Title: Controlled atmospheres for air pollution studies
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00098017/00001
 Material Information
Title: Controlled atmospheres for air pollution studies
Physical Description: viii, 154 leaves. : illus. ; 28 cm.
Language: English
Creator: McCaldin, Roy Oeland, 1923-
Publication Date: 1958
Copyright Date: 1958
Subject: Air -- Pollution   ( lcsh )
Sanitary Engineering thesis Ph. D
Dissertations, Academic -- Sanitary Engineering -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis -- University of Florida.
Bibliography: Bibliography: leaves 146-152.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
General Note: Vita.
 Record Information
Bibliographic ID: UF00098017
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000570751
oclc - 13728446
notis - ACZ7733


This item has the following downloads:

PDF ( 5 MBs ) ( PDF )

Full Text






JUNE, 1958


It is a pleasure to acknowledge Dr. E. R. Hendrickson, Supervisory

Committee Chairman, for suggestions and constructive criticism that he

has given during this investigation.

Dr. A. P. Black and Professors A. L. Danis, F. W. Gilcreas, and

J. E. Kiker have all from time to time furnished assistance and advice

helpful to the conduct of this study. Their assistance is gratefully


Mr. Stanley Pasznik and Mrs. Shirley Weidler have both contributed

to this work with extensive laboratory assistance, and their help is


This work was sponsored by the U. S. Public Health Service for

which appreciation is expressed, and thanks go to Mrs. Margie DuMez

for preparing the manuscript and to Mrs. Dede McCaldin for her assist-

ance in editing the manuscript.







I. Introduction

II. Purpose

III. Historical and Review of Literature


Toxicological Studies

Bacteriological Studies

Physiological Studies

Odor Studies

Chambers for Industrial Hygiene

Chambers for Air Pollution

Test Gases and Aerosols

IV. Design Considerations and Description of Equipment
and Procedures

Design Considerations

Description of Chamber

Gas Feed System

Units of Measure

Flow Measurements












Sampling Instruments

Analytical Methods

Determination of Gas Leakage From Test Chamber

Uniformity of Chamber Content

Gas Losses Due to Adsorption on Chamber Surface



Experimental Test Methods

Results and Discussion

Effects of Aging on Chamber Gas Concentration



I. Gas Meter Calibration

II. Pressure Drop Due to Air Flow Through Samplers

III. Experimental Flowrator Calibration

IV. Computed Flowrator Calibration

V. Calibration Data for Gas Meter Number 2.

VI. Volume of Air Added to Chamber Producing Associated
Increase in Pressure

VII. Experimental Data for Calculating Gas Leakage From

VIII. Chamber Leakage Calculations

IX. Uniformity of Gas Concentration

X. Estimated Per Cent of Chamber Surface Covered With
Sulfur Dioxide Molecules

XI. Adsorption of Sulfur Dioxide



























XII. Adsorption of Sulfur Dioxide (Summary of Calculations) 133

XIII. Adsorption of Hydrogen Fluoride 137

XIV. Adsorption of Hydrogen Fluoride (Summary of
Calculations) 142

XV. Storage Effect on Gas Concentrations 144






I. Range of Pollutant Concentrations Reported in a Number
of Cities 17

II. Relation Between Parts Per Million by Volume and
Micrograms Per Cubic Meter for Some Gases Commonly
Found as Air Pollutants 47

III. Adsorption of Gases by Charcoal 83

IV. Adsorption of Gases on the Chamber Surface 92



I. Gas Concentration Remaining After Withdrawal of
Sample and Replacement with Fresh Air 28

II. Air Pollution Chamber 30

III. Cross Section of Door Seal 33

IV. Cross Section of Sample Port 33

V. Cross Section of Replaceable Plate 35

VI. Gas Meter Calibration 49

VII. Pressure Drop Due to Air Flow Through Samplers 51

VIII. Flowrator Calibration 53

IX. Calibration Curve for Gas Meter 55

X. Volume of Air Added to Chamber and Associated Increase
In Pressure 65

XI. Gas Leakage Rate from Chamber for Different Pressures 67

XII. Location of Mixing Fan Blade and Probable Air Flow
Pattern 70

XIII. Schematic Diagram of Simultaneous Sampling System 72

XIV. Schematic Diagram of Cyclical Sampling System. 72

XV. Uniformity of Gas Concentration with Respect to Mixing
Time 74

XVI. Relation Between Adsorption and Gas Pressure bO

XVII. Experimental SO2 Concentration Compared with the
Calculated Concentration 88

XVIII. Experimental HF Concentration Compared with the
Calculated Concentration 93

XIX. Storage Effect on Gas Concentration 96



I. Air Pollution Chamber 29

II. Portion of Door Seal 34

III. Replaceable Plate 36

IV. Air Drying Equipment 38

V. Outlet Pipe and Valve 40

VI. Gas Dispensing and Injection System 43

VII. Hendrickson Air.Samplers 56



Air pollution has been defined as the presence in the air of sub-

stances put there by the acts of man in concentrations sufficient to in-

terfere with the comfort, safety, or health of man, or to deprive him of

the full use and enjoyment of his property.23 The most common air pollu-

tants include various industrial vapors, sulfur oxides, industrial dusts,

and smoke.55 Since these pollutants are largely a by-product of either

fuel utilization or industrial activity, it is evident that the extent of

air pollution will continue to grow as the nation's industrial activity

and fuel usage grow. Problems of health, safety, or nuisance arise at

some point in the constantly increasing usage of the atmosphere as a waste

receiver, and it becomes necessary to control air pollution.

Sampling is an integral part of any control effort. By this means

the degree of contamination and types of properties of contaminants can

be determined., Adequate sampling can furnish the information necessary

for planning remedial measures. It can be used to determine compliance

with regulations, to evaluate effectiveness of control measures, and to

serve as a base line on which to measure over-all improvement or degra-

dation of atmospheric quality.

The need for adequate air sampling instrumentation and the need to

develop test atmospheres for study purposes is attested by a report of

the Engineering Committee at the Air Pollution Research Planning Seminar

which was held in December, 1956.72 The report said in part:


"It became apparent in discussing engineering research
that instrumentation plays an important role, both in iden-
tification of a source and its magnitude and in the evalua-
tion of control equipment. Some of the needed engineering
research requires the prior development of suitable instru-

The report contained a tabulation of needed research including study

on the preparation of test gases and aerosols. To this end the inves-

tigation of simulated atmospheres has been made.


This research deals with the design and operation of a static

chamber to contain atmospheric pollutants. It concerns a specific

method of achieving known gas concentrations, and it evaluates the

various means by which these concentrations become reduced within the

c chamber.

The study was stimulated by problems encountered in air pollu-

tion sampling and is concerned with the physical limitations of

samplin6 devices. 'Jith this controlled atmosphere it is a simple

matter to determine the efficiency of air pollution sampling equip-

mett .:hen operated at varied flow rates and when sampling pollutants

at differing levels of concentration.

A controlled atmosphere in the laboratory that simulates field

conditions can be useful in performing fundamental studies in air

pollution. It is of value in calibrating and testing existing air

sampling instruments and in developing new instruments. This should

make possible a solution of problems dealing with the distribution of

materials between the gaseous and particulate form, and should also

help in determining the alteration that occurs in samples during

collection and storage.




The concept of a controlled atmosphere is not new. Devices have

been prepared for controlling practically every aspect of man's environ-

ment, and test subjects have included human beings, animals, bacteria,

and viruses, as well as vegetation and inanimate objects such as equip-

ment and materials.

Extremes of pressure conditions have been simulated for studying

effects of high altitude flying and deep sea diving. Controlled condi-

tions of temperature and humidity have been employed in studying the

physiology of man under climatic stress, in developing protective cloth-

ing and shelter, in study of bacterial decay rates, and in the study of

the relationship of foreign particles in the atmosphere which serve as

condensation nuclei to precipitation processes.60

Toxicological Studies

A review of the chamber studies that have been conducted in a

number of fields shows the development of some ingenious equipment to

achieve the particular goals of the researcher, and in some cases the

methods and apparatus developed show promise for use in air pollution

studies. Perhaps the largest body of literature on gassing and dusting

chambers belongs to the field of toxicology. Reference to studies

dating back to mid-nineteenth century are to be found on exposure of

animals to toxic atmospheres in enclosed spaces.37,40

The case for the use of dust chambers in toxicological studies

was summarized by Haynes.25 Experiments on the effects of dust on the

lung, and on the manner in which the lung disposes the dust had been

conducted along separate avenues. Animals were exposed to dust clouds

as well as injected with dust suspensions both intratracheally and

intravascularly. It was observed that the pulmonary reaction obtained

by exposure to dust clouds most closely approximated that found in

human pneumoconiosis. Consequently, physiological studies on the

effects of air-borne dusts and mists could be done most effectively by

use of dust or vapor chambers.

A number of dusting boxes have been described in which a variety

of small animals have been exposed to various concentrations of toxic

dusts and mists.3,16,27,31,58,59,75 The test dusts, mists, or gases

have been those identified in certain industrial atmospheres, and about

which toxicological data are needed to carry out protective measures

for exposed personnel. For example, Urban73 in 1954 reported the

experience of the Trudeau-Saranac Institute with two types of chambers

for exposing laboratory animals to aerosols. The chamber studies were

conducted to determine the toxicity of dusts or fumes from materials

such as beryllium, expanded perlite, and calcined diatomaceous earth.

The dust chambers were of 14-guage stainless steel. One shape used

was rectangular and another shape had octagonal sides and a pyramidal

top and bottom. The second shape proved more suitable for exposure of

animals. Vapors and dusts admitted at the top of the chamber were more

uniformly mixed in the octagonal chamber. The chamber volumes were 1.6

and 2.5 cubic meters, respectively. Dust or vapor concentrations in


these chambers were determined by sampling the test atmospheres.

Miller46 described a respiratory chamber suitable for chronic

exposure to carbon monoxide. This had a volume of 336 cubic feet, and

its versatility was apparent when it was described as capable of

handling 2 men, or 12 dogs, or 200 rats. This chamber was constructed

of 24-guage sheet metal supported by a wood frame. All nail heads and

joints were soldered to prevent leakage, and an air lock was used to

reduce gas losses when the test subjects entered the chamber. The

carbon monoxide supply was taken from the city gas lines which contained

23.4 per cent carbon monoxide. This was then mixed with the air supply

entering the chamber, and gas concentrations in the chamber were

determined from analysis of air samples.

Henderson26 pointed out that chamber size could be small and yet

suitable for certain work. He described an exposure chamber consisting

of a spray tube into which only the nose and mouth of small animals,

such as rabbits, were admitted. Svirbely70 described a slightly larger

container of 13.5 liters capacity into which one or two small animals

such as guinea pigs or rabbits could be placed. Monkman47 described

a glass chamber only large enough to expose sample slides to various

chemical reagents. This was a cylinder approximately nine inches long

and three inches in diameter.

Spiegel68 pointed out that simplicity could be achieved, He built

a small chamber consisting of a five-gallon battery jar which was bolted

against a board; the board held the metering apparatus. Feed concen-

trations were determined by metering air through a bubbler that con-

tained the test solution which in turn was mixed with a metered quantity

of fresh air. Average atmospheric concentrations were calculated from

the total volume of air used. This apparatus was used for tests with

mice and guinea pigs. Simplicity of construction was claimed as one of

the primary advantages for this type of device.

Irish28 summarized the features to consider when building an animal

test chamber. The discussion included provisions for: air supply,

supply of test material, metering, mixing, the chamber proper, exhaust

system, sampling, analytical methods, and the test animals. Some in-

vestigators, such as Spiegel, have relied on measurements of the air

supply and supply of test materials to determine the exposure concentra-

tion in the chamber. Others, such as Boyland3 and Fairhall16 have

relied on analysis of air samples to determine the chamber concentrations.

Marshal and Knolls40 discussed problems of gas and dust losses

due to adsorption on the fur of test animals. In 1919, they described

an apparatus for administering gases and vapors which consisted of a

glass box of 130 liter capacity. The test animals were placed in the

box and exposed to various gas concentrations. Reduction of gas or

vapor concentration due to surface losses on the animal fur was kept to

a minimum by use of high rates of air change. Approximately two air

changes per minute were used. The gas concentrations were calculated

on the basis of gas and air flow rates. This device was based largely

on the series of investigations conducted by Lehmann from 1884 to 1913,

and the Lehmann apparatus in turn was based on a modified Petenhoffer

respiration apparatus developed at an even earlier date.

