Chemical Engineering Journal 256 (2014) 372379

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

Biodiesel purication in one single stage using silica as adsorbent

Debora L. Manuale a,b, Elisabet Greco a, Adriana Clementz b, Gerardo C. Torres a, Carlos R. Vera a,b,
Juan C. Yori a,b,
a
Facultad de Ingeniera Qumica, Universidad Nacional del Litoral, Santiago del Estero 2829, 3000 Santa Fe, Argentina
b
Instituto de Investigaciones en Catlisis y Petroqumica (INCAPE) (FIQ-UNL, CONICET), Santiago del Estero 2654, 3000 Santa Fe, Argentina

h i g h l i g h t s

 We report a dry biodiesel purication method of one single stage treatment.

 The methodology is a solution for biodiesel rening saving money and process time.
 The treatment is made under conditions of vacuum and temperature.
 The selected adsorbent is silica Trisyl 3000.
 Silica behaves as a non-selective adsorbent of high capacity.

a r t i c l e i n f o a b s t r a c t

Article history: In this paper a simple biodiesel purication method consisting of one single stage treatment with silica as
Received 2 June 2014 adsorbent is described. The presented method seems a solution for biodiesel rening saving water,
Received in revised form 5 July 2014 money and process time. The treatment is made under conditions of vacuum (0.2 bar) and mild temper-
Accepted 7 July 2014
ature (6590 C). The method allows for the removal of the excess of methanol and water (present in the
Available online 14 July 2014
crude biodiesel and adsorbed on the adsorbent material) simultaneously with the adsorption of the
impurities.
Keywords:
Silica Trisyl 3000 retains 23% of its weight when the process is conducted at atmospheric pressure and
Biodiesel
Rening
nearly 235% when performed under vacuum conditions. This greatly improves the utilization of the
Adsorption adsorbent and reduces the cost of the process.
Silica Under these conditions, the silica behaves as a non-selective adsorbent of high capacity that is able to
adsorb different types of impurities, which makes it an excellent adsorbent to purify biodiesel. On the
other hand, the adsorption phenomenon is not limited to the formation of a theoretical monolayer
of adsorbed impurities; it is more complex and includes the formation of multiple layers.
2014 Elsevier B.V. All rights reserved.

1. Introduction effects on engine performance if not removed [1,2]. The crude bio-
diesel requires a subsequent rening process in order to meet the
Biodiesel is obtained from the transesterication of vegetable strict international biodiesel standard specications (ASTM D6751
oils or animal fats with a short-chain alcohol, with or without and EN 14214 Standards) [3,4].
the use of catalyst. After the reaction glycerol (the main by- The process of crude biodiesel rening is complex and it is usu-
product) is separated by settling or centrifuging and a crude ally achieved via two techniques; wet and dry washings. Conven-
biodiesel phase is thus obtained. Many contaminants can still be tionally wet washing is the most widely employed technique to
present in the crude biodiesel depending on the technology of remove these impurities.
transesterication used. Impurities such as residual methanol The wet washing method employs multiple successive washing
and traces of glycerol, soaps, catalyst, phospholipids, water and steps with water in order to remove major impurities such as
unreacted glycerides can be found, that can produce harmful soaps, catalyst, glycerol and methanol from biodiesel. This tech-
nique is based on the afnity of polar compounds with water.
There are some disadvantages in the use of this method such as:
Corresponding author at: INCAPE, Santiago del Estero 2654, 3000 Santa Fe,
Argentina. Tel.: +54 342 4533858; fax: +54 342 4531068.
(i) high cost (representing 6080% of the total processing cost
E-mail address: [email protected] (J.C. Yori). [5]), (ii) a substantial increase of total production time, since water

http://dx.doi.org/10.1016/j.cej.2014.07.022
1385-8947/ 2014 Elsevier B.V. All rights reserved.
 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379 373

