Printer Friendly

VOC and HAP removal from dryer exhaust gas by absorption into ionic liquids.

Abstract

Currently, thermal oxidizers are commonly used to reduce emissions from dryers and presses at forest products facilities. These consume large quantities of natural gas and are expensive to operate. As an alternative, we developed and tested a lab-scale absorption system using a room temperature ionic liquid, tetradecy(trihexyl)phosphonium dicyanamide, as the absorbant. Tests were conducted on simulated exhaust containing only methanol and [alpha]-pinene, kiln dryer exhaust, and veneer dryer exhaust. Methanol, [alpha]-pinene, and formaldehyde were successfully removed from the exhaust. The total hydrocarbon reduction was 73 to 78 percent.

During the manufacture of wood composites and lumber, an estimated 45,000 Mg/year (Federal Register 2003) of volatile organic compounds (VOCs), such as terpenes, are emitted. Some of the compounds are also hazardous air pollutants (HAPs), such as methanol and formaldehyde. HAP emissions are estimated at 17,000 Mg/year (Federal Register 2003). The air pollutants produced in the forest products industry are difficult to manage because the concentrations are generally low and the volume of gas high. Presently, thermal oxidizers (TO) are commonly used for the destruction of VOCs and HAPs. TOs consume large amounts of natural gas to heat air and moisture. The combustion of natural gas generates increased C[O.sub.2] and N[O.sub.x], which have negative implications for global warming and air quality.

As an alternative to TOs, biofilters have received the most attention and development because they use little energy (Boswell et al. 2002). The temperature of the exhaust from wood products facilities can be high enough to inhibit the growth of microbes in biofilters, thus reducing the microbial degradation rate of VOCs. Biofilters can also be difficult to operate because mill shutdowns disturb the life cycle of the microbes. Non-polar compounds, such as [alpha]-pinene, are difficult to absorb into the aqueous biofilter media. Adsorption of VOCs and HAPs onto activated carbon, silica gel, polymeric beads, or zeolites has been studied, but their application to dryer and press exhaust may be limited because of moisture in the gas or the deposition of fatty acids and resins on media blocking adsorption sites and making regeneration of the adsorbents difficult. In a process similar to what we propose, Xia et al. (1999) studied absorption of VOCs from the petroleum industry into silicone oil through a hollow fiber membrane and regenerated the silicone oil under vacuum. Other techniques, such as degradation of VOCs and HAPs with plasma, corona, or uv are still in development. Any destructive method eliminates the possibility of chemical recovery.

Many of the challenges of VOC removal are addressed by an absorption system containing a room-temperature ionic liquid (RTIL) as an absorbent. RTILs are salts, but unlike NaCl, which melts at over 800 [degrees]C, RTILs are in the liquid state at room temperature because their irregular structures inhibit crystallization (Brennecke and Maginn 2001). RTILs are receiving much attention as replacements for organic solvents in industrial processes with significant cost and environmental benefits. Some of these processes include organic synthesis (Blanchard et al. 1999; Branco and Afonso 2001, Carmichael et al. 1999; Shelton 2001), extraction, electrolytes for batteries (Kim et al. 2004), and metal deposition.

RTILs would be excellent absorbents for exhausts from wood products facilities because of their unique properties. They have no measurable vapor pressure; thus, there would be no loss of RTIL to the atmosphere. They can be made to have high solubility for a wide range of organic compounds; thus, VOCs such as methanol, formaldehyde, and terpenes can be absorbed in RTILs. They may have high thermal stability to 200[degrees]C (almost 400[degrees]F) and high chemical inertness so they could withstand the conditions in industrial processes.

The objective of this work was to remove pollutants from exhaust gas at an industrial source using all RTIL absorption system and clean and recycle the RTIL.

Procedure

Tetradecy(trihexyl)phosphonium dicyanamide was selected as the RTIL absorbent. It has good absorption properties for methanol and [alpha]-pinene, is thermally stable, and is relatively easy to synthesize (Wang et al. 2007). It has a density of 0.89 g/mL at 20[degrees]C and a molecular weight of 549.9 g/mol.