Walton76 also dealt with the problem of reduced gas or dust concen-

tration due to surface losses on the hair of the test animals and in

1920 described a small gassing chamber to be used for exposures of short

duration. This chamber was designed to reduce the surface effects of

test animals on concentrations in small chambers and to avoid building

large expensive chambers. Its salient features included provision for

high flow-rates and a small movable air lock for facility of placement

and removal of the test subject.

Silver62 sLmnarized a number of the problems which investigators

faced when they exposed animal subjects to test atmospheres. In 1946

he described the principles influencing the design and operation of

constant flow gassing chambers for use in animal experimentation. He

stated that a dynamic cha.iber is one through which air is continuously

drawn at a fixed rate, and the agent to be studied is introduced into

the influent in the form of smoke, dust, or fog. After the desired

concentration has been established, the test animals are quickly

introduced through a door.

The theoretical concentration in a dynamic chamber has been cal-

culated as follows:
Agent flow in milli-
Concentrationi in nilligramiis per liter = grams per minute
Air flow in liters
per minute

This is true only after the chamber has reached equilibrium, and

if there are no surface or leakage losses. When a vapor or gas is

introduced at a uniform rate into a chamber through which there is a

continuous flow of air, the concentration within the chamber increases

until it is practically constant. If the mixing occurs uniformly

throughout the chamber, the concentration at any time may be calculated

as follo\.6:

W -bt
C = (1-e -)
b a

where C concentration in milligrams per liter at time t

W milligrams of agent introduced per minute

a = chamber volume in liters

b = volume of air passing through chamber each minute

t = time in minutes

Surface losses were considered for two coating materials and a

marked difference was shown between them. With a "Cotoid" lacquer

(made by Lithgrow Corp., Chicago) less than a 10 per cent reduction of

chlorovinyl dichloroarsine was encountered. With a cellulose acetate

coating more than an 80 per cent reduction occurred for the same gas

concentration and exposure. It was further shown that even small

quantities of exposed rubber within the chamber would markedly reduce

concentrations of organic vapors.

Test subjects, both animal and human, were shown to reduce the

concentration established in the chamber. Surface loss due to subjects

could be reduced percentage-wise by keeping the chamber volume large

in relation to the volume and surface of the subject. Entry of subjects

into the chamber could also upset the equilibrium conditions. This

problem was held to a minimum by using high air flow rates, large

chamber volumes, and rapid entry into the chamber through small openings.

Bacteriological Studies

A controlled atmospheric environment has been used on many occasions

for bacteriological studies. The ability to maintain a large air mass

under constant conditions makes it possible to evaluate the effects of

different atmospheric states on air-suspended bacteria and virus. Of

especial importance to the bacteriologist is the effect of temperature


and humidity on the viability of bacterial aerosols. Robertson53

described the construction of a test room built within a larger room so

that a wide range of temperature and humidity conditions could be main-

tained. Kaye29 described an apparatus in which the effect of vapors

upon the viability of bacterial aerosols could be determined. In this

apparatus the aerosol was generated in one chamber and then blown equal-

ly into two spheres. One sphere was used as a control and the other

was used to measure the effect of the study vapor. The spheres were of

clear plexiglass and approximately two feet in diameter.

A simple device to store air-borne bacteria for measuring bacterial

decay in relation to time, was described by Ferry and Maple.17 Storage

was performed in a 400 liter balloon. This was a Darex Kytoon balloon,

type K125, made by Dewey and Almy Chemical Co., Cambridge, Mass.

Edwards15 performed investigations of aerosol systems formed by

atomizing suspensions of virus into the air. Atomization was accomp-

lished with an Aerograph Pencil Spray with air supplied at a pressure

of 35 pounds per square inch. These tests were made with a glass carboy

containing mice which were exposed to the virus of influenza.

The use of bacterial aerosols in air pollution studies was empha-

sized by Kethley and others.30 It was pointed out that bacterial

aerosols could be used to study filter efficiency, distribution, and

rate of distribution of bacteria in an enclosed space, the operation of

aerosol collection devices, and the effect of electric charge on aero-

sol behavior. A high degree of sensitivity and accuracy could be

obtained by employing standard microbiological techniques for bacteria

collection and analysis. A practically monodisperse concentration


could also be produced by atomizing a solution containing a pure strain

of bacteria. The aerosol was created in a pre-chamber and large clumps

of bacteria were allowed to settle out before the aerosol was admitted

to the test chamber.

Full books on air-borne bacterial infection have been prepared by

Wells,77 Rosebury,54 and Bourdillon.2 Topics related to controlled atmos-

pheres include methods of sampling air for bacteria, methods of disinfect-

ing air, measurement of air contamination, and animal tests with infective

aerosols. In carrying out tests with infective aerosols exhaustive

safety precautions have had to be followed which in some respects compli-

cated the experiments.

In practically all bacteriological studies the aerosol concentra-

tions have been determined from cultured samples of the test atmospheres.

Physiological Studies

Exposure chambers have also been used as therapeutic devices.

Church and Ingram8 described a chamber for administering aluminum

hydroxide dust to advanced silicotics. This dust was created by scrap-

ing a powder into the air stream at a fixed rate. The dust mixture

flowed through an airtight 64 cubic foot box which had two-way diaphragm

valves leading to exhaust hoses. Patients could sit outside the chamber

and inhale aluminum dust through hoses connected to the chamber without

exhaling back into the chamber. The dust box made it possible for a

number of patients, or test subjects, to breathe from an atmosphere of

uniform mixture. Previous to this experiment, hand aspirators and other

devices were used to prepare individual dosing without any adequate means

of controlling the dust size and concentration administered.


Landahl and Hermann33 described an auxiliary use for cloud chambers

in conducting physiological studies. They studied the degree in which

different solid aerosols are retained in the lung. The human test

subjects breathed through a hose system which supplied the test air

from a chamber. The chamber or cloud reservoir in this case served to

balance the flow of air-borne dust as subjects inhaled and exhaled.

nIpactors were used to collect dust samples from the inhaled and

exhaled air, and the difference between the inhaled and exhaled dust

concentration represented the amount retained in the lung.

Odor Studies

An odor test room was described by Deininger and McKinley.11 The

test room had a 500 cubic foot volume and could hold four to six people.

Walls were made of one-quarter inch thick asbestos-cement and were coated

with a strippable plastic film, tygon. The interior surfaces were

also lined with aluminum foil with the polished side exposed. Sodium

silicate was used as the adhesive for the aluminum surfacing. Flooring

was made of asphalt tile but aluminum was recommended. The room was

arranged so that a measured quantity of odorant could be administered

to the room and circulated by fan. The panel of observers would

rapidly enter the room from a closed anteroom and record their obser-

vations regarding description and strength of odor observed.

Chambers For Industrial Hygiene

Atmospheric test chambers have found a wide use in the field of

industrial hygiene. As reported earlier many toxicological studies

have been made using test chambers. From the engineering standpoint

chambers have proven useful in fundamental studies on gases and aerosols


and for development and evaluation of control equipment and instruments.

Drinker13 in 1924 described one of the first test rooms used in this

country for engineering studies on air-borne industrial dusts.

Silverman64 reported on methods to calibrate air sampling instru-

ments for industrial hygiene. Features of both flow-through and static

chambers were discussed. It was stated that for static chambers the

room volume should be large in relation to the volume of air removed

by sampling in order to prevent significant dilution of gas concentra-

tions. Chambers for this purpose have been built that vary from 100 to

3,000 cubic feet in volume. Roberts52 used a chamber of this type for

experimental dust studies. His static chamber had a volume of 513

cubic feet, and was used to perform various studies on the physical

properties of quartz dust.

In the case of dynamic or flow-through chambers there is no

limitation on the sampling volume or rate. Dynamic chambers can be

used to produce gas vapor concentrations for most contaminants and

for many ranges of concentration. One of the advantages of the dynamic

chamber is that less space is necessary although more metering equip-

ment is required.

One common industrial hygiene engineering problem has been that

of making routine calibrations of sampling equipment. To meet this

problem a number of relatively simple gas and mist chambers have been

developed. Sallee, Elgin, and Miller56 reported a technique for obtain-

ing known vapor concentrations by crushing a glass ampule loaded with

a measured quantity of solvent within a 500 liter tank. The tank was

fitted with a heating element to volatilize solids of low vapor pressure,


a fan to mix the vapor, a thermometer, and suitable openings for sampl-

ing and servicing. The vapor concentration in the chamber was reduced

gradually as samples were withdrawn and the vapor-air mixture was re-

placed by fresh air. Sampling instruments were calibrated by comparing

the theoretical sample strength in the chamber with the results deter-

mined from analysis of samples taken from it. The difference between

the two results was used to determine sampler efficiency.

Stead and Taylor69 described a somewhat similar device made from

a five gallon jar. Stead described in more detail the limitations

associated with this form of vapor chamber. If outside air is not

admitted to the chamber, the pressure decreases as sampling proceeds

and corrections must be employed on meter flow readings. If outside

air is admitted, the gas concentration is reduced. The mixing rate for

the dilution air is important because theoretical calculations for

reduced concentration assume that there -is instantaneous mixing. For

the particular arrangement of equipment used, it was found that a

flow rate of at least 1.5 liters per minute was necessary to afford

reasonably rapid mixing of the diluted batch. It was reported that for

flow rates of 2 liters per minute and up, actual and theoretical con-

centrations of vapor in the container agreed within about 10 per cent.

However, sampling had to be carried out immediately after the gas

mixture was created in order to avoid errors thought to be caused by

condensation and adsorption. Test gases used were benzol and carbon

monoxide. Benzol was used in concentrations between 400 and 1,000 ppm

by volume, and carbon monoxide between 250 and 500 ppm. Mercury vapor

was also used at a concentration of 0.75 mg per cubic foot.


An article in the "Encyclopedia of Instrumentation For Industrial

Hygiene"l describes a way of minimizing the effects of dilution that

occur when operating a static chamber. Four stainless steel tanks of

34.7 liters each were filled with a known vapor mixture. The vapor

was withdrawn from one tank and used to calibrate industrial hygiene

instruments. The dilution effects were minimized by arranging the

tanks in series and causing the diluting air to enter the first tank

when a sample was withdrawn from the fourth tank. The diluted gas-air

mixture in the first tank caused a smaller dilution in the second tank,

and the slightly diluted mixture in the second tank produced even less

dilution in the third, and the mixture reaching the fourth tank caused

only very slight dilution.

A calibration curve was shown for this system of chambers in which

methanol was used at about 300 ppm by volume in the original concentra-

tion. Average recovery as related to theoretical recovery was reported

to be 91.3 per cent for a fritted glass bubbler and an impinger placed

in series. It was reported to be 79.1 per cent for either the bubbler

or impinger alone. It is of interest to note that if the samplers in

series each operated at 79.1 per cent efficiency as they did individ-

ually, the total efficiency for series operation would have been 95.6

per cent instead of 91.3 per cent. This suggests then that the

efficiency of the second sampler in series was considerably less, or

about 58 per cent, possibly due to the decreased concentration of gas

in the air when it reached the second sampler.

The atmospheres of mines, mills, and industrial plants, as well as

those in a test chamber were used by Brown and Schrenk4 to perform


comparative studies of dust sampling equipment. Silica and coal dusts

were used in both eight and sixteen cubic meter test chambers. When

using silica or bituminous coal as the test dust it was found that micro-

scopic light field counts between simultaneous impinger samples agreed

within plus or minus 5 per cent. Corresponding results between midget

impinger samples agreed within plus or minus 10 per cent. Microscopic

dark field counts for simultaneous midget impinger samples agreed only

within plus or minus 20 per cent.

Chambers For Air Pollution

With the rather wide use that the dust and gas chambers have had

in industrial hygiene and bacteriology, it is only natural that they be

used to solve a number of air pollution problems. The greatest single

difference between the controlled atmosphere used for industrial hygiene

study and that for air pollution study is in gas and dust concentrations.

Silverman63 has reported that most gas concentrations for industrial

hygiene study vary from 10 to 500 ppm by volume. The atmospheric levels

of pollutants of air pollution interest vary from a trace up to about

3 ppm by volume.

