J O U R N A L O F M A T E R I A L S S C I E N C E 3 9 (2 0 0 4 ) 3705 – 3716

Topological organization and textural changes of

carbon macro-networks submitted to activation

with N2 and CO2

J. MATOS∗, M. LABADY, A. ALBORNOZ, J. LAINE, J. L. BRITOLaboratorio de Fisicoquımica de Superficies, Centro de Quımica, Instituto Venezolano deInvestigaciones Cientıficas (I.V.I.C.), Apartado 21827, Caracas 1020-A, VenezuelaE-mail: jmatos@ivic.ve

Polygonal macro-networks of carbon in the form of films were synthesized by thecontrolled pyrolysis of saccharose with or without potassium in the form of KOH. Thetopological organization of these random two-dimensional networks differ significantlyfrom other two-dimensional cellular structures. The pentagon was the most abundantpolygon in the film networks, which had ring areas in the range of 4 to 12 mm2. In additionto two-dimensional films, we also obtained irregular three-dimensional sponge-like ballsfeaturing a random network structure similar to that obtained in the films. These spongesdeveloped in the form of coiled tubes. We attribute the formation of these carbonmacro-coils to the presence of potassium particles inside the original carbonaceous matrix.A study of textural changes of the macro-networks of carbon by activation under flow of N2

and CO2 has shown that activated carbon with surface areas as high as 980 m2 · g−1 can beobtained from non-activated macro-networks. From our study of textural changes, wepropose a detailed mechanism that explains the role of KOH as a catalysts for the activationof the macro-networks of carbon. C© 2004 Kluwer Academic Publishers

1. IntroductionNetworks of cells having randomly distributed areasand numbers of sides constitute a large class of nat-ural materials including polycrystalline structures ob-served in metals, ceramics, polymers, Langmuir mono-layers, magnetic froths, flame cells, biological tis-sues, soap froths, etc. [1–10]. On the other hand, welldefined, molecular networks such as fullerenes andnanotubes have been some of the most remarkable re-cent discoveries in the area of carbon research [11–17]. Another ill-defined type of carbon network, corre-sponds to the microporous structure of activated car-bons (a material with many applications [18]), as-sumed to consist of polyhexagonal sheets of carbon,forming micropores and mesopores of widths between10 and 500 A [19, 20]. In addition, electron mi-croscopy has revealed other networks in activated car-bons [21]. In this case, the macropore (>500 A width)structure observed originates from the preservation af-ter carbonization of the biological tissue skeleton ofthe raw material (wood, coconut shell, etc.). Thus,at least three carbonaceous network structures format different length scales: the molecular (fullerenes,nanotubes), the micro (microporous carbon), and themacro (macroporous carbon). Despite extensive in-vestigation [11–27] of synthetic routes to form bothmolecular-network and micro-network carbon materi-

∗Author to whom all correspondence should be addressed.

als, few studies discuss the synthesis and characteri-zation of macro-networks [28]. The study of the topo-logical and textural characteristics of carbon films is ofconsiderable interest since they relate closely both tothe physical and chemical properties of these materials[29].

We can consider carbon films to be two-dimensional(2D) random cellular networks, which we treat as ran-dom partitions of the plane into cells (two-dimensional(2D) polygons with three or more sides) [1, 2, 5, 30].The topology of natural and simulated cellular networksconstrains the possible configuration of the cells, sincean infinite space-filling two-dimensional network withtrigonal vertices obeys the Euler topological relation[1, 2, 31–33]:

〈n〉 =∑

n

[nρ(n)] = 6, (1)

for the first moment of the cell-side distribution, whereρ(n) is the probability of ocurrence of a cell with nsides, which corresponds to the trivial constraint onprobabilities, given by:

∑

n

[ρ(n)] = 1 (2)

0022–2461 C© 2004 Kluwer Academic Publishers 3705

References [3, 5–7, 34, 35] present detailed descrip-tions of some of the concepts involved in the topologicaldefinition of such random cellular structures.

This communication presents, first, the method ofsynthesis and the resulting topological organization oftwo carbonaceous macro-network 2D-films obtainedby a step-wise heat treatment of saccharose with orwithout potassium hydroxide. Second, we study exper-imentally the texture changes of these carbonaceousnetwork materials when activated at up to 800◦C withN2 or CO2, to elucidate the potential applications ofthese carbonaceous materials. Our study addreses theproduction of network films of activated carbon thatcould serve as fixed co-adsorbents in heterogeneousphotocatalytic processes such as the photomineraliza-tion reactions of organic pollutants in waste water [36].

2. Experimental2.1. Network processingWe placed a solution of either 1 mL of water or 1 mLof aqueous solution (34 wt%) of potassium hydroxide(Merck, analytical grade) plus 2 g of D(+)-saccharose(Merck, bacteriology grade) in a 50 mL pyrex beakerand stirred it at 80◦C until the solution became viscousand light brown. Immediately, we remove the heat andleft the solutions until the saccharide recrystalized asan homogeneous and transparent sheet at the bottom ofthe beaker.

We then applied two consecutive heat treatments tothe sheets: first, a low temperature stabilization treat-ment inside an oven (static air) at the following tem-peratures: 60, 90, 110, and 130◦C, for 10 min at eachtemperature using heating rates between steps of about3◦C · min−1. This initial heat treatment aims to promotevery slow and homogeneous expansion of the saccha-ride film along the beaker walls up to its exterior border,since the final stabilization temperature corresponds tothe beginning of the glass transition of saccharose. Thesecond heat treatment consists of carbonization underan inert flow of nitrogen (∼=100 mL · min−1) using thefollowing step temperatures: 50, 100, 150, 200, 250,350◦C, soaking 30 min at each temperature, and finally1 h at 450◦C, using heating rates between steps of about5◦C · min−1. The final 450◦C carbonization tempera-ture is optimal for the production of low-temperature(L-type) activated carbon from lignocellulosic raw ma-terials [37, 38].