The viscosity of the RTIL was measured using a Brookfield DV-2 viscometer with a LV-1 spindle. The RTIL was first dried by heating to 137[degrees]C, then allowed to cool overnight. It was then heated from 20[degrees]C to 120[degrees]C while the viscosity was measured. The RTIL was again allowed to cool overnight, 3 percent by weight water was added, and viscosity measurements were made from 20 to 80[degrees]C, after which the water content was 2.7 percent.

Absorption system description

An absorption-desorption system was constructed using 25-mm inside diameter by 1-m-long glass columns that were electrically heated and insulated (Fig. 1). Stainless steel packing (Aceglass 6624-04) was added to each column to a depth of 50 cm.

[FIGURE 1 OMITTED]

A metering pump circulated RTIL from the desorber to the top of the absorber where it dripped onto the packing and flowed through it at 4 mL/min. Gas at a temperature of 60[degrees]C containing contaminants entered the bottom of the absorption column, passed through the packing, countercurrent to the RTIL, at 400 to 500 mL/min. The absorption column was heated to 60[degrees]C to better simulate what might happen in a large industrial column. Without heat, the small-diameter column would have readily cooled to ambient temperature even if the gas entering the column were hot. The absorption column was operated at approximately atmospheric pressure. A control valve on the outlet line maintained the RTIL in the bottom of the absorber at a level below the packing.

The desorber was heated to 120[degrees]C and operated at subambient pressure, approximately 1 kPa. The combination of vacuum and high temperature caused the contaminates to transfer from the liquid to the gas phase. Ports to a vacuum pump were located at the top of the column and approximately half way up the packing. The low pressure in the desorption column drew RTIL from the bottom of the absorption column, through the control valve and a preheater. In the desorber, the RTIL dripped onto the packing and flowed through it. The metering pump removed the RTIL from the bottom of the absorber for reuse. A small amount of ambient air, 3 mL/min, was allowed to enter the bottom of the desorber through a critical orifice.

Testing

Testing of the device occurred in three stages. First, we used gas generated in the laboratory. Second, we used gas from a laboratory lumber kiln. The kiln provided a convenient source of actual wood dryer exhaust that was close to other lab equipment. Finally, the device was taken to a commercial facility and gas was sampled from the exhaust of veneer dryers.

Tests on laboratory-generated gas

The laboratory-generated gas contained methanol at 25 to 60 ppmv and [alpha]-pinene at 150 to 170 ppmv as surrogates for the VOCs in dryer and press exhaust. The balance was air. Methanol was added from a permeation tube (VICI Metronics, Poulsbo, Washington), the temperature of which was controlled in a GC-oven. A second air stream was saturated with [alpha]-pinene at a controlled temperature of 40 to 80[degrees]C. A third air stream was saturated with water at a controlled temperature. These streams were mixed and diluted to obtain the desired concentrations in the gas stream. Concentrations were verified based on the weights of methanol and [alpha]-pinene evaporated. Transfer lines were heated to 60[degrees]C to prevent condensation.

A 0.2 mL sample of the gas entering and leaving the absorption column was sampled using a 1 mL gas tight syringe. The syringe was stored at 33 kPa and 60[degrees]C to remove contamination. The GC responses to the methanol and [alpha]-pinene were converted to concentrations using response factors obtained by injecting 1 [micro]L of dichloromethane containing 4, 40, or 400 mg/L of methanol and [alpha]-pinene. Solution injections were made daily to check for drift in the response factors. The GC was operated with a 60 m long, 0.53 mm inside diameter, type 624 fused silica capillary column. The carrier gas was helium at 40 cm/sec. The oven temperature ramped from 45[degrees]C to 200[degrees]C over 10 minutes. The injection port was held at 200[degrees]C and contained a 2-mm inside diameter glass liner. The FID was held at 250[degrees]C and used hydrogen at 47 mL/min, air at 400 mL/min, and makeup helium at 25 mL/min.

Samples of the ionic liquid were taken before and after the absorber and analyzed using headspace analysis. Headspace analysis was used because the RTIL does not evaporate when directly injected into the GC injection port, even at 200[degrees]C. A 0.5 mL liquid sample was placed in a 4 mL vial. After 8 hours of equilibration at 60[degrees]C a 0.2 mL sample of the vial headspace gas was injected into the GC as described above. The liquid phase concentrations were determined from the gas phase concentrations based on the Henry's Law constants for the respective compounds in the RTIL (Wang et al. 2007).