Wohlers and Bell79 in 1956 conducted a literature review and made

extensive tabulations of measured concentration of atmospheric pollutants

found in metropolitan atmospheres. It was reported, for example, that

the average atmospheric sulfur dioxide concentrations for some 47 cities

generally varied between 0.01 and 0.18 ppm by volume. The minimum

concentrations reported were as low as zero and the maximum concentra-

tions reported were as high as 3 ppm in isolated circumstances.

Table I summarizes these data on concentration of atmospheric





Pollutant No. of Cities Average Range of Concentra-
Reporting tions (Parts per million by
lower upper

Sulfur dioxide 47 0.01 0.18

Hydrogen sulfide 4 0.00 0.11

Nitrogen oxide,
(expressed as NO2) 8 0.02 0.89

Ammonia 7 0.01 0.22

Aldehyde, (expressed
as formaldehyde) 8 0.04 0.80

Chloride 9 0.02 0.09

Carbon monoxide 8 2. 10.

Oxidants, (expressed
as ozone) 15 0.01 0.35

Fluoride, (expressed
as HF) 7 0.00 0.02

Carbon dioxide 8 330. 390.

Dustfall, (in tons
per square mile) 31 11. 156.


pollutants. It shows the lower and upper ranges of concentration re-

ported for various substances. The terms "lower" and "upper" are not

defined mathematically. They merely indicate that a great majority of

the reported data falls between the figures shown. Obviously by choice

of sampling location one could collect samples of greater or smaller

concentrations than those shown.

A number of smog chambers have been built to synthesize and study

one or more contaminants found in the atmosphere. Much of the recent

air pollution chamber work has been done in conjunction with problems

of the Los Angeles smog. A chamber used by Magill39 was designed to

measure the sensory response of human subjects to various smog conditions.

This chamber of about 1,000 cubic foot volume was tested both dynamically

and statically. In dynamic tests the measured and theoretical concen-

trations were within 5 per cent of one another. In static tests using

formaldehyde the analytical concentration was found to be 30-40 per

cent lower than the theoretical nominal concentration. This discrepancy

was thought to be due to the surface effects of the chamber. The

static tests were considered only as exploratory, and no corrections for

nominal concentration of test substances were made.

Mader38 employed a plexiglass chamber with an eight cubic foot

capacity to study the effects of solar radiation on various hydrocarbons.

Elaborate precautions were used to flush the chamber of foreign gases

and insure that the air in the chamber was pure before admitting the

study hydrocarbons. Measured quantities of hydrocarbons were added to

the chamber air with hypodermic needles which were adapted to fit glass-

stoppered openings. It was reported that relatively high gas concentra-


tions had to be employed in this study in order to compensate for both

surface loss and other losses in this relatively small chamber.

Substantially larger smog reaction chambers were employed by

Goodwin, Bolze, and Morriss.21 A pair of 2,100 cubic foot glass reaction

chambers were used to contain automobile exhaust at the relatively low

concentrations found in Los Angeles on smoggy days. Due to the large

volume of the chambers, substantial sample quantities could be removed

without causing measurable reduction in smog concentration. In this

study measurements were made of temperature, relative humidity, light

intensity, concentration of oxidants, nitric oxide, nitrogen dioxide,

and extent of rubber cracking. Also, a panel of 20 students entered

the chamber to observe the degree of eye irritation caused by different


Morriss and others49 further described characteristics of this

chamber. The surface was 80 per cent glass and 20 per cent white alkyd

resin paint. The chamber was caulked and considered to be fairly air-

tight. It was found that under static conditions nitric oxide, nitrogen

dioxide, and various hydrocarbons had a half life of ninety minutes

during absence of sunlight. Losses were thought to be due partially to

surface adsorption.

Cummings and Redfearn9 also reported the effects of adsorption on

container walls when they described two ways of achieving 0.5 ppm con-

centration of sulfur dioxide in air. The first method, that of injecting

a measured quantity of sulfur dioxide into a 20 liter container, failed.

This was thought to be due to adsorption and desorption of sulfur dioxide

on the walls of the vessel. The second and more satisfactory method of


preparation was performed by successive dilutions of the sulfur dioxide

and air mixture. This represented a dynamic flow of gas and air mixture

whose theoretical and analytical concentrations were found to agree


The adsorption of gases and vapors on aerosol particulates was

discussed by Gordieyeff.22 This extensive theoretical discussion, how-

ever, was limited to adsorption of volatile materials by solid particles

and to the toxicity of particles before and after exposure to a vapor.

Experimental results suggested that toxicity was due primarily to the

dust per se rather than to its adsorptive action.

Middleton45 used a six cubic foot fumigation chamber to determine

the response of plants to air pollution. The chamber air was pre-

cleaned by passing it through activated charcoal. Ozone and hydro-

carbons were added to the chamber in measured quantities, and damage

to vegetation within the chamber was related to duration of exposure

as well as gas concentration.

Cadle and Schadt6 described a plan for correlating theoretical

with actual collection efficiencies for a number of coranonly used field

instruments. This plan dealt only with aerosol collecting devices.

Later Schadt and Cadle61 reported the results of this work. Six partic-

ulate sampling devices were used in the study which related theoretical

to experimental sampling efficiencies for collection of a mono-disperse

aerosol. A ten cubic meter test chamber was used for most of this work.

Test Gases and Aerosols

Ability to prepare and introduce the proper concentration of test

gas or aerosol to the chamber is as important as the ability to main-


tain the mixture after it is in the chamber. Silverman,63,64,65 and

Silverman and Billings66 have thoroughly described a number of methods

for creating solid and liquid aerosols and gaseous pollutants. They

have also published extensive tabulations of commonly used aerosols.

Since this topic is well covered in the literature, it seems desirable

only to outline the dust, vapor, and gas production methods and mention

some of the problems attendant with their preparation.

Aerosols may be formed by crushing or grinding of solids, conden-

sation of the vapors of materials that have been volatilized, combustion

of organic or inorganic substances, or evaporation of a volatile liquid

solution which contains the aerosol. Gaseous concentrations have been

produced by injecting measured gas quantities into chambers by a number

of methods. Gas burettes, pipettes, and hypodermic syringes have all

been used for this purpose. Desired gas concentrations have also been

produced by successive dilution of gases, by quantitative chemical

reactions,48 and by controlled vaporization from the liquid state.

Chen and others7 described a method of dust generation capable of

producing uniform clouds with a particle size ranging from 1-30 microns.

The dust was dispersed by sonic vibration and then transported in a

metered air stream. Particles in the dust cloud were evaluated by

measuring the amount of light scattered at ninety degrees to the incident

light beam. The light scattering was reported to be a function of the

surface area of the aerosol present.

The use of an aspirator for the production of heterogeneous aero-

sols was reported by Lauterbach and others.36 The aerosol concentration

was found to be proportional to the concentration of soluble material


in the aspirated solution. Mean particle size values for the aerosol

were also dependent upon the concentration of the ablution. In a test

with sodium chloride solution the mass median diameter of the aerosol

was doubled when the original salt concentration was increased by a

factor of ten.

A simple and compact fume generator was described by Glauberman

and Breslin.20 This was designed for use as a standard test instrument

to measure the dust retention of filter media. Uranium chips were burned

in an argon and oxygen gas stream. This generator produced solid par-

ticles of 0.05 to 0.2 microns. Dautrebande, Alford, and Highman10

achieved a mean particle size for quartz and willemite that was substan-

tially under one micron in diameter. This was done by passing the dust

stream through successively smaller cyclones, the smallest of which was

less than one inch in diameter. It was also found that mixing the dust

with water droplets increased the mean particle diameter by 50 to 100

per cent due to liquid coating on the particles. Dust size was deter-

mined from calculations based on settling rates in a still air.

Aerosols of a very uniform size have been produced by Sinclair67

and Lamer.32 The principle involved is one of slow and uniform conden-

sation of vapor that contains condensation nuclei. The mass of the

condensable vapor compared to the number of nuclei determines the size

of particles produced. When this operation is carefully controlled, it

is reported that a particle size can be produced which does not vary

more than 10 per cent from the average. Another method of controlling

particle size was mentioned by Druett and May.14 Droplets of any chosen

size down to about ten microns could be discharged from the tip of a


micro-pipette. This was done by applying a momentary high voltage to

the liquid. The major problem with this method of mist production lay

in pipette blockage from foreign materials.

A variety of solid and liquid aerosol production devices have been

employed by Cadle and Magill5 in doing smog studies. Powders were con-

tinuously dispersed at a uniform rate by drawing the powder through a

glass trough beneath an air operated aspirator. The dusty atmosphere

was blended in a one cubic meter baffled premixing box before being

sent into the ten cubic meter test chamber. Volatile liquids were also

prepared by adding liquid drop-wise with a mechanically driven syringe

to a hot plate which volatilized the liquid. The vapor was then swept

into the test chamber with nitrogen gas.

Another vaporizing device was used by Mavrodineanu and Coe41 in

order to maintain a given concentration of fluosilicic (H2SiF6) or hydro-

fluoric (HF) acid in greenhouse studies on plant damage. Dilute solutions

(.01 to 5 per cent) of commercial preparations of these acids were atom-

ized through a platinum nozzle in an electrically heated tube. The tube

was heated to 150 degrees centigrade which adequately transformed the

liquids to the gaseous state. The vapor was then mixed in the supply air

stream which went to the study greenhouse. It was found that with this

dynamic flow system any required concentration could be readily maintained.

Diffusion of the vapor of a liquid into a gas stream has been employ-

ed by McKelvey and Hoelscher43 to produce extremely dilute gas mixtures.

The method was stated to be most useful when the low concentration com-

ponent was a liquid at ordinary temperature and pressure, but the method

has also been used for materials that could be readily liquified in the




Design Considerations

In order to carry out the study objectives which were stated at

the beginning of this report, it was desirable to build the chamber

along flexible lines. Therefore, the chamber was so constructed

that it could be operated as a static chamber as well as a dynamic or

flow-through chamber for both gaseous and particulate air contaminants.

It was designed to maintain a reasonably constant atmosphere during

repeated or continuous sampling, and to achieve the minimum possible

loss of gas or aerosol during production of the test atmosphere and

the subsequent storage of that atmosphere in the chamber. Provisions

were made to allow flexibility both for various feed arrangements and

for removal of samples.

The test conditions for which the chamber has be&n designed in-

clude the following: The range of gas concentrations will vary from

zero to generally not more than three ppm by volume with the most

frequently used concentrations being about one-half of a ppm. The

test gases will vary from relatively stable gases such as dry sulfur

dioxide to highly reactive gases such as hydrogen fluoride. The chamber

must be suitable to retain both gases and dusts although in this study

only gases are evaluated. Moisture content in the chamber will vary

from ambient humidity down to a relative humidity of about 5 per cent.

Lower humidities will be used on preliminary work with the more



reactive materials. Pressure in the chamber at all times will closely

approximate the ambient atmospheric pressure, and only laboratory

temperature conditions will be used.

The shape and size of the chamber constituted the first design

consideration. Since surface losses due to adsorption or chemical

reaction could constitute a significant loss, it was desirable to have

the maximum chamber volume with the least chamber surface. A sphere

offers the optimum shape for this requirement, but it presents con-

struction difficulties. As a compromise, a cubical shape was chosen

for maximum volume to surface ratio and minimum construction difficulties.

The volume to surface ratio is also related to the size of the

chamber. For example, a cubic chamber with a side of one foot would

have a volume to surface ratio of 0.167; a cube with a six foot side

would have a ratio of 1.000; and a cube with a ten foot, side would have

a ratio of 1.67. The actual numerical ratio depends on the units of

measure which are chosen, but the volume to surface ratio does increase

with size. Consequently, for a reduced surface effect it would be

desirable to use as large a chamber as possible. Another factor favor-

ing a large chamber is the facility of achieving the desired gas or

aerosol concentration with a high degree of accuracy. For example, a

one ppm concentration of a gas can be achieved in a 1,000 cubic foot

(28.3 cubic meter) chamber by adding 28.3 cubic centimeters of gas.