washing requires several cycles of washing and one stage of centri- The aim of this work was to develop a simple dry method for
fugation and/or drying, which are time intensive processes, (iii) the biodiesel rening, using silica as adsorbent material, involving
difculty of meeting normed specications regarding nal water the fewest possible steps in order to produce biodiesel that meets
content, (iv) product loss, because some of the biodiesel is lost with strict international quality standards. Rening operating condi-
aqueous efuents, and, (v) the amount of waste water that is gen- tions were sought to ensure: (i) adsorption of the main impurities
erated as process efuent, which must be treated before the nal of biodiesel, particularly glycerides, (ii) reduction of the rening
discharge, introducing an additional cost to the process. costs, and (iii) optimization of the silica adsorption capacity.
In the wet washing process, for every 100 L of biodiesel
produced some 20 L of wastewater are produced [5]. The water
content of the biofuel after the washing step (mentioned in point 2. Experimental section
iii), is around 0.15% [6,7]; this value is above the maximum set
by the standards (0.05%) and its therefore necessary to introduce 2.1. Materials
a later stage of centrifugation and/or drying under vacuum condi-
tions. The wastewater due to its physicochemical characteristics, The raw material employed to produce biodiesel was a
should be treated before the nal discharge in order to reduce degummed soy oil provided by a local producer. Methanol
the environmental impact. The washing water is characterized by (99.9%) and pure glycerol (>99.5%) were supplied by Sintorgan
high values of pH, COD (chemical oxygen demand), BOD (biological SA. Sodium estearate (99%) was chosen as a model soap and was
oxygen demand), fats and oils, needing a complicated and expen- supplied by Serisa Qumica SRL. Glyceryl monoestearate (98%)
sive biological treatment and thus increasing operating costs [5,8]. was used as a model monoglyceride and was supplied by Cloretil
Problems associated with wet washing have led to the develop- SACIF. Silica samples were TriSyl 3000, 300B and 450 (provided
ment of the dry washing process to purify crude biodiesel. This by W.R. Grace Argentina SA).
process involves the use of adsorbents such as carbon, silica,
bleaching clay, and activated bleaching clay, zeolite, ion exchange 2.2. Biodiesel preparation
resins (amberlite or purolite), cellulosics or sawdust, diatomaceous
earth and magnesium silicate powder (magnesol) [9]. The process Reference Methyl soyate was prepared according to the stan-
requires a mixing unit and the spent adsorbent must be generally dard procedure (catalytic low-temperature method using NaOH
discarded [10]. Cooke et al. [11] reported that water washing may as catalyst) described by Noureddini and Zhu [15].
be substituted in industrial plants by dry washing with Magnesol Many batches of the solution of the biodiesel product and unre-
(provided by Dallas Group of America, Inc.). However up to now acted methanol were synthesized and mixed together in order to
the use of the dry washing technique has not spread much at provide a common biodiesel stock. This biodiesel stock sample
industrial level, mainly due to the following disadvantages: (i) was spiked with contaminants such as monoglycerides and soaps,
the cost of adsorbents, (ii) the difculty of some adsorbents for up to the values indicated in Table 1. Then it was kept in a dessica-
adsorbing certain types of impurities such as glycerides and free tor for later use.
fatty acids (FFA), and (iii) the difculty of an adequate regeneration
and reuse. For this reason the solid residue has to be disposed of in
landlls or other applications (composting, use as a potential ani- 2.3. Bleaching experiments
mal food additive) [2]. Therefore it is essential to nd a cheap
adsorbent, with high adsorption capacity and capable of adsorbing The spiked biodiesel base sample was rened by performing
all biodiesel impurities. bleaching tests at varying conditions and with different silica
Removal of glycerides from biodiesel is an important step of the adsorbents. Bleaching was performed at different temperatures
process because key aspects of the quality of the fuel strongly and contact times in a gas-tight stirred autoclave (stirring speed:
depend on the content of bound glycerol. Main problem with these 250 rpm). 100 cm3 of spiked biodiesel were put in contact with
compounds is that when heated they tend to polymerize forming the silica (adsorbent weight: 1.0 and 3.0 g) at the selected temper-
deposits. They also increase the cloud point of biodiesel and they ature (50, 65, 80 and 90 C) for a conveniently chosen contact time
complicate the operation of liquidliquid phase splitting units (15100 min) The inuence of vacuum and temperature was
due to their amphiphilic nature. assessed by performing the bleaching procedure at normal pres-
With respect to the ability to adsorb glycerides there are con- sure and in vacuum (0.2 bar) and by varying the bleaching reactor
icting results in the literature. Berrios and Skelton [2] reported temperature. After the treatment the adsorbents were decanted,
that magnesol and ion exchange resins do not retain glycerides ltered and the liquid phase was sampled for analysis. All experi-
(mono (MG), di (DG) and triglycerides (TG)) in a signicant way; ments were performed in triplicates.
consequently, the rening process cannot comply with the limits
set up in the EN 14214 standard. On the contrary, Faccini et al.
[12] reported that Magnesol and silica showed suitable results
Table 1
regarding free and bound glycerol removal. Main properties of the spiked biodiesel.
It is known that silicas efciently remove the main impurities of
Properties Spiked crude biodiesel (%) Normal limitsa (%)
biodiesel, such as: methanol, residual metals, phospholipids and
soaps [13]. It has been previously reported [14] that glycerol has Free glycerol, % 0.20 0.02
Bound glicerol, % 0.84 0.24
a great afnity for the silica surface and is selectively adsorbed
Monoglycerides, % 2.41 0.80
from biodiesel solutions. It was found that adsorption of glycerol Diglycerides, % 0.08 0.20
is not inuenced by the presence of small amounts of water and Triglycerides, % 0.04 0.20
soaps. Conversely the presence of monoglycerides (MG) and/or Acidity, % 0.27 0.50
methanol (MeOH) lowers the adsorption capacity of glycerol due Soaps, % 0.25 0.02
Water, % 0.10 0.05
to the competition for the adsorption sites on silica. Then, a neces- Methanol, % 5.00 0.20
sary rst step of methanol removal previous to the silica treatment
a
is required. As specied by the ASTM D 6751 and EN 14214 quality standards.
374 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379

