Determination of Soluble Silica in Water
A Photometric Method
H. LEWIS KAHLER
W. H. & L. D. Betz, Philadelphia, Penna.
This method measures the molybdenum
blue color developed by reducing the yellow
silicomolybdate complex with sodium sul-
fite at suitable pH. It requires only 4 min-
utes for a test, a 10-ml. sample, and gives
an accuracy on par with gravimetric analy-
sis. Phosphate, alkalinity, tannins, iron,
and other ions present in natural and
boiler waters in addition to color, tempera-
ture, and color progression, offer no practical
interference. The work was completed
with a Klett-Summerson photoelectric
photometer, but the method can be em-
ployed with any similar instrument.
D I E K E R T and IVandenbulke (2) have shown that small
concentrations of soluble silica can be determined colori-
metrically by measurement of the yellow color that develops
on the interaction of soluble silica and ammonium molybdate
in acidic solution. The yellow color is stated to be a complex
of composition H&i(hIo0,)cH20 (2, 6). Since the appear-
ance of this method, Thayer (12) has shown that it is sub-
ject to interference from phosphate and iron, due to forma-
tion of phosphomolybdate and iron silicomolybdate. Xumer-
ous other investigators have used essentially this same method
for the determination of silica, most of them using Nessler
or Hehner tubes for comparison or different types of colorim-
eters (6,7,9) such as Duboscq (IO). An exhaustive survey of
the literature has failed to reveal any mention that tannins in
natural and boiler waters interfere with this method. This
interference was noted in this laboratory while attempting to
apply the Dienert-Wandenbulke method using the Klett-
Summerson photoelectric photometer. Satural and boiler
maters gave higher results than gravimetric analysis.
I n order to investigate this further, tannins of two different
classes, pyrocatechol and pyrogallol, were investigated for
their reaction with molybdate by the Dienert-Wandenbulke
procedure applied to the Klett-Summerson photometer and
the Taylor visual comparator. The results recorded in Table
I show that both types of tannins interfere with this method,
giving high silica equivalence. The pyrogallol type per unit
weight forms a more intense yellow complex with molybdate
TABLE I. INTERFERENCE OF TANNINS WITH THE DIENERT-WANDENBULKE
METHODIN THE ABSENCEOF SILICA
--Pyrocatechol Class Tannina- ,--Pyrogallol
Si02
Si02 equivalence, Si02
equivalence,
Tannin photometer Tannin comparator
Taylor equivalence,
Tannin photometer Tannin comparator
Pp.m. P.p.m. P.p.ni. P.p,m. P.p.ni. P.p.m.
* Commercial grades of silica-free tannins.
than the pyrocatechol type and therefore gives higher silica
equivalence. This 1%-orkshowed conclusively that the Dien-
ert-Kandenbulke test could not be used for accurate work in
x-aters having tannins present.
Isaacs (4) reported that reduction of the silicomolybdate
complex by sodium sulfite to molybdenum blue of composi-
tion Mo808.H20( 8 ) in low acidity is not accompanied by color
from phosphomolybdates. Bertrand (1) criticized the
method, but Foulger (3) showed that phosphate, free from
silica, gives no color with the reagent a t the acidity recom-
mended by Isaacs. The latter procedure was not found suit-
