Hydroxyl-Terminated Polybutadiene. I. A Studyof the Polymerization of Butadiene in the Presenceof Hydrogen Peroxide

K. JANKOVA,1 D. KAMENSKI,1 V. GOTCHEVA,2 P. KOFFINOV 2

1 Prof. Assen Zlatarov University, 8010 Bourgas, Bulgaria

2 Institute of Petrochemistry and Oil Processing, Neftochim Co., 8104 Bourgas, Bulgaria

Received 16 July 1996; accepted 5 December 1996

ABSTRACT: The polymerization of butadiene in a solution of 1-butanol as a solvent andin the presence of hydrogen peroxide as an initiator was studied. The influence of fourfactors—polymerization temperature, concentration of the initiator, polymerizationtime, and stirring rate—on the yield, molecular weight characteristics, and the number-average functionality of the hydroxyl-terminated polybutadienes (HTPBs) were deter-mined. Regression models which adequately fit the experimental data were obtained.On this basis, the optimum values of the most significant factors—polymerizationtemperature and concentration of the initiator for obtaining maximum yield of HTPB—were evaluated. q 1997 John Wiley & Sons, Inc. J Appl Polym Sci 64: 2491–2496, 1997

INTRODUCTION action is also the decomposition of hydrogen per-oxide to HOO• and H•; however, the indicated

The preparation of terminally hydroxylated poly- reaction pathway is favored by a high tempera-butadienes can be accomplished by anionic or ture.8 Consequently, the products are found tofree-radical polymerization.1 The free-radical po- contain low molecular weight monofunctionallymerization of butadiene in a solution and in the polymers as well as high polymers containing twopresence of hydrogen peroxide as an initiator ap- or more hydroxyl groups. In addition to the hy-pears to be a more economical route to these liquid droxyl endgroups, small amounts of other oxygen-rubbers. The reaction medium is a common or- containing species, including aldehydes, ketones,ganic solvent for the monomer and the initiator, carboxylic acids, hydroperoxides, and epoxides,and alcohols are particularly suitable to this pur- have been detected.4,5,9

pose. Polymerizations are usually conducted for The commercial development of this technol-2–4 h at temperatures within 90–1307C, yielding ogy7 requires a specification of the conditions foressentially water-white odorless liquid prod- the synthesis of terminally hydroxylated polybu-ucts.2–7 The polymerization is initiated predomi- tadienes with definite functionality and molecularnantly by HO• radicals. The key to achieving an weight characteristics. However, the technical lit-effective difunctionality is the termination taking erature contains little basic information on poly-place predominantly by a combination of growing merizations of this type. The conditions for prepa-macroradicals. All chain-transfer reactions in- ration of hydroxyl-terminated polymers based onvolving the solvent, initiator, polymer, and mono- dienes of average molecular weight in the ordermer occur as side processes. A significant side re- of a few thousands and an average content of 2.1–

2.3 hydroxyl groups per chain have mainly beenreported.1,3,6,8–13 Polymers containing hydroxylCorrespondence to: K. Jankova.

q 1997 John Wiley & Sons, Inc. CCC 0021-8995/97/132491-06 groups up to 5 or even more have been found only

2491

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied

2492 JANKOVA ET AL.

Table I Levels of Factors and the Responses of the Statistical Design of Experiments forHTPB Synthesis

Factors

z1 z2 z3 z4

Exp Runs (mass %) (7C) (h) (min01) y1 y2 y3 y4 y5

1 5.00 120 4 1000 60.6 5250 6070 3.07 5.622 2.75 120 4 1000 54.3 5450 12,100 2.22 3.623 5.00 100 4 1000 36.5 5380 10,250 1.91 2.624 2.75 100 4 1000 17.5 4090 7900 1.93 2.745 5.00 120 2 1000 58.6 4660 9180 2.05 4.726 2.75 120 2 1000 50.6 6000 15,040 2.50 4.597 5.00 100 2 1000 30.3 4020 8420 2.09 3.008 2.75 100 2 1000 24.7 5390 8700 1.61 3.719 5.00 120 4 600 59.4 4830 11,930 2.47 3.32