If the gas feed can be determined to the nearest 0.01 cibic centimeter,

then the error would be plus or minus 0.035 per cent. If the volume

of the chamber were one-hundredth as great (10 cubic feet or 0.283

cubic meters) and the same feed system were used, the per cent error


in metering the feed gas or aerosol would be one hundred times greater,

or plus or minus 3.5 per cent. Of course successive dilutions could

also be used to achieve low concentrations in a small chamber, but

each dilution of feed entails problems of metering and loss, especially

of dust in the diluting apparatus.

Another factor favoring a large chamber is the decreased effect of

dilution as samples are withdrawn. If the sample which is removed from

the chamber is small in relation to the chamber volume, than the make-up

air which replaces the sample will produce only a small dilution of the

gas or aerosol concentration in the chamber. Since the sample size is

dictated by the particular test being conducted, the original choice of

a large size chamber would be desirable in minimizing changes in test

concentrations due to dilution.

The theoretical dilution that should occur when a sample is removed

from a chamber and is replaced by fresh air has been described in various

ways by Silver62 and Silverman.65 It can be shown in the following

manner: v
C e V

where C = the concentration remaining expressed as
a fraction of the original concentration

v = volume of sample removed

V volume of chamber

This value of C, the theoretical concentration remaining, is based on

the assumption that the mixing of replacement air is instantaneous and

uniform. The above formula is useful in calculating the dilution that

occurs with sample removal and in determining the amount of air required

for purging the chamber before a new test is started. If a 100 liter


sample is withdrawn from a 6.12 cubic meter chamber, the remaining con-

centration will be 0.98 of the original concentration. The same formula

shows that when purging the chamber for a new test, it will take 4.6

chamber volumes to reduce the concentration of any existing contaminant

to 0.01 of its original concentration.

Figure I shows a plot of the theoretical gas concentration in a

216 cubic foot chamber (6.12 cubic meter) for various degrees of dilution.

In the case of a large chamber there is the possibility of thermal

stratification of the atmosphere which would hinder air mixing or pro-

duce dead air spaces in the chamber. This problem would probably not

be significant for the size chamber chosen for this work.

The final and controlling factors in the choice of size were those

of available space, cost, and ease of construction. It was desirable to

build as large a chamber as possible and yet satisfy these last require-

ments. The dimensions finally chosen were six feet on a side which allow-

ed for a volume of 216 cubic feet. This size chamber could be fabricated

from stock building materials and would fit in the available laboratory


Description of Chamber

The chamber was built on a two inch by four inch stud frame. One-

quarter inch thick masonite was nailed to the inside of the frame with

the smooth side of the masonite facing inwards. All joints were sealed

and sanded smooth before painting. The completed chamber is shown in

Plate I, and the various appurtenances are shown in Figure II.

One of the most important considerations with respect to the in-

terior of a chamber is that of surfacing. The ideal surface is one


C z


'- %0 *-
0 0

M u 0
I Io 2

u 0 0o
8 0)
N 0 0

o 0*
o > >

S0 0 0

o/o o On 0

0 aC s s o o a oo co a



'~ f~;6;; ;;:.
~,: ;2. i; ;~~I~I1PI ;T Il'ljlZ




Dimensions 6 ft. x 6 ft. x 6 ft.
Volume 216 cu. ft. 6,120 liters
Surface = 216 sq. ft. 20.1 sq. meters

PVC pipe
Fan motor


1 2 Door

Calcium sulfate drying jars


that is chemically inert, physically smooth, readily cleaned, and that

can be applied without undue difficulty or expense. A number of materials

were considered, including stainless steel, glass, plexiglass, enamel

paint, varnish, and various plastics. Polyvinyl chloride, which is a

plastic polymer of vinyl chloride prepared from acetylene and hydro-

chloric acid, was finally chosen for the surfacing material. This

material, designated by the letters PVC, has been described as resistant

to many of the industrial gases and vapors at both normal and elevated

temperatures. It is available commercially and costs little more than

a good enamel paint. Although a number of the other materials which

were considered are also chemically resistant, the PVC proved through

its flexibility and ease in handling to be the most acceptable material.

Since this material could be put on with a brush, all of the surfaces in

the chamber were coated. This included the walls, fan blade and shaft,

and sampling ports. In addition, the inlet and outlet pipes for use in.

dynamic flow operation were also made of PVC.

Six coats of PVC paint were applied by brush, and the finished

surface was considered reasonably smooth, but not as smooth as the

surface found on a similar enamel paint. The surface is readily clean-

able since it is waterproof.

The door seal constituted a problem since it was desirable to main-

tain the integrity of the PVC surface and also have an airtight seal.

This was accomplished by building a sealing strip around the door frame.

This seal strip was made of a flexible rubber tube encased in a thin

sheath of PVC. The resilient rubber tube provided a tight seal, and

the PVC sheath provided the inert surface. A cross section of this


door seal is shown in Figure III, and Plate II shows a close-up photo-

graph of a portion of the seal.

The door itself comprises one end of the chamber. It is held in

place with twelve C-clamps that can be adjusted to make a tight fit

against the door seal. The door can be removed and replaced in a matter

of minutes. The large size opening made the initial painting easier

and facilitated placement of test materials within the chamber.

Six individual sample ports were placed in one side of the chamber

for use in the gas mixing tests. These ports allow for sample removal

and yet produce only a limited irregularity in the chamber surface.

They were made by drilling one-inch holes in the chamber and cementing

rubber stoppers into the holes. The stoppers contain one-quarter inch

diameter glass tubes which protrude into the chamber one-half inch.

The tubes extend approximately two inches beyond the outside of the

chamber so that sampling tubes can be connected, or seals can be placed

over the ports. Figure IV shows a cross section of one of the sample


In addition to the individual sample ports, two removable plates

were placed in the chamber. One of these is shown in cross section in

Figure V and illustrated by photograph in Plate III. The principal

purpose of the plates is one of flexibility. Since it was not possible

to anticipate at the time of construction all of the chamber openings

that would be required, the plates were installed. If, for example, a

larger opening to the chamber is required for direct collection of aero-

sols on filter paper, the necessary opening can be prepared on a ply-

wood plate, and this can be substituted for the plate in use. The plates




Drawing is 1/2 scale

1/4 in. Rubber tub,

1/4 in. masonite

1/4 in. Masonite door

-2 in. by 4 in. Stud frame-'



Drawing is full scale

1/4 in. Glass tube

Rubber stopper

All inside surfaces
of chamber are
coated with PVC

1/4 in. Masonite wall


-. -(~ -~n m~~
1* ^ -- -

- *
*-- \w .









Drawing is 1/3 scale

2" x 4" stud
chamber frame


3/4" Plywood frame

cut out for
window -

Circular cut
out for probe

strip -

2" x 4" stud
chamber frame

Interior of

-Cellophane tape seal

lass window

All interior surfaces
except window are
coated with polyvinyl


tape seal

1/4" Masonite surface


__ __ __ I





are held in place with easily removed nailing strips, and their surfaces

are made contiguous with the inside of the chamber wall with cellophane

tape over which the PVC is applied. One of the original plates was built

with a small viewing window and three sample ports as shown in Plate III.

The other was similar except that one of the sample ports was replaced

by the outlet connection for use when the chamber is operated dynamic-

ally. The small view windows were installed for possible dust study

through the reflection or refraction of light on the dust in the chamber.

Air used in the chamber is normally furnished from the laboratory

air supply. This is piped to the laboratory from an air compressor in

the building and contains some impurities. Before going into the chamber

the air flows through a bank of calcium sulfate drying jars which are

sold under the trade name of Drierite. The efficiency of this drying

agent has been reported by Dover and Marden.12 The water vapor remain-

ing in air that has been passed over anhydrous calcium sulfate is 1.4

milligrams per liter of air, and this means that the air in the chamber'

contains approximately seven grains of moisture per pound of dry air

and has a relative humidity of approximately 5 per cent at laboratory

temperature. Filter pads at the inlet and outlet end of the drying

salts remove any particulate matter that may come through the air supply.

Plate IV shows the arrangement of the air-drying equipment. The air is

fed to the distributing pipe along the bottom of the drying jars. It

then flows upwards through the jars containing the drying salts and is

collected in the pipe behind the top of the jars. The dried and filter-

ed air is then admitted to the chamber.

In the arrangement for dynamic flow operation the air goes from the


I- O



driers to a large Flowrator, serial number V4-1113/1, and then into a

system of one and one-half inch PVC piping which delivers the metered

air to the chamber through the air inlet at the top center of the chamber.

The Flowrator and pipe connections leading to the top of the chamber can

be seen in Plate I. With this air supply and air cleaning system it is

possible to achieve a maximum air flow of approximately 10 cubic feet

per minute to the chamber. This flow rate would be low for any dynamic

system used to evaluate control equipment, but it is more than adequate

for the modest air requirements imposed when calibrating air sampling

equipment. The principal drawback to this dynamic feed rate is that

it takes considerable time to build up a balanced concentration of the

study gas.

About one foot above the Flowrator in the direction of air flow

there are two orifices placed in the PVC pipe for admitting the desired

chemical feed. The orifices are made of capillary tubes with very

small openings. These can be used to admit small quantities of the

desired gas to the air stream. In one feed system for this purpose the

gas may be displaced at a uniform rate from a gas burette to the air


The inlet and outlet pipes for the dynamic feed system can be closed

near the chamber by use of two disk valves. The outlet valve at one

side of the chamber is shown in Plate V. The inlet valve is of similar

construction and is located at the top center of the chamber. The

valves are made of one-quarter inch thick plexiglass and are coated

with PVC. Each consists of a disk, held between two flanges, which

can be oriented so that holes line up with the inside of the pipe to



''"""~ ""~ ~ .r



allow completely unobstructed flow or partially restricted flow. One

position on the disk is a blank, and when this is placed in the line of

flow of the pipe the valve is completely closed. The principle drawback

to this valving system is that it takes two or three minutes to change

a valve position. As can be seen in the photograph the disk is held

tightly in place by four bolts.

The exhausted air and gas mixture from the chamber is conducted to

a laboratory hood and exhaust system and is eventually disposed to the

atmosphere from a chimney on the roof of the building. Arrangements can

be provided for cleaning the waste gases before disposal. Gas cleaning

before disposal was not necessary, however, due to the low concentrations

and small quantities of pollutants used in this work.

When the chamber is to be operated under static conditions, fresh

air is flushed through it. All openings are sealed and the desired

quantity of study gas is injected into the chamber through a plastic
membrane attached to one of the sample ports.o, The air-gas mixture is

then stirred with the fan blade until the chamber. contents are uniform-

ly mixed.

The fan blade is mounted on a shaft extending down from the top of

the chamber, and the mounting is arranged so that the blade can be

raised or lowered in order to achieve optimum mixing within the chamber.

The fan shaft turns in a Teflon plastic bearing at the point of entry

into the chamber, and this bearing affords a tight seal against gas

leakage from the chamber. The drive motor for the fan was mounted

outside the chamber in order to eliminate any interference the motor

might produce on the study gas concentrations.


Gas Feed System

A volumetric hypodermic syringe with needle has proven to be one

of the simpler devices for admitting a small measured quantity of gaseous

contaminant to the chamber. Plate VI shows the apparatus involved in

transferring sulfur dioxide from the compressed state in the cylinder

to atmospheric pressure conditions in the syringe. The gas is allowed

to flow from the cylinder and regulator through a short length of

tygon tube and then into a series of scrubbers before being allowed

to escape to the exhaust air system. The hypodermic needle is inserted

into the tygon tube and the syringe is allowed to fill with gas. During

the loading operation the plunger is worked back and forth a few times

to insure that air is purged completely from the hypodermic syringe.

The hypodermic syringe is then filled and ready to deliver a measured

quantity of gas to the static chamber. The tygon tubing connected to

the gas cylinder serves as a satisfactory membrane through which the

hypodermic needle may draw gas. The needle makes only a small hole

that practically seals itself when it is removed from the tubing.

Although this gas feed method is simple and convenient, it has a

few dra;.backs. One problem encountered in making an accurate calculation

of the quantity of gas in the hypodermic syringe is that of finding the

temperature of the gas. Although both gas cylinder and static chamber

are at room temperature, the gas is decompressed and cooled on its

release from the cylinder. Actual measurement of the gas temperature

as it enters the hypodermic syringe is difficult, and calculation of

Lhe theoretical temperature drop is complicated by problems of heat

transfer as the gas flows through the valves and tubing. A satisfactory



- A,..






solution to this problem is achieved by loading the syringe with test

gas 60 minutes before use, and allowing both syringe and gas to come

up to room temperature. This 60 minute time was found adequate to warm

a thermometer bulb to room temperature when it was sealed in a 10 cubic

centimeter hypodermic syringe at a temperature differential of 15 degrees

centigrade below room temperature.

