9
 
293.9
109.07
 
48.48
 
35.02
 
33.63
 
28.48
 
33.24
 
48.12
-
 
55.6
 
67.9
 
69.2
 
73.9
 
69.5
 
55.2
3.3    Electrochemical impedance spectroscopy
Further investigation was resumed by EIS after 30 minutes immersion at 303 K. The Nyquist plots of impedance data for aluminium alloy in various concentrations of Trigona sp. and Apis sp. were presented in Figures 4.11 and 4.12. The diameter of capacitive loop increased as the concentration of the inhibitors increased indicating strengthening of inhibitive film. The diameter of capacitive loop represents the resistance to the corrosion (Arockiasamy et al., 2014). The diameter will decrease significantly with the decrease in resistance. Based on the plots given, it was clear that the impedance response changes with the addition of the inhibitor. Higher frequency capacitive loop was due to the adsorption of inhibitor molecules due to the charge transfer reaction (Singh et al., 2014). 
It can be seen from the figures that the Nyquist plots contain depressed semicircle with one time constant also in Bode-phase plot and the size of capacitive loop increased with the inhibitor concentration indicating a charge transfer process mainly controlling the corrosion of aluminium. The depression behaviour is often referred to a frequency dispersion which has been attributed to roughness and other inhomogeneity of solid surface during the corrosion process (Singh and Ebenso, 2013). The Nyquist plot consists of a large capacitive loop at high frequencies (HF) and an inductive loop at low frequencies (LF). The capacitive loop at high frequencies (HF) was attributed to the charge transfer of the corrosion process and the double layer capacitance behaviour (Emran, 2014) while the low frequencies inductive loop may be related to a process of relaxation obtained by adsorption of compounds such as H+ and PO43- on the electrode surface. Besides that, at low frequency it can be attributed to the re-dissolution of the passive surface (Verma et al., 2014).
The Nyquist plots were analysed by fitting the experimental data to the simple equivalent model (Figure 4.10). The fitting circuit of was done by ZSimpWin software version 3.20. It was shown in Tables 4.7 and 4.8. The list of the impedance parameters of the Nyquist plots in both inhibitors in various concentrations was also shown in the tables.  It consists of solution resistance (Rs) and charge transfer resistance (Rct) which was placed in parallel to the constant phase element (CPE) as in the equivalent model. The measurement of electron transfer across the surface is inversely proportional to the corrosion rate from the Rct value. Randles- CPE is generally used in the circuit instead of Randles- Cdl circuit to fit the semicircle more accurately and defined by two values of components Yo and n which was expressed by the following relation in Equation 4.1 (Dadgarinezhad and Baghaei, 2012; Eldesoky, 2015): 
                                                                                                      (4.1)
Where Yo is the admittance, ω is angular frequency (ω = 2πf where f is the AC frequency), j is the imaginary unit, and n is used for the phase shift (CPE exponent). The general unit for CPE is in F cm-2 (Farad cm-2).
            The parameters of electrochemical impedance of the Nyquist plot of stingless bee honey and natural honey in various concentrations were all listed in Tables 4.7 and 4.8. The Randles model (Figure 4.10) which was the simulation of the Nyquist impedance plots containing constant phase element (CPE) than the capacitance double layer (Cdl) showed excellent agreement with the experimental data. Both tables reveal that as the concentration increased the value of Rct also increased. It can be seen from the tables that the Rct value for 1 M H3PO4 solution without the presence of both inhibitors (1 M H3PO4) was at 177.10 Ω cm2 while for stingless bee honey inhibitor, the highest value of Rct was found to be at 1100 Ω cm2. However, the value of Rct for natural honey was found to be at 779.80 Ω cm2. This revealed that more inhibitor molecules were adsorbed on the metal surface at higher concentrations. Thus, it will lead to greater surface coverage. The percentage of inhibition efficiency (IE %) may be calculated from the value of Rct and it was observed that for stingless bee honey the IE % was 83.9 % whereas for natural honey the value was 77.3 %. The highest inhibition efficiency (IE %) of both inhibitors obtained at optimum concentration was at 1500 ppm and remained almost constant with further increase of concentrations.
           
Rs
Rct
CPE
 
 
Figure 4.10: The Randle’s CPE circuit model
            The CPE values decrease with the increasing concentrations for both inhibitors. The reduced value of CPE may increase the thickness of the electrical double layer (Arockiasamy et al., 2014). This revealed that both honey molecules affect the adsorption at the metal-solution interface (Khaled et al., 2010). This process leads to the formation of a surface film and the extent of the metal dissolution will decrease.
            From the table, it was clear that by increasing the inhibitor concentration, the CPE values tend to decrease and the Rct value increases. The increasing value of Rct leads to an increase in percentage of inhibition efficiency. It can be seen from the table that both honeys proved they were good corrosion inhibition for the aluminium alloy in 1 M H3PO4. The results obtained from EIS method was in good agreement with the weight loss measurement and potentiodynamic polarization method as well. Similar situation can be observed that stingless bee honey showed better performance with the highest percentage of inhibition efficiency (IE %) until the optimum values of concentration in these three methods even though the values were not exactly the same. The trend of IE % was still similar in all methods. This differences could be due to the immersion time and surface area of the aluminium specimen, different size of the cell that was used and lastly the total volume of the electrolyte (Curioni et al., 2015).