Two tests are reported. The gas entering the absorber was dry and contained 24.7 ppmv methanol and 150.9 ppmv [alpha]-pinene in the first. In the second, the gas contained 20 percent water vapor by volume, 59.4 ppmv methanol, and 169.5 ppmv [alpha]-pinene. The tests lasted 9 days and 3 days respectively. Sampling occurred at 12- and 24-hour intervals during tests, respectively.

Tests on laboratory lumber kiln gas

The absorption and desorption columns were unchanged from the description above. A heated box containing a pump, filter, and flowmeter was constructed to move an exhaust sample from the kiln through the columns. The sampling methods were changed so that the performance of the absorber was measured using a total hydrocarbon analyzer and two chilled impinger trains. Valves in the heated box allowed gas from before or after the absorption column to be sampled. All transfer lines were heated to 120[degrees]C.

A total hydrocarbon analyzer (THA) (JUM, 3-200) was used to detect the concentration of organic compounds before and after the column. The detector was calibrated to propane in air each time the gas was sampled. The THA sample flow rate was 2.5 L/min, which exceeded the column flow rate of 500 mL/min. The balance of air to the detector was supplied using ambient air that had passed through a charcoal filter. This was approximately a 6:1 dilution ratio. To maintain the same system pressure and not recalibrate the analyzer, the same dilution ratio was used for samples taken prior to the column. To provide paired measurements, the gas stream leaving the column was measured, then flow to the column was stopped for approximately 30 seconds and the gas stream was sampled prior to the column.

Two impinger trains were operated to sample for methanol and formaldehyde according to NCASI method 98.01 (NCASI, 1998). Simultaneous operation on pre- and postcolumn gas provided paired measurements. The flow rate through each train was 110 to 250 mL/min. The impinger aqueous samples were stored in 100 mL bottles and refrigerated prior to analysis. Gas chromatography and visible spectrophotometry were used according to the NCASI method to determine the concentrations of methanol and formaldehyde, respectively. The instrument detection limits were 0.26 and 0.10 ppmw in the condensate, respectively. The method detection limits were 0.11 ppmv for methanol and 0.05 ppmv for formaldehyde in the gas phase.

Sampling occurred during two kiln charges over 2- and 3-day periods. The kiln was operated at a 78[degrees]C dry-bulb temperature and a 68[degrees]C wet-bulb temperature. The kiln contained western hemlock for the first charge and a mixture of hemlock and white spruce for the second. The type of wood and operating conditions were not the subject of study. The kiln was used to provide a source of emissions containing a broader array of compounds than the lab-generated source.

Test on commercial veneer dryer gas

A 3-day field test was conducted at a commercial site where the combined exhaust from two veneer dryers was fed to a regenerative catalytic oxidizer. The absorber and sampling equipment were not changed except that the 500 mL/min sample was drawn from a duct prior to the pollution control device instead of a kiln. The device was set up and allowed to operate overnight. Sample collection began on the second day; however, the dryers were operating at reduced volume giving very low levels of contaminants in the exhaust gas. On the second afternoon, both dryers began redrying hemlock and on the third morning both dryers began drying green Douglas-fir heartwood. The sampling and analytical procedures were the same as described for the kiln gas.

Results and discussion

The viscosity of the dry RTIL varied from 525 mPa x s at 20[degrees]C to 26 mPa x s at 80[degrees]C, and 14 mPa x s at 107[degrees]C. When the RTIL contained 3 percent water, the viscosity was lower, 380 mPa x s at 30[degrees]C and 22 mPa x s at 80[degrees]C. This is important in the operation of equipment. The RTIL absorbs water in the absorber and it operates at the lower viscosity, at least in the lower portion. In the desorber the water flashes from the RTIL before it reaches the packing so the entire desorber operates at the higher viscosity.

Tests on lab gas

We initially tried to operate the desorber at 60 to 90[degrees]C and a pressure of 4 kPa. This was not adequate to remove [alpha]-pinene from the RTIL. The efficiency of the desorber decreased over several days of operation because the RTIL entering it contained an increasing amount of [alpha]-pinene indicating that the desorber was not removing contaminants from the ionic liquid. We therefore increased the temperature and reduced the pressure to those described in the procedure, 120[degrees]C and 1 kPa.

