10.5% of the beryllium present. The statistical error could be reduced using larger samples or repeating the analysis several times. The method has been adapted to solution analysis using the large 16-cc. sample container b y making the appropriate correction for the neutron moderating effect of the \vater.
Ibid., Report S o . P.G.-171, pp. 55772. (4) U. 6. Dept. of Health, Education, and Welfare, “Radiological Health Handbook,” P B 121784R, p. 133 (Office Tech. Services, September 1960). (5) V‘agner, C. D., Lukens, H., Otvoa, J. IFT., Intern. J . Appl. Radiation Isotopes 11, 30-7, (August 1961).
LITERATURE CITED
(1) Aidarkin, B. C., Gorshov, G. V.,
Grammakov, A. G., Kolchina, A. G., Tr. Radievogo Inst., Akad. Nauk SSSR, 8993 (1957); C.A. 54, 18175 (1960). ( 2 ) Iredale, P., United Kingdom Atomic Energy Authority Research Group, Report No. AERE-EL/M-108, Atomic Energy Research Establishment, Harwell, Berks, England (1960). (3) Milner, G. W. C., Edwards, J. W.,
RECEIVEDfor review June 18, 1962. Accepted August 21, 1962.
Condensation Nuclei, a New Technique for Gas Analysis FRANK
W. VAN LUIK, Jr., and RALPH E. RIPPERE’
General Engineering laborafory, General Electric Co
b A technique for measuring low concentrations of gas involves conversion of gas molecules to liquid or solid submicroscopic airborne particles which act as condensation nuclei. Water vapor is caused to condense on the particles. The opacity of the vapor is then measured and related b y electrical signal to gas concentration. This procedure has many promising applications in air pollution control, toxic vapor detection, and continuous monitoring of chemical processes. technique for measuring low gas concentrations by conversion of gas molecules to condensation nuclei has been developed. An understanding of the technique is dependent upon a knowledge of condensation nuclei and the capability of the condensation nuclei detecting instrument. A general description of the technique is given here, with a few of the possible instrument configurations that have been developed in the authors’ laboratory and some performance test data on stdected gases. NET\
., Schenecfady,
N. Y THE DETECTOR
Figure 1 indicates the amount of saturation required to cause a “particle” of water to act as its own condensation nucleus ( 1 ) . While the major proportions of condensation nuclei are not water particles, once the growth process is triggered by a nucleus and water molecules start to attach themselves to it, the growing particle acts as if i t \\ere completely composed of water. Therefore, Figure 1 can serve as a guide in suggesting the amount of supersaturation required to condense water on particles of a given size. The time required for the growth process-from nucleation a t about 0.001micron diameter to visible droplet about 1.0 micron in diameter-is about 7 milliseconds (’7). This speed of growth is an important factor in the instrumentation, and the growth in size of the particle provides an amplification factor which can be utilized to give extreme sensitivity t o this method of measurement.
The condensation nuclei detector
(CND) developed b y General Electric Co. is shown schematically in Figure 2 (6). The power supplies, rotor driving mechanisms, and auxiliary equipment are omitted froni the figure.
The air iainple containing the particles to be detected and measuredj is drawn through the inctrument by the vacuum pump. ‘l’lie siimple first enters the humidifier, n-iic I e its relative humidity is raised io 100% d h Ratt‘r. The sample then passes through a rotating valve into the cloud chamber, where it is expanded adiabaticallj- by the valving and transport system, causing the saniple to cool and the relative humidity to rise to any desired supersaturation up t o 400% ( 5 ) . This unstable condition can be lelieved b y three prosesses: Condensation in nhich the nucleus of the droplet is a single water molecule;
’OOr
CONDENSATION NUCLEI
Condensation nuclei are liquid or solid submicroscopic airborne particles, each of nhich can act as the nucleus for the formation of a water droplet. Their size may vary from 0.001 to 0.1 micron in diameter. Such nuclei are found in nature in concentrations ranging from a few hundred to hundreds of thousands per cubic centimeter of air. The nucleus derive3 its name from its triggering role in causing supersaturated water vapor t o condense into droplets. Water vapor in air that contains no particles will not start t o condense into droplets until the air is SO070 supersaturated. Prcsent address, Computcr Depart, Sunnlvale, Calif. ment, Gcneral Electrir Co
2
300
I
w
Figure 1. Relative humidity necessary for condensation
E
200
01 10-6
.om1
I
I
IO-’ .OOl
10-6
I
10-6
I
10-4
SIZE CM RADIUS .I 1.0 S I Z E MICRONS
.01
GAS MOLECULE CONDENSATION NUCLEI ARC NUCLEI TOBACCO SMOKE
I
10-3
10.0
J
IO-*
100.0
RANGE OF PARTICLE SIZES
FOG
t?zzzzz
VOL. 34, NO. 12, NOVEMBER 1962
1617
33 &
, , EXPANSIOII CHAMBER
READ CUT
'
Figure 2. Condensation nuclei detector this does not take place until the saturation has reached about 800%. Condensation on the walls of the chamber; this affects relatively few of the water molecules in the sample because most of them are far away, in terms of molecular distances, from the walls. Condensation on airborne particles in the sample; this is the primary and dominating effect. The cloud of water droplets so formed, each containing a n airborne nucleus, is measured in a dark-field optica! system by scattering light on a multiplier phototube. The amount of light scattering is proportional to the number of droplets in the sample. This relationship holds because all the droplets after growth are the same size, a uniformity which eliminates the effect of particle size variation on the scattering of light (7'). The sample is then flushed out of the chamber through the rotation valve and discharged in the exhaust of the vacuum pump. This complete cycle may be varied from three t o 10 times per second. Normal operation is a t 5 cycles per second. The complete sequence of events for a n y sample-from intake and humidifping, to expansion, measurement, and discharge; plus the time needed for instruments to respond-can be rapid, 1.5 seconds. That is the time constant of the detector to obtain 82% of full scale reading of a step input of condensation nuclei. The electrical signal produced b y the multiplier phototube (an RCA 931X tube) first passes through a filter designed to remove the 60-cycle flicker of the light source and to pass the 3- to 10-cycle frequency generated by the CND's own cycling rate. The filtered signal passes through a peak-reading voltmeter, is amplified, and comes to the readout device as a direct current proportional to the number of particles present in the sample. The C N D can detect concentrations as small as 10 or as large as lo7 particles per cubic centimeter. Sample flow can be varied from 20 to several hundred cubic centimeters of gas per second by adjusting a critical orifice in the rotating valve. Characteristics of Detector. Characteristics of t h e CND t h a t are important t o its use as a primary sensing element in gas analysis or detection are : Extreme sensitivity. On a molecular basis this is about lo4 molecules 1618
0
ANALYTICAL CHEMISTRY
of material in 1019 molecules of air, or 1 part in 10'5. Good repeatability and stability of detection. Short-term stability is better than 5% of point and unaffected by the temperature of the sample or its moisture content. An electrical output signal of 1 ma. into 1500 ohms. Power s u*m_l v : 60-cvcle ax., 115 volts, 300 watts. Volume about 4 cubic feet and weight about 70 pounds.
AMBIENT AIR
FILTER
VOLUME
WATER
I
Gas to Particle Conversion. Before t h e nuclei detector can be used t o measure gas concentrations, t h e gas must be changed in form into a substance t h a t will have a vapor press u r e a t ambient temperature and p r e s s u r e s u c h t h a t the substance will b e either a liquid or solid submicroscopic particle. Experiments have indicated that the particles formed do not have t o be hygroscopic, polarized, or ionized to serve as condensation nuclei. Many processes meet these criteria. Nine of these are: photochemical, pyrolysis, hydrolysis, acid-base reactions. reverse photochemical, ammonolysis, electrochemical, oxidation, and chemical. Many permutations and combinations of these processes are possible. One such combination is to obtain controlled conversion by one of the techniques and suppress this conversion by the introduction of another This is in effect a reverse converg". sion which can be used for detection. Perhaps the most readily understandable process is that used to measure the concentration of sulfur dioxide in air (3, 4). Here the conversion process is a photochemical reaction that takes place when SO, gas is subjected to ultraviolet light of a wavelength of 2537 A. The SO2 molecules break u p and recombine as SOI, Jvhich in turn reacts with mater vapor in the air and forms liquid particles of sulfuric acid-fine condensation nuclei. The conversion processes may be classified into two groups: those which do not go to complete equilibrium (this may be a desirable factor for fast response), and those which reach equilibrium, as hydrolysis. I n some cases a combination of reactions could be used.
Figure 3. Test apparatus for measurement of NO2 The function of the filter is most important, for without control of ambient nuclei consistent results would not be possible. The gas to be analyzed may be carried in air, in a carrier gas such as nitrogen, or in some other gaseous compound or mixture. Obviously, the less complex the carrier gas the more straightforward d l be the construction of the filter and conversion e q u i p ment. The filter must not only be capable of removing ambient particles. but i t must also allow the gas to pass without undue adsorption or chromatographic effects. It is relatively easy to provide a filter which will reduce the ambient condensation nuclei to less than 50 particles per cc. Therefore, the sample entering the converter will contain essentially only the carrier gas and the gas to be measured. The converter can use any of the nine processes listed above or combinations thereof, but its prime function is t o produce consistently a number of nuclei proportional t o the concentration of the gas to be measured. The reaction need not go t o completion because of the extreme sensitivity of the CND, and as long as the cycling time is so arranged that all samples enter the detector at the same relative point in the reaction. I n the detector itself, the sample is processed as previously described, producing an electrical signal proportional to the gas concentration, that can be calibrated in a variety of readout units, such as parts per million. Response time is in no way dependent on the concentrations being sampled. The converter and detector can be adjusted t o measure concentrations all the way from fractions of a part per million up to several per cent.
MEASUREMENT SYSTEM REQUIREMENTS
INTERPRETATION OF RESULTS
A practical instrumentation system for measuring the concentration of a gas is composed of: a particle filter t o remove any ambient condensation nuclei present in the sample, a reaction section where t h e nucleogenic process takes place, the C N D with a n appropriate readout device, a meter or recorder, and a pump which ejects the used sample and draws in succeeding ones.
The problem in the quantitative interpretation of results lies in the fact that the sensitivity and repeatability are dependent upon the size distribution and concentration of particles produced in the nucleogenic process. Both factors need to be known for a quantitative calibration to be made. Such a calibration has been considered, and some of the conditions which affect i t are: type of conversion process being used,
0.1
10
I P.P.Y. OF NO2
Figure 4. Calibration curve for NO? detection b y hydrolysis
converter pressure and temperature, gas flow rate and dwell time in converter, and hygroscopicity of nuclei created. One of the more important conditions is the dwell time between the creation of nuclei and the measurement of their concentration in the CND. The number of nuclei produced in the converter is generally large. Thousands of particles per cubic centimeter are normal. Since individual particles are continually growing b y coagulation with adjacent particles, the size distribution is a rapidly changing factor unless steady-state condition can be obtained by controlling the dwell time within the converter. Tests utilizing diffusion chambers and electrical mobility measurements indicated that the nuclei produced are from 0.0008 to 0.02 micron in diameter for residence times up to a few seconds. Therefore, precise size distribution measurements are very difficult. Because of the influence of such conditions, which are either not known or are difficult to measure, direct calibration has been used exclusively t o obtain an empirical relationship between gas concentration and electrical output of the CND. Tests have indicated that when the converter temperature ( l t 2 O C.), pressure (* 10 mm. of Hg), and gas flow rate (=t2%) are controlled, consistent results are obtained. As an example, with KO2 detection at the 5-p.p.m., level calibration over many days is within zt5% of point. The effect of other gases in the atmosphere on the calibration is also of importance. Many tests have been conducted in ambient atmosphere and the difficulty encountered depends upon the gas being detected. SO, and NOz detection is influenced very little b y other ambient gases present. However, a n unsym-dimethylhydrazine detector will also detect ammonia or the amines. I n general, this technique can be utilized to detect classes or related families of gas and, as are most techniques, is not completely selective in its response. Care in picking the type of conversion and controlling the operation of the converter can minimize the influence of other ambient gases. Some of the practical difficulties encountered in using this new technique
stein from the extreme sensitivity of the detector. One must be very careful that the detector is not affected by nuclei other than those being produced in a controlled manner in the converter. Also, extraneous reactions in the converter must be carefully avoided; the carrier gas must contain no substance or impurity that is also nucleogenic. The adsorption of certain gases on the tubing passages or on the surface of the converter chamber may cause some difficulty, as it reduces the sensitivity of the instrument and delays the response time. If large quantities are adsorbed, it may take some time for the signal to diminish and stabilize for maximum sensitivity. The best design provides for having the gas enter the reaction zone immediately upon entrance into the apparatus and to have the reaction period as short as possible. INSTALLATIONS
Despite the uncertainties, practical systems have been developed in the authors' laboratorv. and several units have been installed in the field. Nitrogen Dioxide Detection. Nitrogen dioxide may be readily detected after i t has been converted to nitric acid particles b y hydrolysis. Figure 3 is a schematic diagram of w c h a system. The NOz gas is injected into a 1-cu. meter air-filled chamber in a known concentration b y volume. T h e XOZ and air pass through a Gelman Type E fiber glass filter and into the converter, a 500-ml. flask containing approximately 200 ml. of water a t 76" F. The K O z reacts with the water to produce particles of nitric acid in the order of 0.002 micron in diameter. The sample containing these particles is then passed into the condensation nuclei detector, which is adjusted to produce a supersaturation of about 375%. The number of condensation nuclei so produced is indicated in Figure 4, where the two curves are for two different concentration levels. To make the transition in level, one merely has t o change the sensitivity control on the detector. Total flow rate through the filter and converter was 1 cu. foot per minute. ils a n indication of the response times obtained, Figure 5 represents a typical response when approximately 6 feet of '(-inch Tygon tubing was used to connect the filter-converter and the nuclei detector. A delay of about 10 seconds was observed from the time the NOz was injected into the intake chamber until the detector indicated full scale. This response time was primarily a function of the flow rate and the total volume of chamber, tubing, and converter. It can be changed b y adjusting the system
flow rate. This does not affect the operation of the detector, since moat of the air bypasses the instrument. Ammonia Detection. Detection of ammonia gas (2) involves an acid-base reaction t o form particles. The tests n-ere conducted with the apparatus sketched in Figure 6. The ammonia gas was carried in dry nitrogen. The converter was composed of a Fisher-Milligan gas mashing bottle filled lvith a solution of HC1 of varying molarity, The ammonia reacted with HCI to produce ammonium chloride. -4 known ammonia solution !vas placed in a similar gas mashing bottle and its temperature measured. With these tiyo values the vapor pressure was determined and the concentration of
Table 1.
Some Types of Gases Detected
- Substance Aluminum iodide Ammonia Benzene Carbon dioxide Carbon monoxide Chlorine Ethyl alcohol Freon 12-21 Fuming nitric acid Hydrochloric acid Hydrocarbons Methyl mercaptan hlonoethylamine Mercury Metallic carbonyl Nitrogen dioxide ?;aphtha Octane Sulfur dioxide Sulfur hexafluoride Sulfuric acid Solvesso 100 Toluene Unsym-Dimethylhydrazine
Type of conversion
Concentrations utilized, p.p.m.
Hydrolysis
0.01
Acid-base Photochemical Electrochemical Chemical
0.005 2 5 1
Chemical Reverse photochemical Pyrolysis Hydrolysis
1
5 2 0.5
Hydrolysis
0.5
Photochemical Oxidation
0.1 0.01
Chemical
0.5
Photochemical Hydrolysis
0.001 0,001
Hydrolysis
0.5
Reverse photochemical Photochemical Photochemical Pyrolysis
5 2
0.001 1
Hydrolysis Reverse photo- 1 chemical Reverse photo- 1 chemical 0.1 Chemical
VOL. 34, NO. 12, NOVEMBER 1962
9
1
1619
=D I S C H A R G E
4
’ 0PRESSURE REGULATOR -X P N C H CLAMP
Figure 6. Test apparatus for measurement of ammonia
$2 FLOWMETER i i
01
02 03
05
IO
2
3
5
IO
PPY A Y Y O N I A
Figure 7. Concentration of ammonia vs. nuclei reading
TC THERMOCOUPLE
saturated anmonia gas lenving the bottle mas determined. Further detection \vas accomplished b y dilution with N,. The flow was adjusted to match the requirement of the condensation nuclei detector (100 cc. uer second). After dilution., theammonia gas ~ ’ a spassed through the HC1 bubbler. The concentration oc ammonia in the gas entering the HCI bottle ~vaqcomputed from the vapor pressure and the flowmeter data and plotted against the nuclei reading (Figure 7 ) . The HCI concentration i:, given in terms of the molar concentration of the liquid. When the HC1 molar concentration was changed from to s%, great increase in sensitivity was noted. The cause for this is unknown; howvei., a t 8ycmolar,
the stoichiometric ratio was exceeded and a n excess of HC1 n-as present in the converter. Additional work is now under m y to study this further. OTHER GASES DETECTED
‘I‘ahle I indicates some type3 of gases whirh have been detected and the conversion mechanisms used. The semitivities indicated are those n hich nere readily obtained nithout evtensive tests and do not indicate ultimate performance. LITERATURE CITED
(1) Dm Gupta, s.E.,C;hosh, 8 , K,, R ~hfod. ~ . phys. 18, 225-90 (1946). (2) Dunham, S. R., General Electric Co.,
Schenectady, S . T.,private correspondence, 1960. (3) Dunham, S. B., S a t c u e 188, 51-2 (Oct. 1. 1960). (4j Gerhard, G. It., Johnstone, H. F., I n d . Eng. Chern. 47, 972-6 (1955). ( 5 ) Nolan, P. J., Pollack, L. IT., Proc. Roy. Irish Acad.,SlA, 9-31 (1946). (6) Rich, T. -4.,‘Continuous Recorder for Condensation. Nuclei,” 4th International Srmuosium on .\tniosDheric Condensation huclei, Heidelburg; Germany, May 1961. (7) Skala, G. F., General Electric Co., Schenectady, K. Y., private correspondence, 1959. RECEIVEDfor review April 23, 1962. Accepted September 7 , 1962. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March 6, 1962.
General X-Ray Spectrographic Solution Method for Analysis of iron-, Chromium-, and/or Manganese-Bearing MateriaIs BETTY J. MITCHELL and HUGH J. O’HEAR Technology Department, Union Carbide Metals Co., Division o f Union Carbide Corp., Niagara Falls, N. Y.
b A simple and accurate x-ray spectrographic method i s described for the determination of iron, chromium, and manganese in samples which can be put into solution. For 1 gram of sample per 100 ml. of solution, these elements plus other heavy elements present may be determined in the range from 0.1 to 99.9% of the original sample. The method has been applied to the analysis of manganese and chromium alloys, slags and ores, Alnico, steels, and refractories for combinations of iron, chromium, manganese, nickel, copper, and cobalt. Fluorescent intensity changes due to variations in the volume, temperature, acid, and acid concentration of the solution sample are corrected b y ratioing against the intensity of an added control element. Manganese, nickel, and copper are discussed as controls for samples in perchloric, nitric, and sulfuric acid solution. The method may be classified as a uni1620
ANALYTICAL CHEMISTRY
versal spectrographic method in the respect that only a single calibration system per element i s required for all materials and acid matrices.
X
fluorescence methods have been applied successfully to the analysis of a variety of materials for elements in widely varying concentrations. Specialized sample preparation and interelement effect correction are necessary for accurate analysis. Because of their physical properties. many ferroalloys, carbides, and steels are not amenable to accurate x-ray analysis in powder or chip forms; complex ore samples require sets of similar standards. Solution methods (4, 6) have been developed for the analysis of several types of steels, but require careful control of the acid matrix. Since chemical methods for the determination of iron, chromium, and manganese are among the most rapid in use, an x-ray spectrographic method should be simple and -RAY
accurate to compare favorably. This paper describes such an x-ray method. It may be used to make significant analytical savings for the analysis of large numbers of similar samples or samples requiring multiple determinations. The general technique or variations thereof may be eltended to include heavy elements in any sample which can be put into solution form. INSTRUMENTATION
An x-ray Industrial Quantometer (Applied Research Laboratories’ XIQ) (6) was programmed for the simultaneous measurement of iron, chromium, manganese, nickel, and copper. The K a line of each element is measured using LiF crystaIs and Multitron detectors, operating at 1400 and 1800 volts, for concentrations ranging from 0.1 to 10 and 0.01 to 0.1 mg. per ml., respectively. I n the X I Q the sample is excited by the end-window x-ray tube located above it, and the sample or sample