The amount of adsorbed impurity on the solid Q (%) was deter- 3. Results and discussion
mined from a balance of the impurity in the liquid phase, measur-
ing the variation in the concentration of the impurity in solution 3.1. Physicochemical characterization of silicas
before and after the treatment.
The results of surface area, pore volume, mean pore diameter
C o  C f  100 and pore size distribution are shown in Table 2 and Fig. 1. An
Q % ;
W adsorbent material must have high specic surface area to enable
the adsorption of greater amounts of impurities and a suitable pore
where: structure to allow the access of all impurities to its internal surface
Q (%): amount of adsorbed impurity on the solid (in g of without difculty. This latter topic is critical for large molecules
adsorbed impurity per 100 g of adsorbent). (soaps, mono, di and triglycerides) with sizes similar to those of
Co: concentration of the impurity in the liquid phase before the the pores of the adsorbent.
treatment (in g of impurity per 100 g of biodiesel). Isotherms obtained for the three silicas (not shown) can be clas-
Cf: concentration of the impurity in the liquid phase after the sied as type IV, according to IUPAC nomenclature [17]. This type
treatment (in g of impurity per 100 g of biodiesel).w: adsorbent of isotherm is characteristic of mesoporous solids (pore diameters
concentration (in g of adsorbent per 100 g of the biodiesel). between 20 and 200 ) [18]. Silicas have high specic surface area
values (approximately 600 m2 g1) and silica Trisyl 3000 presents
2.4. Biodiesel quality assessment the highest value of surface area. The pore size distribution is uni-
modal for all samples (see Fig. 1) with pore diameters between 25
The characterization of the biodiesel stock solution and rened and 125 . Therefore the structures are purely mesoporous. The
biodiesel was made according to the ASTM 6751 and EN 14214 wide pore network of the silica was thought to present no prob-
norms [1,2]. lems for diffusion of the adsorbates. The approximate length of
The data reported here are averages of duplicate determina- some molecules was included in Fig. 1. It can be seen that the sizes
tions. Reproducibility was very good. of soaps and TG are smaller than the average pore diameter of the
The water content of the samples was assessed by Karl-Fischer network (Table 2). Then, it can be concluded that the silicas have
potentiometric titration, according to the EN ISO 12937 and ASTM high specic surface area and a suitable porous structure for use
D 2709 standards. The Karl-Fischer reagent was supplied by Merck in biodiesel purication processes by selective adsorption.
and the titration was performed with an Altronix apparatus with a The results of TGA are shown in Fig. 2. A signicant mass loss
glass electrode. The soap content was indirectly assessed by means can be seen in the three cases, which starts at temperatures near
of the ASTM D874 test for sulfated ash content. The FFA (free fatty 40 C and ends at temperatures of 110120 C. The loss of mass
acid) content was determined by volumetric titration with a solu- is greater in samples 450 and 300B (40 and 50% respectively)
tion of NaOH as indicated in the ASTM D664 test. and lower in sample Trisyl 3000 (10%). This loss is attributed to
The concentration of glycerol and glycerides was determined by the release of water adsorbed on the surface. A subsequent
means of the method indicated in the EN 14105 an ASTM D 6584.
Pyridine (99.9%, Merck) was used as a solvent and as silylation cat-
alyst for the derivatization of the samples. These were derivatized Table 2
with N-Methyl-N-(trimethylsilyl) triuoroacetamide (MSTFA, Results of surface area (Sg BET), pore volume (Vp) and mean pore diameter (Dp
Sigma Aldrich). 1,2,4-Butanetriol (>95%, Sigma Aldrich) and trica- Wheeler) of the silicas.

prin (>99%, Sigma Aldrich) were added as internal standards. After Adsorbent Dp wheeler () Sg BET (m2 g1) Vg (cm3 g1)
15 min at room temperature the silylated samples were dissolved Trisyl 3000 71 631 1.1
in n-heptane (99.5%, Merck) and injected in the GC for their anal- Trisyl 450 67 598 1.0
ysis. The analysis was performed with a Varian Star 3400 CX gas Trisyl 300B 74 592 1.1
chromatograph equipped with an on-column Supelco injector
and a FID detector.
The methanol content of the samples was assessed according to
the EN 14110 standard, Methanol was determined using a Varian
0.016 TriSyl
Star 3400 CX gas chromatograph equipped with a FID detector 300B

using the head space method with 2-propanol as the internal stan-
dard (>99%, Ciccarelli).
TriSyl
450

2.5. Adsorbents characterization TriSyl

3000
dV/dD

The composition of the silicas was determined by inductively 0.008

coupled plasma-optical emission spectroscopy (ICP-OES, Perkin
Elmer, Optima 2100 DV) after digestion in an acid solution of
hydrochloric acid and further dilution in water.
The adsorbents were characterized by means of nitrogen
adsorption at 194.6 C. Before the adsorption, the samples were
degassed in vacuo at 150 C for 2 h. Adsorption isotherms were
Soap

TG

measured in an automatic Accusorb Micromeritics equipment.

