Competition among Mixed Adsorbates Affecting the Adsorption of Gaseous Methyl Ethyl Ketone by Hydrophobic Molecular Sieve

Internal circulation cabinets, equipped with adsorbents for emitted volatile organic compounds (VOCs), are extensively used to store organic solvents in Taiwanese university and hospital laboratories. This work evaluates how ketone and isopropanol (IPA) respectively influence the methyl ethyl ketone (MEK) adsorption capacity of a hydrophobic molecular sieve (MS) with a high Si/Al ratio of 1000, HiSiv 1000, manufactured by UOP LLC Corporation, USA. Experimental results for MEK at concentrations from 113 to 2815 ppm reveal that the MEK adsorption capacity is linearly expressed as the natural logarithm of adsorption time (ln (t)) at the constant temperature 27C and relative humidity 72 ± 3%. The variations in the simulated MEK equilibrium adsorption capacity (Qe, g MEK/g MS) and concentration (C, ppm) are closely fitted using a Langmuir isotherm. Lower VOC concentration ratios less than 0.17 for mixing acetone and IPA respectively with MEK rarely affect the MEK adsorption by HiSiv 1000. However, the effect of competitive adsorption, which evidently reduces the adsorption capacity of MEK, is related to the higher mixture concentration ratio (> 0.80) of IPA to MEK. The experimental results show that the hydrophobicity characteristics of MEK, acetone and IPA were responsible for variation in adsorption capacity of hydrophobic molecular sieve. Working parameters of adsorbent materials in the internal circulation cabinet have been also provided for preventing breakthrough of VOC adsorption by the molecular sieve.


INTRODUCTION
Many organic solvents have been used in laboratories of Taiwanese universities and hospitals.These volatile organics with low boiling points are typically stored in sealed storage cabinets to prevent the unhealthy effects of gaseous volatile organic compounds (VOCs) on workers.Mobile cabinets with internal circulating adsorption system (Fig. 1) are widely employed to store solvents in indoor area.The adsorbent, currently molecular sieves (MS) in Taiwan, must be periodically renewed to prevent the over-adsorption of VOCs by the adsorbent under breakthrough.However, the detailed adsorption characteristics of molecular sieves, used by local manufacturers of storage cabinets, have not been thoroughly examined, nor has the adsorption performance been well evaluated.Zhao et al. (1996) defined that the pore size of the microporous molecular sieve as under 20Å.Giaya et al. (2000) and Erdem-enatalar et al. (2004) reported that increasing the SiO 2 to Al 2 O 3 (Si/Al) ratio by dealumination would change the material from hydrophilic to hydrophobic.Experimentally, the adsorption capacities of chloroanilines on montmorillonite and kaolinite sieves depend on the hydrophilicity and lipophilicity of adsorbates; that is, the extent of adsorption on hydrophobic adsorbents increases with lipophilicity of adsorbates (Angioi, et al., 2005;Polati, et al., 2006).Additionally, the effects of moisture competition and high temperature desorption, which reduce the VOC adsorption capacities of adsorbents, have been extensively investigated (Keener and Zhou, 1990;Chou and Chiou, 1997;Lercher and Seshan, 1997;Yun, et al., 1998;Gawry et al., 2001;Bilgiç and A kin, 2003;Manjare and Ghoshal, 2006).
Generally, various solvents (such as ketones and alcohols) are gathered together in an internal circulation cabinet.This investigation evaluated the competitive adsorption of mixed gaseous methyl ethyl ketone (MEK), acetone and isopropanol (IPA) by a hydrophobic molecular sieve.The experimental data for MEK adsorption capacity (Q, g MEK/g MS), which is altered by mixed acetone and IPA, have been provided to local manufacturers to improve the application of adsorbent materials used in an internal circulation cabinet.

Fundamental adsorption theory used in this study
The VOC adsorption capacity (Q, g VOC/g MS) by molecular sieve increases with the concentration of VOC (C, ppm) (Erdem-enatalar et al., 2004;Yun, et al., 1998).Additionally, Scahill et al. (2004) used the Nelson-Harder correlation to relate the breakthrough time variations (t 1 and t 2 ) of the different inlet contaminant concentration (C 1 and C 2 ): where, n is a constant.According Scahill et al.'s findings, the VOC adsorption capacity by molecular sieve is proportional to the natural logarithmic value of adsorption time.

Q ln (t)
(2) Since zeolite molecular sieves rely on strong physical forces rather than chemisorption, the shape of the isotherm for single component adsorbate is characterized by a Langmuir-type (Bhatia, 1989).Langmuir model are most frequently used to describe adsorption equilibrium on homogeneous solids, like HiSiv 1000 used herein (Himeno, et al., 2007).The Langmuir (Eq.( 3)) isotherm models are used to fit the experimental equilibrium capacity data (Q e ) with various inlet concentrations (C) for adsorption of a single VOC at a constant temperature (Chou and Chiou, 1997;Yun, et al., 1998;Manjare and Ghoshal, 2006;Choudary, et al., 2002;Yoo, et al., 2005;O'Connor, 2006)).
where a is mass of adsorbate required to saturate completely a unit mass of adsorbent; and b is the Langmuir adsorption constant (Reynolds and Richards, 1995;Lin et al., 2006).Values of a and b are constants to be determined by performing a series of isothermal continuous-flow column tests (Benefield, et al., 1982).
Rudzi ski et al. (1995) used the exponential formula, like the Freundlich equation, to simulate the competitive adsorption between methane and ethane in binary gas mixtures.The modified relationship used to find the competive adsorption capacity of component A (Q A ), based on the work of Rudzi ski et al., is: where Q A0 represents the adsorption capacity for the single component A. According to the competitive adsorption between concentrations of components A and B, if n is negative, the adsorption capacities of omponents A and B reduce each other, because Q A and C B are inversely proportional.

D
rmined by setting q VOC , q w and q d for a series of adsorption tests, using the following alculations.

c esign of adsorption system
The adsorption apparatus, presented in Fig. 2 and taken from Chou and Chiou's study (1997), was adopted to assess exactly the effect of mixed VOCs on the adsorption capacity of MEK by molecular sieves (Q, g MEK/g MS) during the period of adsorption (t, min).This system was based on the method of air-stripping using a low flow rate regulating air compressor (part 3 of Fig. 2), and transferred VOC (q VOC ) with a gaseous concentration of C g * at a constant rate (5 mL/min).The tested VOC was MEK.
MEK separately mixed with various volume ratios of ketone and IPA was also tested by mixing liquid VOCs in the tube (part 6 of Fig. 2).The other air blower (part 2 of Fig. 2) passed a saturated water stream (q w = 0.3-0.8L/min) with a relative humidity (RH) about 100%, mixed with dry air (q d = 0.7-1.2L/min).The gaseous flow rates satisfied q VOC << (q w + q d ).The ranges of VOC concentration (C) and RH were dete variation of the adsorption capacity (Q) with various VOC co The totally gaseous stream (q VOC + q w + q d ) was maintained at 1.5 L/min in all test runs.Influent MEK concentrations (C i ) typically ranged from low (113-416 ppm), through middle (759-1429 ppm) to high (2815-4564 ppm), to evaluate the ncentration (C) and adsorption time (t).
A gastight glass tube (part 11 of Fig. 2) was filled with around 10 g of hydrophobic molecular sieve (HiSiv 1000, UOP LLC Company, Des Plains, IL, USA).The space velocity (v S ) of VOC-rich air stream through the adsorbent zone was approximately 1.3 L/s, and the empty bed residual time (EBRT) was around 0.76 s.The adsorbent tube was placed in an oven (part 13 of Fig. 2) at a constant temperature of 27 o C. The VOC adsorption periods ranged from 20 to 720 min.The RH value of the VOC-carrying stream was measured using a humidity analyzer (testo 400, TESTO, Germany, part 12 of Fig. 2).The RH of the gas streams was maintained at 72 3% in all test runs.Operation of the system at a constant RH and temperature prevented moisture and heat from affecting the VOC dsorption capacity. a Air blower 3.

Calculating VOC adsorption capacity
When the adsorption test was completed, the glass tube and molecular sieve were placed in a high-temperature oven to desorb VOCs in a dry stream of 0.5 L/min at 250 o C. The desorbed VOC-rich gas was collected in 10 L Teflon gas sampling bags (SKC, USA).The desorption period took 2.5 hr, in which the 99.99% of adsorbed VOCs were removed from the molecular sieve within preliminary desorption tests.For checking the performance of molecular sieve reuse and equilibrium adsorption achievement (Reynolds and Richards, 1995), the remained VOCs on molecular sieve after desorption procedures were examined less than 0.01% adsorption mass.The molecular sieve could be reused for the next experimental runs but not exceeding 5 runs.All desorbed gases were sampled using a gastight syringe (Hamiltom, USA) and then analyzed using a GC-FID (6890N, Agilent, USA).The internal volume of the Teflon bags was determined by the method of buoyancy using draining water.Multiplying the analyzed concentration by the volume of the bag yielded the mass of VOC adsorbed by the molecular sieve.Summing all of the VOC masses in a desorption run and then dividing that sum by the mass of molecular sieve in grams yielded Q.The calibration curves for VOC analysis by GC-FID ere interiorly replicated examination using samples of known concentrations once per week.
1. Typical Properties iv TM 100 nit w

perimental materials
The tested hydrophobic molecular sieve (HiSiv 1000) was produced by UOP LLC.HiSiv adsorbents are synthetically produced silicon-aluminum oxides that have an Si/Al ratio of from 10 to 1000.HiSiv 1000 differs from types A and X zeolites for its significantly higher proportion of Si/Al ratio in the molecular structure.Table 1 presents the typical characteristics of HiSiv 1000.UOP LLC states that HiSiv 1000 is preferred for adsorbing large molecules (0.6-0.8 nm), such as alcohols, aldehydes, ketones and ethers.Larger molecules such as multiple-ringed aromatic compounds are not adsorbed because they are too large for the pore openings.Nota der a humid condition of RH higher than 70%.Target VOCs, MEK, acetone and IPA, were of analytic reagent grad

Variations of adsorption capacity with VOC concentrations
MEK adsorption tests were performed using the molecular sieve at an inlet concentration ranging from 113 to 4564 ppm, over periods of 60-600 min at 27 o C. Fig. 3 shows the variations of MEK adsorption capacity (Q) with concentration (C) at various adsorption times, and Table 2 presents formulae (Eqs.( 7)-( 11)) in Fig. 3. Table 2 shows

Determining the VOC equilibrium isotherms of adsorption
The VOC adsorption curves (Fig. 3) have not yet reached at equilibrium for the adsorption period of 600 min.Equations ( 7)-( 11) were adopted with various mean values of inlet concentration (113, 416, 759, 1429 and 2815 ppm) using extended adsorption period of 3200 min to determine the equilibrium adsorption capacity Q e (Table 3).The equilibrium adsorption times drop from 3200 min at C = 113 ppm to 2000 min for C = 2815 ppm.Note: Calculation based on an increment rate of adsorption capacity less than 0.01 per 100 min.
Fig. 4 presents the simulated equilibrium adsorption isotherms obtained using Langmuir formulae with the data of Q e and C in Table 3. Notably, the reciprocal values of Q e and C, 1/Q e and 1/C in Fig. 4, must be adopted to find constants a and b in Eq. (3) (Benefield, et al., 1982).Note: Based on the results in Table 3.

Breakthrough experiment of MEK adsorption by molecular sieve
Gaseous VOC breakthrough tests for molecular sieve adsorption have frequently been utilized to evaluate the equilibrium adsorption capacity (Q e , g VOC/g MS) (Yoo, et al., 2005;Lin, et al., 2005;Manjare and Ghoshal, 2005).Breakthrough tests of MEK adsorption (C = 1448 (1471 ± 319) and 440 (483 ± 183) ppm) were performed herein to verify the feasibility of the simulated equilibrium capacity data in Table 3.The initial (t = 0-100 min) and complete adsorption curves are shown as Fig. 5.According to the simulated equilibrium adsorption times of 2200 min for 1448 ppm and 2600 min for 440 ppm in Table 3, the experimental values of C 0 /C i at 2200 and 2600 min are satisfactorily 0.990 and 0.994 in Fig. 5, which are close to equilibrium adsorption condition of C 0 /C i = 0.990.If further calculating equilibrium adsorption capacity, using the formula [(C 0 -C i ) x (q VOC + q w + q d ) x t] ( t = VOC detection period deviation), the simulation errors for calculation in Table 3 are 3

Effect of mixture of VOCs on adsorption capacity
Based on Rudzi ski et al.'s competitive adsorption theory (1995) (Eq. ( 4)), Fig. 6 presents the effects of mixed VOCs on the adsorption capacity of MEK for a serious of 300 min adsorption experiments of various acetone/MEK and IPA/MEK concentration ratios.The inverse proportionality between the adsorption capacity ratios of binary-component to single-component (Q/Q 0 ) and the concentration ratio is fully demonstrated for n < 0. ), then the MEK adsorption capacities are close to those obtained using Eq. ( 9).The MEK adsorption capacities gradually drop below the simulated value as the acetone/MEK and IPA/MEK concentration ratios increase.Notably, when the IPA/MEK concentration ratio approached 0.80 ( 796 .0 842

670
), competitive adsorption of mixed IPA occurs at an actual adsorption capacity of as low as 79.4% of estimated value.However, the strong adsorption competition between mixed adsorbates does not dominate in a mixture of acetone and MEK, even when the concentrations of MEK and acetone were close to each other (730 ppm vs. 795 ppm).a. Estimation using Eq. ( 9) and adsorption time 300 min.
Besides the electrostatic force between the adsorbent and adsorptive sites of molecular sieves, which is primarily due to the cations within the crystal lattice structure of molecular sieves, the affinity between adsorbates and adsorbent for mixed VOCs is essential.The n-octanol-water partitioning constant (K ow ) is a universal parameter used in determining the affinity of organic compounds to the lipophilic/hydrophilic surface.A VOC with a higher K ow is more easily adsorbed by the hydrophobic adsorbents on their adsorption sites.Angioi et al. (2005) and Polati et al. (2006) proposed the lipophilicity/hydrophobicity theory, based on the examination of the adsorption capacities of chloroanilines on montmorillonite and kaolinite sieves.According to the K ow values in the Syracuse Research Corporation PhysProp's database, K ow of MEK, acetone and IPA are 1.95, 0.575 and 1.12, respectively (2006).Among target VOCs, MEK is the most hydrophobic, and has the strongest affinity for hydrophobic HiSiv 1000.Table 5 reveals that when the acetone/MEK and IPA/MEK concentration ratios are under 0.17, both acetone and IPA result in limited competitive adsorption of MEK; however, as the acetone/MEK and IPA/MEK concentration ratios increase, IPA reduces the adsorption capacity of MEK more than does acetone.Given this concentration competition, hydrophobicity or K ow values of hydrocarbons should be further considered if VOCs of different K ow are presented in a sealed cabinet with hydrophobic molecular sieve adsorbents.

