Richard J . Countess and Julian Heicklen
444
Kinetics of Particle Growth. I I . Kinetics of the Reaction of Ammonia with Hydrogen Chloride and the Growth of Particulate Ammonium Chloride' Richard 4 . Countess and Julian Heicklen" Department of Chemistry and Center for Air Environment Studies. The Pennsylvania State University. University Park. PeflflSyiVania 16802 (Received August 15. 79721
NHa and HC1 at 60 ppm each in N2 were allowed to react in a flow apparatus a t 25" and atmospheric pressure. Solid NH4C1 was produced and the particle size distribution measured for various reaction times. Under the conditions of the experiment both wall losses and particle coagulation were negligible. Monomer NH4C1 is produced by the gas-phase reaction of NH3 and HC1 with a rate constant of 1.9 x 10-l7 cm3/sec. The monomer can then nucleate particles, though the predominant reaction is condensation of the monomer on the particles already present. The particle size distribution follows the law Fjll = N { t {exp( - / 3 l 9 , where Ffl}is the number of particles/cc containing a t least 1 monomer units, N(t1,is the N f t }first rises porportionately total number of particles/cc at any reaction time t , and /3 = 6.24 X with reaction time, but its rate of increase falls as the reactants, NH3 and HC1, are consumed. The rate constant, hill, for condensation of the monomer on any particle with 1 monomgr units follows the relationship kll - I } = ( k ( l ) s)ffl)/f(l- 11, where 6 = kp3/6, f{I} = pF{1)/312/3,and h i s the average value of kill for all I.
+
Introduction In an attempt to learn about the dynamics of particle growth, the formation of NH4C1 from gaseous NH3 and HCI was investigated. This reaction is well known and is a classical general chemistry experiment used to demonstrate Graham's law of gaseous diffusion. Unlike other investigators2-5 who studied the NH4Cl aerosol generation by low-temperature volatilization, atomization of an alcohol solution, irradiation by X-rays, electrical discharges, or by controlled condensation of vapor upon suitable nuclei (the La Mer generator), we studied the generation of NH4C1 from the chemical interaction of NH3 and HC1 vapors to circumvent some of the problems of the other studies ( e . g , wall losses, large concentrations of electrical charges). One important aspect that has not been resolved is the rate of reaction between the two reactant gases. The main reason for this is that the formation of NH4Cl is more complicated than simply the homogeneous bimolecular gas-phase rieaction of NH3 and HC1 vapors. In addition particles are produced which might act as sites for condensation of the vapors resulting in gas-solid reactions. In the experiments reported here NH3 and HC1 are allowed to react In a flow apparatus, and the particle count and size distribution measured a t various reaction times. Our results gave information both about the gas-phase reaction between NET3 and EiCl and the mechanism of particle growth. Experiment a1 Section Anhydrous iirnrnonxa (>99.99% purity) and Electronic Grade hydrogen chloride (>99.99% purity) gas cylinders were purchased from Matheson; prepurified nitrogen ( >99.997% purity) was purchased locally from Phillip Wolf. Special equipment to handle the corrosive reactant gases included a stainless steel regulator with check valve containing a filter, Matheson No. 12-240, for the ammonia and a mond regulator with purge assembly and check The Journal of Physical Chemistry, Vol. 77, No. 4, 1973
valve (also with filter), Matheson No. B15-330 K, for the hydrogen chloride. Before and after each experiment the hydrogen chloride regulator was purged with dry nitrogen. Prior to mixing the reactant gases, each gas was dried (-60" liquid N2-propanol mixture and drierite trap in the HC1 line, -30" liquid N2-propanol mixture and glass wool trap in the NHa line), filtered through glass wool, and diluted with N2 carrier gas that had also been dried ( - 196" liquid N2 and drierite trap) and filtered (glass wool). A measurement of the particulate concentration in the individual filtered gases with a Gardner Condensation Nuclei Counter indicated less than 100 particles/cc equal to or larger than 0.002-~diameter. Gases were metered with stainless steel Matheson N.R.S. high accuracy needle valves, Models 4176-2101 and 4181-2505, and flow rates were measured with rotameters, Matheson Models 610 and 600 (R-2-15AAA) for the reactant gases and a Matheson Model 602 and Brooks Model R-6-15A for N2, supplied with a calibration for each gas and subsequently verified in our lab with either an American Meter Co. dry test meter (1 cf/rev) or a bubble meter (100-cc burret). The desired ratio of carrier to reactant gas was obtained by a primary dilution of the reactant gases with N2 followed by a secondary dilution with another source of dry N2. Excess gases from the primary dilution were exhausted into the hood while the two gas mixtures horn the secondary dilution were fed through two separate mixing jets into a Pyrex reactor tube and then exhausted into the hood. Flow rates were adjusted so as to maintain laminar
CAES Report No. 264-72. R. Whytlaw-Gray and H, S, Patterson, "Smoke," Edward Arnold, London, 1932 J. M DallaValle. C. Orr, and B L Hinkle, Brit d Appl Phys Suppi 3. S198 11954). W. B. K u n k e l . >. Appl. Phys.. 21, 833 (1950). G. 0. Langstroth. Can J. Res.. 25, 49 (1947).
Kinetics of the Reaction of NH3 with HCI
R
.. .
. . I 2 9 aec.
445 TABLE I: Particle Size Distribution Parametersa 0 (SMALL REACTORI
\
(SMALL REACTOR. 1.71 mtc. (LARGE REACTOR, t.77soc.) 0 (SMALL REACTOR.2.57rec. (LARGE REACTOR.2.67*ec.}
2'62
?v .\ I\B\%\\L
-4 . 4 7 aec. X
(LARGE REACTORI
I
10
- 6 ~ [ f ] particles/cc ,
24.2 16.0 13.4 12.3 10.4 12.5
1.42 2.7 5.0 8.0 8.7 9.0
1.29 1.74 2.62 4.47 7.77 11.67
10 - 4 c ~ qcm-'
small carbon-coated copper E.M. grids with the precipitator run in the intermittant mode so as to precipitate a homogeneous sample of the entire aerosol smoke on each grid. Grids were stored in a desiccator until ready for analysis. Infrared analysis with a Beckman IR 10 spectrophotometer using both a 10-cm Pyrex cell with NaCl windows and a Beckman Multipass cell (10 cm to 10 m) with KBr windows was used to follow the disappearance of the NH3 or HCl ir bands for the chemical reaction
;r h
t, sec
i
lo4 -
NH3
+
k
HCl--NH4CI
in an attempt to obtain the rate constant k .
.\
\o \
i~30 ~
A/
0.I
0.2
*.,
0
0.3
A
0.4
( 0
as
_-1
0.6
0.7
PARTICLE DIAMETER (d),MlCRONS
Figure . Plot of F l d l , the number of particles per cc groater t h a n t h e diameter d , vs. d for various reaction times at 25". The initial reactant Concentrations were 60 ppm of NH3 and 60 ppm of HCI in 1 atm of Nz. flow in the reactor. All connections were of Pyrex, stainless steel, Teflon, or polycarbonate tubing and fittings. A positive tank pressure, slightly above 1 atm (2 psig), allowed the operation of the flow experiments without the emcumhrance of vacuum pumps. Also most of the particle counters operated at 1-atm pressure. Contact time of the reacting species in the flow tube could be varied by one of two methods. In the first rnethod, one of the gas inlets was movable along the axis of the reactor tube, with the other one rigid and thus the distance between the mixing zone of reacting species and the exit port c o d d be varied. The second method wa!j by varying the flow rate. Two different size reactor vessels were constructed out of Pyrex tubing, one with an internal cross-sectional area of 3.0 crn2 and 1.0 m long, the other 8.5 cm2 internal cross sectional area and 2 0 m in length. Also two different inlet heads were built for the movable inlet; the first consisted simply of' a Pyrex tube, 6 mm o.d., and the second one (used only with the larger of the two reactor tubes) consisted of a circular Pyrex ring (2.3 cm 0.d.) with a series of finely drilled holes around the ring and pointing downstream toward the exit. The particle size distribution was measured by precipitating the NI14CI particles from the gas stream with an electrostatic precipitator, T.S.I. Model 3100, and obtaining the particle sizes and particle number by electron microscopy using a Phillips Model 300 transmission electron microscope. The particles were precipitated onto
Results Experiments were done in which 120 ppm of NH3 in N2 was flowed into the reactor in the outer tube and 120 ppm of HC1 in NZ was flowed into the reactor through the inner tube. The two streams had equal flow rates to that the initial NH3 and HCl concentrations were both 60 ppm after mixing. Preliminary experiments with the HC1 stream on the outside and the NH3 stream on the inside gave identical results. Particles of NHlCl were produced and after a given reaction time, t , the particles were captured by the electrostatic precipitator and counted. No significant accumulation of NH4C1 occurred on the reactor walls. The total number of particles per cc larger than a given diameter, d , was obtained, and this quantity is called F l d J , The particle size distribution for various reaction times are shown in Figure 1. In this figure log F l d } is plotted us. the particle diameter for six different contact times ranging from 1.29 to 11.67 sec. The semilog plots are all reasonably linear. Reactions carried out in the large and small reactors gave similar results for the same contact times. The reliability of each data point in Figure 1 may be no better than a factor of 2 . However, the lines .fitting the points and their intercepts are probably reliable to 4=25%. Since the plots in Figure 1 are linear, the expression describing F { d l is F ( d J = iV{ t ) exp( - a d )
(1)
where N { t ) is the total number of particles/cc a t time t , and in general N might also be a function of t . The values for NItl and N are the intercepts and slopes, respectively, of the plots in Figure 1. They are listed in Table I for the various reaction times. The values for Nit} increase from 1.42 X 106 to 9.0 X 106 particles/cc as t increases from 1.29 to 11.67 sec. The data are plotted in Figure 2 . Initially N { t }is proportional to t with a proportionality constant of 1.6 X lo6 particles/ cc sec. At longer times, the rise is much slower. The Journal of Physical Chemistry, Vol. 77, No. 4 , 1973
Richard J. Countess and Julian Heicklen
446
the rates of particle production were too slow to obtain meaningful data for the larger particle sizes, whereas at significantly higher concentrations, the NH4CI smoke became dense and gravitational settling became important. Because wall losses are negligible, the total number of NH3 (or HCl) molecules, n{tl,removed from the gas phase can be computed from the total particle mass. The number of particles containing 1 NH4C1 monomers per cc, fill, is given by fll) =
- dF{l}/dl= ( & V ( t 1 / 3 1 2 ' 3 ) exp( - P P 3 )
(3)
Then
nItI t, sec.
Figure 2. Plot of N(t1, the total number of particles/cc, vs. reaction time at 25". T h e initial reactant concentrations were 60 ppm of NH3 and 60 ppm of HCI in 1 atm of N2.
On the other hand cy is relatively independent of t . Except for the earliest reaction time, a = (13 f 3) x lo4 cm -1. Since the density of NH4Cl is 1.53 g/cc (at 17"),6 eq 1 can be recast as
F { l J=
X I ~eJx p ( - P P )
(2)
where 1 is the number of NH4CI monomers in the particle, and (I= 6.24 X 10- 3; FI1) is then the total number of particles/cc containing at least 1 monomer units. Equation 2 presupposes that the solid particles are entirely NH4Cl. It is possible that either NH3 or HC1 could be trapped in the particles. To check this point, deposits were obtained from reactions in which either NH3 or HCI were in excess and the C1- to NH4+ ratio analyzed. In the analysis for NH4+, excess HCl was added to convert any free NH3 to MH4+. For a reaction with 500 ppm of HCl the Cl- to NH4+ ratio was 1.04; and 125 ppm of "3, whereas for a reaction with 2500 ppm of NH3 and 250 ppm of HCI, the C1 - to NH4* ratio was 1.06. Both analyses are essentially stoichiometric.
Discussion The experiments were designed to minimize wall losses of particles and axial diffusion. Yet it was necessary that reaction times be sufficiently long so that the initial mixing time and sampling times were minimal. For this reason the reaction times were kept between about l and 12 sec. From simple kinetic molecular theory considerations, the rate constant for diffusion to the walls is - D / x 2 where D is tho diffusion coefficient and r is the distance of travel. For it particle 20 A in diameter, L) is about 0.012 cm*/sec. Since x is about 1 cm, the lifetime for loss to the wall is about 80 sec. Thus for times