able for rapid photometric work, as it employs acetic acid a t
relatively high pH, which gives color progression too great
for accurate work.
I n the present study, the use of hydrochloric acid for de-
velopment of the silicomolybdate complex a t p H 2.4 to 2.7,
followed by reduction by sodium sulfite, was found to be very
satisfactory, giving adequate silica sensitivity and no prac-
tical interference from phosphate and tannins. The effect of
pH on the formation of phosphomolybdate before reduction
and the subsequent silica equivalence after reduction is given
in Figure 1. As the acidity decreases above pH 2.7, the effect
of phosphate becomes negligible, but the color development
when silica is present requires more time and is accompanied
by considerable color progression. The p H range of 2.4 to 2.7
before reduction is considered optimum and is the basis for
this present method.
EFFECT OFACIDITY ON
REDUCT ION OF
PHO SPHOMO LY BDATE
BY S U L F I T E I N
ABSENCE OF S I L I C A
FICURE 1
/
I /
/
30 2.5 2.0
PH OFSOLUTION BEFORE REDUCTION
Reagents
Hydrocliloric acid reagent, 0.248 N , 20 mi. of 38
per cent grade (1.19 specific gravity) acid per liter,
Ammonium molybdate reagent, 102 grams of am-
Class Tannino-
monium molybdate (grade 81 per cent Mo03) per
Si02 liter. Sodium sulfite reagent, 170 rams of sodium
equivalence sulfite (grade 97 per cent, anhyjrous) per liter.
Taylor Sodium silicate solution, 102.0 mg. as silica per
P.p.711. P.p.nz.
liter.
Procedure and Conditions of Test
The procedure uses 10 ml. of sample for analysis,
5 ml. of hydrochloric acid reagent, 5 ml. of
ammonium molybdate reagent, and 10 ml. of sodium
sulfite reagent. The sample is treated with the
536
August 15, 1941 ANALYTICAL EDITION 537
of disodium phosphate as PO4exhibit 0.8 to 2 p. p. m. of silica
TABLE11. EFFECTOF PHOSPHATE between p H 2.7 and 2.4. I n Table 11, 300 p. p. m. of POr in
Disodium 7
- Silica Found b y Photometer the presence of silica give as high as 0.2 p. p. m. of silica in low
Phosphate Present, Present, Present,
Present 9.2 p. p. m. 24.7 p. p. m. 42.0 p. p. m. silica content and give low silica results on the order of 1
P . p . TIL. PO, P. p , Wk. P . p . ni. P . p . n1. p. p. m. in high silica content. The reason for this phe-
20 9.3 24.6 42.3 nomenon is not clear. It cannot be explained on the basis that
40 9.4 24.8 42.2
100 9.4 24.6 42.2 the phosphate buffers the solution away from the correct p H
200 9 .6 24.6 41.5 range, as the bufferingeffect of this concentration of phosphate
300 0.0 23.0 40 8
was found to be only 0.1 pH. It must in some way interfere
with either the complete formation of the silicomolybdate
TABLE111. EFFECTOF TANNINS~ complex or the reduction of t’his complex to molybdenum
7
- Tannin Present blue.
Si02 Si02 Si02 Pyrocatechol Pyrogallol Pure tannic
Present Found Difference type type acid Tannin of two commercial varieties and pure tannic acid
P.p,m. P.p.in. P.p.m. 1’. 11. 711. P . 11. 111. I’. p . l i i . offer no practical interference, as Table I11 reveals. The ef-
fect of the tannins is different in the presence and absence of
silica. The yellow tannin molybdate complex is not reduced
by sulfite to molybdenum blue, but its color gives a n additive
18.5 18 6 +0 1 effect to the silica present. Even in the most highly colored
19.7 +1.2 io0 ..
21.2 f2.7 400 .. boiler maters, the tannin concentration is rarely over 200
21.6 +3.1 600
‘2.5 +4 0 1000 p. p. m., which offers no practical interference to the method.
Snell (IO) states that iron interferes with the Dienert-
Kandenbulke met’hod. The effects of both ferrous and ferric
ions on the suggested method are reported in Table IV.
Ferric iroii gives no interference throughout the concentra-
lS.5 1X.i +0.2
19.3 +0.8 200 .. tion investigated, and the effect of the ferrous iron is negligible
21.4 +2.8 400 .. in low silica content and not serious in high silica concentra-
22.1 +3.6 BOO
23.0 +4.5 800 tion. Average natural and boiler n-aters contain far less
23.7 +5.2 1000 .. concentration of these ions than is required to give any d i c a
100
600 interference.
1000 The temperature of sample and reagents is a factor in this
lS..5 20.5
23.0
+2.0
+4.5
200
600
method and should be maintained within * 5” F. of the solu-
25.0 +6.5 .. 1000 tion temperature at time of standardization. Temperatures
0 All products were SiO.-free. Pyrocatechol a n d pyrogallol types are a few degrees helom or above this limit give no serious error.
i o % tannins; pure tannic acid is 100%. High temperatures promote color development with at-
tendant high silica results, while lower temperatures have the
opposite effect.
acid andmolybdatereagentsand allowed to stand for approximately
1 minute. This gives the silicomolybdate complex time to de-
velop completely, after which it is stable for 5 minutes. During AXD FERRIC
IV. EFFECTOF FERROUS
TABLE IONS
this time interval, another sample is treated with the acid and
sulfite reagents in addition to distilled water equivalent to the
Present
--Silica as Si()?-
Found
--- Ihfference
Fe+++
Present
FeT-
Present
molybdate reagent, and is used to adjust the photometer to zero
reference. When this operation is complete, the test sample is 1’. p . 1,i. 1’. p. 7,l. I’ p I , ( . P . p . m. I’. p . !,<.
reduced by the sulfite reagent and tested a t the end of 1 minute. 6 4 6.3 -0 1 ..
The Klett-Summerson photometer has two photocells of the 6 4 0 0 O’i ..
6.3 +0.1 1. 3 ..
blocking type, a standard source of illumination, a sensitive 6.3 -0.1 2.7 ..
galvanometer, color filters of average bands, a 21-cm. (14-inch) 6,; +0.1 6.7 ..
logarithmic scale, and operates with 110 volts alternating or di- 6.3 -0.1 13.4
6.3 -0 1 26 8
rect current. Summerson (11) gives a detailed discussion of the 2.5.5 25.2 -0.3
photometer. The work u-as completed with a 13-mm. test tube to 26.3 -0.2 ti:7
facilitate rapid routine analysis. The instrument also accommo- 25.2 -0.3 13.4
dates sample cell depths of 2.5, 10, 20, and 40 mm. The blue re- 25.2 -0.3 26.8
duction color of the final silica test was found to absorb best in G.4 6.3 -0 1 .. p i
the spectral region 600 to 700 mp. In this work, a 620 mp filter 6.3 -0.1 .1
(590 to 680) was employed. t i .3 -0 1 .. 6.,
i.6 +I 2 .. 13 4
0 4
With this procedure a standard curve n-as established from 0 1
1 ,
a series of points of known silica content. The curve is linear :3 . :3
throughout the concentration range of 0 to 50 p. p. m. of silica 19.2 - li :i .. 13 4

used in this study. The curve was checked by other standard

solutions prepared from pure silica fused in sodium carbonate, TABLEV. EFFECTOF ALKALINITY
and hydrolyzed silicon tetrachloride. Hard-rubber bottles 7- Silica as SiOz- 7 dlkxlinity t o Alkalinity t o
Present Found Difference Phenolphthalein Methyl Orange
were used throughout this study to prevent contamination P . p . iri. P. p . m. P . 11. m. P. p . m. as CaCOz
from glass bottles. 5.1 ,5 2 +O. 1 9j 35
.5 0 -0.1 240 440
4 7 -0.1 ,520 880
Effects of Ions and Conditions on the Method
25.5 25 6 +0.1 30 60
25.7 +0 2 170 :13u
Orthophosphate, a common constituent in boiler waters, 25 4 -0.1 320 350
does not interfere with the method, as shown in Table 11. 22.5 -3 0 li20 1120
A comparison of results given in Figure 1 and Table I1 shows 75
280
t h a t phosphate in the absence of silica gives some silica 450
550
equivalence, while in the presence of silica this effect is ob- 1195
served only with low silica content. I n Figure 1, 300 p. p. m.
538 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 13, No. 8

Advantages of Photometric Method

TABLEVI. STABILITY
OF YELLOW
SILICOMOLYBDATE
COMPLEX
This photometric method gives accuracy comparable to
7
-
Present
Silica as Si01
Found Difference
. Time Silicomolybdate Complex
Stood before Sulfite .4dded gravimetric analysis and requires only a 10-ml. sample and
P.p.m. P.p.m. P.p.ni. Mifl . less than 5 minutes for testing. It is sensitive enough to de-
5.1 5.0 -0.1 1 tect 0.1 p. p. m. as silica. The average time required for the
5.0 -0.1 5
5.2 +0.1 8 gravimetric analysis in this work was approximately 5 hours
5.2 +o. 1 15 per duplicate sample, and large volumes oT sample were nec-
5.2 f0.1 30
25.5 19.2 -0.3 Sulfite added immediately after essary to attain this accuracy. For work involving lower
molybdate silica concentration below 50 p. p. m. of silica, this method
25.2 -0.3 1
25.3 -0.2 0.5 can be modified simply by use of a larger cell depth, and
25.2 -0.3 3 underthese conditions silica can be determined more accurately
25.2 -0.3 5
25.3 -0.2 10 than with the test tube.
24.5 -1 0 28
45.0 45.3 +0.3 1 Conclusions
44.5 -0.5 4
43.5 -1.5 8 The proposed method measures the blue color resulting
40.8 -4.2 15
from the reduction of silicomolybdate complex and was found
to give acceptable results compared to gravimetric analysis.
OF MOLYBDENUM
TABLEVII. COLORPROGRESSION BLUE The method can tolerate as high as 150 p. p. m. of phosphate
COLOR as PO4 and 350 p. p. m. of alkalinity as calcium carbonate
--Silica Found- Error Caused by hlissing Time without dilution. Tannins show no practical interference.
Regula1 of Reading
Silica as 1-minute 2-minute B y 3 seconds
Si01 reading reading B y 1 minute (calculated)
P. p . ni. P. p . i n .Si02 P . p . m. Si02 P. p . m. Si OI TABLEVIII. GRAVIMETRIC
vs. PHOTOMETRIC
RESULTS
87.0 36.7 38.1 +1.1 0.06
Silica- Sample Dilution,
18.5 18.6 19.: +0.9 0.045 7-
Gravi- Photometric
7.4 7.2 7.t +o 3 0.015 dsniplp Location Sample metric Photometric Procedure
1.8 1.i 1.9 fO. 1 0.005
M1. P. p . m. Si02
Raw Fitchburg, I500 3 1 3.1 ..
Mass. 3.1 3.2
The effect of alkalinity is given in Table V, which shows no Boiler Berlin, K . H. 250 70.0 67.8 1:l
66.8 68.5 1:4
interference in low silica concentration. I n high silica concen- 67.0 1:4
tration total alkalinity can be tolerated to 350 p. p. m. Even Boiler Kooky Mount,
N. c.
500 63.2
65.8
65.4
65.0
1:l
1:l
though many boiler waters exceed this alkalinity, it is not con- Feed Rocky Mount, 1500 1.9 1.6 ..
sidered a disadvantage to the method, as suitable dilution of x. c. 1.4 1.7 ..
sample reduces alkalinity to a level where it offers no effect. Boiler Philadelphia, 300 33.4 33.5
Penna. 34.2 33.8
Dilution is often necessary with boiler samples, as their silica 33.8
contents are beyond the range of the method. Dilution does Raw Philadelphia, 1000 12.1 12.8
not introduce appreciable error, as the data in Table VI11 Penna.a 12.6 12.5
show. Boiler Curwensville,
Penna.
75 48 0 46.8
47.2 iIi
Color progression of both the yellow silicomolybdate and Feed Berlin, ? H. i 1.500 3.1 3 1
the molybdenum blue is of little significance. Approximately 3 2 3 0
a minute is required for the yellow Fortified with sodium silicate.
color to attain its maximum de- ~~

velopment, after which it is com- TABLEIx. ASALYSIS O F S.4MPLES

pletely stable for 5 to 30 minutes, ---Alkalinity---
Hard- Phenol- Methyl Total
depending on its concentration, as Sample SO4 CI n e w phthalein orange POa Solids Organic Fe pH
indicated in Table VI. This com- P . p , n r . P . p , m . P . p . m P . p . n i . P.p,ni. P.p.ni. P . p . m . P , p . m . P . p . n i . P . p . m .
pletely stable time of 5 minutes is 1 .. ..0 .. ...
more than ample for the complete
2
3
96
560
20
48 o
i08
io4
i00
170
io0
io iOio
300
50
:: 1ii:a 0.7
4 8 1 4 0 12 .. ... 0:5 7.1
silica test, so that any change in the 5 608 88 0 176 804 45 iii4 190 .. 11.5
yellow silicomolybdate color beyond 6 20 4 48 4 40 .. ... .. 8,s
7 60 50 120 240 .. li60 240 , . 10 1
5 minutes is of no importance. Table 8 8 0.5 lk 0 12 .. 7 2
VI1 shows that. the color progres- a As CaCOs.

sion of the blue reduction col& is

easily controlled. The results as
shown are not seriously in error, even if the time is exceeded
by 1 minute. Time errors are in the order of 3 seconds
and cause no error in the application of this method.
Results on Natural and Boiler Waters
The new procedure was tested on eight actual samples from
various parts of the country. The comparison of gravimetric
with photometric results is presented in Table VI11 and
shows that the procedure gives acceptable accuracy.
I n the boiler samples, dilution was employed in the photo-
metric method without sacrifice of accuracy. Table I X
gives the analysis of the samples and indicates that sulfates,
chlorides, and hardness, in addition to the other ions in-
vestigated, do not interfere.
Ferric and ferrous iron, chlorides, sulfates, and mineral hard-
ness offer no interference. The Dienert-Wandenbulke
method is not considered suitable for accurate silica deter-
mination in waters containing tannins, which interfere seri-
ously by forming yellow tannin-molybdate complexes.
Acknowledgments
The author wishes to express appreciation to W. H . &
L. D. Betz, in whose laboratories this investigation was com-
pleted. He is also indebted to C. A. h’oll, chief chemist, for
his previous work on the reaction of tannins with molybdate
solutions and to E. C. Feddern for his cooperation. The as-
sistance of the Klett Manufacturing Company, New York,
which markets this instrument, is also acknowledged.
August 15, 1941 ANALYTICAL EDITION 539

Literature Cited (8) Munro, Proc. Trans. Nova Scotian Inst. Sci., 16, 9 (1928).
(9) Schwartz, IND. ENG.CHEM.,Anal. Ed., 6, 364 (1934).
(1) Bertrand, Bull. SOC. chim. biol., 6, 157 (1924). (10) Snell, "Colorimetric Methods of Analysis", New York, D. Van
(2) Dienert and Wandenbulke, Compt. rend., 176, 1478 (1923). Nostrand Co., 1936-37.
(3) Foulger, J . Am. Chem. SOC.,49,431 (1927). (11) Summerson, J . Biol. Chem., 130, 149 (1939).
(4) Isaacs, Bull. SOC. chim. biol., 6, 157 (1924). (12) Thayer, IND.ENQ.CHEM.,Anal. Ed., 2, 276 (1930).
(5) Jolles and Neurath, Z.angew. Chem., 11, 315 (1898).
(6) Mellon, IND. ENQ.CHEM.,l n a l . Ed., 11, 80 (1939). PRESENTED before t h e Division of Water, Sewage, and Sanitation Chemistry
(7) Muller, Ibid., 11, 1 (1939), a t t h e 100th Meeting of the American Chemical Society, Detroit, Mich

Anhydrous Sodium Thiosulfate,

A Primary Standard
HAZEL SI. T03ILINSON AVI FRANK G. CIAPETTA, Temple University, Philadelphia, Penna.

T HE reaction between bromoacetate and thiosulfate ions

has been found by Slator ( 8 ) and by La Mer and his co-
workers (4, 5, 6, 9) to be uniquely free of side reactions. Ac-
cordingly, the measurement of velocity constants using these
reactants offers a critical method of detecting the existence of
decomposition products of either reagent. By employing such
kinetic measurements, Kamner and Tomlinson working with
La Mer (6, 7 ) have shown that the pentahydrate of sodium
thiosulfate decomposes upon standing a t room temperature
for several months after recrystallization. At the suggestion
of La hIer, the investigation has been extended to include
t h e stability of the anhydrous salt.
I n 1904, Young (IO) suggested the use of the anhydrous
salt a s a primary standard but presented only meager support-
ing data. This paper shows that anhydrous sodium thio-
sulfate is stable thermally at 120" C. and meets the re-
quirements of a primary standard so well t h a t it deserves more
extended use.
The methods of purification of reagents and the technique
of reaction velocity measurements were those of La Mer and
Kamner. K a t e r solutions 0.01 h' in reactants were employed.
After two kinetic determinations, sodium thiosulfate penta-
hydrate was placed in a n oven a t 120" C. Portions of the
dehydrated product were used for subsequent reaction rate
measurements. The data in Table I show conclusively that
this thiosulfate was not decomposed by the heat to which i t
was subjected over a period of 11 veeks. The drifts in ve-
locity constants noted by La Mer and Tomlinson using the
pentahydrate which had been kept a t room temperature for
2 years were not' observed.
Samgles of c. P. sodium thiosulfate pentahydrate were dried
of iodine used are in good agreement, even when the c. P.
grade salts were employed without recrystallization. The
analysis of the dehydrated, unrecrystallized sample 2 indi-
cated that it was slightly decomposed. This was confirmed
by kinetic measurements upon both dehydrated and original
samples. The velocity constants so obtained exhibited pro-
nounced drifts which corresponded to those observed in pre-
vious investigations.
When an accuracy greater than 2 parts per thousand is not
obligatory, the data indicate that the c. P. salt of commerce
may not require recrystallization. Upon recrystallization, the
results are in agreement well within the limit of experimental
error.
Arsenious oxide (99.82 per cent As203 us. Bureau of Stand-
ards Sample 83) was used to determine (1) independently the
normalities of two standard iodine solutions prepared by
weight. The excellent agreement between these iodine con-
centrations determined by arsenite and thiosulfate titrations
is shown in Table 111. Moreover, it is evident from the pre-
cision obtained that no marked change had occurred in the
composition of these anhydrous samples which had been
stored in a bright laboratory, but seldom in direct sunlight,
I. EFFECT
TABLE SODIUXTHIOSULF~TE
OF HEATIKG
(;is determined b y rate of reaction with sodium bromoacetate)
Days Kept a t 120' C . Velocity Constant, k
0 0.489 * 0.001
0 0.490 * 0.000
2 0.490 * 0.002
4 0.487 0.002
9 0.492 f 0 . 0 0 2
16 0.492 0.001
f

28 0 490 0.001
a t 120 C. to constant mass and stored over calcium chloride. 79
f

0.488 =t 0.001
Portions of these samples were recrystallized from water belon-
65" and then subjected to dehydration to ascertain the impor- kav. = 0.490 * 0,001
tance of recrystallization. The time required for dehydration de-
pends, of course, upon the frequency of stirring the solution
formed upon heating the hydrated salt and later upon the fre- TABLE
11. ANALYSESO F I O D I K E SOLTsTIONs B Y SODIUM
quency and degree of subdivision. Individual samples of the THIOSULFa4TE
thiosulfate (0.2 to 0.4 gram) were titrated with 40 to 50 ml. of Normality by Normality Calcd. Deviation
standard iodine solutions using calibrated volumetric apparatus; Sample NazSz03 from 1%Content 70
the iodine purification, solution preparation, and titrations were 1 0,05825 0.05828 0.05
carried out after the method of Kamner ( 3 ) . c. P. resublimed io-. 1 recrystallized 0.04045 0.04046 0.03
dine was sublimed from a 1 to 5 potassium iodide mixture and this 2 0.05825 0.05839 0.24
product was sublimed again, dried, and stored over calcium chlo- 2 recrystallized 0.04045 0.04046 0.00
ride. All iodine solutions contained 4 per cent of potassium 3 0. 06825 0.05832 0.12
iodide. Freshly prepared solutions of soluble starch (0.5 per cent) 3 recrystallized 0.05s25 0,05825 0.00
were used and correction blanks determined. 4 0.05s25 0.05822 0.05
4 recrystallized 0.05590a 0,05592 0.04
0.058ljb 0,05825 0.00
From Table 11, i t may be seen that the concentrations of 0.04255' 0.04257 0.05
iodine solutions determined against weighed portions of these Sample a t 120' for 18 days, b 28 days, C 79 days.
anhydrous samples and the values calculated from the masses