When silicon tetrafuoride is absorbed by water, it reacts to form fusilicic acid.

The mechanism of the absorption process has been studied by Whynes (1956),
who suests that the reaction probably occurs in steps, as represented by
e!uations 6"6 and 6"#$
The simple fuosilicic acid probably reacts with additional %i&' or %i() to form a
more comple* form of this compound. The second reaction (e!uation 6"#) is
re+ersible to the point that solutions of fuosilicic acid e*ert a de,nite +apor
pressure of -& and %i&'.
Whynes has presented data on the concentration of fuosilicic acid in
e!uilibrium with %i&' +apors. these are reproduced in &iure 6"1/. The cur+es
indicate that an e*haust as containin 0./1 silicon tetrafuride can produce a
solution containin about /)1 fuosilicic acid at temperatures below #023. This
has been con,rmed in a plant wash"tower by continuously recirculatin the
solution for se+eral hours.
Tail as scrubbers for phosphate operations typically utili4e +ery dilute a!ueous
solutions. 5ata on the +apor pressure of -& and %i&' o+er solutions containin
less than ).5 percent -)%i&6 are i+en in &iures 6"1' and 6"15 which are
based on 6ussian data (7llanione+, 196/) as presented by -ansen and 5anos
&luosilicic acid solutions e*hibit +ery low p- +alues, as indicated by the data in
Table 6"11 presented by %herwin (1955) for industrial acid samples.
9bsorber 5esin
9s both hydroen fuoride and silicon tetrafuoride are +ery soluble in water,
the as",lm resistance would be e*pected to be the controllin factor in their
absorption. This has enerally been +eri,ed by e*perimental e+idence althouh
Whynes (1956) found the %i&' absorption to be complicated by a tendency to
form mist particles in the as. there is also a tendency of the silica (produced
by reaction with water) to form a solid ,lm on the outside of the water droplets,
thus hinderin absorption.
9bsorbers which ha+e been used or proposed for the remo+al of fuoride +apors
from as streams with water include the followin eneral types (see also
3hapter 1 and the ,rst section of this chapter)$
1. %pray chambers (+ertical and hori4ontal)
). 3ounterfow pac:ed towers (typically low"pressure drop pac:in)
/. ;enturi scrubbers, in which hih as +elocities cause atomi4ation of the
'. <=ectors (=et scrubbers) in which hih"+elocity water =ets are used to scrub
and e*haust the as
5. 3ross"fow pac:ed scrubbers
With spray towers, on the other hand, the proposed e!uation is
>(? @ ABCA?
That is the number of transfer units is e*pected to be appro*imately
proportional to the power introduced in sprayin the li!uid. This power would,
of course, increase with an increase in either the !uantify of water or the
pressure drop across the no44les.
>o e!uation is proposed for +enturi or =et scrubbers. howe+er, it should be
noted that, with the former, most of the power must be supplied with the as
stream, which pro+ides atomi4ation and mo+es the as stream.
5ata presented by Bunde for the absorption of -& in se+eral types of
e!uipment are reproduced in Table 6"1) and plotted in &iure 6"16. -e
concludes that for a i+en re!uirement of -& or %i&' remo+al eDciency, rid"
pac:ed towers re!uire the least power and +enturi scrubbers the most, with
spray towers in between. -is analysis of the data of Eerly et al. (195') and &irst
and Warrer (1956), for the absorption of -& in wetted beds of ,ber (wet cells),
indicates that the power consumption of this type of e!uipment is intermediate
between those of the spray and +enturi scrubbers. 5ata on the absorption of
%i&' in spray, pac:ed, and =et scrubbers, presented by Aettit (1951 9, E) and
%herwin (195'), were found to show no correlation with power introduced,
althouh the spray"scrubber points fall reasonably well on the line for spray
scrubbers handlin -&, as shown in &iure 6"16.
%herwin (195') also presents data relatin to the eFect of no44le desin,
number of no44les tried, which included one solid"cone and two hollow"cone
spray"pattern desins, did not appear to i+e appreciably diFerent
performances, nord did any impro+ement result from the substitution of as
many as nine no44le for one centrally located one. 9s would be e*pected,
howe+er, the water"fow rate was found to ha+e a +ery stron eFect. Gsin the
a+erae performance +alues for all no44les, >(? has been calculated for
se+eral water"fow rates from %herwinHs study, and the result are shown in
Table 6"1/.
5ata on which the table is based were obtained in the ,rst of si* towers of a
commercial plant. ?as and li!uid fowed concurrently downward throuh the
unit, which was # ft s!uare by )8 ft hih. 9ppro*imately 10000 cfm of as
containin about 0.61 %i&' entered the unit at a temperature of about 6023.
%uDcient data are not a+ailable to permit plottin the points on &iure 6"16.
howe+er, if power input is assumed to be proportional to water"fow rate, the
data would form a line closely parallelin that plotted for spray scrubbers.
%herwinHs data on the relati+e eDciencies of each of the si* spray towers
operatin in series are also of interest. The si* units were arraned for
downward as fow in the ,rst, then alternately upward and downward fow,
with the as lea+in the top of the last section. Water was fed to the system
throuh )' spray no44les (four per tower), at a pressure of 8 psi and a rate of
/.8 pm per section. <ach unit was # ft s!uare by )8 ft hih. Aertinent results
of this study are tabulated in Table 6"1'.
The hih apparent eDciency of the ,nal stae is belie+ed due to the presence
of some pac:in which was installed in this section for the purpose of remo+in
silica particles from the as.
7t is interestin to note the small diFerence between countercurrent and
concurrent operation in the +arious sections. This is due to the fat that solution
concentration has little eFect on the absorption rate in the ranes
encountered. 7n a separate e*periment to e+aluate this factor, solution from
the ,rst two towers was recirculated continuously o+er the ,rst tower to build
up the fuosilicic acid concentration. The fuoride" remo+al eDciency was found
to be relati+ely constant (60 to 801) for solution concentrations from 0 to 151
fuosilicic acid and then to drop rapidly to about /51 as a concentration of )51
acid was approached.
5etailed compilations of enineerin desin methods for wet scrubbers is
presented by 3al+ert et al. (19#)). The study includes a sample calculation for
the desin of a water scrubber to remo+e %i&' from ases e+ol+ed from a
phosphoric acid manufacturin operation. The calculation is based on
processin as from the diestion of )000 TonCday of phosphate roc: in sulfuric
acid to produce 600 TonCday e!ui+alent. 9ppro*imately '00 lbChr of %i&' are
emitted from the reactor and the scrubber in appro*imately ))0000 ft/Chr of
as saturated with water at 1'' 2&. 7n order to obtain a remo+al eDciency of
99.81, it is concluded that two scrubber staes are re!uired. The ,rst stae is
simply a +ertical dict appro*imately 80 pm of water are sprayed in at the top
of the duct. The second stae is a / ft diam. Ey 1) ft hih cyclonic spray
scrubber employin )) no44les, each operatin with 6 pm of water and a
water pressure of 60 psi. 9n eDciency of 91.8 percent fuoride remo+al is
predicted for the ,rst stae and o+er 99.8 percent for the o+erall system.
5etailed discussions of the desin of wet scrubbers for phosphate plants are
also presented by -ansen and 5anos (1980, 198)).
(peratin 5ata
7n the production of superphosphate fertili4er from phosphate roc:, the roc: is
normally pul+eri4ed, mi*ed with sulfuric acid, and dischared into a den where
the reaction between roc: and acid proceeds. The fresh superphosphate is then
remo+ed from the den by to storae for curin. <*haust ases containin
fuorine compounds are drawn from the mi*in and den operations and, in
some instances, from the ele+ator and other units.
Typical &lorida pebble phosphate roc: contains about /.61 fuoride e*pressed
as elemental fuorine, and appro*imately /)1 of this is released durin the
acidulation process (Aettit, 1951). 9lmost all of the fuoride +apors are e+ol+ed
in the mi*er and den althouh a sliht e+olution of +apor occurs durin
subse!uent handlin and storae operations. The composition of mi*er ases
and total fue ases from this roc: corresponds to appro*imately 11 of the
weiht of the roc: or )51 of the fuorine oriinally present.
3omparati+e data on three types of scrubbers used in superphosphate plants
are presented by Aettit (1951 9, E). 9lthouh no conclusions are drawn in the
study, the data, which are summari4ed in Table 6"16, indicate that the
hori4ontal scrubber oFered the hihest eDciency with the lowest water"fow
rate. The hih eDciency of this unit probably resulted from the use of hih"
pressure no44les and the lon tortuous path which the as stream was forced
to follow.
5ata on 1/ scrubbers handlin superphosphate"den as ha+e been presented
by %herwin (195'). Ten of these are more or less con+entional spray"tower
systems, one is pac:ed tower, and one is a =et"e*hauster system. The spray
towers show +alues for I?a ranin from 0.6) to ).65. the pac:ed tower, I?a
of /.#, and the =et e*hauster, a I?a of 15.6. The +olume for the =et e*hauster is
based on the +olume of the JtowerK which would enclose the +ertical +enture
pipe from the =et le+el to the le+el of the li!uid in the tan: below. The system
ob+iously i+es a +ery hih +olume"coeDcient of performances. howe+er,
power consumption was reported hih and o+erall fuorine"remo+al eDciency
for two units in series was not as hih as that of the ma=ority of the spray
installations. 9 portion of the data on these units is summari4ed in Table 6"1#.
7t will be noticed that the two"stae spray tower (installation ') i+es slihtly
better performance than the si*"stae spray tower. The two ma=or reason for
this appear to be (a) the lower as +elocity which allows the mist formed to
settle out and (b) the appreciably hiher water"circulation rate. %ilica"
deposition problems enerally fa+or the use of a spray tower for this ser+ice
o+er the more compact pac:ed tower.
-ansen and 5anos (198)) report on e*perience with a lare (18 ft * 8 L ft * '6
ft) crossfow scrubber in a phosphoric acid plant. The scrubber consisted of a
spray chamber followed by multiple pac:ed beds of plastic wo+en mesh. With
reard to the spray chamber section of the scrubber, they conclude that a
spray no44le pressure o+er 60 psi is re!uired to attain 801 fuoride remo+al
eDciency (1.5 transfer units). the amount of spray chamber water should be
about )0 to /0 pmC1000 acfm. and full cone spray no44les directed
countercurrent to the as fow are preferred. The plastic wo+en mesh may be
irriated with low"pressure cocurrent sprays. howe+er, the no44les should be
mounted so that they are e!uidis"solutions of these acids at any concentration
and, therefore, corrosion"resistant alloys, oranic materials (wood or polymers),
concrete, or bric: must be employed.
<*haust as containin hydrofuoric acid is normally produced at an ele+ated
temperature, and carbon"steel ducts may be employed to con+ey the as to
the puri,cation unit pro+ided its temperature is well abo+e the dew point. 7t is
usually necessary to precool the as before it enters the actual absorption
e!uipment in order that the latter can be constructed from, or lined with
oranic materials. The as can be most readily coo:ed by the use of water
sprays within the ducts, and the spray section of the con+eyor duct should be
constructed of stainless steel or other material which can stand both hih
temperatures and corrosi+e li!uids. 9fter coolin, the as and dilute a!ueous
hydrofuoric acid can be handled in e!uipment constructed or lined with
resistant materials such as poly+inyl chloride, polyethylene, Iynar and
neoprene. &or lare aluminum"plant scrubbers, hydrofuoric acid"absorption
towers ha+e been constructed of clear heart"redwood sta+es with internal
pipin of poly+inyl chloride and brass, stainless"steel, or Monel no44les. >ew
scrubbers +essels are more commonly constructed of &6A with a 5ynel shield
to protect lass ,bers from the fuorides.
<=ectors for hydrofuoric acid ser+ice may be constructed of cast iron or steel
and lined with a protecti+e ,lm such as neoprene or Iynar, or they may be
fabricated entirely from a hydrofuoric acid" resistant material. 9split J&K, a
modi,ed phenolic resin forti,ed with an inert carbon ,ller, has pro+ed to be
satisfactory for this ser+ice (Erown and Tomlinson, 195)).
7n silicon tetrafuoride operations, the as is usually a+ailable at moderate
temperature (from phosphate"roc: acidulation). howe+er, relati+ely hih
fuoride concentrations may be de+eloped. 5ependin on whether recirculation
of fuosilicic acid solution is practiced, the fuosilicic acid concentration in the
li!uid may rane from a fraction of 11 to o+er 101. Eecause of the wide rane
of operatin conditions and performance re!uirements of fertili4er"plant
e*haust"as purifyin systems, a considerable +ariety of construction materials
ha+e been utili4ed.
<*perience in ?reat Eritain (%herwin, 195') indicates that ood"!uality
enineerin bric: without fros, bonded with a late* hydraulic"type cement , is
the most satisfactory type of massi+e construction material for silicon
tetrafuoride"absorption towers. 7n the Gnited %tates, the towers are more
commonly constructed of &6A or wood, with or without a protecti+e oranic
coatin. Tower basins and sumps are usually constructed of con+entional
Aortland"cement concrete. This material is apparently protected from se+ere
attac: by the precipitation of silica and other compounds in the pores as a
result of the initial reaction between fuosilicic acid and constituents of the
Monel metal and hih chromium"molybdenum"nic:el stainless steel are
enerally the most resistant materials of construction for e!uipment in contact
with fuosilicic acid. -ow"e+er, they are +ery e*pensi+e, and their use is
enerally reser+ed for precision parts in +ery se+ere ser+ice, such as impellers
and no44les, particularly where recirculation is practiced. 3ommon brass has
pro+ed !uite satisfactory for e*terior pipin and, in some cases, for spray
no44les. Bead is not satisfactory for handlin fuosilicic particles.
7n addition to causin erosion of solution"handlin e!uipment, silica particles,
which precipitate from both aseous and li!uid phases, complicate the desin
of lines and +essels. The silica coats the inside of ducts and columns, plus
pac:ins, and settles out of the solution stream where+er its +elocity is
reduced. The problem is best handled by desinin the e!uipment so that all
surfaces in the absorber are fushed with ade!uate !uantities of water and the
resultin solution is :ept mo+in as rapidly as possible in lines and collection
basins. 7n spite of such precautions, it is common practice to fush out silicon
tetrafuoride"absorption systems with hih"pressure water once e+ery ) or /
days and to physically scrape deposited silica from the walls and ducts se+eral
times a year.
5isposal of 9bsorbed &luoride
7n both -& and %i &' absorption operations, the product water may be too acid
and to*ic for direct disposal to sewers. 7t is, therefore, common practice to
neutrali4e the eNuent with limestone (or lime) in a separate tan:. The fuoride
ion is precipitated as calcium fuoride, and where other components such as
silica and iron are present, these may also be precipitated as a result of the p-
chane. The solids can be readily separated by settlin or ,ltration and
disposed of separately, althouh occasionally the neutrali4ed mi*ture, which is
relati+ely innocuous, can be disposed of as a dilute slurry.
<Forts ha+e been made in some installations to reco+er the fuorine e+ol+ed in
phosphate"roc: processin as cryolite or other mar:etable products. The
!uantify of fuorine e+ol+ed in such operations is +ery lare so that steps to
reco+er it would appear warranted. 9 process de+eloped by the Tennessee
;alley 9uthority (T;9) is claimed to accomplish this (9non., 195#). 7n the T;9
process, the water used to scrub noduli4in":iln e*haust as is maintained at a
p- of 5 to 6 by the continuous addition of ammonia, and the resultant rich
li!uor is treated to precipitate impurities and yield a +aluable >-'& solution.
The absorber solution is recirculated to brin its fuorine content up to about /5
Cliter. The rich solution is then treated with suDcient ammonia to raise the p-
to 9, thus precipitatin iron, silica, and part of the phosphorus. The precipitate
is ,ltered oF, and the solution is then used to ma:e cryolite (by addin sodium
sulfate and alum at a p- of 6) or aluminum fuoride.
9dditional studies conducted by T;9 on the remo+al of fuorine from a!ueous
scrubber eNuents and the production of useful fuorine compounds ha+e been
reported by Tarbutton et al. (1958) and by Eraber and &arr (19#0). The latter
study describes cryolite reco+ery from eNuents from phosphorus furnaces.