To improve desorption, we added an airflow of 3 mL/min into the desorber at the bottom. At a given level of vacuum, this small airflow resulted in a decreased partial pressure of the contaminants in the desorber which, in turn, lowers the concentration of contaminates in the liquid phase in equilibrium with the gas.

After the modifications, the absorber removed 92 to 95 percent of the methanol and nearly 100 percent of the [alpha]-pinene from the contaminated gas (Table 1). The values in Table 1 are averages of two measurements for test one and four measurements for test two. The high absorption of [alpha]-pinene was expected because of the difficulties in removing it from the RTIL. Similarly, the methanol was relatively easy to clean from the RTIL and was more difficult to remove from the air. Moisture in the gas in the second test, 20 percent by volume, had no noticeable effect on the absorption. This is consistent with past work (Wang et al. 2007) in which the Henry's Law constant was not significantly different when the RTIL contained 2 percent water compared to dry.

After approximately 20 days of operation, the ionic liquid had changed from a pale yellow to a dark brown color. Infrared and nuclear magnetic resonance analysis on the brown ionic liquid indicated that the ionic liquid was stable and that the brown color comes from the breakdown of the [alpha]-pinene upon heating (data not shown). We concluded that the most likely cause of the brown color is a small amount of verenol and a larger amount of verbenone. This did not appear to affect the absorption characteristics of the ionic liquid.

Tests on kiln exhaust

The concentrations and removal efficiencies for the absorption column on kiln exhaust are shown in Table 2. The total hydrocarbon, a relative measure of all the organic compounds in the gas, was reduced by approximately 80 percent. This is lower than expected based on the relatively high methanol and [alpha]-pinene reductions on the lab-generated gas. Methanol and formaldehyde removal efficiencies were variable with the average removals near 90 percent. Again, this is somewhat lower than for the lab-generated gas. The source gas contained 39 percent water by volume which was almost twice as high as for test two with lab generated gas. The ionic liquid at the bottom of the absorption column contained approximately 4 percent water by weight during these tests. The water content of the gas did not appear detrimental to the operation of the device. However, water in the RTIL might interact, or hydrate with the ionic liquid and reduce the absorption of other polar compounds. This level of water, 4 percent by weight, is approximately 60 percent on a molar basis.

Test on commercial veneer dryer gas

The concentrations and removal efficiencies for the absorption column on veneer dryer exhaust are shown in Tables 3 and 4. The total hydrocarbon was reduced by approximately 73 percent. It was suggested that the lower removal was because this was a direct-fired dryer the gas may contain gases from the combustion unit that may be difficult to absorb. Again, as the inlet concentration increased, the removal efficiency tended to increase. While absorber efficiency should not be a function of concentration if all flows are constant, the desorber might become the limiting part of the system. At a given total pressure, the partial pressure of the contaminants can only be brought to a certain level. In turn, the contaminate level in the RTIL cannot be lower than its equilibrium concentration based on Henry's Law.

The methanol concentration in the dryer exhaust was lower and the formaldehyde concentration was higher than in kiln exhaust. The combustion unit probably contributed to the high formaldehyde concentration. Reductions in excess of 95 percent were obtained for both compounds. The zero values for methanol mean that the gas was cleaned to less than the method detection limit of 0.11 ppmv. If we used one-half of the method detection limit for the exit concentration (as required in some environmental testing), the methanol removal would have been 98.9 percent. The MC of the gas from the veneer dryer was lower than that from the kiln, low enough that we did not accumulate water in the impingers during the NCASI 98.01 method. Although the NCASI method is only validated for methanol, formaldehyde and phenol, we were also able to observe reductions of ethanol (100%), acetone (83% to 100%), and acetaldehyde (36% to 100%) based their concentrations in the impinger catch,

The concentration reduction required in the EPA rule for maximum achievable control technology (MACT) (Federal Register, 2005) for wood processing is 90 percent for total hydrocarbon, methanol, or formaldehyde. Alternatively, outlet concentrations of less than 20 ppmv for total hydrocarbon or 1 ppmv (if > 10 ppmv uncontrolled) for methanol or formaldehyde are acceptable. The absorption column with RTIL met the MACT for percent reduction of methanol and formaldehyde and the concentration rule for total hydrocarbon.