We denote these carbonaceous macro-network sam-ples as ANW-1 and ANW-2, for the saccharose dissolvedin water or in the KOH solution, respectively.

2.2. Statistical characterizationof macro-network films

Our statistical analysis of photographs of the macro-network films ANW-1 and ANW-2 (three samples foreach set of conditions) is followed as in previous stud-ies of other cellular structures [1, 2, 5]. We analysed atotal of 750 and 558 cells, respectively, for the ANW-1and ANW-2 samples, and the error in measuring cell ar-eas was less than 5%. We found polygonal cells with

from three to nine sides. We measured the probabilityof ocurrence of cells with n sides ρ(n), the distribu-tion of areas of n-sided cells, ρ(An), and the averagepolygon areas for each topology 〈An〉 by an analyticalmethod that we have previously employed [28]. We ob-tained the average polygonal areas 〈An〉 as the productof the distribution of areas of n-sided cells ρ(An) withthe actual geometric size of each of the photographedsamples according to the magnification factor of thephotographies.

2.3. Activation processing of the networksBesides the macro-network films we also obtainedtwo types of carbonaceous three-dimensional (3D) net-works, that we used to study textural changes dur-ing activation processing. We performed the activa-tions in a tubular furnace, under either inert N2 flow(∼=100 mL · min−1) with several final residence times:5, 60, 120 and 300 min at constant final tempera-ture of activation equal to 800◦C; or under CO2 flow(∼=100 mL · min−1) at the following temperatures of ac-tivation: 700, 750 and 800◦C, with a constant residencetime of 5 min.

2.4. Textural characterizationof the networks

We used N2 adsorption at 77◦K to characterize the tex-ture of the macro-networks (ANW-1 and ANW-2) and theactivated network materials in terms of the B.E.T. sur-face areas. We measured the full isotherms in the rangeof 0.03 to 630 Torr using an ASAP-2010 apparatus fromMicromeritics. For experimental details see [39, 40].

3. Results and discussion3.1. Statistical analysis

of macro-network filmsA previous communication [28] described the topolog-ical organization of the ANW-1 macro-networks. Thepresent work describes the topological organizationof both macro-networks films ANW-1 and ANW-2, butshows photographs only for the ANW-2 samples.

In both samples, two- and three-dimensional macro-networks were formed simultaneously during car-bonization. The macro-network films adhered both tothe bottom and the cylindrical beaker walls, coveringapproximately one third and one half of the beakerheight, respectively, for the ANW-1 and ANW-2 samples.Fig. 1a shows a typical ANW-2 macro-network, consist-ing of a random net of polygonal ring cells, each formedby grains of carbonaceous material. Both ANW-1 (notshown) and ANW-2 (Fig. 1a) had polygons from threeup to eight or nine sides. After thermal treatment, theANW-2 sample has carbonaceous material between thecells while the ANW-1 film sample does not. The insideof the ANW-2 network cells is not empty as in the ANW-1films. Fig. 1b shows a particular ANW-2 sample that has“T ” type vertices with an internal angle θ = 180◦.In several types of random network, particularly so-lidified polymer films with topological organization

3706

(a)

(b)

Figure 1 Images of the carbonaceous macro-network film from the ANW-2 sample adhering to the pyrex beaker: (a) Typical random pattern showingthe predominance of five-sided cells and (b) A subsection showing the presence of “T ” type vertices with internal angle θ = 180◦.

very similar to the present macro-networks, the T -typevertices results from the dynamics of cell formation[1, 2].

Fig. 2 shows the abundance of n-sided polygonsρ(n) obtained from the statistical analysis of 750 and558 cells, respectively, for ANW-1 and ANW-2 macro-network films. Both distributions are similar with pen-

tagons most abundant and:

〈n〉ANW-1 = 5.155 (3)

〈n〉ANW-2 = 5.161 (4)

Although the low value of 〈n〉 occurs in microscopicsolidified polymer films [1, 2], most macroscopic

3707

Figure 2 Polygonal distributions of the macro-networks from the ANW-1 (Sac-H2O) and ANW-2 (Sac-KOH) films.

cellular structures have hexagon as the most abundantpolygon as shown in Table I.

Fig. 3 shows the fraction of the total area in n-sidedcells ρ(An) of both macro-networks. The distributionsare practically identical and at n = 5, ρ(A5) > ρ(A6).Again, these results agree with microscopic polymerfilms noted above [1, 2], and differ from those inother micro- and macroscopic random networks, suchas polycrystalline SCN [3–6], polycrystalline metals[4, 5], soap froths [7], flame cells [8], Langmuir mono-layer [9], and magnetic froths [10]. Therefore, two-dimensional carbon macro-networks geometric prop-erties differ intrinsically from those in other types ofnetwork.

T ABL E I Comparison of peak values in cell-side distributions ρ(n)for various materials

System Peak n-value ρ(5) ρ(6) Reference

ANW-1 5 0.334 0.206 Present workANW-2 5 0.353 0.233 Present workPolymer polygrain 5 0.392 0.25 [1]Flame cells 5.4 0.36 0.34 [8]Soap froths 5.5 0.3 0.3 [7]Langmuir monolayer 6 0.225 0.345 [9]Polycrystalline metal 6 0.267 0.4 [4]Polycrystalline SCN 6 – – [3, 6]Magnet froth 6 0.25 0.55 [10]

Fig. 4 shows that for both macro-networks, the av-erage cell areas of an n-sided cell, 〈An〉, is linear in n(with correlation coefficients close to 1 in both cases):

〈An〉 = (0.007 mm2) + (1.235 mm2)n (5)

〈An〉 = (−0.0011 mm2) + (1.2988 mm2)n (6)

for ANW-1 and ANW-2, respectively. Although this lin-ear behaviour is consistent with Lewis Law [1, 2, 41],the constants in Equations 5 and 6, differ from thosereported for other types of random network [1–10].Table II shows that most cell areas fall inside the range 4to 10 mm2 and 4 to 12 mm2, for ANW-1 and ANW-2, re-spectively. Therefore, the average values, 〈a〉, for eachcase are: 6.7993 and 7.7921 mm2, indicating that theLewis constants (λ) for the present work are: 0.1816and 0.1667, which are similar to the values reported(λ = 1/4) [1, 2, 41].

