Development of HD silicas for tires - processes, properties, performance.

A tire has to fulfill several demands: good dry and wet grip, secure braking, reasonable service life, good treadwear and low rolling resistance. The development of tires with these desirable properties was made possible (ref. 1) by using newly developed highly dispersible silica together with silanes and special solution SBR polymer (refs. 2-4).

The "green tire" with a silica tread, introduced in 1992, provides a low rolling resistance. A study by Kline International (ref. 5) from 1998 indicates an annual market growth of silicas of approximately 4% for passenger car tires.

Another increasingly important field for the use of highly dispersible silica is the truck tire market. Truck fleet owners will save a lot of money if their tracks are equipped with low rolling resistance tires. Wolff reported (ref. 6) that rolling resistance was reduced by 30%, wet traction remained nearly stable and the treadwear index was decreased only by 5% when a silane-modified silica was used to replace N220 carbon black in a natural rubber truck tread. A decrease of the rolling resistance by 30% indicates fuel savings of about 6%. To maintain the treadwear level, it is necessary to use a highly dispersible silica.

Formation of precipitated silicas

The formation of precipitated silicas (ref. 7) is usually a chemical reaction between sodium silicate (water glass), and an acid, sulfuric acid in most cases.

The product of this reaction between sodium silicate and sulfuric acid is a precipitated silica. By-products of the reaction are sodium sulfate, which must be washed out, and water. To control this process, it has to be taken into account that the formation of precipitated silica is reversible. Under special conditions, for example high pH value and temperature, the reaction will be on the side of the raw materials. The chemical reaction is an equilibrium which can be influenced by the process parameters.

The process of silica formation starts with the raw material storage. One tank is used for sodium silicate, the other for sulfuric acid. The chemical reaction mentioned above is carried out in precipitation tanks. The precipitation tanks are equipped with agitator systems to achieve excellent mixing of the components. Typical reaction times are between one and four hours at temperatures between 50 and 90 [degrees] C. After the precipitation, the suspension is pumped into filter presses to separate water and dissolve sodium sulfate. The obtained filter cake can be dried in different types of short and long time dryers. The dried silica can optionally be milled or granulated and then shipped to the customers.

For the development of new products, a fundamental knowledge about the influence of the parameters of each process step on the product properties is necessary. The most important parameters in the precipitation are the pH-value and the temperature of the reaction, the concentration of the components, the reaction time, as well as dosing and mixing of water glass and sulfuric acid. The most important properties of the silica which can be controlled in the precipitation step are surface area, structure of the silica, dispersibility of the silica, silanol group density, which means chemical activity of the surface, and particle size distribution.

In order to learn how to create special silica structures it is helpful to monitor the reaction carefully. In the very beginning, small isolated particles, called primary particles, are formed. The concentration of these particles is low, the particles are far away from each other. The particle size is in the range of a few nanometers. With ongoing time, the amount and size of particles increases. When the concentration is sufficiently high and the particles are close enough together, a reaction between the primary particles can occur. Si-O-Si bonds between primary particles are formed. The resulting larger particles are called aggregates. The still ongoing process results in a continuous growth of the number and size of the aggregates. If the concentration reaches a certain limit, the aggregates are close enough for the formation of greater units, called agglomerates. These agglomerates are connected via Van-der-Waals forces and hydrogen bonds. The formation of agglomerates is reversible. They can be destroyed, for example, by mechanical treatment.

Development of highly dispersible (HD) silicas means to find the best combination of aggregate and agglomerate sizes and numbers under very special conditions. On the one hand, the silica structure must be well prepared to be penetrated by the polymer during the beginning of the first mixing stage (incorporation). On the other hand, the structure which is partly penetrated by the polymer must be decomposed easily during the further mixing process.

During the filtration of the silica, the most important process parameters are filling and washing time, solid content and electrical conductivity. The obtained filter cake can be dried on various types of dryers. The easiest way to classify the different types is to divide them into short and long time dryers. The period the silica stays in short time dryers is in tire range of seconds, for long time dryers in the range of hours.

The most important process parameters are, as already mentioned, type of dryer, drying temperature, solid content and time. The properties which can be influenced in the drying step are the water content, the surface area, the structure and the dispersibility of the silica, as well as the particle size distribution.