10 2.75 120 4 600 59.0 5560 12,320 2.22 5.2111 5.00 100 4 600 46.5 4180 8430 2.02 2.4312 2.75 100 4 600 37.6 5920 10,470 1.77 2.9613 5.00 120 2 600 60.6 4100 8240 2.01 4.0014 2.75 120 2 600 38.6 5700 10,360 2.03 5.2315 5.00 100 2 600 27.1 4890 8260 1.69 3.2716 2.75 100 2 600 19.4 5110 8460 1.65 3.40

in some isolated cases.14 The purpose of this arti- use by evaporation under a vacuum at 407C over-night. The experiments were carried out by usingcle was to evaluate the significant factors and

their optimal values for obtaining hydroxyl-termi- a factorial design of the first order and the secondorder, respectively.nated polybutadiene (HTPB) in the polymeriza-

tion of butadiene in the presence of hydrogen per- The number-average (MV n ) and weight-average(MV w ) molecular weights of polymer samples wereoxide.determined by GPC. The measurements were per-formed in THF solvent at 357C in a 1 mL/minflow, using a Waters 200 gel permeation chroma-EXPERIMENTALtograph equipped with four Ultrastyragel col-umns (1 1 106, 2 1 104, 1 1 104, and 1 1 103 AHTPB was prepared by free-radical polymeriza-pore size). The calibration curve was obtainedtion of butadiene initiated by commercial H2O2

with polystyrene standards having an MV w /MV n(32–35%) in 1-butanol as a solvent. Both the sol-õ 1.1 in the molecular weight range of 470,000 tovent and monomer were obtained from Neftochim1800. The hydroxyl groups’ content was deter-Co. (Bulgaria), whereas H2O2 was a product ofmined by acetylation.10,16 The polymers were ace-Fluka (Germany). For the polymerization, 85–tylated by excess acetic anhydride in dry pyridine102 cm3 (94.8–113.8 g) H2O2 (32.2%) was dis-(3 h, 1007C). The excess anhydride was hy-solved in 1210 cm3 (980 g) 1-butanol in a 5 Ldrolyzed with water (1 h, 1007C) and the aceticmetal autoclave in a nitrogen atmosphere. Then,acid received was titrated with alcoholic KOH us-2260 cm3 (1410 g) butadiene was condensed un-ing phenolphtalein as the indicator. The number-der pressure. The stirring speed of the reactionaverage functionality ( fU n ) was calculated frommixture was adjusted within 600–1000 min01 .the MV n value and the hydroxyl content (xOH, massThe polymerization temperature was 95–1207C.%) using the following formula: fU n Å MV nrxOH/After 2–4 h, the autoclave was cooled down andMOHr100, where MOH is the molecular weight ofthe reaction product was drained from the bottom.the hydroxyl group.The product was stored overnight followed by de-

cantation of the solvent. Liquid antioxidant WingRESULTS AND DISCUSSIONStay K (Goodyear Co.) was added to the oligomer

and the latter was concentrated in a liquid falling The effects of the initiator concentration (z1) , tem-perature (z2) , polymerization time (z3) , and stir-film.15 HTPB was, consequently, purified before

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied

HYDROXYL-TERMINATED POLYBUTADIENE. I 2493

ring speed (z4) on the yield of the oligomer (y1) ,the number-average molecular weight MV n (y2) ,weight-average molecular weight MV w(y3) , indexof polydispersity MV w /MV n (y4) , and number-aver-age functionality fU n (y5) were studied.

The experiments were based on a 24 factorialdesign with z1 , z2 , z3 , and z4 , each at two levels.The values of these factors and the responses ob-tained in various experimental runs are listed inTable I.

These data were fitted to the response function:

yi Å f ( x1 , x2 , x3 , x4) Å b0 / b1x1

/ b2x2 / b3x3 / b4x4 / b12x1x2

/ b13x1x3 / b14x1x4 / b23x2x3 / b24x2x4

/ b34x3x4 / b123x1x2x3 / b124x1x2x4

/ b234x2x3x4 / b1234x1x2x3x4 (1)

where variables stand for the following ratios:

x1 Å (z1 0 3.9)/1.1 x3 Å (z3 0 3)/1

x2 Å (z2 0 110)/10 x4 Å (z4 0 800)/200 (2)

The constants b0, b1, . . . , b1234 were evaluatedby the least-squares method. The mathematicalmodels obtained which include statistically signifi-cant coefficients only are presented in Table II.

The analysis of the model equations (Table II)shows that the temperature is the most signifi-cant factor for the yield of the oligomer and itsfunctionality. The next important variables arethe concentration of the initiator and the polymer-ization time. The stirring rate appears to be a lesssignificant factor. On the other hand, the interac-tions between the factors do not affect the yieldof HTPB.

A relatively more complicated equation is ob-tained for the number-average molecular weighty2 . The negative sign of the coefficient at x1 (con-centration of initiator) shows an inversely propor-tional dependence between x1 and y2 . The temper-ature is the significant factor and the stir numberis the insignificant one in this case, too. However,the interactions between the factors have a strongeffect on the number-average molecular weight.Perhaps, this is the reason for the impossibilityof obtaining a low molecular weight polymer,solely by increasing the concentration of the initi-ator. The temperature also has influence on theweight-average molecular weight (y3) and the

Tab

leII

Reg

ress

ion

Mod

els,

Sta

nd

ard

Dev

iati

ons,

and

F-c

rite

rion

Fcr

it

Res

pon

seF

un

ctio

nM

ath

emat

ical

Mod

els

Sta

nda

rdD

evia

tion

Fca

l(aÅ

0.05

)

y 1(D

,%

)y 1Å

42.5

81/

4.86

9x1/

12.6

31x 2/

3.84

3x3

6.2

1.87

8.74

y 2(M

V

n)

y 2Å

5021

.880

383.

13x 1/

149.

38x 20

128.

13x 1

x 2/

208.

13x 1

x 3/

135.

62x 2

x 420

1.8

2.55

8.89

/15

5.62

x 1x 4/

291.

88x 1

x 2x 40

230.

62x 1

x 2x 3

x 4

y 3(M

V

w)

y 3Å

10,3

800

288.

7x1/

1518

.7x 2/

803.

7x3/

571.

2x40

267.

5x1x

2/

775.

0x1x

350

2.8

4.96

9.55

/30

5x1x

4/

402.

5x2x

3/

615x

2x4/

676.

2x1x

2x30

283.

7x1x

2x4/

788.

7x1x

3x4

/23

0x1x

2x3x

4

y 4(M

V

w/M

V

n)

y 4Å

2.07

75/

0.24

38x 2/

0.12

38x/

0.09

50x 4/

0.11

63x 1

x 2x 3/

0.10

38x 1

x 2x 3

x 40.

1786

2.66

8.79

y 5Å

3.77

75/

0.76

13x 2/

0.31

75x 1

x 4/

0.33

88x 1

x 2x 4

0.46

353.

558.

74y 5

(fn)

polydispersity (y4) , respectively. It may be con-

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied

2494 JANKOVA ET AL.

Table III Levels of Initiator Concentrations and Temperature

Factors Upper Level /1 Base Level 0 Lower Level 01 Unit

Initiator concentration z1 (x1 in coded form) (mass %) 7 6 5 1Temperature z2 (x2 in coded form) (7C) 115 105 95 10

cluded that one of the most significant factors in- x1 Å (z1 0 6)/1 x2 Å 0 (z2 0 105)/5 (4)fluencing all characteristics of the polymerization

The constants A0 , A1 , A2 , etc., were estimatedproduct is the temperature.by using the least-squares method. By substitut-The optimum levels of the significant variables,ing them in eq. (3), we get the regression modelsi.e., z1 and z2 , were determined from experimentspresented in Table V. The standard deviations asbased on a 3n statistical design. The values ofwell as the calculated and critical values of the F-factors z3 and z4 were z3 Å 4 h and z4 Å 1000criterion are also shown there. In all cases, Fcalcmin01 , which were accepted from the preliminaryis lower than Fcrit and it can be concluded thatexperiments.the mathematical models adequately describe theThe 3n factorial design was chosen because theprocess under study.information obtained from it concerns both linear

The dependence of the initiator concentrationand quadratic components of the effects of the fac-on the number-average molecular weight is pre-tors. The quadratic component may imply a maxi-sented in Figure 1. As can be seen, the number-mum or minimum response at the same interme-average molecular weight (y2) drops rapidly whendiate factor combination or at a point outside theboth the concentration of initiator (z1) and therange examined for some or all the factors, indi-temperature (z2) increase in the studied experi-cating a need for further work at the differentmental range. This behavior is in correspondenceset of levels.17 Moreover, this design is quasi D-with the theoretical foundation of radical poly-optimum and it allows one to obtain more precisemerization kinetics. The maximum amount of ini-estimations of the regression coefficients.tiator which is necessary for obtaining a minimumThe actual values of the variables correspond-molecular weight appears to be 7 mass %.ing to the coded values are listed in Table III.

Calculations for the different yi ( i Å 1, 2,rrr5)Table IV shows the responses yi of statistical de-values can be performed by assigning arbitrarysign experiments. The results for the (yi ) at differ-values for x2 in the equations given in Table Vent values of z1 and z2 were fitted to the responseand then by solving the resulting quadratic equa-function:tions for x1 . This is shown in Figures 2–4.

It is interesting to note that the molecularyi Å f ( x1 , x2) Å A1x1 / A2x2 weight distribution (MV w /MV n ) becomes wider when/A12x1x2/A11x2

1/A22x22 , iÅ 1, 2, 3, . . . , 5 (3) the temperature and concentration increase (Fig.

2). This may be the result of the stronger influ-ence of these two factors on MV n than on MV w .where x1 and x2 are the coded forms for

Table IV The Responses of Factorial Design 32

Exp Runs z1 (mass %) z2 (7C) y1 (%) y2 y3 y4 y5

1 7.00 115 36.8 2500 7780 3.11 3.892 6.00 115 67.9 2950 8650 2.93 2.953 5.00 115 75.3 3740 8660 2.32 2.074 7.00 105 61.3 3330 10,130 3.04 2.105 6.00 105 47.0 3400 9350 2.75 3.926 5.00 105 69.5 4150 8860 2.13 4.447 7.00 95 37.3 3730 9400 2.52 3.248 6.00 95 40.9 4220 10,100 2.39 3.929 5.00 95 29.2 4760 10,900 2.29 4.54

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied

HYDROXYL-TERMINATED POLYBUTADIENE. I 2495

Table V Regression Models, Standard Deviations, and F-criterion

Response Standard Fcrit

Functions Mathematical Models Deviations Fcal a Å 0.05

y1 (D, %) y1 Å 59.5 0 6.4x1 0 0.37x21 / 12.1x2 124.8 6.07 9.28

0 11.6x1x2 0 11.4x22

y2 (MV n) y2 Å 3447 0 515x1 / 215x21 31,769 1.99 9.28

0 586.7x2 0 52.5x1x2 / 60x22

y3 (MV w) y3 Å 9499 0 185x1 0 78x21 0 885x2 714,630 0.09 9.55

/ 155x1x2 0 198x22

y4 (MV w/MV n) y4 Å 2.72 / 0.32x1 0 0.12x21 / 0.19x2 0.02428 2.02 9.55

/ 0.14x1x2 0 0.05x22

y5 Å 3.6 0 0.3x1 0 0.2x21 0 0.46x2 0.80664 0.92 9.55y5 ( fn)

/ 0.78x1x2 0 0.04x22

Figure 3 shows that the yield of HTPB slightly yield with increase of the initiator concentration.These results can be explained by the formationdepends on the initiator concentration at low tem-

perature levels. Any increase of temperature over of low molecular products diluted in 1-butanol,whose amount was not taken into consideration.1057C results in an insignificant decrease of theAt the same concentration of the initiator, theyield of the oligomer increases and the number-average molecular weight decreases with increas-ing polymerization temperature (Figs. 1, 3,and 4).

Figure 4 also shows that the combination of thelowest number-average functionality and molecu-lar weight values can be reached at the highestvalues of x1 . Under these conditions, the number-average functionality is about 3. It should benoted that a value of the number-average func-tionality close to 2 can be obtained when the tem-perature and initiator concentration are 1057Cand 7%, or 1157C and 5%, respectively. Underthese conditions, the values of MV n are 3330 and3740, and at MV w /MV n , 3.04 and 2.32, respectively.

The optimum values of z1 (concentration of the

Figure 1 Number-average molecular weight MV n (y2)as a function of the concentration of the initiator (z1)at different values of temperature. Figure 2 Constant level curves for y4(MV w /MV n ) .

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied

2496 JANKOVA ET AL.

Figure 3 Constant level curves for y1 . Figure 4 Constant level curves for y2 ( interruptedline) and y5 (continued line).

initiator) and z2 (temperature of polymerization) scribe appropriately the polymerization process.were determined by finding the maximum of the These models correlate reasonably well with theresponse function y1 subject to the inequality con- experimental data and they were used for esti-straints: mating the significant factors and their optimum

values.

REFERENCES

Optimize y1 Å f1(z1 , z2) r maximum

Subject to 3000 õ y2 õ 10,000

1 õ y4 õ 3

2 õ y5 õ 31. D. M. French, Rubb. Chem. Technol., 42, 71 (1969).2. J. D. Daury, Caoutch. Plast., 571, 85 (1977).3. Russ. Patent 193,715 (1965).4. J. A. Verdol and R. W. Ryan, U.S. Pat. 3,714,110The maximum of this objective function was found

(1973).by using the simplex method of Nelder and5. W. D. Vilar, Braz. Pat. 7,707,285 (1982).Mead.18 As a result, the following ultimate opti-6. Jpn. Pat. 58-127,711 (1982).mum values of variables were obtained:7. K. Jankova et al., Bulg. Pat. 41,562 (1985).8. C. Pinazzi, G. Legeay, and J.-C. Brosse, J. Polym.

Sci. Symp., 42, 11 (1973).9. J.-C. Brosse, M. Bonnier, and G. Legeay, Makro-

mol. Chem., 183, 303 (1982).

z1 Å 00.89(5.1% H2O2) y1 Å 75.9

z2 Å 0.95(114.57C) y2 Å 3620

y4 Å 2.65

y5 Å 2.63

10. W. D. Vilar, S. M. C. Menezes, and L. Akcelrud,Polym. Bull., 33, 556 (1994).

11. S. Poletto and Q. T. Pham, Makromol. Chem. Phys.,195, 3901 (1994).

These results were additionally experimentally 12. A. K. Bhowmick and H. L. Stephens, Handbookconfirmed. of Elastomers. New Developments and Technology,

Marcel Dekker, New York, 1988.13. Jpn. Pat. 57-8844 (1983).14. B. Veruovic, Sb. VSCHT Prace, 9, 217 (1983).CONCLUSIONS15. Chr. Karagjozov, K. Jankova, and Iv. Mladenov,

Chem. Tech., 43, 61 (1991).The polymerization of butadiene with hydrogen16. S. Siggia and J. Hanna, Quantitative Organic Anal-peroxide for obtaining hydroxyl-terminated poly- ysis via Functional Groups, Wiley, New York, To-

butadiene was studied. Based on the experiments ronto, 1979.and the modeling of the process, the significant 17. J. E. P. Box and N. R. Draper, Empirical Model-factors and their optimum values for obtaining Building and Response Surfaces, Wiley, Toronto,the maximum yield of the HTPB were deter- 1987.

18. R. S. Rao, J. Chem. Sci., 14, 23 (1993).mined. Regression models were obtained to de-

4164/ 8E65$$4164 04-29-97 14:29:07 polaa W: Poly Applied