3.2. Morphological characteristicsand chemical composition of thecarbonaceous macro-networks

In addition to two-dimensional films, irregular sponge-like balls form in the bottom of the beakers. Fig. 5 showsan image of a part of the carbonaceous bulk in an ANW-2

3708

Figure 3 Distributions of areas in the macro-networks from the ANW-1 (Sac-H2O) and ANW-2 (Sac-KOH) films.

sample. Fig. 5a shows that this three-dimensional fea-ture resembles the 2D random network in the same ma-terial (Fig. 1), a random network of polygonal carbona-ceous grains of different sizes. For both types of sample,the sponges connect loosely to the film network. More-over, the noticeable separation of the polygons (Figs 1and 5) suggest weak linking, which most probably leadsto the easy disconnection of the sponge balls from thefilms adhered to the beaker’s wall.

The carbonaceous grain structure observed in Fig. 5bdoes not satisfy Lewis’ Law [2, 41], suggesting thatthe nucleation time for the different grains in the ma-trix is not the same. This is explained as follows: Bycomparing several images of the two- and the three-dimensional ANW-2 macro-networks, we found severalinterfaces where the carbon grains developed into a con-tinuous 3D polygonal net of macrotubes (Fig. 5b). Per-haps the carbonaceous macro-networks, form in a waysimilar to those of carbon nanotubes [42].

T ABL E I I Average areas of n-sided cells, 〈An〉, of the ANW-1 and ANW-2 macro-network films as a function of n

n sided-cells n = 3 n = 4 n = 5 n = 6 n = 7 n = 8 n = 9

[〈An〉] for ANW-1 (mm2) 3,708 4,948 6,186 7,416 8,656 9,882 –a

[〈An〉] for ANW-2 (mm2) 3,896 5,194 6,493 7,792 9,090 10,39 11,69

aPolygonal cells of 9-sides were not present in the ANW-1 sample.

Fig. 5b shows a surprising morphology. Some of thecells have coiled shapes [26, 43, 44] (denoted mcs inFig. 5b). Several authors have described the formationof coiled carbon tubes and their growth mechanism[26, 43, 44]. Carbon tube nano-coils form due to thepresence of iron or indium and tin oxide films actingas catalysts, mainly in thermal chemical vapor depo-sition. Certain metal species tend to migrate throughcarbonaceous deposits formed during catalytic vapor-chemical decomposition of hydrocarbons to producecarbon nano-tubes, filaments and vapor-growth fibers[45]. We attribute our carbon macro-coils to the pres-ence of potassium microparticles in the carbonaceousmatrix, indirectely confirmed by the absence of macro-coiled carbon filaments in the ANW-1 sponge whichlacks potassium [28]. In macroscopic morphology theANW-1 sponge [28] resembles typical reticulated glassycarbon foams with a pentagonal dodecahedron struc-ture as those obtained in polystyrene and polyurethane

3709

Figure 4 Average polygonal areas in the macro-networks from the ANW-1 (Sac-H2O) and ANW-2 (Sac-KOH) films.

foam insulations [46] or from thermal processing ofmesophase pitches [47].

Compairing Figs 1 and 5, shows that 3D cells havevoid spaces between cells indicating that the bulk ofthe material was consumed by pyrolysis because theadhesion forces to the pyrex wall reduced volatilizationin the 2D films.

During carbonization the material volume increasesremarkably, particularly, in the saccharose sample dis-solved in KOH solution (ANW-2). Table III shows thatthe large increase in volume during carbonization in the3D ANW-2 sample results in a lower density than for theANW-1 sample (0.0156 g/cm3 against 0.0223 g/cm3).We estimated the apparent densities in Table III fromthe ratio between the weights of different masses ofthe samples (mechanically compressed into a graduatedcylinder) and their volume after substracting the theori-cal remnant weight of potassium (as K2O) in the ANW-2

T ABL E I I I Apparent density, ρapp, of the 3D ANW-1 and ANW-2

macro-network materials

Network ρapp (g/cm3)

ANW-1 0.0223 ± 0.0009ANW-2 0.0156 ± 0.0006

sample. As we described above, for 1 mL 34 wt% ofaqueous KOH solution, the weight of K2O is 0.4501 g.

Since our carbonaceous materials are a complexbiphasic arrangement of crystalline lamellar and amor-phous phases, we expect topological effects at bound-aries between crystalline-crystalline and crystalline-amorphous phases [2]. Since the present networks aresaccharose-based carbons, we expect textural changesduring thermal activation at temperatures higher thanthose employed for the carbonization [27, 48].

Several results of Laine and co-workers [28, 38, 49]reinforce the above suggestion. Thus, elemental analy-sis of the “blank” macro-network (ANW-1) has shown[28] a chemical composition of: C = 61%, O = 36%,and H = 3% (wt%), which corresponds to C8H4O4.Compared to the saccharose base C12H22O11 the hy-drogen and oxygen decrease by 4 and 12 wt% respec-tively, and the carbon increases by 16 wt%. On theother hand, nitrogen adsorption in the ANW-1 and ANW-2macro-networks was undetectable, indicating negligi-ble B.E.T (Brunauer-Emmett-Teller) surface areas [28],unlike activated carbon which normally has around1000 m2/g surface area [18]. As will be seen below, thepresent 3D macro-networks have relatively large cellswhich are more than 100 times wider than the macrop-ores found in carbons derived from lignocellulosic raw

3710

(a)

(b)

Figure 5 Image of bulk carbonaceous macro-network from the ANW-2 sample: (a) 3D random structure of a subsection of the sponge and (b) Subsectionshowing macrocoil shapes (mcs).

materials (agreeing with the B.E.T. surface areas forANW-1 and ANW-2 samples listed in Table V). Finally, atemperature programmed reduction (TPR) analysis ofthe network blank material [28, 49] has shown a TPRtrace resembling those reported [38] for L-type acti-vated carbons; i.e., those prepared by low temperature-

“chemical”-activation, a synthesis method that usesa chemical additive (normally H3PO4, ZnCl2, KOH,etc.). In that work, after a first positive peak in theTPR spectrum attributed to H2-reduction of the car-bonaceous material (consumption of hydrogen), a hightemperature negative peak (which is not observed in

3711

T ABL E IV Summary of the experimental conditions employed for the study of texture changes of the 3D macro-networks under activation by N2

and CO2

Activation ActivationNetworks temperature (◦C) residence time(min) Other experimental conditions

AN2-2 800 60 N2-activated from ANW-1, without H2O pre-washingACO2-2 800 60 CO2-activated from ANW-1, without H2O pre-washingAN2-3 800 60 N2-activated from ANW-2, with H2O pre-washingACO2-3 800 60 CO2-activated from ANW-2, with H2O pre-washingAN2-4A 800 60 N2-activated from ANW-2, without H2O pre-washingAN2-4B 800 120 N2-activated from ANW-2, without H2O pre-washingAN2-4C 800 300 N2-activated from ANW-2, without H2O pre-washingAN2-4D 800 5 N2-activated from ANW-2, without H2O pre-washingACO2-4A 800 60 CO2-activated from ANW-2, without H2O pre-washingACO2-4B 800 5 CO2-activated from ANW-2, without H2O pre-washingACO2-4C 750 5 CO2-activated from ANW-2, without H2O pre-washingACO2-4D 700 5 CO2-activated from ANW-2, without H2O pre-washing

H-type activated carbons; i.e., those prepared by hightemperature-“physical”-activation) was assigned to hy-drogen evolution from the macro-network material [28,49].

The earlier results indicate activating the ANW-1 andANW-2 samples is possible to obtain activated carbonsshowing large B.E.T. surface areas but also that theymight be usable for hydrogen storage [45].

3.3. Textural changes in the networkscaused by activation

3.3.1. General discussionTo confirm that high temperature heat treatmentschange the topology and texture of carbon macro-networks we studied activation with N2 and CO2 ofboth 3D networks. Table IV summarizes the samplesand experimental conditions.

The first section of Table IV, includes two samples ofANW-1 activated under N2 (AN2-2) or CO2 (ACO2-2) at800◦C for 1 h. Since ANW-1 has no potassium the blankwas not washed with H2O before activation.

The second section of Table IV includes two sampleswith potassium (ANW-2) activated under N2 (AN2-3) orCO2 (ACO2-3) at 800◦C for 1 h. Before activating ANW-2we washed it with H2O until it reached a neutral pH.To compare the AN2-3 and ACO2-3 samples with theAN2-2 and ACO2-2 samples, we substract the potassiumoriginally in the ANW-2 carbonaceous matrix.

The third section includes four samples with potas-sium (ANW-2) activated under N2 at 800◦C for differentactivation times (5, 60, 120, and 300 min). We des-ignate these samples AN2-4i. To study the influence ofthe potassium in activation, we did not wash the ANW-2with H2O before activation.

Finally, the last section of Table IV shows four sam-ples with potassium (ANW-2) activated under CO2 atactivation temperatures from 700 up to 800◦C and withactivation residence times from 5 to 60 min. We desig-nate these samples as ACO2-4i. As in the previous Sec-tion, to study the influence of potassium on activation,we did not wash with H2O before activation. We hopeto compare the influence of both the choice of gas andof potassium on total burn-off of saccharose and theB.E.T. surface in the two types of activation.

TABLE V Total burn-off of saccharose and B.E.T. surface areas ofthe 3D networks

Saccharose burn-off a B.E.T. surface areaNetworks (%) (m2· g−1)

ANW-1 85 10ANW-2 66 15AN2-2 92 8ACO2-2 91 270AN2-3 70 79ACO2-3 72 137AN2-4A 72 558ACO2-4A 100 –b

aExcept for the values for AN2-3 and ACO2-3 which were washed withH2O, all the saccharose burn-off (%) values were estimated after sub-tracting the theoretical reminant weight of potassium as K2O (≈0.4501g).bThe B.E.T. surface area of the ACO2-4A sample is not included since itwas totally pyrolyzed.

The B.E.T. surface areas of the ANW-1 and ANW-2samples listed in the Table V (10 and 15 m2 · g−1) sug-gest a non-porous texture, in line with the morphologyobserved in Fig. 5, which shows that the 3D macro-networks of carbon have relatively large cells morethan 100 times wider than the macropores found in car-bons derived from lignocellulosic raw materials. Theoptimum temperature for L-type activated carbon syn-thesis is 450◦C [37, 38], which is the same final tem-perature we employed to prepare our macro-network.Table V shows that the burn-off in our experimentswas relatively high (about 85%) for ANW-1 comparedto normal chemical activation (40–60%). However,the burn-off (Table V) in the ANW-2 sample (withpotassium) is only slightly higher than normal (about66%).

The total burn-off obtained in the AN2-2 and ACO2-2samples is slightly higher (respectively, 92 and 91%)than that of ANW-1 (85%). In addition, the total burn-offfor the AN2-3 and ACO2-3 samples is slightly higher (re-spectively, 70 and 72%) than that of ANW-2 (66%), onlyslightly higher than the normal range for chemical acti-vation. This is in line with reports by Laine and cowork-ers [20, 21, 38, 39, 50] which propose that the chemicalactivator (H3PO4 and/or KOH) supresses volatilizationand pyrolysis that would destroy the fine microporous

3712

which causes the high surface area of L-type activatedcarbon. In high oxidation activation with CO2 (ACO2-2sample) the B.E.T. surface area obtained from ANW-1was moderate (270 m2 · g−1) while it was negligiblewhen activated with N2 (AN2-2 sample). However, theB.E.T. surface areas for the AN2-3 and ACO2-3 sam-ples show a different trend. Activation of ANW-2 un-der N2 (AN2-3) produced a higher B.E.T. surface areathan for the AN2-2 sample obtained from ANW-1 (79against 8 m2 · g−1). Activated under CO2, the ACO2-3sample produced a lower B.E.T. surface area than didthe ACO2-2 sample (137 against 270 m2 · g−1). Thesesurface areas seem to indicate that although the ANW-2sample was previously washed with H2O, it still con-tained a small amount of potassium, sufficient to inhibitexcesive pyrolysis and hence to increase the B.E.T. sur-face area for activation under nitrogen compared to ac-tivation of ANW-1 under N2 (79 � 8 m2 · g−1). In otherwords, KOH is an effective catalyst for the activationof the carbonized ANW-2 macro-network, both in thepresence of N2 and CO2.

KOH at the surface of the carbonized lignocellulosicmaterial decomposes according to the reaction [50]:

2KOH → K2O + H2O, (7)

producing gaseous water that due to its small weightand molecular diameter diffuses easily into the carbonpores, contributing to gasification. The basic reactionof carbon with water vapor is endothermic and of thefollowing stoichiometry [21, 51]:

H2O + C → CO + H2, �H = +117 kJ · mol−1.

(8)

In addition, the abundant oxygen available from thecarbonaceous matrix can consume the yield productsof Equation 8 (CO + H2):

CO + H2 + 2O → CO2 + H2O, (9)

further increasing the concentrations of activators (CO2+ H2O).

Yuhn and Wolf [52] have shown that other metalalkaline salts, like sodium nitrate and hydroxide, in-teract with carbon to form surface carbonate species.Accordingly, the formation of a layer of oxides (sim-ilarly to the layer of polyphosphate reported by Laineand Calafat [50]) over the developing pore structurecould prevent the excessive gasification found for catal-ysis with potassium hydroxide, due to interaction withgaseous CO2.

Like the gasification reaction of Equation 8, the re-action of carbon with carbon dioxide can be expressedas:

CO2 + C → 2CO, �H = +159 kJ · mol−1. (10)

On the other hand, above 800◦C the water-gas shiftreaction is at equilibrium [51]:

CO + H2O � CO2 + H2 (11)

with an equilibrium constant between 0.5 and 1. Bothreactions (8) and (10) are important. Though super-heated steam is usually the primary activater, it is notin our case, so we suggest that reactions (7) to (9) pre-dominate during activation of macro-networks with N2,while reactions (7) to (11) and particularly reactions(10) and (11) are all involved during the activation ofmacro-networks with CO2.

The last section of Table V (AN2-4A and ACO2-4A sam-ples), suggest that this mechanism not only is in line(via reactions (7) to (9)) with the moderate burn-off ofsaccharose (72%) and relative high surface area (558m2 · g−1) of the ANW-2 sample (with potassium) acti-vated with N2 at 800◦C for 1 h (AN2-4A), but also, (viareactions (10) and (11)) with the maximum burn-offof saccharose (100%) observed for the ANW-2 sample(with potassium) activated with CO2 at 800◦C for 1h (ACO2-4A). In neither AN2-4A and ACO2-4A was themacro-network (ANW-2) washed with water before ac-tivation so they contained all the remnant potassium.The last section of Table V suggest that the presenceof potassium as KOH initiates, by means of reaction(7), the catalytic activation of the carbonaceous net-work. This effect is stronger for the sample activatedunder CO2 (ACO2-4A) where the macro-network sam-ple (ANW-2) was totally pyrolyzed probably by the in-creased amounts of activators (CO2 + H2O) producedby reaction (11).

3.3.2. Influence of the activation timeFig. 6 shows the B.E.T. surface areas and the to-tal burn-off of saccharose estimated after substractingthe theoretical remnant weight of potassium as K2O(≈0.4501 g). ANW-2 samples activated under N2 at

Figure 6 Surface B.E.T. areas and burn-off behaviour of networks acti-vated with N2 (at 800◦C) as a function of residence time.

3713

800◦C as a function of the residence activation timewere designated AN2-4i. The burn-off tends to increasewith the activation time, a well-known trend reportedfor several types of lignocellulosic materials [53–60].The burn-off of the AN2-4i samples shown in Fig. 6 in-creases from 68% at 5 min to close to 80% at 300 min,principally through reactions (7) to (9). The quantitiesof activators (CO2 + H2O) produced by reaction (9),may increase strongly at longer activation times leadingto greater activation and more pyrolysis of the saccha-rose. The B.E.T. surface areas for the AN2-4i samples,in Fig. 6 show that the texture of the macro-network(ANW-2) changes as a function of the residence time,the surface area increasing from 565 m2 · g−1 at 5 min(AN2-4D) to a value of 558 m2 · g−1 at 60 min (AN2 − 4A)to a maximum of 850 m2 · g−1 at 120 min (AN2-4B); thendrastically falling to 540 m2 · g−1 at 300 min (AN2-4C).For activation times longer than 2 h, activation seems tobecame less efficient, probably due to the well-knowndestruction of the fine micropore structure which pro-duces the high surface area of L-type activated carbons[20, 21, 37, 38]. The results in Fig. 6 agree with thismechanism, so reactions (7) to (9) are probably the maincause of texture changes during activation of the ANW-2macro-network under N2 at 800◦C.

3.3.3. Influence of the activationtemperature

We now discuss the ANW-2 macro-network samplesdesignated ACO2-4A activated for 5 min under CO2 asa function of the activation temperature. Fig. 7 showstheir B.E.T. surface areas and total burn-off of saccha-rose. The burn-off tends to increase with the activatingtemperature. This burn-off dependence resembles theence of the temperature on yields of carbons produced

Figure 7 Surface B.E.T. areas and burn-off behaviour of networks acti-vated with CO2 (by 5 min) as a function of temperature.

from different lignocellulosic materials [37, 38, 55–57,59]. The saccharose burn-off increases from a moder-ate value of 72% at 700◦C (ACO2-4D) to a high value of94% at 750◦C (ACO2-4C) up to a maximum of 100% at800◦C (ACO2-4D). This trend is in line with the mecha-nism suggested above using reactions (7) to (11). Themost important reactions are Equations 10 and 11, par-ticularly the last, which begins to be important when theactivation temperature approaches 800◦C. The higherthe activation temperature the more saccharose pyrol-ysis. In line with this behaviour, the B.E.T. surfacearea should decrease with increasing final activatingtemperature. Fig. 7, shows that these values decreasesfrom 981 m2 · g−1 at 700◦C (ACO2-4D) to 717 m2·g−1 at750◦C (ACO2-4C) to zero near 800◦C (ACO2-4B) when themacro-network pyrolyzes completely. This drastic tex-tural change shown in B.E.T. surface area results fromthe destruction of the micropore texture created duringthe initial stages of activation by the strong oxidationconditions. The presence of potassium as KOH (thatproduces H2O via reaction (7)) and the high tempera-tures (via reaction (11) which promotes the productionof CO2) both increase this destruction. Both oxidizingagents and activators (H2O and CO2) can pyrolyze thecarbonaceous matrix via the two gasification reactions(8) and (10).

3.3.4. Influence of the KOH concentrationFinally, we discuss the influence of the KOH concen-tration on the textural changes of the macro-networks(ANW-2) activated under N2 at 800◦C for 1 h. Fig. 8shows the B.E.T. surface areas obtained from the ad-sorption isotherm of N2 at 77◦K and the total burn-off ofsaccharose estimated after substracting the theoretical

Figure 8 Surface B.E.T. areas and burn-off behaviour of networks acti-vated with N2 (at 800◦C, by 1 h) as a function of KOH concentration.

3714

remnant weight of potassium as K2O, as describedabove. Fig. 8 shows that the burn-off decreases mono-tonically from a maximum of 87% at 1 wt% of KOH to aminimum of 72% at 34 wt% of KOH and then slightlyincreases to 76% at 67 wt% of KOH. This burn-offbehaviour is very similar to that reported by Do andAhmadpour [54] for the preparation of activated car-bon from Macadamia nutshells by chemical activationwith ZnCl2. Fig. 8 also shows that the B.E.T. surfaceareas for the activated macro-networks have a volcanoshaped dependence on the KOH concentration. Thisprofile shows a low B.E.T. surface area of 96 m2 · g−1

for 1 wt% of KOH, in line with the water washed macro-networks AN2-3 under the same experimental activationconditions (N2, 800◦C, 1 h) with a B.E.T. surface areaof around 79 m2 · g−1 (Table V). This result suggeststhat at least, some potassium as KOH is required tocatalyse macro-network activation. For KOH concen-trations higher that 1 wt%, the B.E.T. surface areas ofthe activated macro-network increase exponentially upto a maximum value of 558 m2 · g−1 at 34 wt%. Coinci-dentally, this point corresponds to the concentration ofKOH which has the minimum saccharose burn-off. ForKOH concentrations higher than 34 wt%, the B.E.T.surface area decreases drastically to a moderate valueof 276 m2·g−1 at 67 wt% of KOH.

These results are in line with the mechanism pro-posed above via reactions (7) to (9). The presence evenof very low concentrations of potassium as KOH in-hibits the pyrolysis of the macro-network, indepen-dently of the concentration of the hydroxide. In termsof the B.E.T. surface areas, the texture changes in theactivated macro-networks in which the microporosityat low KOH concentrations increases to a maximumwith increasing KOH then decreases for higher con-centrations of KOH, results from the stronger oxida-tion conditions, favoured by very high concentrationsof potassium as KOH. As reaction (7) indicates, KOHthermally decomposes to produce K2O + H2O, andgaseous water is a very efficient activator that via re-action (8) gasifies the macro-network. Another expla-nation of the loss of B.E.T. surface area in Fig. 8 isthe well-known capacity of the alkali metals, in par-ticular potassium, to intercalate between the parallelplanes of graphite-type carbons [61–63]. Given the ini-tial macroscopic morphology of the non-activated net-works (Fig. 5), high concentrations of KOH may favourthe intercalation of potassium into the macro-networksheets during activation, concominantly blocking mi-cropores and so decreasing the surface areas as in Fig. 8.In addition, potassium-intercalation in graphite influ-ences its hydrogen adsorption capacity [63] in linewith the results in Section 3.2, that the temperatureprogrammed reduction profile of the present macro-network of carbon shows a large negative peak at veryhigh temperatures that is due to hydrogen evolution[28].

4. Summary and conclusionIn summary, controlled pyrolysis of saccharose pro-duces polygonal macro-network films adhering to thebottom and walls of pyrex flasks. Our data, analysis

and correlations indicate that the topological proper-ties of these random 2D networks, composed princi-pally of carbon aggregated grains (carbon polygrains),differ significantly from other two-dimensional cellularstructures. The pentagon was the most abundant poly-gon in the films, with ring areas in the range of 4 to10 mm2 and 4 to 12 mm2, for networks originatingfrom saccharose pre-dissolved in water or in aqueoussolution of 34 wt% KOH, respectively.

Irregular 3D sponge balls with random networkstructures similar to those in the films also formed at thebottom of the beakers for both samples. These spongesare assemblies of carbonaceous grains of different sizeswith an interface where the carbon grains develop intoa continuous 3D net of macrotubes with coiled shapes,due to the presence of potassium microparticles in theoriginal carbonaceous matrix. The properties of thesesponges indicate that they might be useful for hydrogenstorage.

The activation process we present can produce ac-tivated carbons with B.E.T. surface areas as high as850 and 981 m2 · g−1 for activation with N2 and CO2,respectively, corresponding to surface areas 85 and98 times higher than those of non-activated macro-networks. We propose that KOH catalyzes the activa-tion of macro-networks.

Future work, identifying the effects of sample prepa-ration, molecular weight and chemical composition ofthe saccharide precursors on the random cell structuresof these materials will enhance our understanding of thephysics of carbon networks, and probably such studiesmay permit us to find some connection between thetopological and texture properties.

Since activated carbon materials can induce photo-catalysis [64–68], and since our carbon films are trans-parent, we are testing the influence of our carbonaceousmacro-networks on titania-photocatalyzed reactions.

AcknowledgementsAuthors acknowledge to Referee of Journal of MaterialScience, and financial support by FONACIT.

References1. T . H U A N G, M. R. K A M A L and A. D. R E Y , J. Mater. Sci.

Lett. 14 (1995) 220.2. M. R . K A M A L, T . H U A N G and A. D. R E Y , J. Mater. Sci.

32 (1997) 4085.3. V . E . F R A D K O V, M. E . G L I C K S M A N, M. P A L M E R,

J . N O R D B E R G and K. R A J A N , Physica D. 66 (1993) 50.4. V . E . F R A D K O V, A. S . K R A V C H E N K O and L. S .

S H V I N D L E R M A N , Scripta Metall. 19 (1985) 1291.5. D . W E A I R E and N. R I V I E R , Contemp. Phys. 25 (1984) 59.6. J . S T A V A N S , Rep. Prog. Phys. 56 (1993) 733.7. J . S T A V A N S and J . A . G L A Z I E R , Phys. Rev. Lett. 62 (1989)

1318.8. D . A . N O E V E R , Phys. Review A. 44 (1991) 968.9. K . J . S T I N E, S . A . R . A U S E O, B . G. M O O R E, J . A .

W I S E and C. M. K N O B L E R , ibid. 41 (1990) 6884.10. D . W E A I R E, F . B O L T O N, P . M O L H O and J . A .

G L A Z I E R , J. Phys. Condensed Matter. 3 (1991) 21012.11. H . W. K R O T O, J . R . H E A T H, S . C . O’B R I A N, R . F .

C U R L and R. E . S M A L L Y , Nature 318 (1985) 162.12. I . M O C H I D A, M. E G A S H I R A, Y. K O R A I and K.

Y O K O G A W A , Carbon 35 (1997) 1707.13. E . V . O S I P O V and V. A. R E Z N I K O V , ibid. 40 (2002) 955.

3715

14. P . J . D E P A B L O, C . G O M E Z-N A V A R R O, A. G I L , J .C O L C H E R O, M. T . M A R T I N E Z, A. M. B E N I T O, W. K.M A S E R, J . G O M E Z-H E R R E R O and A. M. B A R O , Appl.Phys. Lett. 79 (2001) 2979.

15. Y . Z H A N G, A. C H A N G, J . C A O, Q. W A N G, W. K I M,Y. L I , N . M O R R I S , E . Y E N I L M E Z, J . K O N G and H.D A I , ibid. 79 (2001) 3155.

16. A . C A O, X. Z H A N G, C. X U, J . L I A N G, D. W U and B.W E I , Appl. Surf. Sci. 181 (2001) 234.

17. M. R . D I E H L, S . N . Y A L I R A K I , R . A . B E C K M A N,M. B A R A H O N A and J . R . H E A T H , Angew. Chem. Int. Ed. 41(2002) 353.

18. R . C . B A N S A L, J . -B . B O N N E T and F . S T O E C K L I , “ActiveCarbon” (Marcell Dekker, New York and Bansel, 1988) p. 1.

19. F . S T O E C K L I , Carbon 28 (1990) 1.20. J . L A I N E and S . Y U N E S , ibid. 30 (1990) 601.21. J . L A I N E, S . S I M O N I and R. C A L L E S , Chem. Eng. Commun.

99 (1991) 15.22. L . V . G O V O R, M. G O L D B A C H, I . A . B A S H M A K O V,

I . B . B U T Y L I N A and J . P A R I S I , Phys. Rev. B 62 (2000) 2201.23. L . J . D U N N E, P . F . N O L A N, J . M U N N, M. T E R R O N E S,

T . J O N E S , P . K A T H I R G A M A N A T H A N, J . F E R N A N D E Z

and A. D. H U D S O N , J. Phys. Condens. Matter 9 (1997) 10661.24. H . T . N G, M. L . F O O, A. F A N G, J . L I , G . X U, S .

J A E N I C K E, L . C H A N and S . F . Y . L I , Langmuir 18 (2002)1.

25. R . I M A M U R A, K. M. J . O Z A K I and A. O Y A , Carbon 37(1999) 997.

26. L . P A N, T . H A Y A S H I D A and Y. J . N A K A Y A M A , Mater.Res. 17 (2002) 145.

27. J . N . R O U Z A U D and A. O B E R L I N , Carbon 27 (1989) 517.28. J . M A T O S and J . L A I N E , J. Mater. Sci. Lett. 17 (1998) 649.29. J . S C H W A N, S . U L R I C H, T . T H E E L, H. R O T H, H.

E H R H A R D T, P . B E C K E R and S . R . P . S I L V A , J. Appl.Phys. 82 (1997) 6024.

30. M. F . M I R I and N. R I V I E R , Europhys. Lett. 54 (2001) 112.31. M. A. P E S H K I N, K. J . S T R A N D B U R G and N. R I V I E R ,

Phys. Rev. Lett. 67 (1991) 1803.32. B . D U B E R T R E T, T . A S T E and M. A. P E S H K I N , J. Phys.

A 31 (1998) 879.33. B . D U B E R T R E T, T . A S T E, H. M. O H L E N B U S C H and

N. R I V I E R , Phys. Rev. E 58 (1998) 6368.34. J . A . G L A Z I E R, M. P . A N D E R S O N and G. S . G R E S T ,

Phil. Mag. B 62 (1990) 615.35. N . R I V I E R , ibid. 52 (1985) 795.36. J . M A T O S , “Effect of Activated Carbon in Heterogeneous Photo-

catalytic Reactions,” Ph. D. thesis, Venezuelan Institute of ScientificResearch, I.V.I.C., 1999.

37. M. J A G T O Y E N and F . D E R B Y S H I R E , Carbon 31 (1993)1185.

38. J . L A I N E, A. C A L A F A T and M. L A B A D Y , ibid. 27 (1989)191.

39. J . L A I N E, M. L A B A D Y, F . S E V E R I N O and S . Y U N E S ,J. Catal. 166 (1997) 384.

40. J . M A T O S, J . L . B R I T O and J . L A I N E , Appl. Catal. A: Gen.152 (1997) 27.

41. T . H U A N G, T . T S U J I , M. R . K A M A L and A. D. R E Y ,J. Mater. Sci. 34 (1999) 4551.

42. Z . Z H A N G, B. W E I and P . M. A J A Y A N , Chem. Commun.9 (2002) 962.

43. S . A M E L I N C K X, X. B . Z H A N G, D. B E R N A E R T S, X. F .Z H A N G, V. I V A N O V and J . B . N A G Y , Science 265 (1994)635.

44. D . B E R N A E R T S, X. B . Z H A N G, X. F . Z H A N G,S . A M E L I N C K X, G. V A N T E N D E L O O, J . V A N

L A N D U Y T, V. I V A N O V and J . B . N A G Y , Philos. Mag.A 71 (1995) 605.

45. G . G . T I B B E T T S , G. P . M E I S N E R and C. H. O L K ,Carbon39 (2001) 2291.

46. J . K U H N , Int. J. Heat Mass Transfer 35 (1992) 1795.47. J . K L E T T, R . H A R D Y, E . R O M I N E, C . W A L L S and T .

B U R C H E L L , Carbon 38 (2000) 953.48. A . B U R I A N, P . D A N I E L , S . D U B E R and J . D O R E , Phil.

Mag. B 81 (2001) 525.49. A . A L B O R N O Z, M. L A B A D Y, M. L O P E Z , J . M A T O S,

F . R U E T T E and J . L A I N E , in “36th IUPAC Congress, Fron-tiers in Chemistry and New Perspectives for the 2000’s” (Geneve,Switzerland, 17–22 August, 1997).

50. J . L A I N E and A. C A L A F A T , Carbon 29 (1991) 949.51. T . W I G M A N S , ibid. 27 (1989) 13.52. S . J . Y U H N and E. E . W O L F , Fuel 63 (1984) 1604.53. F . R O D R I G U E Z-R E I N O S O, M. M O L I N A-S A B I O and

M. T . G O N Z A L E Z , Carbon 33 (1995) 15.54. A . A D M A D P O U R and D. D. D O , ibid. 35 (1997) 1723.55. H . T E N G and H.-C . L I N , AIChE Journal 44 (1998) 1170.56. A . C . L U A and J . G U O , Carbon 38 (2000) 1089.57. C . -F . C H A N G, C.-Y. C H A N G and W.-T . T S A I , J. Coll.

Interf. Sci. 232 (2000) 45.58. C . M O R E N O-C A S T I L L A, F . C A R R A S C O-M A R I N , M.

V. L O P E Z-R A M O N and M. A. A L V A R E Z-M E R I N O , Car-bon 39 (2001) 1415.

59. D . L O Z A N O-C A S T E L L O , M. A. L I L L O-R O D E N A S, D.C A Z O R L A-A M O R O S and A. L I N A R E S-S O L A N O , ibid. 39(2001) 741.

60. A . R O B A U S A N C H E Z, A. A G U I L A R E L G U E Z A B A L andL . D E L A T O R R E S A E N Z , ibid. 39 (2001) 1367.

61. L . B E R N A S C O N I and P . A . M A D D E N , J. Phys. Chem. B 106(2002) 1161.

62. N . N A R I T A, S . N A G A I and S . S U Z U K I , Phys. Rev. B 64(2001) 5408.

63. H . C H E N G, G. P E Z , G. K E R N, G. K R E S S E and J .H A F N E R , J. Phys. Chem. B 105 (2001) 736.

64. T . T O R I M O T O, S . I T O H and S . K U W A B A T A , Env. Sci.Technol. 38 (1996) 1275.

65. J . M A T O S, J . L A I N E and J . -M. H E R R M A N N , Appl. Catal.B: Env. 18 (1998) 281.

66. Idem., Carbon 37 (1999) 1870.67. J . -M. H E R R M A N N, J . M A T O S, J . D I S D I E R, C .

G U I L L A R D, J . L A I N E, S . M A L A T O and J . B L A N C O ,Catal. Today 54 (1999) 255.

68. J . M A T O S, J . L A I N E and J . -M. H E R R M A N N , J. Catal. 200(2001) 10.

Received 13 March 2003and accepted 10 February 2004

3716