Finally, the silica can be granulated in different ways to obtain a dust free product, an important demand of the robber industry. The important parameters are feed rate and granulation pressure. During granulation, the silica structure, particle size distribution, dispersibility and dust content can be influenced. Table 1 shows the influence of process variations on the properties of the final product. The two main process steps are shown.

A precipitation at a high pH-value and the use of a long time dryer lead to a poor dispersibility in rubber. The resulting product is a conventional silica. If the pH-value during the precipitation is changed from a high to a lower level (all the other parameters are comparable to the first process) by using the same type of dryer, the quality of the dispersibility changes from poor to medium. The product is a semi-HD silica like U3370.

Using a combination of a precipitation at a high pH-value together with a short time dryer leads to an HD silica with a good dispersion behavior. An optimized precipitation, with some other changes in process parameters and the use of a short time dryer, result in an HD silica with excellent dispersion. Sometimes even small modifications in process parameters lead to major changes in product properties and in rubber behavior.

Properties of the silica

The best pre-condition for excellent dispersion is the design of a tailor-made silica structure. The classical structure test method, which is frequently used for carbon black, is the DBP-value. This value describes the available void volume in the silica for dibutylphthalate (DBP). The higher the DBP-value, the higher the structure. For example: A conventional silica has a DBP-value of 175 g/100g, the highly structured silica U7000 a DBP-value of 210 g/100g.

Another method is the measurement of the void volume while increasing the pressure on the silica from 0 to 30 MPa. This demonstrates that the structure of the HD silica U7000 is higher and less fragile than that of the conventional silica and U3370 GR (figure 1).

[FIGURE 1 OMITTED]

The improved dispersion characteristics of U7000 can be explained as follows: During the beginning of the mixing process when very high shear forces exist, the initial voids in the silica U7000 survive for a longer time than those of conventional silica. Therefore, the polymer has more time to penetrate into the voids of U7000. During the following de-agglomeration, a decomposition of silica agglomerates (which were penetrated with polymer) occurs. This is reflected in the excellent dispersion of U7000. In comparison, the conventional silica is initially strongly compacted and difficult to disperse. This measurement of the void volume is a good method to explain and therefore predict large differences in the dispersion behavior of a silica.

The easiest way among different methods to determine the dispersion is the optical microscopy (Phillips method). The conventional silica shows a lot of undispersed particles in the green tire tread compound, while fewer undispersed particles are observed by using the semi-HD silica U3370. Only the use of the recently developed HD-silica U7000 results in a very small amount of undispersed particles of a smaller size (figure 2).

[FIGURE 2 OMITTED]

In another test with the surface analyzer, a tiny needle scans the vulcanizate surface and monitors the surface roughness. The comparison of the topography of the conventional silica and the newly developed silica U7000 (ref. 8) shows quite clearly the different dispersion behaviors.

The dispersion coefficient has been developed in order to detect even small differences in the dispersion behavior: Testing was carried out according to the method described in the literature (ref. 9). Using a razor blade, the vulcanizates are cut into thin pieces which then have a glossy appearance. These test pieces are investigated under an optical microscope. The resulting photos with a wide scale of grey tints are then reduced to black and white. The black areas represent the silica, the white area the background. The dispersion coefficient can be calculated on the basis of the size of the black and white areas, usually on 40 different photos. For the calculation of the dispersion coefficient, the filler volume and a correction with the Medalia factor are taken into account.

D(%) = sum of particle area/number of photos - 10,000 - Medalia factor/ filler volume area of the photos

factor of Medalia = filler volume/100 + 0.78/2

Figure 3 shows such dispersion coefficients for various silicas. The conventional silica gives a poor dispersion coefficient of only 73%; the best dispersion would have been 100%. U3370 shows improved dispersion behavior of about 86%. Only the use of the recently developed U7000 in the tire tread leads to a dispersion coefficient of 97%.

[FIGURE 3 OMITTED]

Monitoring of the quality of U7000

The quality and especially the dispersibility of a high-tech product like U7000 has to be monitored carefully. A new method was developed to predict the dispersion behavior of a silica in rubber compounds with a single analytical measurement (ref. 2). The use of the laser diffraction measurement for the determination of particle size distribution allows the detection of small differences in the dispersion behavior (figure 4).

[FIGURE 4 OMITTED]

This method uses an ultrasonic treatment of the silica suspension to de-agglomerate the silica particles, monitored by simultaneous laser diffraction. It is possible to detect particle sizes between 40 nm and 500 [micro]m. The resulting particle size distribution of U3370 and of U7000 shows a bimodal distribution. The main peak in the range of about 10[micro]m describes the initial structure with large agglomerates of silica. This higher structure is partly decomposed during the ultrasonic treatment which simulates the same energy input, as during the mixing process of the rubber compound. The easier this decomposition is, the larger and bigger the first peak in the range of 0.5 [micro]m. At a constant ultrasonic energy input the deagglomeration (dispersion) of U7000 is more advanced than with U3370, the portion of de-agglomerated small particles is significantly higher. The relation between the peak height of the original agglomerates and the peak height of the decomposed agglomerates is defined as the WK-coefficient (figure 4).

There is a good correlation between this WK-coefficient and the dispersion coefficient. The lower the WK-coefficient (which means a larger and higher peak of decomposed agglomerates) the better the dispersion. This has become now a suitable analytical method to monitor the quality of U7000 production.

Performance of HD silicas

A tire has to fulfill three main properties: Rolling resistance, wet grip and abrasion resistance. This is shown in the so called "magic triangle" of the tire performance (figure 5). The use of a conventional silica instead of carbon black in tire treads improves the rolling resistance and the wet grip (refs. 3, 4, 6 and 10); the treadwear is problematic.

[FIGURE 5 OMITTED]

The use of a highly dispersible silica is necessary to obtain improved rolling resistance and wet traction and a good abrasion resistance on an equal level as with carbon black.

The significantly improved abrasion resistance can be measured in the laboratory (refs. 11 and 12) or in an extensive road test. Figure 6 illustrates the improvement of the treadwear by using different silica grades at various speed and force indices.

[FIGURE 6 OMITTED]

The reliability of the prediction of the abrasion resistance with the dispersion coefficient is demonstrated by comparison with the road test results (figure 7). Changing from a conventional to an HD silica leads to an improvement of the treadwear index of 20%.

[FIGURE 7 OMITTED]

Trends in tire performance

Additionally, the automotive industry asks for a further reduction of the rolling resistance by 20% which would lead to further fuel savings of approximately 5%. This demand is not only valid for passenger car tires but also for truck tires. The key properties of a truck tire are treadwear and rolling resistance. Due to the heavy load and relatively slow driving speed, the fuel consumption of trucks depends very much on the rolling resistance (ref. 12). Any improvements here are highly welcome and fleet owners who carefully monitor fuel consumption as one of their major expenses request low rolling resistance tires. The fuel consumption of a truck alone contributes to approximately 60% of the overall truck operating costs.

Another important key property of a truck tire is a low heat build-up to ensure a long service life of the carcass as well as greater driving safety. If the heat build-up is low, the tire can be retreaded more often.

The silica technology is not yet established in the truck tire due to the fact that it makes the price calculation more difficult and because the polymer system is different.

For a passenger car tire which already has a silica tread. the situation is different. The possibility of achieving the reduction of rolling resistance just by a further optimization of the tread compound is very limited.

This is not only valid for the tire tread, but could also be extended to the whole tire body. The different parts of a passenger car tire contribute to the rolling resistance in different amounts (figure 8). The compounds of the truck tire body influence the rolling resistance of truck tires to a much greater degree (ref. 10) than in passenger car tires (figure 8).

This means that the tire body contributes to the rolling resistance of a passenger car tire for approximately 50% and to that of a truck tire for 70%. Traditionally, most compounds of the tire body are filled with carbon black.

Carbon black and silica have a different influence on the rolling resistance (measured as tan [delta] 60 [degrees] C) and the service life (measured as heat build-up [DELTA][T.sub.Center]) (figure 9).

[FIGURE 9 OMITTED]

To choose a silica for the tire body, it is not necessary to take the resulting abrasion resistance into account. The treadwear is not only influenced by the dispersion behavior of the silica used, but also by its surface area. This can be demonstrated with the Grosch measurement (refs. 11 and 12). The higher the surface area of the silica, the better the abrasion resistance (figure 10).

[FIGURE 10 OMITTED]

It is therefore possible to select silicas which reach the highest reduction in tan [delta] 60 [degrees] C and the highest reduction in heat build-up, but also good reinforcing and dispersion behavior in certain tire body compounds. As shown in figure 9, silicas with a lower BET surface area of approximately 120 [m.sup.2]g fulfill these requirements.

Low surface area silica for the use in the tire body

EXP 7006 was developed based on the targets shown in table 2. The WK-coefficient is very low. This suggests that the dispersion behavior of the silica inside the rubber compound is better than 95%.

EXP 7006 was tested in a basic carcass compound (NR/E-SBR) (table 3). The newly developed silica was compared respectively to traditionally used carbon black types and conventional precipitated silicas. Each compound was mixed in three stages (table 4), and for the determination of compound and vulcanizate properties, standard ISO, DIN or ASTM test procedures were adopted. The dynamic testing was conducted on MTS 831 test equipment according to DIN 53513. Figure 11 shows the result of the NR/SBR compounds.

[FIGURE 11 OMITTED]

The use of a silica instead of a carbon black results in an increase in the Mooney viscosity during the mixing process due to the higher filler-filler interaction. The change to the HD silica EXP 7006 leads to a significant reduction of the Mooney viscosity which indicates a better processibility. This behavior is typical for an HD silica because of its weaker filler network. In comparison to the conventional compounds (carbon black, as well as silica), the compound with EXP 7006 shows clear advantages in heat build-up (increase of the temperature during dynamic strain), in the dispersion behavior and the tan [delta] 60 [degrees] C (which correlates with the rolling resistance). The static data (not shown here) are on the same level.

Further in-rubber tests were carried out to monitor the properties of different tire body compounds. EXP 7006 was tested in a sidewall compound and compared respectively with a silica with a higher surface area (150 [m.sup.2]/g) and carbon black. The used sidewall compound is shown in table 5, which is also mixed in a three stage mixing process according to the one described in table 4.

Sidewall compounds are normally produced with carbon blacks of the N 300 series or with the 300 series ill a blend with a soft black. The use of EXP 7006 leads to a clear reduction of the heat build-up rolling resistance and also shows advantages in comparison with the higher surface area silica.

The role of the tack of the green compound to guarantee an easy tire building is a further important property. The tack of a steelcord adhesion compound with the newly developed silica EXP 7006 was tested against a carbon black sidewall compound (table 5).

In comparison to a commonly used compound, the green compound with EXP 7006 shows clear advantages in the tack behavior.

Use of EXP 7006 in winter tires

Another application shows the great potential of HD low surface area silicas. The winter tire market has been growing steadily over the last five years (refs. 14 and 15). The German winter tire sales increased between 1994 and 1999 from 8.2 million to 15.3 million tires. The second and third biggest markets are in Japan, with a total consumption of 12.1 million, and in the U.S. with 10 million winter tires sold in 1999. The key property of such a tire is the grip. The grip must be maintained consistently on wet, snow-covered or even icy surfaces. This means that a winter tire must have a wide safety margin for unexpected extreme conditions, must be durable and maintain its good properties for the duration of its service life. The abrasion resistance does not play such an important role as in summer tires.

Skid resistance was identified as the critical property of the winter tire tread compound (ref. 16). The skid resistance in the temperature range of zero to -15 [degrees] C is related to the compliance of the tread material. The higher the compliance 1/E* or, to put it another way, the lower the complex modulus E*, the better the skid resistance. It was also found (ref. 17) that softer compounds with higher resilience show a higher fiction on ice in the temperature range of zero to -20 [degrees] C.

A winter tire tread compound contains a great amount of natural rubber. The HD low surface area silica EXP 7006 was tested in a basic NR/DCP compound (table 7) in comparison to a conventional low surface area silica. The HD silica U7000, and the conventional silica were also taken into account. The compounds were investigated with an Eplexor which records the complex modulus over a wide temperature range (from -40 [degrees] C to 80 [degrees] C) (figure 12).

[FIGURE 12 OMITTED]

The conventional low surface area silica shows a lower complex modulus than the higher surface area silicas due to the differences in the surface areas and therefore to the reduced total amount of silanol groups. This results in a lower filler-filler interaction. The HD low surface area silica EXP 7006 shows the lowest filler network which leads to the lowest complex modulus E*. Other studies in different polymer blends confirm this result. Therefore, the use of EXP 7006 results in a high compliance which indicates a very good skid resistance when used in a winter tread compound.

Conclusion

At the beginning of the 1990s, the requirements of automobile manufacturers for the simultaneous improvements of rolling resistance, wet grip and service life were met using tread compounds with special polymers and a high silica loading (in conjunction with silanes) instead of carbon black. Such compounds require the use of advanced silica grades with excellent dispersibility. The superior abrasion resistance of tread compounds due to the improvement in the dispersion behavior of the silica used was described with the comparison of a conventional silica, the semi-HD silica U3370 and the HD silica U7000. A new method was developed to monitor the quality of the HD silica U7000. With the special modified laser diffraction measurement, it is possible to predict the silica dispersion in rubber with its important consequences for tire wear.

Further reductions in rolling resistance without influencing the safety aspects can presently only be achieved by using the silica/silane system in tire body compounds. In comparison to the carbon black references, the newly developed low surface area HD silica EXP 7006 leads to a reduction of about 40% of the tan [delta] 60 [degrees] C (which correlates with the rolling resistance), as well as to a reduction of about 30% in the heat build up (which correlates with the service life). Compared to conventional silicas, a further reduction of 10% of tan [delta] 60 [degrees] C and heat build-up, respectively, could be reached by using the tailor-made type which exhibits excellent dispersion behavior.

For the development of a new generation of winter tires, the low surface area HD silica EXP 7006 offers a lot of possibilities. Winter tread compounds with EXP 7006 show a low complex modulus at 0 [degrees] C and a high compliance at -20 [degrees] C which indicates a high skid resistance on wet, snow-covered and even on icy surfaces.

Figure 8 - influence on rolling resistance

Casing        10%
Sidewall      10%
Innerliner     5%
Bead           5%
Tread         50%
Belt          20%

Casing        24%
Belt          20%
Tread         30%
Bead          16%
Sidewall      10%

Note: Table made from pie chart.
Table 1 - process variations - changes in
properties

Precipitation             Drying   Dispersibility          Product

High pH                Long time             Poor     Conventional
                                                            silica
Lower pH               Long time           Medium   Semi HD silica
High pH               Short time             Good        HD silica
Optimized             Short time        Excellent        HD silica
precipitation   (different type)
Table 2 - comparison of the analytical data of
U7000 with EXP 7006

U7000: Use in tire tread
EXP 7006: Use in tire body

                           U7000    EXP 7006

  CTAB ([m.sup.2]/g)        160       120
Surface area
  BET ([m.sup.2]/g)         170       125
WK-coefficient              0.5       1.0
(WK< 3[DELTA]
   HD-silica)
Table 3 - formulation of the basic NR/E-SBR
compound

                                            1 (phr)   2 (phr)

NR                                            60        60
E-SBR (styrene 23.5%, vinyl 16%, oil 27%)     55        55
N660                                          50        --
Silica                                        --        50
TESPT/N330 (50:50)                            --         3
ZnO                                            3         3
Stearic acid                                   1         1
Aromatic oil                                   6         6
Resin (ethyl-butyl-phenol                      4         4
  condensation product)
DPG                                           --        1.5
CBS                                           --        1.5
TBBS                                          0.6       --
Sulfur                                        2.2       22
Table 4 - mixing conditions

Mixing procedure
Internal mixer GK-N (1.5l)
Ram pressure: 5.5 bar
Loading: 0.70 (stage 1), 0.67 (stage 2); 0.65 (stage 3)
Cooling system temperature: 70 [degrees] C (stage 1); 90 [degrees]
C (stage 2); 50 [degrees] C (stage 3)

Stage 1
0' - 1.5'            Polymer
1.5' - 3.5'          Other components of stage 1
3.5'                 Cleaning
3.5' - 5.5'          Mixing
5.5'                 Cleaning
5.5' - 6.5'          Mixing; dump
Batch temperature:   <155 [degrees] C

Stage 2
0' - 2'              Batch stage 1
2' - 5'              Temperature about 150 [degrees] C
5'                   Dump
Batch temperature:   <155 [degrees] C

Stage 3
0 - 2'               Batch stage 2, accelerators, sulfur
2'                   Dump
Batch temperature:   <110 [degrees] C
Table 5 - sidewall formulation