0.000
The specic area and pore size distribution were estimated by 0 50 100 150 200 250 300
the BET and the BJH methods, respectively [16]. TGA analysis of
Pore diameter,
the adsorbents was performed with a Shimadzu TG-1000 equip-
ment. The samples were tested without previous treatment and Fig. 1. Pore size distributions of the silicas: (N) TriSyl 300B, (d) TriSyl 450, (j)
were considered to be in equilibrium with room humidity. TriSyl 3000.
 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379 375

12 Table 3
Adsorption capacity of impurities. Conditions: contact time = 90 min., P = 1 bar,
adsorbent concentration = 3.6 wt%.

Adsorbent T (C) QFG (%) QMG (%) QDG (%) QTG (%) Qsoap (%)
10
TriSyl 3000 50 8.0 16.4 n,d. n.d. 0.99
TriSyl 450 7.0 12.1 2.3 1.1 0.90
TriSyl 3000 TriSyl 300B 8.5 17.9 2.3 1.1 0.5
8 TriSyl 3000 65 8.4 13.6 n.d. 1.0 0.79
W, mg

TriSyl 450 7.6 14.7 n.d. 0.7 0.58

TriSyl 300B 8.1 14.7 n.d. 1.0 0.44
TriSyl 3000 80 8.0 12.3 n.d. 0.2 0.54
6 TriSyl 450 TriSyl 450 8.6 18.0 n.d. 1.1 0.45
TriSyl 300B 7.1 13.3 1.5 1.0 0.32
TriSyl 300 B n.d.: not detectable.
4

0 300 600 900 afnity for adsorption of FG and MG can be explained con-
T, C sidering the presence of surface OH groups on the silicas,
capable of adsorbing and retaining polar groups.
Fig. 2. TGA traces of the silicas. (iii) The treatment temperature does not appear to have a signif-
icant effect on the adsorption capacities of the different
evolution of structural water or decomposition processes at higher adsorbents for the analyzed impurities.
temperatures were not observed.
Adsorbed water can affect the adsorption process in several As the silicas have a high initial surface water content the pos-
ways: (i) blocking the access of impurities to the adsorption sites, sibility of needing a drying treatment prior to rening in order to
(ii) preventing the diffusion of these impurities by increasing the increase the adsorption capacity is therefore raised. Nock postu-
intraparticle resistance, and (iii) changing the surface afnity of lated that the high moisture content trapped within the silica
the adsorption sites. Feng et al. [19] studied the rate of adsorption structure is benecial and creates the ideal environment for
and desorption of water on silica and they concluded that the silica attracting polar impurities [20]. In this sense a drying pretreatment
surface becomes hydrophobic after heating in vacuo but it of silicas before rening could be detrimental and affect their
becomes hydrophilic immediately after contact with water at adsorption capacity. Anyway it is necessary to assess the adsorbed
low temperature. water effect. Thus, a previous conditioning step of the adsorbent
Moreover the presence of impurities in signicant amounts in would be required in addition to removing the methanol excess
the materials used as adsorbents may affect the adsorption sites from the crude biodiesel. If the rening treatment were performed
and consequently their adsorption capacity, therefore is very under conditions of vacuum and temperature, it would be possible
important to know the chemical composition of the adsorbents. to reduce the number of process stages, removing the excess of
The results of the chemical analysis by ICP of the three silica methanol and water (present in the crude biodiesel and also
samples indicated that all of them had essentially the same compo- adsorbed on the adsorbent material) simultaneously with the
sition. Silica samples were composed of SiO2 of high purity impurities adsorption. This would eliminate the need to add previ-
(>99.0%) containing impurities such as Mg, Al and Na, in contents ous stages of removal of excess methanol and dehydration of the
lower than 0.2%. adsorbent.
In this research study a single rening step was performed by
3.2. Inuence of bleaching conditions treating the crude biodiesel directly with the adsorbent, without
a previous drying step and under temperature and vacuum
Preliminary experiments were performed at three different conditions.
adsorption temperatures (50, 65 and 80 C) under atmospheric It is necessary to adjust the operating conditions in order to
pressure and using the untreated adsorbent in order to choose remove water and methanol fast enough so that they do not inter-
the best silica for the next exhaustive study. Prior to the adsorption fere with the adsorption of other impurities, thus lengthening the
tests, the spiked biodiesel was subjected to a treatment under vac- required process time. It is intended that in one step most of the
uum at 90 C for 1 h in order to remove residual methanol. This impurities present in the crude biodiesel are removed in a reason-
step is necessary since the presence of signicant quantities of able short contact times, if possible similar to that used in dry
polar compounds, such as methanol, affects the afnity of the most rening processes which employ other adsorbents. The treatment
hydrophilic compounds (glycerol) for the adsorbent, reducing its conditions (pressure, temperature, mass of adsorbent) to be
adsorption [13]. After this treatment the methanol content in the selected as optimal should ensure both adsorption of FG, DG, TG
spiked biodiesel was less than 0.01% while the water content and soaps, as well as the elimination of methanol and water. In this
was reduced to 0.04%. way a product that meets quality specications would be obtained.
Table 3 shows the results of adsorption capacity: free glycerol Table 4 shows the results of tests carried using silica Trisyl 3000
(QFG), monoglycerides (QMG), diglycerides (QDG), triglycerides as adsorbent. The choice between silicas was made considering
(QTG) and soaps (QS) for the different samples at three operating that silica Trisyl 3000 has the highest surface area and contains
temperatures. The results indicate that: the lowest amount of adsorbed surface water.
The temperatures evaluated were 65, 80 and 90 C. Assays were
(i) The three silicas adsorb glycerides. performed at the absolute pressure of 0.2 bar. The concentration of
(ii) The silica samples adsorb similar amounts of impurities adsorbent used was 1.1%. Contents of soaps, water, methanol, FG,
despite having different water contents. The order of afnity MG, DG and TG, ash point and FFA were then measured on the
for the impurities is: MG > FG  Soaps, DG, TG. The greater rened biodiesel.
376 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379