The concentration of methanol in the RTIL at the end of testing was 0.1 ppmw entering and 1.6 ppmw leaving the absorber. A mass balance between the two phases indicates that the calculated mass of methanol entering the liquid phase was 22 percent less than that leaving the gas phase. While this is not as good as hoped, we consider it acceptable given the number and complexity of the measurements. Applying the methods described in Treybal (1980) for the concentration and flow data shows the number of theoretical stages in the column to be 3.1. Thus, the height of a theoretical plate for the absorber as tested is 16 cm. The assumptions in this estimate are a linear equilibrium curve and dilute concentrations.

In all, we used the RTIL for approximately 40 days of testing with the temperature of the ionic liquid cycling from 60 to 120[degrees]C. Other than darkening, which we attributed to some breakdown of the contaminants, the RTIL at the end of testing had similar characteristics to clean RTIL. The Henry's Law constant for the used ionic liquid was within the range of data described in Wang et al. (2007) for clean ionic liquid indicating that the absorption properties are similar.

Conclusions

Contaminants from kiln and veneer dryer exhaust can be absorbed into an RTIL, and the RTIL can be recycled after cleaning with vacuum and heat. The reductions for HAPs are likely to meet the requirements of the MACT rule. Extended testing on commercial exhaust is needed to determine if the removal efficiency of the RTIL absorption system will continue.

Literature cited

Blanchard, L.A., D. Hancu, E.J. Beckman, and J.F. Brennecke. 1999. Green processing using ionic liquids and C[O.sub.2]. Nature 399(6731):28-29.

Boswell, J.T., P. John, B. Adams, K. Sturman, K. Fry, and M. Tracy. 2002. Biofiltration of particleboard press vent emissions. In: Proc. of the AandWMA's 95th Annual Conf. and Exhibition Conf.: Baltimore, Maryland.

Brennecke, J.F. and E.J. Maginn. 2001. lonic Liquids: Innovative fluids for chemical processing. AIChE J. 47(11): 2384-2389.

Branco, L.C. and C.A.M. Afonso. 2001. Ionic liquids as recyclable reaction media for the tetrahydropyranylation of alcohols. Tetrahedron 57(20):4405-44l0.

Carmichael, A.J., M.J. Earle, J.D. Holbrey, P.B. McCormac, and K.R. Seddon. 1999. The Heck reaction in ionic liquids: A multiphasic catalyst system. Org. Lett. 1(7):997-1000.

Federal Register. 2003. National Emission Standards for Hazardous Air Pollutants. Plywood and Composite Wood Products. 68(6):1275-1339.

--. 2005. National Emission Standards for Hazardous Air Pollutants. Plywood and Composite Wood Products. 70(145):44011-44036.

Kim, K.S., S. Choi, D. Demberelnyamba, H. Lee, J. Oh, B.B. Lee, and S.J. Mun. 2004. Ionic liquids based on N-alkyl-N-methylmorpholinium salts as potential electrolytes. Chem. Commun. 2004:828-829.

National Council for Air and Stream Improvement, Inc. (NCASI). 1998. Method CI/WP-98.01 Chilled Impinger Method for Use at Wood Products Mills to Measure Formaldehyde, Methanol, and Phenol. Methods Manual (03.B.004). NCASI, Res. Triangle Park, North Carolina.

Shelton, R. 2001. Catalytic reactions in ionic liquids. Chem. Commun. 2399-2407.

Treybal, R.E. 1980. Mass-transfer Operations. McGraw-Hill, New York. 784 pp.

Wang, F., M. Milota, P. Mosher, K. Li, and M. Yankus. 2007. Henry's Law constants for methanol and [alpha]-pinene in ionic liquids. Wood and Fiber Sci. (In press).

Xia, B., S. Majumdar, and K.K. Sirkar. 1999. Regenerative oil scrubbing of volatile organic compounds from a gas stream in hollow fiber membrane devices. Ind. Eng. Chem. Res. 38(9):3462-3472.

The authors are, respectively, Professor, Research Assistant, and Associate Professor, Wood Sci. and Engineering, Oregon State Univ., Corvallis, Oregon (Mike.Milota@OregonState.edu, Paul. Mosher@OregonState.edu, Kaichang.Li@OregonState.edu). The authors appreciate support from the Dept. of Energy, Agenda 2020 Program through the Office of Industrial Technologies (Award No. DE-FC07-03 ID 14432), Weyerhaeuser Corporation, Tacoma, Washington, Louisiana Pacific Corporation, Portland, Oregon,, and Boise Cascade Corporation, Boise, Idaho. This paper was received for publication in July 2006. Article No. 10233.

* Forest Products Society Member.

(C) Forest Products Society 2007. Forest Prod. J. 57(5):73-77.

Mike Milota * Paul Mosher Kaichang Li *
Table 1.--Concentrations of methanol and [alpha]-pinene in the
gas entering and leaving the absorber. Test 1 is with dry gas
and in test 2 the gas contains 20 percent water by volume.
Removal is the reduction in concentration as a percentage.

 Test 1

 Concentration

 Enter Exit Removal
 ppmv ppmv %

Methanol 24.7 1.7 92.9
[alpha]-pinene 150.9 0.2 99.9

 Test 2

 Concentration

 Enter Exit Removal
 ppmv ppmv %

Methanol 59.4 3.0 94.9
[alpha]-pinene 169.5 0.3 99.8

Table 2.--Concentrations in the gas entering and leaving
the absorber for kiln exhaust. Removal is the reduction
in concentration as a percentage.

 THC

 Concentration

 Enter Exit Removal
Test Sample ppmv ppmv %

 1 1 36.9 6.8 81.7
 2 27.4 5.0 81.8
 3 23.5 3.6 84.5
Average 82.6

 2 1 44.9 8.3 81.5
 2 48.2 9.7 79.7
 3 38.7 10.2 73.5
Average 78.2

 Methanol

 Concentration

 Enter Exit Removal
Test Sample ppmv ppmv %

 1 1 51.0 6.7 86.9
 2 86.9 15.0 82.7
 3 64.3 1.88 97.1
Average 88.9

 2 1 28.0 0.7 97.7
 2 27.9 0.6 97.9
 3 25.9 0.0 100
Average 98.5

 Formaldehyde

 Concentration

 Enter Exit Removal
Test Sample ppmv ppmv %

 1 1 0.67 0.11 83.7
 2 0.67 0.05 92.8
 3 0.82 0.05 94.5
Average 90.3

 2 1 0.39 0.09 76.6
 2 0.21 0.01 95.7
 3 0.18 0.00 93.7
Average 88.7

Table 3.--Total hydrocarbon concentrations in the gas
entering and leaving the absorber for veneer dryer exhaust.
Removal is the reduction in concentration as a percentage.

 Concentration

 Enter Exit Removal
Day Time ppmv ppmv %

 1 14:22 81.5 14.4 82.3
 19:08 71.6 8.1 88.7
 19:21 45.9 18.8 59.1
 2 8:30 76.3 13.2 82.7
 15:49 30.0 11.6 61.2
 3 9:11 38.1 14.0 63.3
Average 72.9

Table 4.--Methanol and formaldehyde concentrations in the gas
entering and leaving the absorber for veneer dryer exhaust.
A concentration of zero indicates no detection on the
chromatograph. Removal is the reduction in concentration as a
percentage.

 Methanol

 Concentration

Day Time Enter Exit Removal

 (ppmv) (%)

 1 15:17 3.9 0.0 100
 19:40 7.5 0.0 100
 2 9:10 4.3 0.0 100
 15:30 5.4 0.0 100
 3 9:01 5.8 0.0 100
Average 100

 Formaldehyde

 Concentration

Day Time Enter Exit Removal

 (ppmv) (%)

 1 15:17 4.0 0.31 92.1
 19:40 12.8 0.11 99.1
 2 9:10 7.7 0.21 97.3
 15:30 11.9 0.17 98.5
 3 9:01 9.9 0.5 94.9
Average 96.4
COPYRIGHT 2007 Forest Products Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:volatile organic compounds; hazardous air pollutants
Author:Milota, Mike; Mosher, Paul; Li, Kaichang
Publication:Forest Products Journal
Geographic Code:1U9OR
Date:May 1, 2007
Words:4430
Previous Article:Determination of moisture content and density of fresh-sawn red oak lumber by near infrared spectroscopy.
Next Article:Monitoring of abrasive loading for optimal belt cleaning or replacement.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters