Effect of some heating parameters on
volume expansion of borax
B. Kahraman and M. Ozel
Mining Engineering Dept. and Geophysieal Engineering Dept., respeetively,
Dokuz Eylul University, Buea, Izmir, Turkey
O.F. Ozguven
Eleetrieal & Eleetronies Engineering Dept., Inonu University, Malatya, Turkey
T. Batar
Mining Engineering Dept., Dokuz Eylul University, Buea, Izmir, Turkey
N. Erdogan
Ministry of Energy & Natural Resourees, Eleetrieity Generation Co. Ine., Ankara, Turkey
Z. Diele, S. Safak and S. Kurt
Eleetrieal & Eleetronies Engineering Dept., Mining Engineering Dept. and Department of Statisties,
respeetively, Dokuz Eylul University, Buea, Izmir, Turkey
Abstract
About 70% ofthe world's total boron reserves are located in Turkey. Boron minerals are typically concentrated by
attrition scrubbing followed by screening and classification to remove clay minerals and other impurities. These wet
concentration methods, however, result in a considerable amount of accumulated tailings in large tailings dams. The
tailings are approximately 28% solids and contain 14% B 2 0 3. Problems caused by these tailings can be minimized
by employing dry beneficiation techniques involving heat treatment. This method has been successfully tested on
laboratory and pilot scales. During heat treatment, borax undergoes expansion in volume. The amount of expansion
is dependent on the particle size, temperature and du ration of heat treatment. Because the optimum temperature
and the duration of the heat treatment are not weil defined, the boron ore acts as a calcined clay mineral and is discarded along with the gangue. It is important to predict such parameters using mathematical formulations. Thus, a
mathematical model is used to predict the required temperature and duration period prior to heat treatment. In the
experimental study presented here, aseries of systematic thermal tests was carried out in a muffte furnace. Two variables, temperature and time, were studied in a response surface design. A quadratic regression model was chosen and
regression coefficients were calculated. Using a regression equation, the volume expansion of the particles in various
circumstances can be predicted. The mathematical test results are then correlated with the experimental results co
relate the optimum expansion of borax with heating time and temperature and with borax grades and recoveries.
Key words: Calcination, Expansion, Boron minerals
Introduction
Boron eompounds are essential ingredients in the manufaeture of a variety of industrial produets. Together, Turkey
and the United States possess about 90% of the world's
boron reserves, with two-thirds ofthese loeated in Turkey
(Celik et al. , 1993). Turkey has a total of224 Mtofproven
boron reserves, and 339 Mt of probable reserves (Kose et
al., 2002; Helvael, 2005).
The most eommereially important boron minerals are:
•
•
•
•
tineal (Na2B40i IOH 20) ,
eolemanite (Ca28 60 ll '5H 20),
ulexite (NaCa8 50 9'8H 20) and
kernite (Na2840i4H20).
Equally, boron minerals and brines rieh in boron are
subjeeted to preeoneentration during the benefieiation
proeess. To remove unwanted clay minerals, this step
usually eonsists of a serubbing operation. Unfortunately,
Paper number MMP-07-024. Original manuscript submitted June 2007. Revised manuscript received and accepted for
publication January 2008. Discussion of this peer-reviewed and approved paper is invited and must be submitted to SME
Publications Dept. prior to Feb. 28, 2009 . Copyright 2008, Soeiety for Mining, Metallurgy, and Exploration, Ine.
MINERALS & METALLURGICAL PROCESSING
169
Val. 25, No. 3 • August 2008
ROM
Figure 1 - A general flowsheet of processing of tincal
ore (borax).
because the boron minerals are rather friable, they tend to join
the fine fraetions that are lost in the tailings. Thus, those fines
discarded as waste usually eontain large amounts of valuable
boron minerals. While Turkey provides a substantial percent
of the world's eoncentrated boron needs and while Turkey's
proeessing plants yield 32% B 20 3 concentrates from 24%
B 20 3 tincal ore, the fine tineal tailings eontain from 14% to
18% B 20 3 ,mostly less than 0.2 mmin size (Bataretal., 1995).
More than a quarter of a million metric tons of tineal in the
saturated brine has now aecumulated in vast tailing ponds, and
this is available for a new remediation technology (Akar et al.,
2005). For discarded slurries obtained from wet processing
methods, new tailing dams are needed every four years, eonsidering a plant life of250 years, the seriousness of the problem
is apparent. As a eountry that supplies a great amount of the
world's boron eoneentrate, it is essential for Turkey to apply
an alternative method. Because tincal mineral is dissolved in
water, the reeommended method eannot be included water in
any stage of proeessing.
Although there are many laboratory, pilot and semi-pilot
applieations for the heat treatment of boron, application on
an industrial-seale is limited. Heat proeessing of colemanite
by deerepitating followed by air tabling has been applied at
the Ameriean Borate Corporation near Lathrop Wells, Nevada
(Miles, 1973; Wilson et al., 1975; Smith and Walters, 1980).
Other industrial-scale borax production is from rasorit (kernit),
known as eommercial boron in the United States, which eonsists
of the precalcination of the rasorit. Produets containing 65%
and 46% B 20 3 , wh ich are referred as Rasorit 65 or 46, are
marketed after calcination either in powder form or grindable
produet (Civelekoglu et al., 1987). In Turkey, researehers have
conducted a number of laboratory-scale, pilot-plant-seale and
semi-pilot-plant-seale experiments supported by universities
with the eooperation of government institutions. Experiments
August 2008 • Vol. 25 No. 3
on laboratory-seale applieations were supported by TUBITAK
and were performed by Aytekin at al. (1992). The test results
showed that a eoneentrate containing about 60% B20 3 with a
waste eontaining about 15% B 20 3 eould be obtained from feed
assaying 26% B 20 3 (Aytekin et al., 1992). In a study on a semipilot-scale application carrled out by Akdag et al. ( 1993), a 54.8 %
B20 3 eoneentrate with a 16.4% B 20 3 waste was achieved from
a tineal ore assaying 26% B 20 3 . In pilot-seale studies, 52.67%
and 56.63% B 20 3 eoneentrates were obtained from eolemanite
ores eontaining 30.49% and 31.80% B 20 3 , respectively, by
means of heat treatment (Buyuran and Ylldmm, 1984). All
experimental studies carried out by various researchers have
nearly followed the same flowsheet shown in Fig. 1.
Etimaden Klrka Borax Enterprise, wh ich is the first industrial-scale heat proeessing application in Turkey, produeed 500
t of ealcined concentrate in 2006. A coneentrate assaying 56%
B 20 3 was produced with a tailings containing 11 % to 14%
B 20 3 . Therefore, with a high B 20 3 eoncentrate final product
and a low B 20 3 waste, this environmentally friendly method
is expeeted to be applied to ulexite and eolemanite ores, whieh
are eurrently concentrated using wet methods.
It is expeeted that the ideal ealcination parameters for the
production of tineal coneentrate will be determined as the wet
methods are abandoned. Because the expansion eharacteristies
of ulexite and eolemanite ores during heat proeessing are totally
different than that of tineal, the results of this study cannot be
applied to boron minerals such as ulexite and eolemanite.
The majority of boron minerals exist in the hydrated form
and they lose their water of crystallization on heating to a certain
temperature. Boron minerals undergo different alterations when
subjected to different temperatures. As eolemanite undergoes
deerepitation, another boron mineral of eommercial importance,
ulexite, exhibits shrinkage under heat treatment with no significant fragmentation (Sener et al., 1992; Batar, 1996; Batar et al.,
1998). Tineal deserves special attention beeause of the different
structure it aequires while loosing its erystal water. Its volume
gradually expands and pops up in a manner similar to popcorn.
Solids expand in all directions and volume expansion depends
on the change in temperature, the original volume and the type
of substance (Celik et al., 1998). This behavior may allow for
the separation of gangue minerals,mainly montmorillonite-type
clays and carbonates from lineal.
The literature on the proeessing of boron minerals is vast,
but a mathematical interpretation of the volumetrie changes
of the expanded tincal particles exposed to thermal treatment
is not mentioned. A mathematical expression is needed for
multiple independent variables affecting the volumetrie expansion of the tineal partieles treated with heat at different times
and temperatures . Of those applications, multiple regressions
are widely used in the industrial world, partieularly in situations where several input variables potentially influenee some
performance measure or quality charaeteristie of the product
or proeess. In many areas, either a first-order or a second-order
model has been used. The seeond-order model is very flexible
and would likely be useful as an approximation to the true
response surfaee in a relatively small region. For this reason,
the second-order model will be used to develop an appropriate relationship between the process variables x and Y and the
volume expansion z as follows
z = ßo + ß,x+ ß2Y+ ß3 X2 + ß4l + ß5XY+E
(1)
where
eis the error term and determines the parameters (the ß's)
by the least squares method (Meyers and Montgomery,
1995).
170
MINERALS & METALLURGICAL PROCESSING
The model eontaining a seeond-degree funetion derived by
the determination of parameters provides volumetrie expansions at different times and temperatures. Equation (1) yields
the volume expansion ofthe tineal mineral (NazB40i lOH 20)
sized between 25.0 and 12.5 mm (8 g) at any time and any
temperature by determining the parameters.
The results of ealeined produets of four different fraetions are
given in Table 1. The 25.0 to 12.5-mm (8 g) fraetion ealcined
data were only used for estimating the regression equation.
Similar regression equations ean be obtained eorresponding to
the other three fraetions. Thus, it is possible to build the eorrelation among the temperature, time and volume expansion
as weIl as for other fraetions.
Table 1 - Volume expansion of tineal ore (zi) with respeet to temperature (7j) and time (ti)'
Time (t;),
Fraction
Experimental
Samples for the experiments taken from Klrka Borax Works
were erushed below 25 mm and then classified by sieving to
-25.00+ 12.50, -12.50+9.51 and -9.51 +4.76 mm size fraetions.
In the tests, particle groups that have simi1ar shape to a sphere
were seleeted and subjeeted to heat treatment. To determine
eaeh volumetrie expansion that eorresponds to different heat
and time, three different particles with identieal size were
used. Subsequently, the particle that was least deformed was
taken as the basis.
Aseries of heat treatment experiments were performed in
a muffte fumaee to determine the ideal ealcination time and
temperature. The temperatures for the tests were kept between
200° and 450°C for aIl the size fraetions at time intervals of
5, 8, 11, 14, 17 and 30 minutes to examine the behavior of the
borax samples. The results of experiments for the four fraetions
are given in Table 1.
Beeause of the porous strueture of the ealcined produets,
the approximate volumetrie ealculations for the expanded
ealcined produets were made using mieronized quartz particles
instead of eonventional methods sueh as image analysis, the
Arehirnedes Prineiple and nitrogen gas measurement. In the
volumetrie determination of expanded ealcined produets,
where the surfaee irregularity and roughness ends and the
porosity begins is very erueial for eaeh material. Therefore,
this material was immersed in the free-ftowing mieronized
quartz particles (approximately -30 !Am) with a eertain volume,
and the displaeed volume was measured as the volume of the
expanded ealcined produet.
In the ealcination tests, low temperature and short duration periods were found to be suffieient for the fine particle
fraetions but not for the eoarse fraetions. For example, at low
temperatures «300°C) and espeeiaIly with smaIl grains sizes,
fuIl ealcination is aehieved, and the grade and effieieney rate is
therefore high. When using coarser grains sueh as -25.00+ 12 .50
mm, ete., at the same temperature, the inner part of the grains
are not affeeted due to insufficient heat. During the heat treatment, higher heat and longer duration of the heat treatment, or
inversely, lower heat and shorterduration ofthe heat treatment,
the mineral will harden by melting or it will aet as a clay mineral beeause of insuffieient ealcination, respeetively. Images
obtained throughout the heat proeess are shown in Fig. 2.
It is of great importanee to know the ideal temperature and
heat duration beforehand to avoid a ehange in volume eausing
exeessive or insufficient ealcination. In ideal ealcination eonditions, borax undergoes an inerease in size, and this enables the
dry separation ofborax from its aeeompanying gangue minerals.
Determination of the ideal ealcination eonditions takes longer
for the different size fraetions. Therefore, it is very important
to determine the optimum heat and the optimum duration of
heat treatment for a minimum amount of waste. In this study,
MINERALS & METALLURGICAL PROCESSING
min
Temperature (Ti),
oe
200
250
300
350
450
0.9 g, 9.51-4.76 mm
5
8
11
14
17
30
0
5
10
10
5
10
15
10
15
15
10
5
20
20
15
10
5
10
30
25
15
10
15
10
0
0
0
0
0
0
1.8 g, 12.5-9.51 mm
5
8
11
14
17
30
0
10
15
10
20
20
20
15
50
30
20
15
5
30
25
10
20
15
15
25
20
10
5
10
0
0
0
5
5
0
5
5
10
25
15
15
10
20
40
35
40
30
35
25
35
30
20
40
50
10
60
55
65
60
50
5
5
5
15
0
5
5
10
10
25
35
25
25
30
35
45
45
20
30
60
90
50
65
45
35
85
30
70
80
65
5
5
5
25
20
10
5 g, 25-12.5 mm
8
11
14
17
30
8 g, 25-12.5 mm
5
8
11
14
17
30
a quadratie regression model was ereated to find the optimum
heat value and duration of heat treatment by eonsidering the
data for the 25 to 12.5 mm fraetion. The relationships among
volume expansion (Zi) , the duration of heat treatment (t) and
the fumaee temperature (T) ean be determined by the equation
obtained from the ealculation of the regression eoeffieients of
the model developed (Eq. (1».
Statistical analysis offitting a second-order model
of the volume expansion of the borax mineral
In this study, 30 observations involving volume expansion
values of borax for the 25.0 to 12.5 mm (8 g) fraetion is eonsidered. The test results are given in Table 2. Table 2 presents
the response variables and the data resulting from an investigation into the effeet oftwo variables, time (ti) and temperature
(1j). The table eontains values for the eorresponding eoded
variables Xi and Yi
x;
=
t; - (maxt; + mint; )/2
(max t;
_
min t; ) 12
=
t; -17.5
12.5
(2)
and
Y;
=
I; - (max I; + min I; ) 1 2 I; - 325
(maxI; -minI;)/2 = 125
(3)
where
Zi denotes the volume of the particle eorresponding to time
(ti) and temperature (T;).
171
Val. 25, No. 3 • August 2008
(e)
(b)
(a)
Figure 2 - Different images of borax particles exposed to heat process: (a) with insufficient calcination time and temperature, (b) at the ideal calcination time and temperature and (c) with excessive calcination time and temperature.
Table 2 -
Data for calcined borax experiment.
Run
order
1ime It;),
min
2
3
4
5
5
5
5
5
5
6
7
8
9
10
Temp
oe
IT;),
Process variables
x;
Y;
Z;
200
250
300
350
450
-1.00
-1.00
-1.00
-1.00
-1.00
-1.0
-0.6
-0.2
0.2
1.0
5
25
30
35
5
8
8
8
8
8
200
250
300
350
450
-0.76
-0.76
-0.76
-0.76
-0.76
-1.0
-0.6
-0.2
0.2
1.0
10
30
60
85
5
11
12
13
14
15
11
11
11
11
11
200
250
300
350
450
-0.52
-0.52
-0.52
-0.52
-0.52
-1 .0
-0.6
-0.2
0.2
1.0
10
35
90
30
5
16
17
18
19
20
14
14
14
14
14
200
250
300
350
450
-0.28
-0.28
-0.28
-0.28
-0.28
-1.0
-0.6
-0.2
0.2
1.0
25
45
50
70
25
21
22
23
24
25
17
17
17
17
200
250
300
350
450
-0.04
-0.04
-0.04
-0.04
-0.04
-1.0
-0.6
-0.2
0.2
1.0
35
45
65
80
20
26
27
28
29
30
30
30
30
30
30
200
250
300
350
450
1.00
1.00
1.00
1.00
1.00
-1.0
-0.6
-0.2
0.2
1.0
25
20
45
65
10
17
August 2008 • Vol. 25 No. 3
In this response surface experiment, two factors (time and
temperature) were used . The effects of these factors were
studied in a response surface design.
First of all, a regression model is fitted for the experimental
data given in Table 1 for the 25.0 to 12.5 mm (8 g) fraction.
This is a full quadratic model and includes all linear terms, all
squared terms and all two-way interactions. Using the coded
variables, the model is given as
Z;
= ß" + ßIX ; + ß2Y; + ß3 X/ + ß4Y/ + ß5 X ;Y; + E;
(4)
The regression coefficients are obtained as follows
Z
= 68.665 + 5.47 x -
(5)
0.438 y -
19.607 x 2 - 44.075/ - 0.457 xy
This can be converted into an equation using the natural variables tj and T; by substituting the relationships in Eqs . (2) and
(3), x and t, and Y and T as follows
Z;
= 68.665 + 5.47 ( t; -17.5) _ 0.438(T; -
325)_
125
12.5
19.607C;-17.5)2
12.5
_44.075e~-325)2
125
(6)
0.457 ( t; -17.5) (T; - 325)
12.5
125
or
Z;
= -275.89293 + 4.924456t; + 1.8351344T; 0.12548t;2 - 0.0028208T;2 - 0.OOO29248t;T;
(7)
The response surface has also been analyzed to see the
adequacy of the fitted model and the degree of fit of the error
term by using Minitab DOE. The estimated regression coefficients, the analysis of variance and unusual observations for
Z are given in Tables 3,4 and 5, respectively. For this model,
the Minitab printout is given as: Response Surface Regression:
Z versus x;y. The analysis was made using coded units.
It is essential to verify the adequacy of the fitted model
Z; = 68.67 + 5.47 x; - 0.44 Y; -
(8)
19.61x; - 44.08yJ- O.46x;y;
172
MINERALS & METALLURGICAL PROCESSING
Table 3 -
Estimated regression coefficients for z.
eoef
SE eoef
T
p
68.6646
5.4703
-0.4380
-19.6074
-44.0752
-0.4574
5.329
4.173
4.151
6.556
6.084
5.936
12.885
1.311
-0.106
-2.991
-7.244
-0.077
0.000
0.202
0.917
0.006
0.000
0.939
Term
Constant
x
Y
x*x
y*y
x*y
••
(a) 99
1!
.
e
CU
~
50
Q.
10
1
-40
S = 14.45
R-Sq = 72.3%
R-Sq(adj) = 66.5%
0
Residual
-20
(b)
Table 4 -
Analysis of variance for z.
Source
DF
Seq
55
Adj.
MS
F
P
Regression 5 13,075.3 13,075.3
255.6
Linear
2
366.2
Square
2 12,818.5 12,818.5
Interaction
1.2
1.2
2,615.07
183.12
6,409.24
1.24
12.53
0.88
30.71
0.01
0.000
0.429
0.000
0.939
Residual Error
5,008.8
5,008.8
208.70
29
18,084.2
24
Total
Table 5 -
Adj
55
o
Obs
9
13
14
30
9
13
14
30
"'ijGi:
z
Fit
residual
SE
fit,
Residual
St
85.000
90.000
30.000
10.000
51.401
58.795
58.715
9.557
4.674
4.146
4.388
11.793
33.599
31.205
-28.715
0.443
2.46 R
2.25 R
-2.09 R
0.05 X
20
0
-20
I
-30
-20
-10
0
10
Residual
•
• -• •••
• ••
o
15
20
n
30
• •
.•:...
30
45
Fitted Value
•60
•••
•
(d) 40
R denotes an observation with a large standardized residual.
X denotes an observation whose X value gives it large influence.
No evidence of lack of fit (p '" 0.1).
20
Because an incorrect or under-specified model can result in
misleading conclusions. For this quadratic model, the p-value
for lack offit is >0.10, suggesting that this model adequately fits
the data. It is also seen that the adjusted-R2 is 66.5%, and this
can be accepted as a high value for such designs .An adjusted-R2
of 66.5% indicates that the predictors (time and temperature)
explain 66.5% ofthe variance in response variable. The error
term for this model, S = 14.45, is smaller because the error
variability was reduced by the model used.
Table 3 gives the coefficients for all terms in the model. The
great p-values for the linear terms (p = 0 .202 and p = 0 .917) suggest there is no significant linear effect in the response surface.
But the small p-values for the squared terms (p = 0.006 and p =
0.000) suggest there is curvature in the response surface. Small
p-values for time squared and temperature squared suggesting
these effects may be important.
Minitab generates residual plots (Fig. 3) that can be used
to examine how weIl the data fit the model. The normal probability plot shows an approximately linear pattern with a norMINERALS & METALLURGICAL PROCESSING
I
(c) 40
Unusual observation for z.
Std
order
2
1
40
20
173
-20
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Observation Order
Figure 3 - Residual plots for z: (a) normal probability plot
of the residuals, (b) histogram of the residuals, (c) residuals VS. the fitted values and (d) residuals vs. the order of
the data.
mal distribution. The plot of residuals versus the fitted values
shows that residuals get greater as the fitted values increase,
which may indicate the residuals have nonconstant variance.
The plot of residuals versus observation order shows that
residuals are independent.
Vol. 25, No. 3 - August 2008
Conclusions
The concentrate grade and recovery of B 2ü 3 increase at the
maximum expansion values, while the grade ofB 2ü 3 decreases.
Thus, in this study, to lower the amount of tailings produced,
a quadratic model was chosen that predicted the applied temperature and duration of the material placed in the oven.
With this model, the coefficients and the relation to the volume
expansion (z), period of heat treatment (t) and temperature (T)
were determined for the fraction of 25.0 to 12.5 mm (8 g).
The adequacy of the fitted model and the degree offit ofthe
errorterm are regarded as encouraging. Thus, by means ofthis
quadratic model it seems possible to predict volume change of
a borax particle in the muffle fumace. Determining the similar
parameters for any scale fumace, it will be possible to make
new modeling.1t is hoped that this procedure will contribute a
novel concentration process in the field of tincal beneficiation
particularly in Klrka Plant, Turkey.
References
Akar, A, Mordogan, H., Kemal, M., Batar, T, Kahraman, B., and Helvacl, C.,
2005, "Bor Cevherine Uygulanan ISII 1~lemler," I. Ulusal Bor C:all~taYI,
TAEK-ANKARA, Nisan.
Akdag, M., Batar, T, et al., 1993, "Ozel Dizaynll Flnnlarda Boraksln Ekspansiyon
Yöntemiyle Zenginle~tirilmesi i.;:in Belirlenen Optimal Sartlann Denenerek
Tesise Uygunlugunun Ara~tlnlmaSI," Tübitak MiSAG 21 Proie Sonu~ Raporu,
Proie Ara~tlrmaclsl, izmir, Türkiye.
Aytekin, A, Akdag, M, et al., 1992, "Tinkai (Boraksl Cevherinin Patlatma
Yoluyla Zenginle~ebilirliginin ve Bu Yöntemin Bilinen Mevcut Yöntemler
Yerine ikamesinin Ara~tlnlmaSI," Tübitak MAG 838, MiSAG 9, Proie Sonu~
Raporu, izmir, Türkiye.
August 2008 • Vol. 25 No. 3
Batar, T, 1996, "Beneficiation of Borax by Calcination," Doctoral dissertation,
Dokuz Eylul University, Izmir.
Batar, T, Akdag, M., Kahraman, B., and Celik, M.S., 1995, "Dry processing of
boron minerals for the abatement of environmental pollution," 6th Balkan
Conference, pp. 467-472, Ohrid-Macedonia.
Batar, T, Kahraman, B., Cirit, E., and Celik M.S., 1998, "Dry Processing of Borax
by Calcination as an Alternative to Wet Methods," International Journal of
Mineral Processing, Vol. 54, pp. 99-110.
Buyuran, S., and Ylldlrlm, G., 1984, "Dü~ük Tenörlü Bigadi~ Kolemanit Cevherinin Pilot Tesis Döner Flnnda Kalsinasyonu ve Zenginle~itirilmesi," TÜBiTAK
Gebze, February.
Celik, M.S., Atak, S., and Onal, G., 1993, "Flotation of boron minerals," Minerals
& Metal/urgical Processing, pp.149-153, August.
Celik, M.S., Batar, T, Akin, Y, and Arslan, F., 1998, "Upgrading schemes for
boron minerals through calcination," Minerals & Metal/urgical Processing,
Vol. 15, No. 1, February.
Civelekoglu, H., Tolun, R., and Bulut~u, N., 1987, "Inorganik Teknoloiiler I," iTÜ,
Vakfl kitap yaYlnlan No. 17, istanbul.
Helvacl, C., 2005, "Borates," in Encyclopedia of Geology, Selley R.C., Cocks,
L.R.M and Plimer, I.R., eds., Elsevier, Vol. 3, pp. 510-522.
Köse, H., Batar, T, Kahraman, B., Ediz, N., and Erdogan, N., 2002, "Boron
strategy of the world and its importance for Turkey," 1st International Boron
Symposium, Int. Journal of Mineral Processing, Kutahya, Turkey.
Miles, D.E., 1973, "Rotary Apparatus for Treating Colemanite Ore," Stansteel
Corp., U.S. Patent No. 3712598, Los Angeles, California, January.
Myers, R.H., and Montgomery, D.C., 1995, Response Surface Methodology:
Process and Product Optimization Using Designed Experiments, John Wiley
& Sons, New York.
Sener, S., and Ozbayoglu, G., 1992, "Determination of calcination parameters
of ulexite and possibility of separation from colemanite," in Proceedings
of IVth International Mineral Processing Symp., Ozbayoglu, G., ed., pp.
538-548, Antalya.
Smith, P.R., and Walters, RA, 1980, "Production of Colemanite at American
Borate Corp.'s Plant Near Lathrop Wells," Nevada.
Wilson, D., and Burnwell, T, 1975, Method for Processing Colemanite Ore,"
Tenneco Oil Company, U.S. Patent No. 3865541, Houston, February.
174
MINERALS & METALLURGICAL PROCESSING
Turan Batar