During the temperature equalization period a cap is placed over the

hypodermic needle to reduce leakage. The temperature increases in the

hypodermic and causes an expansion and release of gas from the syringe.

This expansion helps prevent any dilution of the test gas. Finally, a

measured volume of test gas at atmospheric temperature and pressure is

injected into the chamber through a plastic membrane.

The measured volume of gas should be translatable into weight units

by use of the ideal gas equation:

PV = &- RT
where P = gas pressure in atmospheres

V = volume of gas in cubic centimeters

g weight of gas in grams

MW = molecular weight of gas

R gas constant in units of cubic centimeter atmospheres
mole degrees K
T temperature in degrees Kelvin

In the case of sulfur dioxide the calculated weight of a measured volume

of gas was found to agree very closely with the weight determined by

analytical means. When hydrogen sulfide was used, the weight determined

by analysis was somewhat less than the calculated weight. This difference

may possibly be due to use of an impure gas or to dilution of gas during


the test procedure.

It was found that the weight of hydrogen fluoride found by analyti-

cal methods did not agree with the calculated weight for a given volume.

This discrepancy is most likely related to the closeness between the

boiling point of this gas and the temperature at which it is used. When

it is under one atmosphere of pressure, hydrogen fluoride boils at 19.4

degrees centigrade, and it was used at temperatures varying from 21 to

25 degrees centigrade. It is possible that during loading and use of

the hypodermic syringe a slight pressure change could liquefy a portion

of this gas. A pressure of less than three pounds per square inch above

atmospheric pressure is sufficient to liquefy the gas when it is at 25

degrees centigrade. Consequently, it is necessary to take more than

customary care when using this gas.

In order to have a reasonably accurate determination of the hydrogen

sulfide and hydrogen fluoride admitted to the chamber, the gas feed

method was modified as follows. After the gas had adjusted in the

hypodermic syringe to room temperature, the first two cubic centimeters

were injected through a plastic seal into a 500 millileter volumetric

flask containing a scrubbing solution. The flask was shaken vigorously

for a few minutes to allow the gas to react with the solution. The

solution was then diluted to the propel volume and analyzed. The next

portions of gas in the hypodermic syringe were injected into the static

chamber for experimentation, and the last two cubic centimeter portion

was treated exactly as the first. The gas injected into the chamber was

assumed to have the same weight to volume relationship as the average of

the first and last samples analyzed from the hypodermic syringe. The


difference between these two samples was found to be negligible, which

suggests that practically no change occurs in the gas concentration

during its storage of less than two hours in the hypodermic syringe.

However, it was also observed that the weight of hydrogen fluoride

found by analysis of a given volume varied from one hypodermic syringe

loading to the next. It may be concluded that a reproducible volume

of hydrogen fluoride may be metered into the static chamber with a

hypodermic syringe, but it is necessary to analyze a sample of each

hypodermic syringe load to determine the weight of gas contained


Units of Measure

Practically all of the experimental and theoretical work in this

study deals with quantitative measurements of recovery and loss of

various test materials. When test materials are used that are normally

in the gaseous state, concentrations are reported in terms of micrograms

of contaminant per cubic meter of air unless otherwise specified. This

system of units is recommended by Terraglio and Sheehy,71 and is readily

usable. In chemical analysis, contaminants are normally reported in

micrograms, and the sample volume can be readily converted to suitable

metric units of volume. Since the units of measure of parts per million

by volume (ppm) are also used in this study, Table II is presented

to show a comparison between these two systems of measure for a number

of common gases.




Temperature = 25 degrees Pressure = 760 millimeters
centigrade of mercury
Gas Parts per million Micrograms per
by volume cubic meter

Sulfur dioxide 0.1 262

Hydrogen sulfide 0.1 139

Hydrogen fluoride 0.1 82

Nitrogen dioxide 0.1 188

Carbon monoxide 0.1 115

When a gas volume is used, it is necessary to describe the volume

further with regards to temperature and pressure. Since this work is

done exclusively under laboratory conditions, the most commonly occurring

pressure and temperature are used as standard conditions. These are a

pressure of 760 millimeters of mercury and a temperature of 25 degrees

centigrade. Any deviation from these figures can, of course, produce

a change in reported concentrations. However, a three degree temperature

change or a 7.6 millimeter change in barometric pressure would produce

only 1 per cent change in results. Although temperature and pressure

conditions are recorded during each experiment, correction for these

conditions are not employed unless the aggregate deviation becomes

significant with respect to the total experimental error.

In order to report findings in micrograms per cubic meter it is

necessary to determine weight of contaminant admitted to the chamber,

volume of chamber, and volume of sample removed. It is also necessary


to perform a quantitative analysis of the contaminant in the sample. A

method for metering the contaminant to the chamber was described earlier,

and the chamber was determined by construction and measurement to contain

216 cubic feet or 6,120 liters.

Flow Measurements

The volume of sample removed from the chamber can be determined in

a number of ways. Both dry gas meters and variable orifice flow meters

are used in this study.

The dry gas meters used are Zephyrs manufactured by the Sprague Meter

Company of Bridgeport, Connecticut. These meters were calibrated in the

laboratory by displacing a measured quantity of water from a container.

Figure VI shows a calibration curve for one of the meters used in this

study. Performance of this calibration was useful in determining the

expected accuracy to be achieved under various operating conditions with

this type of meter.

The data in Figure VI represent tests in which the meter was in

steady operation before and during time of measurement, and the volume

of air measured was 26.9 liters or 0.947 cubic feet. By having the

meter in motion at the beginning of a test, lag due to mechanical inertia

of the meter mechanism was eliminated. Test results under these condi-

tions are reproducible within about plus or minus 1 per cent.

Temperature apparently does not have a marked effect on this type

of meter. Some test runs were performed with the air temperature at

five degrees centigrade and reproducible results were achieved that

varied 3 per cent from those achieved at normal laboratory conditions

of about 26 degrees centigrade. The comparisons were made after the air



2 3
> i
4 :E
-4 3
N~ <



.-' co o
co 0


O co

S*H *r4
S < <







(1o3o.0 uoToaiaio3) auD1oA paPa91i/aumn OA lenjoV


014 Q0 cO
ON 0 0C -
C;\ 1\ O O O -
O 0 rO re i


volumes in both tests were adjusted to standard conditions.

Variable orifice flow meters produced by Fischer and Porter under

the trade name of Flowrator were also used in this study. Their advant-

ages are several in that they are easily placed in service, can be used

over a very wide range of flow, have predictable calibrations, and can

be used to measure small volumes. One drawback is that it is necessary

to know accurately the temperature and pressure conditions under which

the Flowrator readings are made in order to predict an accurate cali-

bration curve. Another is that the Flowrator gives a non-cumulative

measure, and it is necessary to maintain the float at a set position for

a measured time in order to be able to calculate the total quantity of

air which has passed through the Flowrator.

The normal sampling sequence is for the air and contaminant mixture

to flow from the chamber through the sampler, then through the meter,

and finally through the exhaust pump. Determination of the pressure and

temperature conditions of the air at the Flowrator is complicated by the

use of this sampling sequence. The drop in air pressure that occurs when

air flows through various sample trains was determined experimentally.

Figure VII shows the pressure drop that occurs for two different types

of samplers when operated over their usual ranges in this study. It can

be seen that the pressure drop for air flowing through the sampler plots

as a straight line when compared with the flow rate. This is only true

for the flow rates shown, and as the flow becomes greater the plotted

line curves to the right representing an increasingly great pressure drop

for a uniform increase in flow.

It was also determined experimentally that air passing through the




-4 w

0 w



t 4J


CC "

o0 0
C1 4- "


(9-OZ-W/l-aZ Iaora-oMOl) BuTpvag ale-as aqnj,


Hendrickson sampler at an intermediate flow of seven liters per minute

was reduced in temperature approximately three degrees centigrade. This

represents the net temperature change due to reduction of the air

pressure and due to the increase in moisture content of the air after

passing through the aqueous scrubbing solution. It is considered that

the air is saturated with moisture after it passes through the sampler.

Figure VIII shows the calculated calibration curve for Flowrator

2F-1/4-20-5. The temperature and pressure conditions prevailing at the

Flowrator were used in the calibration calculations. These conditions

were found by applying the above-mentioned corrections to a standard

laboratory atmosphere.

The points immediately to the left of the curve represent experi-

mental calibration data. These data are based on results achieved by

placing the Flowrator in series between the vacuum pump and sampler,

and by comparing Flowrator readings with volume determinations furnished

by a gas test meter ahead of the sampler. A number of individual tests

were run to develop the experimental curve, and on the basis of these

tests it appears that individual results deviate from the average by

about plus or minus 1 per cent. In this study the calculated calibra-

tion curve was used for all Flowrator readings performed under condi-

tions similar to those described for the curve.

When a gas meter is used in the sampling train between the vacuum

pump and the sampler, results are found to vary from those achieved

under laboratory atmospheric conditions. This is due to the reduced

pressure existing at this point in the sample train. Test meter number

2 was calibrated in the sampling train and the correction factor to be



- v4

a 3 u
n w

H -

H fH 9

u -1

c P. 0

4- -o

1 a a4

OQ) -

*4 4 >
*0 *

04 4 4) t*
I a -41

S Ow

00 0
t-o 3 *o 5o

:I 0 0 .
o, 00 a
C-u to \
'JO 4.4

0 fl 0 0a

* 9 LA i ) ".4 -

00 o0'
c ^ cO

8 ^ra

0 co t0 t N 0
CN -4 -4 -4

SuTpeOa alTVS aqnj;


applied when using this meter is shown in Figure IX. It can be seen

that the correction factor increases with increased flow, and this is

in agreement with the results of Figure VII which show an increased

pressure drop with increased flow.

Sampling Instruments

The principal sampling instrument used in this work is the Hendrick-

son sampler. This consists of a cylindrical glass frit which is sub-

merged in the scrubbing solution and through which the sample must pass.

Plate VII shows two views of this device, and the desirable features can

be readily seen. The side arm cup allows for easy and rapid addition

of scrubbing solution, and the funnel shaped bottom with hose and pinch

clamp allows for rapid and complete removal of sample. This sampler is

ideally suited for laboratory work, but the amount of glassware appended

to it makes it rather delicate, and it requires careful handling for

field work. The device was developed by Dr. E. R. Hendrickson at the

University of Florida. The particular sampler used is identified by

number in this work because each piece of equipment has its own partic-

ular characteristics regarding flow and pressure drop. This is due to

the variation in glass frit used, although all of the samplers are

fitted with coarse frits with pores averaging 90 to 150 microns in size.

The collection efficiency of Hendrickson sampler number 3 was

determined experimentally for the test gases used in this study. These

efficiency determinations were made by placing a number of samplers in

series and determining the per cent recovery in the first sampler com-

pared to the total recovery in all of the samplers in the train. The

determination was made at only one flow rate of seven liters per minute



9 co


N u 0 z
0 a ..

0 0

:o 0 o

u 1.4 0 V4
c0 w B10 >

2.CO N \4U
q *U 0 \0
S a < O \ 4 5

0 w 00

d r. @ \
N O i a \

* *.P .\
0 O \ u N
0 1
0 0 0 \

a E10 0

(B 0 a

0 CO 00 c0 cc o

(.Ioloa uoTU39i0o3) a)mlOA palasa/;unnlOA irnlVo

tI .







since this is the flow rate used almost exclusively throughout this

study. The volume of scrubbing solution in each test was 100 milliliters,

and the total volume sampled varied from 70 to 105 liters.

When 0.1 molar sodium tetrachloromercurate was used as the scrubbing

solution for sulfur dioxide, there was a 99.7 per cent recovery in the

first sampler. When a 2 per cent solution of zinc acetate, acidified

with three drops of acetic acid per liter of solution, was used to

collect hydrogen sulfide, complete recovery was achieved in the first

sampler. When 0.1 molar sodium hydroxide was used to collect hydrogen

fluoride, there was a 98 per cent recovery in the first sampler, 2 per

cent in the second, and nothing in the third.

Analytical Methods

Sulfur dioxide, hydrogen sulfide, and hydrogen fluoride are used

in this study as test contaminants. The choice of contaminants was

based on both availability of the gas and availability of suitable

analytical techniques, as well as the variation in chemical properties

offered by these gases. They also represent gases frequently encount-

ered as air pollutants.

The analytical method used for determining sulfur dioxide is

similar to that described by West and Gaeke.78 In this method sulfur

dioxide is withdrawn from the chamber through a fritted glass sampler.

The scrubbing solution is 0.1 molar sodium tetrachloromercurate, and

the sulfur dioxide reacts to form a relatively stable, nonvolatile

solution of disulfitomercurate. This reaction as stated by West and

Gaeke is as follows:

HgCl4'-+ 2S02 + 2H20 = Hg(S03)2 + 4 Cl1 + 4 H+


The determination of disulfitomercurate is based on the red-violet color

that is developed when P rosaniline hydrochloride'- hydrochloric acid

mixture and formaldehyde are added to the sampling solution. Samples are

analyzed colorimetrically and values are compared against a standard


The standard curve was made by measuring the light transmittance

through 10 milliliter samples containing sodium tetrachloromercurate

solution plus known quantities of sulfur dioxide and one milliliter

each of the dye and formaldehyde. When a standard curve was made from

freshly prepared sodium tetrachloromercurate solution, unsatisfactory

results were achieved. In the main body of the curve the light trans-

mission was proportional to the amount of sulfur dioxide present, but

the curve did not pass through the zero axis of the graph as it should.

After the sodium tetrachloromercurate solution had aged for three or

more days a satisfactory standard curve could be achieved which plotted

through the zero coordinates on the graph. Evidently it is necessary

to age the solution a few days in order to allow for complete equilibrium

to be achieved between the reagents (sodium chloride and mercuric

chloride) employed in the solution.

The sulfur dioxide analysis is easily performed. The only common

interfering substance is nitrogen dioxide in concentrations greater

than 2 ppm, and results appear to be reproducible within about plus or

minus 2 per cent. The color development of samples has been found to

vary with temperature, so that as far as possible the samples are all

analyzed at the same temperature. All readings are made after 30

minutes of color development.


Hydrogen sulfide concentrations were determined by the methylene

blue method. The procedure described by Fogo and Popowsky18 was first

employed but only a small percentage of the hydrogen sulfide calculated

to be in the chamber was actually recovered. The procedure described

by Sands and others57 was then employed with considerably more success.

The principal difference in the two procedures lies in the pH of the

scrubbing solution. In the Fogo procedure the zinc acetate solution

is made basic with sodium hydroxide, and in the Sands method the

scrubbing solution is slightly acidified.

The Sands procedure may be described as follows: Hydrogen sulfide

is removed from the air sample in a fritted glass sampler by bubbling the

gas through a solution of 2 per cent zinc acetate to which has been

added three drops of acetic acid per liter of solution. Hydrogen sul-

fide reacts with the zinc acetate to form a fine precipitate of zinc

sulfide. At the end of the sample run the lower portion of the sampler

which contains 100 milliliters of zinc acetate solution is placed in an

ice bath for five minutes. After the solution has cooled, 10 milliliters

of P aminodimethylaniline sulfate in acid solution are admitted to the

sampler, and 2 milliliters of ferric chloride in acid solution are

added. The dye reacts with the acid-liberated hydrogen sulfide to

form methylene blue. The solution is drained from the sampler, and the

sampler is rinsed with distilled water. The sample plus rinse water are

then diluted to 250 milliliters. After color development is complete,

optical density of the methylene blue is measured on the spectrophoto-

meter, and the weight of sulfide in the sample is read from a standard



The solution was originally cooled in order to have the final so-

lution achieve a temperature of approximately 25 degrees centigrade.

This is brought about by liberation of heat when the acidified reagents

are admitted to the scrubbing solution. The reaction is temperature

dependent, and it was found that by timing the exposure to the ice bath

the final temperature could be readily controlled.

The composite reproducibility of the sampling and analytical

technique was determined experimentally. When three successive samples

were taken from the chamber and analyzed, the average deviation for the

samples was found to be 3 per cent. This evaluation was performed on

100 milliliter samples containing approximately 60 micrograms of hydro-

gen sulfide.

There are a number of factors that could have bearing on the accu-

racy of this method. Since hydrogen sulfide recovery was complete in

the first of two samplers in series, sample efficiency should not affect

accuracy. Another possible limitation on the accuracy of this method

lies in the ability to produce an accurate standard curve. In the

standardization procedure hydrogen sulfide is bubbled through distilled

water, and a small aliquot of the hydrogen sulfide water is put into

the zinc acetate solution. This comprises the standard solution.

Immediately after removing the aliquot a measured quantity of iodine

solution is admitted to the sulfide water, and this in turn is back

titrated with sodium thiosulfate. The equivalents of hydrogen sulfide

are then calculated. In this procedure any delay between removal of

the 10 milliliter aliquot and titration of the hydrogen sulfide water

can allow for escape of hydrogen sulfide from the water. Each time the


standard solution is handled there is a real possibility of loss of zinc

sulfide due to adherence of the precipitate to the walls of the glassware.

As the precipitate was found to adhere tenaciously to glass surfaces, the

sample analysis was performed in the original sampling device. Finally,

reagent addition to the zinc sulfide must be done with the minimum amount

of aeration, because the sulfides can be volatilized by aeration as well

as oxidized, thus eliminating them from the quantitative results of the


The analytical method used for determining fluorides is similar to

that described by Megregian.44 Although this is a colorimetric method,

it is unlike those mentioned earlier for sulfur dioxide and hydrogen

sulfide. The fluoride ion is not directly involved in the color develop-

ment. A standard color is developed with eriochrome cyanine R dye in

the presence of zirconyl chloride octahydrate in acid solution. The

fluoride ions present react with the zirconyl ions to form a complex,

and this in turn restrains some of the zirconyl chloride octahydrate

from entering the color-forming reaction. The amount of color reduction

from the standard color is proportional to the amount of fluoride present

in the solution.

Since there are a number of ions that can interfere with results,

field samples must undergo a distillation process to separate fluoride

from these substances. The distillation may be performed by distilling

hydrofluosilicic acid from a solution of the sample in a higher boiling

acid. Some of the common interfering ions are sulfate, chloride and

magnesium. In the current tests hydrogen fluoride gas of known purity

is used, and since there is little chance for contamination with


interfering substances, the distillation step is omitteI.

In this study 100 milliliter samples of 0.1 molar sodium hydroxide

are used to scrub the fluoride in a fritted glass sampler. Fifty-

milliliter aliquots are then treated with three milliliters of eriochrome

cyanine R solution and three milliliters of zirconyl chloride octahydrate

solution. The color developed is read on the spectrophotometer and

plotted against a standard curve. Readings are taken after 30 minutes

of color development and temperatures are maintained at 25 degrees

centigrade plus or minus two degrees.

It is of interest to note that considerable rinsing of the sampler

frit is necessary in order to recover the fluoride in the sample solution.

The hydrogen fluoride used in these tests probably reacts very rapidly

with the sodium hydroxide on the upstream side of the wet frit in the

sampler, and when sampling is complete a large proportion of the sodium

fluoride is entrapped within the frit.

Seven samples were taken one after another from the chamber and

were analyzed in order to evaluate the composite reproducibility of

the sampling and analytical method. When samples were taken from the

chamber containing a concentration of 310 micrograms of hydrogen fluoride

per cubic meter, the average deviation was found to be 4.8 per cent.

The extreme deviation was 13 per cent. Consequently the reproducibility

of-the hydrogen fluoride determination is not as good as that for

sulfur dioxide and hydrogen sulfide.


In order to maintain a known atmosphere in the chamber it is neces-

sary to control or know the extent of the various losses that may occur.

The first potential loss to be evaluated is that of leakage when the

chamber is maintained at a pressure different from that of the ambient


The information to be sought in this test is the volume of air

lost during a given length of time when the chamber is maintained at

a given pressure. By careful control of entering air and control of

samples withdrawn from the chamber during regular operation it is

possible to maintain nearly atmospheric pressure in the chamber, and

consequently most operations are conducted with a pressure differential

of not more than plus or minus 0.2 inches of water. The leakage tests

were carried out with pressure differentials of 0.1 to 0.7 inches of

water, and this range more than adequately covers the expected limits

of chamber operation.

A water-filled manometer was used to measure the difference in

pressure between the chamber and the laboratory, and a metered supply

of air was used to build up the chamber pressure. Fluctuations in

barometric pressure during periods of the test were taken into consider-

ation. Air temperature variations during the tests also had to be con-

sidered and temperature readings were taken from two thermometers. One

was placed approximately six inches from the top of the chamber, and

the other was ten inches from the bottom. Both were four inches in from



one side wall. The air was circulated by a fan blade within the chamber

until the two thermometers agreed to the nearest 0.2 degrees centigrade.

When the first leakage test was run, it became apparent that even

a slight increase in the chamber pressure produced a measurable displace-

ment of the chamber walls, causing the chamber to act something like a

large bladder, Figure X shows the increase in chamber pressure that

occurs with a given addition of air. It can be seen that the addition

of one cubic foot of air to the chamber produces a pressure increase of

only 0.42 inch. It is calculated by the ideal gas equation that the

addition of one cubic foot of air to the chamber would produce a pressure

increase of almost two inches if there were no bulging of the walls.

An important feature of the chamber flexibility is the ability to expand

or contract and accommodate sudden gas withdrawal or entry without pro-

ducing such large pressure changes as to create marked metering inaccu-

racies or to increase leakage pressures.

The break in the lines in Figure X occurred at points where there

was an audible buckling of the chamber due to pressure. These wall

deformations occurred abruptly, but were small and did not cause damage.

The amount of leakage which occurred over a given period was deter-

mined by substituting experimental data in the ideal gas equation.

This states that for a given mass of gas the pressure multiplied by

the volume and divided by the absolute temperature is equal to a con-

stant. This can be expressed as follows:

PVl1 P2V2
TI T2 (I)

o 0

0" :0
o0 \ 44
-o \ 0 e r

o\ o
o\ f





(aqureqD uT dn jiTng asnsgaj





C !1





I I "

(alazt jo saxul)


where P = absolute pressure in inches of mercury

V = volume of gas in cubic feet

T = absolute temperature in degrees Kelvin

subscript I initial condition

subscript 2 final condition

By rearranging,

V2 V1 x 1 x
P2 T1 (2)

The initial and final experimental data are selected such that the cham-

ber volume is the same initially and finally. This is done by selecting

data for comparison that show the same guage pressure on the manometer.

In other words the data are selected so that V2 is equal to V1. If the

mass of gas is the same initially as it is finally, a substitution of

experimental data in equation (2) should verify that V2 is equal to V1.

However, if leakage occurs, the mass of gas is no longer a constant,

and substitution of experimental data in equation (2) should show that

V2 is not equal to V1. The calculated difference between V2 and V1 is

equal to the volume of air that has leaked from the chamber.

Figure XI shows the results of this leakage experiment. It shows

the rate of gas leakage from the chamber for varying pressures. As is

to be expected, the leakage rate goes up with increased chamber pressure.

If operating conditions are carried out as planned at pressure heads of

less than 0.2 inches of water, then loss due to leakage should be on the

order of 0.1 cubic foot per hour, or 1 per cent of the chamber volume

per day.

o 0 0 0

(.aeiuM Jo sequuI) amssa9id zaquinq3








a N



This series of tests was conducted to determine the time required

for a uniform gas concentration to be achieved in the chamber. The

problem is that of injecting a known quantity of gas into the chamber

for study and then mixing the gas with the chamber air until the mixture

achieves a uniform concentration. After a uniform mixture is achieved,

the chamber contents are ready for use.

The problem of mixing may be considered for both dynamic flow con-

ditions and for static chamber conditions. It may also be considered for

particulate matter as well as aerosols. In this particular test series

only a gas-nixing rate in the static chamber is evaluated. Of course

aerosols would exhibit a loss rate due to settling, agglomeration, and

impingement on surfaces that would be different from gases, but this

problem should not invalidate an assumption regarding the time required

to achieve a uniform mixture in the chamber. Consequently, it can be

considered that an aerosol will mix in the chamber as rapidly as will

a gas.