                                     1     2     3

NR                                  50    50    50
BR 96%, 1.4 cis                     50    50    50
N660                                20    20    20
N375                                30    --    --
Silica (CTAB 150 [m.sup.2]/g)       --    30    --
EXP 7006                            --    --    30
TESPT/N330 (50:50)                  --     3     3
ZnO                                3.5   3.5   3.5
Stearic acid                         2     2     2
Aromatic oil                         7     7     7
Cumarone-indene resin                3     3     3
Microcristalline wax                 3     3     3
Antioxidant (dimethylbutylphenyl   2.5   2.5   2.5
p-phenylendiamine)
DPG                                 --     2     2
CBS                                 --     1     1
TBBS                               0.8    --    --
Sulfur                             1.8   1.8   1.8
Table 6 - steelcord adhesion compound

                                                  1      1

NR                                               60     60
E-SBR (styrene 23.5%, vinyl 16%, oil 0%)         40     40
N326                                             50     --
EXP 7006                                         --     50
Conv. Silica                                     15     15
TESPT/N330 (50:50)                               --      7
ZnO                                               8      8
Stearic acid                                      1      1
Cofill 11 GR (conv. silica/resorcinol (1:1)       5      5
Aromatic oil                                      3      3
Antioxidant (IPPD)                              0.5    0.5
Antioxidant (pol. trimethyldihydroquinoline)    0.8    0.8
Hexamethylentetramine                           1.5    1.5
TBBS                                            0.8    1.5
Sulfur unsoluble                                  4      4
DPG                                              --      2
Cure temperature: 160 [degrees] C/t95%
Tack (N])                                      13.8   26.7
Table 7 - mixing process of the NR/DCP
compounds

Rolling mill 3.0, Schwabenthan
Cooling system temperature: 25 [degrees] C/30 [degrees] C
Upm: 17/20

Stage 1              Phr                 Min.

NR                   100            0 - 15 polymer
Silica             Variable       2 - 12 1/2 silica
Zinc stearate        1.0          12 - 20 1/2 silica
Dicumyl peroxide     2.0       20 - 24 other components
(98%)                         24 - 30 processing on mill
                                      sheet off

Acknowledgements

"Development of HD silicas for tires - processes, properties, performance," is based on a paper given at the April, 2001 meeting of the Rubber Division.

"Carbon-silica dual phase filler application to passenger tread compounds," is based on a paper given at the April, 2001 meeting of the Rubber Division.

"Reinforcement of BIIR with silica," is based on a paper given at the April, 2001 meeting of the Rubber Division.

References

(1.) EP 0501 227, US 5.227.425.

(2.) A. Blume, S. Uhriandt, GAK 2, 116 (1999).

(3.) S. Wolff, "The influence of fillers on rolling resistance," presented at the 129th Meeting of the Rubber Division American Chemical Society, New York, April 8-11 (1986).

(4.) G. Agostini, J. Berg and Th. Mateme, "New compound Technology," Oct. 1994, Akron, OH.

(5.) European Rubber Journal, 10,20(1998).

(6.) S. Wolff, "'Silica based tread compounds; background and performance," Tyre Tech., Basel, Switzerland 10 (1993).

(7.) R.K. Her, Chemistry of Silica, John Wiley & Sons (1979).

(8.) A. Blume, paper 73, presented at a meeting of the Rubber Division, American Chemical Society, Chicago, Illinois, April 13-16 (1999).

(9.) R.H. Schuster, H. Geisler and D. Bubmann, 2nd conference on Carbon Black, Mulhouse (France), 9 (1992).

(10.) G. Heinrich, Materialforschung am Reifen, presented at the workshop, Reifen of the Deutsches Institut fiir Kautschuktechnologie, May 25-26 (1998).

(11.) K.A. Grosch, Kautsch. Gummi, Kunstst. 50, 841 (1997).

(12.) K.A. Grosch, ACS Rubber Div. Meeting (Cleveland) 119 (1997).

(13.) W. Niedermeier and B. Freund, paper 28, presented at a meeting of the Rubber Division, American Chemical Society, Nashville, TN, Sept.29 - Oct.2 (1998).

(14.) Gummi Bereifung II, 34 (1998).

(15.) NeueReifenzeitung 10, 88 (1999).

(16.) S. Futamura, Rubber Chem. Technol. 69, 648 (1996).

(17.) A. Ahagon, T. Kobayashi, M. Misawa, Rubber, Chem. Technol. 61, 14 (1988).