Published: August 17, 2011

r 2011 American Chemical Society 7194 dx.doi.org/10.1021/ac201786n |Anal. Chem. 2011, 83, 7194–7197

ARTICLE

pubs.acs.org/ac

Gravimetric Analysis of CO2 Adsorption on Activated Carbonat Various Pressures and Temperatures UsingPiezoelectric MicrocantileversYusung Jin,† Dongkyu Lee,† Sangkyu Lee,‡ Wonkyu Moon,‡ and Sangmin Jeon*,†

†Department of Chemical Engineering, Pohang University of Science and Technology, Pohang 790-794 South Korea‡Department of Mechanical Engineering, Pohang University of Science and Technology, Pohang 790-794 South Korea

bS Supporting Information

Carbon dioxide is a major greenhouse gas that inducesundesired global climate changes.1 International efforts

toward reducing CO2 generation have been undertaken, but itis challenging to achieve meaningful CO2 reduction unless ourlifestyle, based on fossil fuel-derived energy, is adjusted to relyinstead on renewable energy sources, such as solar and windpower. Because changes are realized over time, it is essential todevelop CO2 capture technologies to manage atmospheric CO2

concentrations in the short term.Pressure swing adsorption (PSA) is a promising process that

could be applied immediately for the removal of CO2 emittedfrom industrial plants because it is energy efficient and suitablefor a large-scale CO2 capture system.2,3 The PSA process is basedon the preferential adsorption of a target gas on a porous sorbentat high pressures and the desorption of the gas at low pressures.The adsorption capacity of a sorbent toward CO2 is generallycalculated from measurements of the pressure changes duringadsorption of CO2 using an equation of state. This approach isconvenient, although indirect.

In contrast, mass sensors, such as microcantilevers and quartzcrystal microbalances (QCM), can directlymeasure the adsorbedmass of CO2 with better than 1 ng sensitivity. Wu et al. usedQCM to investigate the adsorption of CO2 on quartz crystals.4

They found that the resonance frequency of quartz crystals wasaffected not only by the mass of the adsorbed gas but also by theviscosity and density of the surrounding medium and the surfaceroughness of the quartz crystals. Quantitative analysis using

QCM results, however, is not straightforward unless the surfaceroughness of quartz crystals is determined after sorbent coating.4

In contrast, we reported in a previous study that the CO2

adsorption behavior (kinetics and adsorption capacity) was notaffected by the surface roughness of the cantilever if the sorbentwas coated on the free end of the cantilever.5 Furthermore, themass sensitivity of the microcantilevers was found to be farsuperior to that of the QCM sensors. However, the previousstudy used a conventional optical beam technique to measurechanges in the resonance frequency of the cantilever, which wasnot easily integrated into a high-pressure system. This problemcan be solved by using piezoelectric or piezoresistive microcan-tilevers because they detect changes in the resonance frequencyelectrically. In spite of this advantage, piezoelectric and piezo-resistive microcantilevers have mainly been used so far as simplesensors for the detection of gases and biomolecules underambient conditions.6�9

In this study, we fabricated piezoelectric microcantilevers andinvestigated the in situ adsorption�desorption behavior of CO2

onto the activated carbon at various temperatures and pressures.The kinetics of CO2 adsorption�desorption on the sorbent weremeasured based on the changes in the resonance frequency of thecantilever during adsorption or desorption at various pressures.

Received: July 11, 2011Accepted: August 17, 2011

ABSTRACT:We investigated the adsorption and desorption ofCO2 on activated carbon using piezoelectric microcantilevers.After coating the free end of a cantilever with activated carbon,variations in the resonance frequency of the cantilever weremeasured as a function of CO2 pressure, which is related tomasschanges due to the adsorption or desorption of CO2. Thepressure-dependent viscous damping effects were compensatedin the calculation of the CO2 adsorption capacity of theactivated carbon by comparing the frequency differencesbetween the coated and uncoated cantilevers. The mass sensi-tivity of the piezoelectric cantilever was found to be better than 1 pg. The fractional coverage of CO2 agreed with a Langmuiradsorption isotherm, indicating that a submonolayer of adsorbed CO2 occurred on the surface of the activated carbon under theexperimental conditions. The heat of adsorption was determined using the Clausius�Clapeyron relation and the fractional coverageof CO2 at various temperatures and pressures.

7195 dx.doi.org/10.1021/ac201786n |Anal. Chem. 2011, 83, 7194–7197

Analytical Chemistry ARTICLE

The heat of adsorption of CO2 onto the sorbent was additionallydetermined based on the CO2 adsorption capacities at differenttemperatures. This study presents, to the best of our knowledge,the first use of piezoelectric microcantilevers for the investigationof the adsorption�desorption of CO2 onto solid sorbents at highpressures.

’EXPERIMENTAL PROCEDURES

Experimental Setup. Piezoelectric microcantilevers werefabricated as described elsewhere.9 The dimensions of thetrapezoidal cantilever were 30 μm in width at the clamped endof the cantilever, 100 μm in length, and 5 μm in thickness. ThePZT (lead zirconate titanate) film was deposited on the clampingregion of the cantilever, and its thickness was 2.5 μm. The typicalresonance frequency and Q factor of the cantilevers weremeasured to be 1.17 MHz and 400, respectively, at reducedpressure (15 Torr) using a function generator (NI-PXI 5422,National Instrument Co., Texas) and a home-built LABVIEWprogram. Whereas the theoretical mass sensitivity of the canti-lever was calculated to be 2.1 fg/Hz,9 in practice, only 0.1 pgchanges in mass could be measured due to the noise of thecantilever (∼100 Hz).Activated carbon was selected as a CO2 sorbent due to its high

surface area, thermal and chemical stability, and hydrophobicsurface properties.10 A working cantilever with an activatedcarbon coating and a reference cantilever without the coatingwere mounted inside a home-built pressure cell (cell volume11 mL). The reference cantilever was used to avoid commonmode noises, such as viscous damping effects. A multiplexer(NI-PXI 2593, National Instrument Co., Texas) was used tomeasure the resonance frequencies of each cantilever every 5 s.The differential frequency changes between the reference andworking cantilevers were used to calculate the mass changes dueto the adsorption or desorption of CO2. The temperature of thecell was controlled using a resistance heater with a programmabletemperature controller (Hanyoung, Incheon, Korea).Measurements of CO2 Adsorption. Activated carbon

(untreated activated carbon, C3345) was purchased fromAldrich(St. Louis, MO) and used as received. The specific surface areaof the activated carbon was determined to be 1060 m2/g bythe Brunauer�Emmett�Teller (BET) method. The cantileverwas coated with activated carbon by immersing a piezoelectricmicrocantilever in activated carbon-dispersed ethanol solu-tions for 1 min. The immersion depth was controlled to coatthe activated carbon only on the free end of the cantilever toavoid stress-induced modulus changes due to the coating.11 The

cantilever was then heated at 100 �C for 30 min to evaporateaway the ethanol and fix the sorbent onto the cantilever surface.

’RESULTS AND DISCUSSION

Figure 1 shows a scanning electron microscopy (SEM) imageof an activated carbon-coated piezoelectric microcantilever,showing that the activated carbon coating is present only onthe free end of the cantilever. Assuming that the cantilever is asimple harmonic oscillator, the frequency changes (Δf) of thetrapezoidal cantilever due to the coating of activated carbon arerelated to the mass change (Δm) according to9

Δf ¼ f0 � f

¼ 12π

ffiffiffiffiffiffiffiffiffiffiffiffik

0:07m

r1� 1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1 þ Δm=0:07mp

" #ð1Þ

where k is the spring constant and f0 (1.174 28 MHz at 15 Torr)and f (1.150 28MHz at 15 Torr) are the resonance frequencies ofthe cantilever before and after coating, respectively. The mass ofthe activated carbon coated on the cantilever was calculated fromeq 1 to be 52 pg.

To investigate the effects of viscous damping on the resonancebehavior of the cantilever, the types and pressures of gas in thepressure cell were varied. Parts a, b, and c of Figure 2 showvariations in the resonance peak of a bare cantilever in thepresence of CO2, N2, and He, respectively. The resonance peaksshifted to lower frequencies at increased pressures due toincreases in the mass and viscosity. The largest shift was observedfor CO2 adsorption, whereas an almost negligible shift wasobserved for He adsorption, indicating that the normalizedfrequency shift was proportional to the molecular mass of thegases as shown in Figure 2d.

With dependence on the damping mechanism of theresonator, the pressure range is divided into three characteristicregions:12 an intrinsic region, a molecular region, and a viscousregion. Because the pressure range in this study was situated

Figure 1. Electron microscopy image of an activated carbon-coatedpiezoelectric microcantilever.

Figure 2. Variations in the resonance peaks of the bare piezoelectriccantilevers as a function of pressure under (a) carbon dioxide, (b)nitrogen, (c) helium (black, 1.5; red, 2; green, 3; blue, 4; cyan, 5; pink, 6;yellow, 7; dark yellow, 8; navy, 9; purple, 10 bar) at 298 K. (d) Variationsin the normalized frequency as a function of molecular mass of the gases.

7196 dx.doi.org/10.1021/ac201786n |Anal. Chem. 2011, 83, 7194–7197

Analytical Chemistry ARTICLE

within the viscous region, the gases acted as a viscous fluid andhydrodynamic inertial forces affected the resonance frequency ofthe cantilever. Assuming that a microcantilever is a string ofspheres, its relative frequency shift in the viscous region due tothe inertial force of the gas on the cantilever is given by13,14

Δf=f ¼ � πr3

3mRTMP þ 9

2

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiμRT=πf

pr

ffiffiffiffiffiffiffiMP

p !ð2Þ

where R,m, T, P, μ,M, and r are the gas constant, the mass of thecantilever, the absolute temperature, pressure, dynamic viscosity,molar mass of the gas, and the radius of one of the oscillatingspheres, respectively. Equation 2 indicates that the frequencychange increased with the pressure due to viscous damping, asshown in Figure 2a�c.

Compared with the resonance frequency, the quality factor of thecantilever (Q factor) was more directly related to the pressure (orviscous damping). TheQ factor was defined as the ratio of the energystored to the energy dissipated and could be calculated according to

Q factor ¼ ffwhm

ð3Þ

where fwhm is the full width at half-maximum. A lower Q factorindicates a higher rate of energy dissipation (i.e., larger viscousdamping) relative to the oscillation frequency. Alternatively, the Qfactor can be obtained as a function of pressure in the viscous regimeusing13

Q ¼ kn2bd2ðFsE=12Þ1=2

6πμrlð1 þ AP1=2Þ ð4Þ

where kn, Fs, E, b, d, and l are a constant for the nth order mode ofresonance, the density, Young’s modulus, width, thickness, and

length of the cantilever, respectively. A is a parameter related tomolecularweight, viscosity, and temperature [A=(πfr2M/μRT)1/2].BecauseAP1/2 > 1 under our experimental conditions (∼100), theQfactor was proportional to 1/

√P. Figure 3 shows that the inverse of

theQ factor increased linearlywith√P. Because the viscous damping

dependedon the density and viscosity of the surrounding gases, largerchanges in theQ factor were observed in the order CO2, N2, andHe.

The viscous damping effects on mass changes were compen-sated using measurements conducted on a reference cantilever.Figure 4 shows the frequency changes of the reference cantileverand activated carbon-coated cantilever as a function of CO2

pressure at 298 K. The measurements were conducted five times,and almost identical results were obtained, indicating that theexperimental results were highly reproducible. The resonancefrequencies of both cantilevers decreased with pressure due to anincrease in the adsorbed mass and viscosity. Larger changes infrequency were observed for the coated cantilever due to theadsorption of CO2 onto the activated carbon. The mass of CO2

adsorbed onto the activated carbon could be calculated from thedifferential resonance frequencies of the reference and activatedcarbon-coated cantilevers using eq 1. Because the surface area ofthe coated activated carbon (5.5 × 10�8 m2/52 pg) was 2 ordersofmagnitude greater than that of the bare cantilever (3 ×10�10m2),the mass of CO2 adsorbed onto the cantilever surface couldbe neglected in calculating the CO2 adsorption capacity of theactivated carbon.

Variations in the pressure-dependent fractional coverage ofCO2 (θ) were measured at various temperatures and are plottedin Figure 5. The fractional coverage could be calculated bydividing the number of adsorbed CO2 molecules (or the massof adsorbedCO2molecules) by themaximum value that could beadsorbed on the sorbent. Assuming that each CO2 moleculeoccupied 0.1 nm2 of the sorbent surface, the number of CO2

molecules adsorbed onto 52 pg of sorbent corresponded to 4.2 ×1011. In contrast, the mass change due to the adsorption of CO2,even at 10 bar, was only 7 pg, which corresponded to 9.5 × 1010

CO2 molecules. This result indicated that the surface coverage ofCO2molecules on the sorbent was less than a monolayer of CO2,even at the highest pressures tested in our experiment. The fractionalcoverage was fit according to the Langmuir adsorption isotherm,

θ ¼ KP1 þ KP

ð5Þ

where P andK are the pressure and Langmuir constant, respectively.The good agreement confirmed the submonolayer adsorption ofCO2 onto the activated carbon.

Figure 3. Variations in theQ factor of the cantilever with pressure (bluecircle, CO2; red square, N2; black triangle, He).

Figure 4. Variations in the frequency and mass of a bare cantilever(black) and an activated carbon-coated cantilever (red) as a function ofpressure at 298 K.

Figure 5. CO2 adsorption isotherms on activated carbon at differenttemperatures: (a) 298 K, (b) 307 K, (c) 323 K, and (d) 358 K. Theexperiments were conducted three times, and the data were fit to theLangmuir adsorption isotherm.