Table 4
Bleaching treatment. Content of impurities. Conditions: adsorbent = silica Trisyl 3000, pressure = 0.2 bar, adsorbent concentration = 1.1%.

T (C) Time (min) Soaps (%) Water (%) MeOH (%) FG (%) MG (%) DG (%) TG (%) FP (C) FFA (%)
Norm values 0.07 0.050 0.2 0.02 0.8 0.2 0.2 >101 0.5
Spiked biodiesel 0 0.25 0.100 5 0.2 2.41 0.08 0.04 45 0.27
65 15 0.25 0.050 0.002 0.03 2.41 n.d. n.d. 163 0.23
30 0.13 0.044 0.001 0.027 2.41 n.d. n.d. 186 0.20
45 0.009 0.042 n.d. 0.027 2.41 n.d. n.d. 190
90 0.004 0.040 n.d. 0.023 0.78 n.d. n.d.
80 15 0.25 0.050 0.0034 0.03 2.41 n.d. n.d. 155 0.17
30 n.d. 0.048 0.003 0.025 0.81 n.d. n.d. 153
45 n.d. 0.045 0.0025 0.02 0.8 n.d. n.d. 0.16
90 n.d. 0.040 n.d. 0.02 0.672 n.d. n.d.
90 15 0.001 0.040 0.005 0.03 0.47 n.d. n.d. 152
30 0.001 0.038 n.d. 0.028 0.45 n.d. n.d.
45 n.d. 0.030 n.d. 0.02 0.4 n.d. n.d. 0.16
90 n.d. 0.030 n.d. 0.019 0.32 n.d. n.d. 0.16

FP: ash point.

n.d.: not detectable.

Based on the results shown in Table 4, the following issues of the other impurities. In fact, it cannot be ruled out that part of
arise: the methanol and water remain adsorbed on the silica surface.
It is possible to achieve impurities concentration values that
(i) It is possible to remove methanol and water from the reac- comply with the quality standards, in one-step under vacuum
tion medium in acceptable treatment times, reaching values (0.2 bars) and temperature (90 C) and at contact times of
complying with the quality standards. 45 min. These contact times are common in the industrial practice
(ii) Contrary to our expectations water elimination does not of oil bleaching.
affect the adsorption of polar impurities on silica. A clear effect of temperature on the adsorption rate of MG is
(iii) It is possible to achieve the limit values prescribed by the noticed, which is larger at higher temperatures. If a sample pre-
regulations for the impurities analyzed, in a one-step treat- sents a greater adsorption rate at higher temperatures, this fact
ment under vacuum (0.2 bars) and mild temperature could be associated to high values of resistance to the diffusion
(90 C), with contact times longer than 45 min and low con- of the impurities within the pores at lower temperatures. Although
centrations of adsorbent (1.1%). The proposed treatment to a lesser degree this also happens with soaps. As to FG the treat-
under vacuum conditions is effective for removing metha- ment temperature does not have a major effect, since it is irrevers-
nol, water, FG, glycerides (particularly MG), FFA and soaps. ibly adsorbed at low treatment times. Then we can conclude that
(iv) Flash point values conrm the low residual methanol con- despite that silica has a suitable porous structure, at the conditions
tents obtained. of treatment there exists a steric hindrance to the free movement
of the molecules of bigger size (glycerides). In a recent paper [21]
The values of the adsorption capacities for some impurities at we reported that during rening with silica powders the mass-
different temperatures are shown in Fig. 3. The comparison of transfer mechanism of FFA is controlled by intraparticle diffusion.
the values obtained under vacuum conditions with those obtained The free acid content in the crude biodiesel was low because of
under atmospheric pressure (Table 3) suggests that the adsorption the characteristics of the feedstock used to prepare the biodiesel
capacity of the silica increases greatly in vacuum. When working at and the process itself in which FFAs react with the alkaline catalyst
90 C and a contact time of 90 min the adsorption capacity of FG to form soaps. Therefore the impurities concentration values after
increases from 79% to 1517%. For soaps it increases from less the treatment are adjusted to the specied quality standards. On
than 1% to 2022% and for MG from 1318% to more than 180%. the other hand the production of biodiesel using alkaline catalysts
The results largely exceeded our expectations. The rise observed provides a high biodiesel yield (>98%) and the main subproducts of
cannot be attributed only to an increase in the number of sites the transesterication are FG and MG. The concentration of DG and
available for adsorption after the removal of surface water. On TG is much lower than that of MG and generally meets the quality
the other hand these results are very promising because the adsor- standards.
bent retains approximately 23% of its weight when the process Table 5 compares the values of adsorption capacities experi-
takes place at atmospheric pressure and nearly 235% when carried mentally obtained for FG, soaps, MG and FFA, with the maximum
out under vacuum (these values are obtained by adding the values theoretical values calculated for each of these impurities, assuming
of Q (%) obtained for each impurity). This greatly improves the uti- that adsorption is reduced to the formation of a monolayer on
lization of the adsorbent, reducing the cost of the process. the adsorbent. For the calculation of the theoretical values the pos-
Another important observation is that the retention of soap is tulates of Langmuir were considered valid [22]. The assumptions
greatly improved under vacuum, equating the amount of glycerol were:
retained. The order of afnity observed was:
MG > FG  soaps > FFA, DG, TG. This leads us to afrm that the sil- (i) The concentration of adsorption sites of silica was 4 OH/nm2
ica after treatment in conditions of mild temperature and vacuum, [23].
behaves as a non-selective adsorbent with a high capacity of (ii) The adsorption stoichiometry was 1:1 for all impurities.
adsorption of different types of impurities. It therefore presents (iii) Each impurity is alone in the liquid medium.
afnity for all impurities in different degrees, which makes it an (iv) All sites are the same for the purpose of adsorption.
excellent adsorbent to purify biodiesel.
Under the conditions used, the methanol and water are Based on the analysis of Table 4 it can be concluded that for the
removed from the medium without interference on the adsorption silica Trisyl 3000 (under vacuum conditions) the adsorption
 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379 377

(a) 200

160
Adsorption Capacity, % Monoglycerides

120

80
(c) 200
Monoglycerides

40
160
Soaps
Free Glycerol

Adsorption Capacity, %
0
0 20 40 60 80 100 120

Time, min.

80
(b) 200

40
160 Monoglycerides Soaps
Adsorption Capacity, %

Free Glycerol
0
120 0 20 40 60 80 100

Time, min.

80

40
Soaps

Free Glycerol
0
0 20 40 60 80 100

Time, min.

Fig. 3. Evolution of the adsorption capacity as a function of time. (a) T = 65 C. (b) T = 80 C. (c) T = 90 C.

Table 5 the surface silica is mostly embedded with water in a multilayer.

Comparison between the theoretically calculated and experimentally determined The authors found differences between the characteristics of the
adsorption capacities (P = 0.2 bar, T = 90 C, contact time = 90 min., adsorbent
concentration = 1.1 wt%.).
rst layer of adsorbed water and the subsequent layers. Yamauchi
and Hondo [25] suggested that water would settle only on part of
Compound Qtheoretical monolayer (%) Qexperimental (%) the silanols at rst as SiOHOH2 complexes by an H-bonding
FG 36.0 15.6 mechanism. A second water molecule would be added on the pre-
MG 140.0 180.1 vious silanol water complexes in the form of SiOHOH2OH2.
DG 244.0 6.9
Garrone et al. [26] compared the adsorption of water on the sil-
TG 348.0 3.4
Soaps 120.0 21.6 ica surface with that of other molecules of similar size (CO, H2O,
FFA 113.0 7.8 N2O and NH3). They found evidences indicating that the main
interactions of silanol groups (SiOH) with the O end of CO, with
the O end of N2O, with the O atom in H2O, and with the N atom
phenomenon is not limited to the formation of a theoretical in NH3, are due to H-bonding. This H-bonding could be extended
monolayer of the adsorbed impurities. This is clear when consid- in the pore space involving more adsorbate molecules. Mizukami
ering that the adsorption of MG alone exceeds the monolayer value et al. [27] proposed the formation of surface molecular macroclus-
(without taking into account the adsorption of the other impuri- ters for adsorption of ethanol on the silica surface and proposed
ties). Then the adsorption phenomenon is more complex and must that the H-bonding between the silanol group and the OH group
include multilayer formation. of ethanol is responsible for adsorption.
The chemistry on the silica surface depends on the nature of The H-bonding adsorption model would allow obtaining
attractive forces existing between the adsorbate and the adsorbent. adsorbed amounts of the impurities higher than the monolayer
This interaction may be due to chemical bonding, hydrogen bond- value. On the other hand it would explain the results obtained in
ing, hydrophobic bonding or van der Waals forces. Hydroxyl the tests at atmospheric pressure and under vacuum. When the
groups play an important role on the silica surface because hydro- rening treatment is performed at atmospheric pressure the sur-
gen bonding is the most common mechanism of adsorption. Parida face of silica is covered by one or more layers of adsorbed water
et al. [24] postulated that due to the presence of silanol groups at so that in principle there would no free silanol groups capable of
378 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379

Table 6 to totally eliminate the methanol and water from the medium
Comparison of the adsorption capacity of different adsorbents. without affecting the adsorption of other impurities. Water
Adsorbent Cads Adsorption capacity (%) Qtotal elimination does not affect the adsorption of polar impurities.
(%) (%) Silica Trisyl 3000 retains 23% of its weight when the process is
FG MG DG TG Soaps FFA performed at atmospheric pressure and nearly 235% when
Magnesola 1 24.0 5.0 6.0 8.0 16.0 5.9 64.9 performed under vacuum conditions. This greatly improves the
Silicaa 2 11.5 2.0 2.5 4.5 8.0 4.7 33.2 utilization of the adsorbent and reduces the cost of the process.
Amberlite 1 15.2 8.5 23.7
resina
At the optimal process conditions silica behaves as a non-
2 2.0 4.7 6.7 selective adsorbent with a high capacity of adsorption of the differ-
Purolite resina 1 14.5 9.0 23.5 ent types of impurities, which makes it an excellent adsorbent to
2 1.0 3.0 4 purify biodiesel. The adsorption phenomenon is not limited to
Trysil 3000b 1.1 15.6 180.1 6.9 3.4 21.6 7.8 235.4
the formation of a theoretical monolayer of adsorbed impurities.
Cads: adsorbent concentration. The phenomenon is more complex and includes the formation of
Qtotal (%): total adsorption capacity (%). multiple layers.
a
Ref. [12]. Pretreatment: (i) the crude biodiesel was heated at 90 C for 10 min to
remove the excess of methanol, (ii) adsorbent: none. Treatment: T = 65 C, P = atm,
contact time = 20 min, stirring. Acknowledgements
b
Our results. Pretreatment: none. Treatment: T = 90 C, P = 0.2 bar, contact
time = 90 min, stirring. This research was sponsored by the Argentina National
Research Council (CONICET, PIP Grant 0398) and Universidad Nac-
ional del Litoral (CAI+D Grant 60-299).
retaining impurities. However this retention indeed happens. The
impurities are linked by H-bonding to adsorbed water molecules.
References
When the treatment is performed under vacuum and mild temper-
ature conditions the water is removed from the surface sites of [1] N. Shibasaki-Kitakawa, K. Kanagawa, K. Nakashima, T. Yonemoto,
adsorption, which are then occupied by impurities. An additional Simultaneous production of high quality biodiesel and glycerin from
adsorption by this mechanism is seen making it possible to obtain Jatropha oil using ion-exchange resins as catalysts and adsorbent, Bioresour.
Technol. 142 (2013) 732736.
values of adsorption greater than the monolayer one. [2] M. Berrios, R.L. Skelton, Comparison of purication methods for biodiesel,
In Table 6 our biodiesel rening results and those reported by Chem. Eng. J. 144 (2008) 459465.
Faccini et al. [12], are compared. These authors used different [3] ASTM D6751, Standard specication for biodiesel fuel blendstock (B100) for
middle distillate fuels, in: Annual Book of ASTM Standards; American Society
adsorbent materials (silica, Magnesol and ion exchange resins) for Testing and Materials, Philadelphia, PA, 1999.
for biodiesel dry rening. It can be seen that: [4] Automotive fuels  Fatty acid methyl esters (FAME) for diesel engines 
Requirements and test methods, Standard EN 14214, Beuth-Verlag, Berlin,
(i) Unlike our methodology, which is of one single stage treat- Germany, 2012.
[5] I.M. Atadashi, M.K. Aroua, A.R. Abdul Aziz, N.M.N. Sulaiman, Rening
ment, that used in reference [12] involves two steps of treat- technologies for the purication of crude biodiesel, Appl. Energy 88 (2011)
ment since it is necessary to remove excess methanol before 42394251.
rening. [6] J. Van Gerpen, Biodiesel processing and production, Fuel Process. Technol. 86
(2005) 10971107.
(ii) The values of total adsorption capacity obtained with our [7] M.B. Oliveira, F.R. Varanda, I.M. Marrucho, A.J. Queimada, J.A. Coutinho,
methodology far exceed those reported by Faccini et al. With Prediction of water solubility in biodiesel with the CPA equation of state, Ind.
our methodology silica Trisyl 3000 retains 2.35 times its Eng. Chem. Res. 47 (2008) 42784285.
[8] P. Jaruwat, S. Kongjao, M. Hunsom, Management of biodiesel wastewater by
weight of impurities while silica and Magnesol retain
the combined processes of chemical recovery and electrochemical treatment,
0.33 and 0.65 times its weight, respectively. Energy Convers. Manage. 51 (2010) 531537.
[9] J. Dugan, A dry wash approach to biodiesel purication, URL:
biodieselmagazine.com/article.jsp?article_id=1918 (2007).
It can be concluded that the proposed method has the following
[10] I.M. Atadashi, M.K. Aroua, A.R. Abdul Aziz, N.M.N. Sulaiman, Membrane
advantages: (i) simplicity, and (ii) it greatly improves the utiliza- biodiesel production and rening technology: a critical review, Renewable
tion of the adsorbent, reducing the cost of the process. Sustainable Energy Rev. 15 (2011) 50515062.
At this point, one last consideration can be made regarding the [11] B.S. Cooke, C. Abrams, B. Bertram, Purication of biodiesel with adsorbent
materials, US Patent 7, 635, 698 B2 (2009).
regenerability of the silicas. Our group is working on this topic [12] C.S. Faccini, M. Espinosa da Cunha, A.M. Aranda Moraes, L.C. Krause, M.C.
using selective solvents. At present we have managed to recover Manique, M.R.A. Rodrigues, E.V. Benvenuttia, E.B. Caramo, Dry washing in
80% of the adsorption capacity of the silicas. The procedure and biodiesel purication: a comparative study of adsorbents, J. Braz. Chem. Soc.
22 (3) (2011) 558563.
the main conclusions will be presented in a next contribution. [13] Grace Davison, TriSyl silica for edible oil rening, Technical Brochure (2005).
[14] V.A. Mazzieri, C.R. Vera, J.C. Yori, Adsorptive properties of silica gel for
biodiesel rening, Energy Fuels 22 (2008) 42814284.
4. Conclusions [15] H. Noureddini, D. Zhu, Kinetics of transesterication of soybean oil, JAOCS 74
(1997) 14571463.
The conventional biodiesel rening downstream transesteri- [16] E.P. Barrett, L.G. Joyner, P.P. Halenda, The determination of pore volume and
area distributions in porous substances. I. Computations from nitrogen
cation reactors, currently performed by decanting/washing or isotherms, J. Am. Chem. Soc. 73 (1951) 373380.
centrifugation/washing, can be conveniently replaced by a simple [17] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouqurol, T.
biodiesel purication method consisting of one single stage treat- Siemieniewska, Reporting physisorption data for gas/solid systems with
special reference to the determination of surface area and porosity, Pure
ment with silica as adsorbent. The method seems a solution for Appl. Chem. 57 (1985) 603619.
biodiesel rening, saving water, money and process time. [18] K.S.W. Sing, Adsorption methods for the characterization of porous materials,
With the proposed method it is possible to achieve a biodiesel Adv. Colloid Interface Sci. 76 (1998) 311.
[19] A. Feng, B.J. McCoy, Z.A. Munir, D.E. Cagliostro, Water adsorption and
composition complying the regulations on the analyzed impurities
desorption kinetics on silica insulation, J. Colloid Interface Sci. 180 (1996)
without a preliminary step to remove residual methanol, as it is the 276284.
case with other adsorbents, and loading the adsorbent without a [20] A. Nock, Silica hydrogel and its use in edible oil processing, URL:
previous drying treatment. Optimum operating conditions are lipidlibrary.aocs.org/processing/silica/index.htm (2011).
[21] D.L. Manuale, G.C. Torres, J.M. Badano, C.R. Vera, J.C. Yori, Adjustment of the
0.2 bar, 90 C, contact times longer than 45 min and 1.1% charge biodiesel free fatty acids content by means of adsorption, Energy Fuels 27
of adsorbent. Working under these conditions makes it possible (2013) 67636772.
 D.L. Manuale et al. / Chemical Engineering Journal 256 (2014) 372379 379

[22] I. Langmuir, The constitution and fundamental properties of solids and liquids. [25] H. Yamauchi, S. Hondo, The structure of water and methanol adsorbed on silica
Part I. Solids, J. Am. Chem. Soc. 38 (1916) 22212295. gel by FT-NIR spectroscopy, Colloid Polym. Sci. 266 (1988) 855860.
[23] S. Ek, A. Root, M. Peussa, L. Niinist, Determination of the hydroxyl group [26] E. Garrone, P.J. Ugliergo, Stepwise adsorption at the same site. A
content in silica by thermogravimetry and a comparison with 1H MAS NMR thermodynamic treatment, Chem. Soc. Faraday Trans. I 85 (1989) 585590.
results, Thermochim. Acta 379 (2001) 201212. [27] M. Mizukarni, M. Moteki, K. Kurihara, Hydrogen-bonded macrocluster
[24] S. Parida, S. Dash, S. Patel, B.K. Mishra, Adsorption of organic molecules on formation of ethanol on silica surfaces in cyclohexane, J. Am. Chem. Soc. 124
silica surface, Adv. Colloid Interface Sci. 121 (2006) 77110. (2002) 1288912891.