Working parameters for the internal circulation cabinet
To prevent the over-adsorption of VOC by HiSiv 1000, the operation parameters are recommended as follows.The accumulated emission concentration was approximately 100 ppm, based on the actual detection in a sealed cabinet, where storing about 20 bottles of 1-L and 3 bottles of 2-L bottles of MEK, acetone and alcohols.Using Eq. ( 11) in Table 2 with a 90% equilibrium adsorption (0.0109 x 0.9 = 0.00981 g MEK/g MS) (see Table 3), the calculated adsorption time is 2312 min.If five times of taking out chemicals from this cabinet, a 2-min preliminary compulsory operation of circulation pump for cleaning the inner air is recommended before workers open the sealed door.The total adsorption period for renewing the adsorbent is estimated as 46.2 weeks [2312 min/(2 min 5 1/day 5 days/week)], which means the molecular sieve effectively adsorbs gaseous MEK, acetone and alcohols within this period.Notably, the design for the volume of adsorption bed and circulation pump must be based on the criteria, v S = 1.3 1/s, and EBRT = 0.76 s.If an unexpected VOC emission of high concentration (i.e.incompletely sealed or rupture of solvent bottles), the VOC adsorption performance of adsorbent must be checked.Additionally, the concentration competitive effect on adsorption capacities of molecular sieve for hydrocarbons with different K ow values should be considered.

CONCLUSIONS
A series of adsorption tests (adsorption temperature = 27 o C, RH = 72 3%) were performed on the hydrophobic molecular sieve, which was used as the adsorbent in an internal circulation cabinet for storing organic solvents.Experimental results show that the MEK adsorption capacity, Q (g MEK/g MS), can be simulated as the natural logarithm of adsorption time for MEK concentrations from 113 to 2815 ppm (Eqs.( 7)-( 11)).This work also established the Langmuir model, which yields the MEK adsorption isotherm between the equilibrium adsorption capacity, Q e , and the VOC concentration, C (ppm), at Q e = [0.000117C/(1+0.00216C)].
A lower concentration ratio (< 0.17) of mixed acetone and IPA with MEK rarely affects the MEK adsorption of hydrophobic HiSiv 1000.However, competitive adsorption, which reduces the adsorption capacity of MEK, results from the higher concentration ratio (> 0.80) of IPA to MEK.However, the higher concentration ratio of acetone to MEK does not cause dominant competitive adsorption.The hydrophobicity of hydrocarbons, specified by k ow , should be further considered when varoius hydrophilic/hydrophobic VOCs are gathered in an internal circulating adsorption system cabinet with molecular sieve adsorbents.If the k ow values of VOCs approach those of the other components, then competitive adsorption is occurring for the mixed adsorbates in the cabinet.It is recommended that specific internal circulating cabinet should be used for the hydrocarbon solvent of a single group to prevent the competitive adsorption.

Fig. 3 .
Fig. 3. Variations of MEK adsorption capacity with inlet concentration at various adsorption times.

Fig. 6 .
Fig. 6.Comparison of concentration competition effect of acetone and IPA for MEK adsorption by molecular sieve.

Table 2 .
Correlations of MEK Adsorption Capacity With Time and Concentrations.
Note: Values of R 2 denote the correction coefficients.
Table4shows that the Langmuir formula, Q e = [0.000117C/(1+0.00216C)], as well as a = 0.0542 g MEK/g MS, and b = 0.00216 1/ppm, more properly specifies the equilibrium MEK adsorption capacities by HiSiv 1000 for lower MEK concentrations under 27 o C and RH = 72 3%.The highest simulation deviation (10.7%) for C = 2815 (2920 ± 947) ppm in Table4may be resulted from the more extreme deviations of inlet concentrations than lower MEK concentration ranges.
Table 5 compares the critical concentration ratios, at which mixed acetone and IPA evidently restrained the MEK adsorption capacities.When the acetone/MEK and IPA/MEK concentration ratios

Table 5 .
Comparison of Estimated and Experimental Adsorption Capacity by Molecular Sieve for Mixed VOCs.