7197 dx.doi.org/10.1021/ac201786n |Anal. Chem. 2011, 83, 7194–7197

Analytical Chemistry ARTICLE

The heat of adsorption of CO2 onto the activated carbon (Q)was determined using the Clausius�Clapeyron equation,

dðln PCO2Þdð1=TÞ ¼ Q

Rð6Þ

Figure 6 shows the changes in pressure as a function oftemperature at various fractional coverages. The equilibriumpressure and temperature for the CO2 adsorption were obtainedfrom Figure 5. The data were fit to straight lines, and the slopes(�1.66 K) were found to be independent of the fractionalcoverage. The negative slope indicated that the CO2 adsorptionprocess was exothermic. The heat of adsorption was calculatedfrom eq 6 to be 14( 0.4 kJ/mol, which was similar to the value inthe carbon-based sorbent reference.15

’CONCLUSIONS

In summary, we used piezoelectric microcantilevers to investigatethe in situ adsorption�desorption of CO2 onto activated carbon atvarious temperatures and pressures with sub picogram sensitivity. Inaddition, the heat of adsorption of CO2 onto the activated carbon wasdetermined, which was found to be similar to the reported values.Althoughnot implemented in this study, high-throughput analysiswithnanogram samples may be possibly achieved using arrayed structuresof piezoelectric cantilevers. This unique feature of piezoelectric canti-levers may be exploited to provide promising analytical tools forinvestigating the gas adsorptionproperties of a variety of nanomaterials.

’ASSOCIATED CONTENT

bS Supporting Information. Variations in the resonancefrequencies of the bare and activated carbon-coated cantileversover time as a function of pressure at 298 K and variations in theresonance frequencies of the activated carbon-coated piezoelec-tric cantilever over time at 298 K. This material is available free ofcharge via the Internet at http://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: jeons@postech.ac.kr. Fax: (82) 54 279 5578. Phone:(82) 54 279 2392.

’ACKNOWLEDGMENT

This research was supported by the Future-based TechnologyDevelopment Program (Nano Fields) through the National

Research Foundation of Korea (NRF) funded by the Ministryof Education, Science and Technology (Grant 20110011246).

’REFERENCES

(1) Metz, B.; Davidson, O.; de Coninck, H.; Loos, M.; Meyer, L.IPCC Special Report: Carbon Dioxide Capture and Storage; CambridgeUniversity Press: New York, 2005.

(2) Na, B. K.; Lee, H.; Koo, K. K; Song, H. K. Ind. Eng. Chem. Res.2002, 41, 5498–5503.

(3) Ebner, A. D.; Gray, M. L.; Chisholm, N. G.; Black, Q. T.;Mumford, D. D.; Nicholson, M. A.; Ritter, J. A. Ind. Eng. Chem. Res.2011, 50, 5634–5641.

(4) Wu, Y. T.; Akoto-Ampaw, P. J.; Elbaccouch, M.; Hurrey, M. L.;Wallen, S. L.; Grant, C. S. Langmuir 2004, 20, 3665–3673.

(5) Lee, D.; Jin, Y.; Jung, N.; Lee, J.; Lee, J.; Jeong, Y. S.; Jeon, S.Environ. Sci. Technol. 2011, 45, 5704–5709.

(6) Thundat, T.; Chen, G. Y.; Warmack, R. J.; Allison, D. P.;Wachter, E. A. Anal. Chem. 1995, 67, 519–521.

(7) Allendorf, M. D.; Houk, R. J.; Andruszkiewicz, L.; Talin, A. A.;Pikarsky, J.; Choudhury, A.; Gall, K. A.; Hesketh, P. J. J. Am. Chem. Soc.2008, 130, 14404–14405.

(8) Maraldo, D.; Rijal, K.; Campbell, G.; Mutharasan, R. Anal. Chem.2007, 79, 2762–2770.

(9) Lee, Y.; Lim, G.; Moon, W. Sens. Actuators, A 2006, 130�131,105–110.

(10) Yang, R. T. Adsorbents: Fundamentals and Applications; JohnWiley & Sons: Hoboken, NJ, 2003.

(11) Lee, D.; Kim, S.; Jung, N.; Thundat, T.; Jeon, S. J. Appl. Phys.2009, 106, 024310.

(12) Newell, W. E. Science 1968, 161, 1320–1326.(13) Blom, F. R.; Bouwstra, S.; Elwenspoek, M.; Fluitman, J. H. J.

J. Vac. Sci. Technol., B 1992, 10, 19–26.(14) Xu, Y.; Lin, J. -T.; Alphenaar, B. W.; Keynton, R. S. Appl. Phys.

Lett. 2006, 88, 143513.(15) Guo, B.; Chang, L.; Xie, K. J. Nat. Gas Chem. 2006, 15, 223–229.

Figure 6. A plot of the logarithm of pressure as a function of the inverseof the absolute temperature at different fractional coverages (9, 0.02;0,0.04; 2, 0.06; 4, 0.08; and b, 0.1). The data are the same as thatpresented in Figure 5.