A series of sample ports were placed in one side wall of the chamber

from which samples could be taken to determine the degree of mixing. The

sample ports were located as shown in Figure II. The locations were

selected to give a representative sample of the gas concentration near

the wall. Since the gas in the chamber was in motion due to the fan

action, it seems reasonable to conclude that when the sample concentra-

tion becomes uniform among the various sample positions, and when the



samples are also uniform in concentration with respect to the same posi-

tion for successive samples, the entire chamber content must be uniform.

The mixing-fan blade is mounted on a shaft extending 28 inches down

from the top center of the chamber. The blades are mounted in a plane

parallel to the axis of rotation with no set pitch, so that their rota-

tion produces a centrifugal force causing an air mixing pattern that is

estimated to look like that shown in Figure XII. The two-blade fan has

a diameter of 16.5 inches and a blade width of 3 inches. By use of a

stroboscopic light it was found that the fan turned about 240 revolutions

per minute. This produced a blade-tip speed of approximately 29 feet per

second. Although the blade motion has no effect on the gas other than

to mix it, the blade could be a source of loss for particulate material

by virtue of impingement of aerosols on the blade. Some researchers

have resorted to large blades in order that adequate mixing can be a-

chieved with blade tip speeds below 15-20 feet per second. Above these

speeds impingement of aerosols may become a problem.

The original test procedure was developed as follows: The chamber

was purged with fresh, dry air, and a measured quantity of the test gas

was admitted to the chamber. In this case sulfur dioxide was injected

with a hypodermic. The fan was started and samples were withdrawn

simultaneously from each of the six sampling ports. Fresh make up air

was admitted as necessary to replace the volume of samples withdrawn.

Additional samples were taken at successive time intervals.

When the samples were analyzed it became apparent that a substantial

variation was reported not only at the initial stages of mixing but also

after the chamber contents had been mixed for two hours before sampling.





scale 1 in. = 1.5 ft.


The average concentration for this mixture was 1.12 ppm sulfur dioxide,

but the lowest and highest figures varied from the average approximately

7 per cent. A further test was made by using only one sampler and meter

and taking successive samples at the different ports. Uniform results

were achieved in this manner. It was then apparent that a substantial

variation existed in the reproducibility of results among the different

sample trains. Extensive meter calibration and sampler recovery rate

studies would be required before this test scheme could be used satis-

factorily. A schematic diagram of this simultaneous sampling scheme is

shown in Figure XIII. The capillary tubes were placed in the system to

help achieve a more equal flow between the different sample trains.

The second and more successful approach was to use only one sampler

and one flow meter operating at a fixed rate. This sample train was

used to collect all of the samples. In this system only the sampling

tube was moved from one sample port to another. Figure XIV shows a

schematic diagram for this system. By checking this procedure on a

one-day old mixture of sulfur dioxide and air that was undoubtedly

uniformly mixed, it was found that for six samples the maximum deviation

from the average concentration of 0.81 ppm was plus or minus 1.25 per

cent. In this case a 0.90 cubic foot sample was taken each time and

the sampling time was 2.5 minutes.

The principal difficulty with this second approach, in which each

position is sampled in sequence, is the time lag involved between samples.

In order to reduce this lag as much as possible, the sample time was

reduced to one minute and the time between samples was reduced to the

minimum required to prepare the equipment. When successive samples were




Sample Ports


Flow Meter

Capillary Tube





Sample Ports


Flow Meter




taken from different ports, the preparation time between samples was

about one and one-half minutes. Since the sampling time was limited to

one minute, the reliability of results was necessarily reduced. This

may be explained by the fact that the timing errors involved in starting

and stopping the sampling operation are of a fixed nature of perhaps

plus or minus one-half second at the beginning and plus or minus one-

half second at the end of the operation. This would account for a

maximum timing error of about one second, which percentage-wise would

be two and one-half times larger in a one minute sample run than in a

two and one-half minute sample run.

S Figure XV shows how this sequence sampling was used to determine

the time required to achieve a uniform gas mixture in the chamber. The

middle set of points in the figure shows the concentration of sulfur

dioxide found at different sampling positions on the chamber and for

different periods of time during the mixing process. It can be seen

from these data that the concentration found at position four during

the tenth minute of mixing was very close to the average concentration

found for the remainder of the samples taken. Further, the concentrations

found at successive ports were found to check fairly closely with the

average concentration.

Additional tests were performed with different concentrations of

sulfur dioxide. These are shown by the upper and lower set of data in

the figure. The time between samples could be reduced from one and one-

half to one minute if the successive samples were all taken from one

port rather than different ports. The data at the bottom of Figure XV

were taken at position number one, and the data at the top of the figure


Ed ^

g <<
5 i

u E



2I 4

I .

1 1





, I

CN 0 c0 0 -4 C 0
N CN r-4 -4 -4






I I .


CO 0 N* C

(.laai" a33 s8ureBSoin:) uotvj~liuuaouoo svD

N 3


4 *




were taken at position number five. In all three tests the results

achieved their average uniformity near the tenth minute of mixing.

From examination of these data in Figure XV it appears that samples

taken after the tenth minute of mixing in the chamber will furnish results

of a uniformity consistent with the accuracy of the sampling technique,

and that the gas concentration is essentially uniform after being mixed

in the chamber for ten minutes.

The slope of the line of best fit for the various runs can be

accounted for in part by the dilution of the gas as samples are replaced

with fresh air. This loss of concentration will be evaluated more

completely in the section of the study dealing with decomposition rates.




It has been found that gas concentrations in the test chamber are

less than the calculated concentration of gas admitted to the chamber.

This loss of gas concentration has been observed by a number of investi-

gators as previously mentioned in the review of literature. If the gas

losses due to leakage and chemical reaction are eliminated or accounted

for, the remaining loss in concentration should be due to adsorption.

This section of the study deals with an evaluation of adsorption. The

variables affecting adsorption are considered, and quantitative estimates

of this phenomenon are made for some gases.

Adsorption in this study refers to the condition in which the con-

centration of a pollutant gas is higher at the chamber surface than it

is in the bulk of the gas mixture.


Adsorption theories have been developed by Polanyi, Williams,

Zrigmondy, Magnus, Langmuir and others.24 The following discussion is

based on the Langmuir monomolecular layer theory and the empirical

Freundlich isotherm. The Langmuir theory covers the principal features

of adsorption of a gas on a solid which are salient to this study.

Langmuir34,35 discovered that when gaseous molecules impinge on a

solid or liquid surface, most of them do not rebound from the surface.

Rather, the molecules condense and are held on the surface by forces



similar to those that hold molecules together. The gas molecules remain

on the surface until the energy of vibration exceeds the amount required

for evaporation. If these surface forces are weak, or if the vibrational

energy of the molecules is great, the average duration of the molecule

on the surface is short. Conversely, if the surface forces are strong,

or the vibrational energy of the molecule is low, the average duration

of the molecules will be long. Adsorption then, is directly related

to the time lag between condensation and subsequent evaporation of mole-

cules from the surface.

Langmuir calculated that the forces which hold gas molecules to the

surface decrease rapidly with increased distance from the surface. Since

these forces become practically nil at a distance slightly greater than

one molecule, it was concluded that for a plane surface the adsorption

layer would not normally exceed one molecule in thickness. Since valency

forces holding atoms together in a chemical compound, and similar forces

holding atoms together in a crystal, emanate from all sides of an atom,

there will be unbalanced or available forces at the crystal surface which

can condense and hold gas molecules that strike the surface.

The Langmuir adsorption isotherm is based on the concept of equi-

librium between gas molecules striking a surface and those evaporating

from the surface in a given time. The derivation of this isotherm is

shown in many books on physical chemistry and will not be repeated here.

The isotherm may be written as follows:

x klk2P
m 1 + k1 p (I)

where x quantity of gas adsorbed


m = quantity of adsorbent

k1 and k2 = constants for a given system

p = pressure of gas

At low partial pressure of the gas, or where the surface is a poor

adsorbent, equation (1) can be reduced to the following expression:19

S, K p (2)
This equation states that the amount of adsorption which occurs is equal

to a constant times the pressure of the gas. This equation is valid at

a fixed temperature (hence the term isotherm) and it applies only at

low gas pressures when the fraction of surface covered with gas molecules

is relatively small. The constant K is specific for a particular ad-

sorbent surface and a particular gas.

It has been calculated in Appendix X that when the test chamber has

a sulfur dioxide concentration of 2,620 micrograms per cubic meter of

air (1 ppm), not more than 7 per cent of the nominal chamber surface is

covered with sulfur dioxide molecules. The nominal chamber surface was

determined from the chamber dimensions. The true surface area available

for adsorption, that is the surface area produced by microscopic and

submicroscopic irregularities in the surface coating, is undoubtedly

greater than the nominal surface used in these calculations. Consequent-

ly, this estimate is conservative, and the actual fraction of surface

covered at the stated concentration is less than 7 per cent.

The Freundlich or "classical adsorption isotherm" has been found

to describe adsorption at intermediate gas concentrations. This ex-

pression has been derived empirically and is expressed as follows:19


x = K pn
m (3)

where n = a constant

For high gas concentrations the expression becomes:19

i = K
m (4)

The numerical value of the constant K depends on the units used and is

not the same in the different expressions.

The range of applicability of adsorption equations (2), (3), and

(4) can be shown graphically in Figure XVI. This shows the relation

between the amount of gas adsorbed and its pressure. The curve is di-

vided into three segments, and the equation which best describes each

segment of the curve is shown. At low pressures, adsorption varies with

the pressure. At intermediate pressures, adsorption varies with a

fractional power of the pressure, and at high pressures adsorption

becomes a constant. This is explained by the change that occurs in the

amount of adsorbing surface covered with gas molecules. At low gas

pressure there is a small quantity of gas adsorbed and consequently a

large fraction of the adsorbent is still available to condense additional

molecules. As the quantity adsorbed becomes greater, the surface avail-

able for further adsorption becomes smaller, and finally the entire

surface is coated with a unimolecular layer of gas molecules. The

particular curve shown in Figure XVI suggests that at this point no

further adsorption takes place, and the quantity of gas adsorbed is

independent of additional increases in the gas pressure. Equation (2)

appears to be most applicable to this study because of the relatively

low gas concentrations which are employed.


.0 0
o *0





M r




aoeins 0o 3Tun aaad paqjospy seo o 0CT3uvnb








The amount of adsorption which takes place is dependent on a

number of factors in addition to the partial pressure of the gas. The

amount adsorbed, almost without exception, decreases with an increase

in temperature. As the temperature increases, the vibrational energy

of the gas molecule increases, and makes it easier for the'molecule to

escape the attracting forces at the gas-solid interface.

Adsorption also depends on the area and nature of the adsorbing

surface. The ability to adsorb is very large per unit of weight for

porous adsorbents such as charcoal. The adsorbing area of the test

chamber on the other hand, has been kept as small as possible, and in

this respect a perfectly plane surface would be ideal. The history of

the surface may have some bearing on its adsorptive qualities. When

glass surfaces have become roughened through use, their adsorbabilities

have been increased. This is most likely due to the increased surface

area created when the glass surface is roughened.

It was considered in this study that repeated and continued exposure

of gas to the chamber surface might condition the surface and alter its

adsorptive properties. In order to evaluate this possibility a large

quantity of sulfur dioxide, 140 cubic centimeters, was admitted and

allowed to remain in the chamber for four days. At the end of this time

the chamber was flushed with fresh air and additional quantities of

sulfur dioxide were admitted. It was found that the ratio between ad-

mitted and recovered quantities of gas were the same before and after

the exposure of the surface to high concentrations of gas.

The nature of the gas plays an important role in determining its

adsorption characteristics. The adsorbabilities of different gases on


a given surface are related to a number of physical characteristics of

the gases. Gregg24 pointed out that adsorbability of a gas is related

to its solubility in water, to its condensability, and in turn to the

critical temperature and attraction constant (a) of the van der Waals

equation. In general, then, the most easily liquifiable and highly

soluble gases are the most readily adsorbed. It was also observed by

Langmuir35 that in a general way larger molecules are adsorbed more

readily than smaller ones.

Table III illustrates this relation between the adsorbability of a

gas and its critical temperature.

The time acquired to reach adsorption equilibrium has been reported

by Gregg24 to be rapid. This is especially true when plane adsorbents

are used. In experiments with mica as the solid surface it was found

that a number of gases reached equilibrium within three minutes. Since

ten minutes are required to achieve uniform mixing in this chamber after

the admission of a gas, and the surface is essentially a plane as opposed

to a porous adsorbent, it appears likely that adsorption equilibrium

would have been reached before samples were withdrawn. This assumption

is supported by experimental findings as follows: When samples are

taken one after another, there is found to be a decreased concentration

of gas in the chamber. If this decrease were due to further gas ad-

sorption on the walls, it would occur at a diminishing rate as adsorption

equilibrium is reached. However, it was found that the gas concentration

in the chamber decreases uniformly with respect to time. Therefore,

adsorption must be complete by the time the first sample is withdrawn

and the steady decrease in gas concentration must be due to something




Gas Volume Adsorbed, Critical
(milliliters of Temperature
gas per gram of (Degrees
charcoal) Kelvin)

Sulfur dioxide 380 430

Chlorine 235 417

Ammonia 181 406

Hydrogen sulfide 99 373

Hydrogen chloride 72 324

Nitrous oxide 54 310

Carbon dioxide 48 304

Methane 16 190

Carbon monoxide 9.3 134

Oxygen 8.2 154

Nitrogen 8.0 126

Hydrogen 4.7 33


other than adsorption.

When more than one gas was present in a mixture, Richardson and

Woodhouse51 reported that the presence of each gas depressed the adsorp-

tion of the other. This reduced adsorption is explained in light of the

Langmuir theory. The molecules adsorbed by a second gas will occupy

some of the adsorption space available on the surface. This in turn

reduces the space available and consequently adsorption of the first gas.

This phenomenon could require additional evaluation in the event that

two or more adsorptive gases are to occupy the chamber at the same time.

Experimental Test Methods

There are a number of methods for carrying out adsorption experi-.

ments. One of the most common of these is the volume method in which

the absorbent is placed in a sealed container. The container is evacu-

ated, and a known quantity of gas is admitted. The pressure of the

unadsorbed gas is measured, and its quantity is calculated. The amount

of unadsorbed gas is subtracted from the total quantity admitted to the

container, and the resulting figure is the amount adsorbed.

In this study known quantities of gas are admitted to the chamber

which contains air. The quantity of gas remaining in the chamber atmos-

phere after adsorption has occurred is determined by chemical analysis

of samples collected from the chamber. The amount of gas adsorbed is

calculated as the difference between the amount of gas admitted to the

chamber and the amount found to be remaining in the chamber after ad-

sorption has occurred. The amount of adsorption calculated in this

manner can be related to the nominal chamber surface and to the equil-

ibrium gas concentration in the chamber. This gas concentration can be


expressed in gas pressure terms and in this study varied from zero up

to a few millionths of an atmosphere.

The principal difference between the volume method and the experi-

mental method used in this study is that in the volume method, designed

expressly for adsorption measurements, the adsorbent surface is large.

with respect to the volume of the container, and consequently the volume

adsorbed is a relatively large fraction of the gas admitted to the con-

tainer. In this experiment the surface is relatively small with respect

to the volume of the chamber, and the quantity of gas adsorbed is a rela-

tively small fraction of the total quantity of gas admitted. Since the

quantity of gas adsorbed is a small number calculated by difference

between two large numbers, the experimental error is magnified in cal-

culating the quantity of adsorption. For example, if 100 units of a

gas are admitted to the chamber with a maximum expected error of plus

or minus 2 units, and 90 units are recovered after adsorption with a

maximum expected error of plus or minus 2 units, the difference between

these figures would be the adsorption, or 10 plus or minus 4 units.

This, then, would represent a maximum variation in the final answer of

plus or minus 40 per cent even though there was only about a 2 per cent

variation in experimental results. This hypothetical example closely

parallels the experimental findings for the sulfur dioxide adsorption


The test procedure for determining gas adsorption on the chamber

surface was as follows: a hypodermic syringe was loaded from a gas

cylinder. The loaded hypodermic syringe was capped and placed beside

the static chamber for approximately 60 minutes to allow its temperature


to come up to that in the chamber. The chamber fan was started, and a

sample was withdrawn from the chamber before any gas was admitted. After

this control sample had been taken a small quantity of gas, generally two

cubic centimeters, was injected into the chamber from the hypodermic

syringe. The injection was made through a thin plastic membrane attached

to one of the sampling ports. The gas was allowed to mix for twelve

minutes following which one or more samples were withdrawn. The sample

gas was drawn from port number four through a tygon tube to a Hendrickson

type sampler where the pollutant gas was scrubbed from the gas-air mix-

ture. The air was then drawn through a valve, a flow meter, and finally

the vacuum pump.

The sampling rate was held as uniform as possible for all of the

adsorption studies. This was done by adjusting the air flow indicator

to the same Flowrator tube scale point for each test. Although the

entire flow range for the Flowrator was calibrated for one set of pres-

sure and temperature conditions (Figure VIII), at this particular scale

reading a number of calibration points were determined for the range of

temperature and pressure conditions expected during the tests. The

actual flow was then determined based on the temperature and pressure

prevailing. The sampling time was varied from fifteen minutes down to

about four minutes depending on the gas concentration in the chamber.

The combination of sampling time and gas concentration was selected so

that the samples collected were of a strength that could be readily


Results and Discussion

The experimental concentration of sulfur dioxide in the chamber was


determined from the analysis of samples taken from the chamber. The

calculated concentration was determined on the basis of gas admitted to

the chamber. Figure XVII shows how the experimental sulfur dioxide con-

centration compared with the calculated concentration. The experimental

concentration was less than the calculated concentration, and the diff-

erence between the two was assigned to surface adsorption.

In order to do this, the implication must be that other possible

losses have been eliminated or accounted for. The losses in concentration

due to leakage and dilution when samples are removed have been evaluated.

The actual weight of sulfur dioxide admitted to the chamber was calcu-

lated by use of the ideal gas equation, and the quantity was also checked

experimentally. This was done by injecting a measured volume of sulfur

dioxide into a flask containing sodium tetrachloromercurate. The solu-

tion was then analyzed to determine the weight of sulfur dioxide in the

given volume. The weight found in this manner checked very closely with

that calculated by the ideal gas equation. Too, the Flowrator for

measuring air sample volumes was calibrated by calculation and checked

experimentally against a gas test meter.

The possibility still existed that a chemical reaction could occur

to account for the loss, or that there was a consistent bias in calcu-

lations or procedure. An experiment was carried out to evaluate these

possibilities and to establish the relationship between the loss in con-

centration and surface area. A frame was constructed to hold a number

of heavy cardboard panels, and the frame and panels were painted with

PVC similar to that on the walls of the chamber. The test scheme was

to determine the gas concentration existing when a measured quantity of



0 .0



-0 U

0 &

(JaaeW 3TqnD aad smeSoloT.i) uoq.ijjuqouoDo Tvuamiaadxz


gas was added to the empty chamber, and to compare that with the concen-

tration found when an identical quantity of gas was placed in the chamber

containing additional surface. If the loss in concentration in the cham-

ber is truly a surface phenomenon, then it should be possible to substi-

tute both sets of data in the adsorption equation (2) shown earlier and

achieve the same value for the adsorption constant K.

Appendix XII summarizes the calculations necessary to solve for

the constant K. The data for the calculations are taken from Appendix

XI, and each step in the summary of calculations is described in the

notes following that appendix.

The K constant in the adsorption equation (2) was solved for both

the empty chamber and the chamber with additional surface. The constant

was found to be 92 x 106 with a standard deviation of 21 x 106 based on

15 tests in the empty chamber. For the chamber with additional surface,

the constant was 95 x 106 with one standard deviation of 22 x 106 based

on 13 tests. On the basis of similarity of the K constant for the empty

and full chamber it appears likely that most, if not all, of the original

loss in sulfur dioxide concentration is due to surface adsorption. This

particular numerical value for the constant is applicable to the system

of sulfur dioxide gas and a plane PVC surface at approximately 25 degrees

centigrade and for sulfur dioxide partial pressures in air of 0 to 1.8

x 10-6 atmospheres. The units for this constant are micrograms of sul-

fur dioxide adsorbed per square meter of surface per atmosphere of


The practical significance of the adsorption constant lies in its

use in determining the quantity of gas necessary to achieve a stated


concentration in the chamber. Suppose, for example, one wished to

achieve a chamber concentration of 262 micrograms of sulfur dioxide per

cubic meter of air (0.1 part per million by volume). It is possible to

substitute this K constant in equation (2) as follows:


ug SO2 adsorbed
20.1 sq. meters surface

ug S02 adsorbed

Desired concentration

= (92 x 106)(0.1 x 10-6 atmospheres pressure)

= 185

= 262 ug x 6.12 cu. meters in chamber
cu. meter

= 1600 ug

Since 2,620 ug S02 occupy one cubic centimeter, it is necessary to add

(1600 + 185) ug
2620 ug SO2 = 6.8 cu. cm. of S02 @ 25 degrees centigrade
and one atmosphere pressure.
cu. cm.

For this particular chamber approximately 90 per cent of the calculated

sulfur dioxide concentration is achieved in the chamber, and there is a

10 per cent loss in concentration due to surface adsorption.

It is interesting to note the more pronounced effect that adsorption

would have for the same gas and surface material if a smaller chamber

were used. If the chamber were a cube with sides of two feet instead

of six as is the case in this study, only 75 per cent of the calculated

gas concentration would be achieved in the chamber. The rest of the gas

would be accounted for by adsorption.

The same general procedure was followed for hydrogen sulfide, but

only two sets of adsorption data and one usable set of aging data were

collected before this test series had to be terminated. Since another


project being carried on in the same laboratory emitted large quantities

of hydrogen sulfide to the laboratory atmosphere, contamination of sam-

ples and reagents occurred and these tests were stopped.

The two adsorption tests on hydrogen sulfide which were completed

showed a recovery of 91 and 97 per cent for an average recovery of 94

per cent. This would suggest that hydrogen sulfide is adsorbed on the

chamber surface to a lesser extent than sulfur dioxide, and this obser-

vation is in agreement with the relative adsorption of the two gases as

shown in Table III. It should be noted, however, that this observation

is based on only two tests and only suggests the fraction of gas adsorbed.

Additional tests would be required before a valid quantitative result

could be established.

Adsorption tests were also run on hydrogen fluoride. Not only is

this a highly reactive gas, but also it is highly adsorbed. Due to the

chemical reactivity of this gas the adsorption data were more difficult

to achieve than for sulfur dioxide.

In order to make the adsorption determinations, a measured quantity

of hydrogen fluoride was admitted to the chamber, and as soon as a mixing

was complete, a sample was withdrawn. The actual concentration determined

from the sample was compared with the calculated concentration of the gas

admitted, and from these two figures the amount of adsorption was calcu-

lated. Since the gas is so reactive and its concentration in the chamber

decreases rapidly, a correction factor had to be applied to the original

calculated concentration to estimate the calculated concentration at the

time of sampling. This correction factor allowed for a decrease in

chamber concentration of 10 per cent per hour. This was determined by


measuring the decrease in gas concentration that occurred in the chamber

over a period of time. This loss rate is discussed more fully in the

last section of the study.

Figure XVIII shows how the experimental hydrogen fluoride concen-

tration compared with the calculated concentration. The difference

between the experimental and calculated concentration was assigned to

adsorption, and it can be seen in the figure that adsorption is consider-

ably greater for hydrogen fluoride than it is for sulfur dioxide. The

average fraction of gas not adsorbed in the chamber was found to be 0.67

of the concentration admitted to the chamber, and the standard deviation

for these results was 0.07. The corresponding adsorption constant is

104 x 106 with a standard deviation of 36 x 106. These findings are

based on data shown in Appendix XIII and XIV.

Table IV summarizes the results of this portion of the study re-

lating to gas adsorption on the chamber surface.



1 2 3 4 5 6 7
Gas No. of Fraction Standard Adsorption Standard Critical
tests of gas not deviation constant deviation temp.
adsorbed K x 106 x 106 degrees C.

HF 8 0.67 0.07 104 36 230.2

SO2 15 0.90 0.02 92 21 157.2

H2S 2 0.94 26 100.4

The third column of Table IV shows the fraction of gas which is not

adsorbed on the chamber surface. Conversely, it may be stated that

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs