4834
GLENA. RUSSELL
VOl. s1
[CONTRIBUTION FROM THE GENERALELECTRIC RESEARCHLABORATORY]
Catalysis by Metal Halides. IV. Relative Efficiencies of Friedel-Crafts Catalysts in Cyclohexane-Methylcyclopentane Isomerization, Alkylation of Benzene and Polymerization of Styrene BY GLENA. RUSSELL' RECEIVEDJANUARY 29, 1959 Friedel-Crafts catalysts have been separated into three categories.
I n the presence of a co-catalyst only the most acidic
of the metal halides will bring about the isomerization of methylcyclopentane to cyclohexane, e.g., gallium chloride-isopropyl
chloride. Less reactive systems which will not isomerize methylcyclopentane at 80" can still catalyze alkylation ofbenzene a t 25", e . g . , ferric chloride-isopropyl chloride. Still less active systems which can neither catalyze the isomerization of ~nethylcyclopentaneor the alkylation of benzene can catalyze the polymerization of styrene, e.g., boron fluoride-isopropyl chloride,
Introduction The results indicate a catalytic sequence of ALBra > In connection with the aluminum halide- GazBr6, GazC&> Fe2C16,SbC16 > ZrC14, BF3, BC13, catalyzed disproportionation of trimethylsilanes, a SnC14,SbC13. measure of the relative effectiveness of Lewis Results acids in catalyzing organic reactions was needed. Methylcyclopentane-Cyclohexane Isomerization. Comparable data from which the relative effective- --Methylcyclopentane was treated with a variety ness of Friedel-Crafts catalysts can be judged are of Friedel-Crafts catalysts in the presence and widely scattered. One of the earliest comprehen- absence of promoters or co-catalysts such as alkyl sive studies concerned the racemization of optically chlorides or water. After the desired reaction active a-phenethyl chloride. Although the re- period the products were analyzed by gas-liquid sults were complicated by concentration and sol- chromatography (g.1.c.). The results of these vent effects the following activity sequence was experiments are summarized in Table I. indicated in the presence of benzene : SbC16 > SnC14 TABLE I > BF3 > HgClz > TiCl,. Calloway summarized ISOMERIZATION OF METHYLCYCLOPENTANE the available data in 1935 in regard to the acylaReaction Cyclohexane/ tion of benzene and concluded that the relative MethylConditions Promoterb > FezC16 > Catalysta 'C. Hr. cyclopentane effectiveness of nietal halides was AU~C16 ZnClz > SnC14 > TiC14, Zrc14.4 0.001 25 18 Sone A more extensive reactivity series of catalytic ,001 25 18 None ability in the acylation of benzene was later pre.OOl 80 18 Xone sented by Dermer, et al., AlzC16 > SbCl, > FezC16 ,001 80C 18 None TeC14 > SnC14 > TiC14 > TeCI4 > BiC13 > Z n c l ~ . ~ ,0085 25 18 CHIC1 Grosse and Ipatieff have considered the catalytic .330 25 18 i-CSH7Cl ability of various metal halides in the reaction be,039 25 18 H2O tween benzene and ethylene in the presence of hy,004 25 18 0 2 drogen chloride.6 They found catalytic ability to ,004 25 18 CH3Br decrease in the order: A12C16> TaClb, ZrCh > CbC14 ,640 25 18 i-CaH7Br > BeC12 > TiCI4. Burk, in a compilation of data, ,395 25 18 HsO suggests the following series of reactivities : A1zB1-6 > < ,001 18 25 0 2 f&C16 > Fed& > ZrCl4 > BF3 > Tic13 > Tic14 > < ,001 26 18 ( CHs)aSiBr SbC15.' 2.60 25 18 CHKOBr In the present work aluminum bromide, gallium 0.187 25 18 i-C3H7Br chloride, gallium bromide, boron trifluoride, boron Catalysis by moist aluminum bromided trichloride, ferric chloride, stannic chloride, an25 48 8.77 timony trichloride, antimony pentachloride and 8.77 25 144 zirconium tetrachloride have been tested as cata50 24 4.60 lysts in the three reactions CI-I.,
+
be+
RX CsHs --f RCsHs Hx C&CH=CH? --f (CsH6CHCHz)n-
50 48 4.76 50 72 4.78 80 6 1.91 2.60 80 12 80 18 2.60 5.0 ml. of methylcyclopentane at 25' used in each experi0.29 mole. E Hetment; 1.87 moles of catalyst as MXs. erogeneous. Approximately 0.50 g. of anhydrous aluminum bromide handled without special precautions. (I
(1) Iowa State University. Ames. Iowa. ( 2 ) G. 4 . Russell, THISJ O U R N A L , 81, 4815, 4895. 4831 (1959). ( 3 ) R. Bodendorf and H. Bohme. A n n . , 616, 1 (1935); H. Buhine and 0 . S e r i n e , Ber., 71, 2372 (1938). (4) S . D Calloway, Chem. R e v s , 17, 327 (1935). ( 5 ) 0. C Dermer, D. M. Wilson, F. hI. Johnson and V. H. Dermer, THISJ O U R N A L , 63, 2881 (1941). (6) A . V. Grosse and V. K. Ipntieff. J . Om. Chem., 1 , 559 (1936). ( 7 ) R . E. Burk, "12th Report of the Committee on Catalysis, h'ational Research Council," John \\"iley and Sons. Inc., K e w Y o r k , N. Y., 1940, p. 251.
In addition to the experiments listed in Table I the following catalyst systems were found to produce less than 1 part in 1000 of rearrangement of methylcyclopentane in 18 hours a t 80': antimony(V) chloride-isopropyl chloride; antimony(111) chloride-isopropyl chloride; ferric chloride-
Sept. 20, 1959
EFFICIENCIES OF FRIEDEL-CRAFTS CATALYSTS
4835
isopropyl chloride ; boron chloride-isopropyl chlo- The present data, the data of Stevenson and Morgan, and earlier data," yield the following ride ; boron chloride-water ; boron fluoride-isoboron equation over the temperature range 25-140'. propyl chloride, boron fluoride-water ; fluoride-isopropyl alcohol; stannic chloride-iso2.098 In K = -4.93 + ___ propyl chloride, zirconium tetrachloride-isopropyl RT chloride. (AH = -4.20, AF2~s.l = 1.25 kcal. in31e-l) The aluminum bromide-catalyzed isomerization Alkylation of Benzene.-A series of experiments of methylcyclopentane has been considered previously by Pines, Abraham, Aristoff and Ipatieff .s were performed wherein benzene, alkyl chloride The results given in Table I are in complete agree- and metal halide were allowed to react in the presment with the conclusions reached by this group. ence of methylcyclopentane. The degree of reThe results summarized in Table I indicate that action between the alkyl chloride and benzene was aluminum bromide, gallium bromide and gallium followed qualitatively by g.1.c. by observation of chloride are superior catalysts for the isomerization the decrease in alkyl halide concentration and deof methylcyclopentane. Even for aluminum and tection of the alkylation product. Results are gallium halides a co-catalyst is necessary. sec.- summarized in Table 11. Alkyl halides are satisfactory promoters, whereas TABLE I1 methyl halides are ineffective. Water is a much O F BENZENE I N PRESEXCE OF METHYLCYCLOmore effective co-catalyst for aluminum bromide ALKYLATION PESTANE' than for gallium chloride. In the absence of ilAlkyllumination oxygen has no effect as a promoter for Cataatingc CsHlt either gallium chloride or aluminum bromide lystb agent CsHgCHI Alkylation products catalysis. AlrBrs CHaBr 0.002 Toluene, CHaBr all consumed Some qualitative statements can be made about AlsBrs i-CsH?Br ,230 Cumene, i-CaH7Br all consumed ,007 Toluene, CHaCl all consumed the fate of the co-catalyst in the experiments sum- G a d X CHsCl ,117 Cumene, i-CaK7Cl all consumed i-CsHC1 marized in Table I. Methyl halides in the pres- GarCls < ,0001 Cumene, i-CsH7Cl all consumed FerClsa i-CsHC1 ence of gallium chloride or aluminum bromide were BF: i-CaH7OH < ,0001 Cumene, i-CaH7OH all consumed only partially consumed whereas isopropyl halides BF, < ,0001 i-CaHC1 not consumed, cumene not i-CaHC1 detected in the presence of these catalysts as well as gallium not consumed, cuniene not bromide were consumed completely with the con- SnClr i-CaHiCI < ,0001 i-CsHC1 detected current formation of a C3-hydrocarbon. Isopropyl SbC18 i-CaHTCl < ,0001 i-CsH7C1 not consumed, cumene not halides were not consumed and little C3-hydrodetected carbon was found a t 80' in the presence of stannic SbClsd i-CaH7C1 < ,0001 Cumene, i-CaH7Cl nearly all consumed chloride, boron trichloride, boron trifluoride, zir- ZrClrd i-C6H7C1 < ,0001 i-CaH7C1 not consumed, cumene nut conium tetrachloride, ferric chloride or antimony detected trichloride. Antimony pentachloride resulted in a Reaction mixture contained 4.0 ml. of methylcpclopenthe partial decomposition of isopropyl chloride a t tane and 1.O ml. of benzene a t 25'; reaction conditions were 80' with the formation of some C3-hydrocarbon. 18 hours a t 25.0'. * 1.87 mnioles as MXs, MXr or MX6. 0.29 mmole. d Heterogeneous. Isopropyl alcohol in the presence of boron trifluoride a t 80' yielded substantial amounts of a The results summarized in Table I1 demonstrate Cs-hydrocarbon, presumably propylene from dethat the presence of an aromatic hydrocarbon rehydration. From the data of Table I, it is possible to obtain tards the isomerization of methylcyclopentane. A F and AH for the isomerization of methylcyclo- This is in agreement with the results of Pines, pentane to cyclohexane. Values calculated for et al.,s and undoubtedly arises from the fact that A F are -1.29 kcal. mole-' a t 25', -1.01 a t 50' benzene removes carbonium ions or incipient and -0.67 a t 80'. Over the temperature range carbonium ions from the reaction mixture. 25-80' AH is -4.70 kcal. mole.-'. These values R'MX4C6H6 +RCbHs 3- HX -I-hlXa can be compared with data obtained from heats of combustion and heat capacities. Thus, the difAll of the catalyst systems which bring about a ference in free energy between methylcyclopentane rapid isomerization of methylcyclopentane also and cyclohexane obtained in this manner is AF' alkylate benzene. Moreover, the systems gallium = -1.16 a t 25' in the liquid phase while AH' for chloride-methyl chloride, alumiunm bromidethe liquid phase isomerization of methylcyclo- methyl bromide, ierric chloride-isopropyl chloride, pentane to cyclohexane a t 25' is -4.27 kcal. boron fluoride-2-propanol and antimony pentam01e-l.~ Stevenson and Morgan, using a dif- chloride-isopropyl chloride, which are ineffective ferent analytical technique, have reported equi- catalysts for the isomerization of methylcyclopenlibrium constants for the isomerization of methyl- tane, will alkylate benzene. Catalyst systems cyclopentane to cyclohexane of 7.59 a t 27', 4.12 which are inefficient in the alkylation of benzene a t 59' and 1.99 a t 100' indicating AH to be -4.1.1° are boron fluoride-isopropyl chloride, stannic chloride-isopropyl chloride, antimony trichloride-iso( 8 ) H . Pines, B. M. Abraham and V. N Ipatieff, THIS JOURNAL. 70, propyl chloride and zirconium chloride-isopropyl 1742 (1948); H. Pines, E. Aristoff and V. N. Ipatieff, ibid., 71, 749 (1949); 72, 4055, 4304 (1950); 76, 4775 (1953). chloride. The inability of boron fluoride to cata-
+
(9) F. D. Rossini, el a t . , "Selected Properties and Thermodynamic Properties of Hydrocarbons and Related Compounds," Carnegie Press, Pittsburgh, P a . , 1953, p. 471-472. (10) D. P. Stevenson and J. H. Morgan, THIS J O U R N A L , 70, 2773 (1948).
(11) (a) A. L. Glasebrook and W. G. Lovell, ibid., 61, 1717 (1939); (b) G.C. S. Schuit, H . Hoog and J. Verheus, Rec. tvau. chim., 59, 793 (1940); (c) S. Mizushima, Y.Morino and R . Huzisiro,J. C h e w . Soc., Japan. 62, 587 (1941).
4S3G
GLENd.RUSSELL
Vol. s1
lyze the alkylation of benzene by an alkyl chloride is well known.12 Polymerization of Styrene. --A series of experiments was performed wherein styrene, metal halide and alkyl chloride were aliowed to react in the presence of rnethylcyclopeiitane. The products were analyzed for cyclohexane by g.1.c. and for polystyrene by precipitation in =ethanol. The results are summarized in Table 111.
faster when catalyzed by aluminum bromide than when catalyzed by gallium bromide.13c I n the disproportionation of ethyltriniethylsilane and other substituted silanes the catalyst sequence observed was hlzBr6> -1lzC16 > Xlz16> GapBrG > BCh, Gad&, ZrC14, SnC14, SbCl,.? The agreement between these catalyst series suggests that they arc both a measure of the acidity of the metal halide. Although an acidity series is not necessarily independent of the reference base, it is suggested that TADLE I11 the most general series of Lewis acidity is B12Br6 > POLTMERIZATIOS O F STYRESE I N PRESENCE O F hIETIIYLA%Cle> - k l h > GazBr6 > Gad& > Fe2C16> SbC15 > CYCLOPESTASE~ ZrCl4, SnC14 > BC13,BFJ,SbClJ. Catahlkyl C5Hlr Catalysts or catalyst systems can be grouped into lysth halideC CHgCH; I'ulymerization products three classes. The more active will readily AliDre CHaDr 0 0027 CIIdBr all consumed, 0.35 K . polyisomerize methylcyclopentane to cyclohexane a t styrene AlrRrs f-CsH7Br .0Oi7 i-CjH;IJr all consumed, 0 . 8 K . ~ ~ c , l y - ?So, alkylate benzene a t 23" and polymerize styrene styrene a t 23". Typical examples of these systems GapCle CIIsCI 0050 CI-IICI all ciinsumetl. 0.lj K . p o l y are aluminum bromide-isopropyl bromide, gallium styrene BI's r-CsH;Cl < 0001 i-CsIIiC1 nut completely corlsumed chloride-isopropyl chloride and aluniinum bromide-water. l 4 0 8 g . polystyrene P ~ ~ C I i-CaIliC1 & < ,0001 i - ~ a ~ laili ~consumed, l 1.1)I.', p o ~ y Less reactive catalysts whicii are extrcmely 5tyrenc inefficient in the isomerization of xiiethylcycloperiStiCli L-cdH7Ci < .0001 i-CaHiC1 not completely consumed, tane a t 23" nevertheless readily alkylate benzene 0.00 K. polystyrene SILL i-CaH7CI < .0001 CsHiC1 not completely conbumed, (1.211 and polymerize styrene. I n this group are galfi. polystyrene lium chloride-methyl chloride, aluminum bromide a Reaction mixtures contained 4.0 nil. uf r n e t h ~ - l methyl bromide, boron fluoride-isopropyl alcohol, taiie and 1 .O m l . of styrene a t 25'; reaction co11ditioils were ferric chloride-isopropyl chloride and antimony 18 hr. at 23'. * 1.87 nimoles as %IS3or MS,. 0.29 pentachloride-isopropyl chloride. mniuk. Heterugeiieous. Finally, a third group of catalysts can be recog.A11 catalysts which alkylated benzene also nized which are ineffective in the isomerization polymerized styrene. In addition, the systems of methylcyclopentane at 80" or in the alkylation boron fluoride-isopropyl chloride, stannic chloride- of benzene a t 25" but which will still polymerize isopropyl chloride and antimony trichloride- styrene. This group contains the catalyst systems isopropyl chloride also brought about the poly- stannic chloride-isopropyl chloride, boron fluoriclemerization of styrene. Styrene inhibited the iso- isopropyl chloride and antimony trichloridemerization of methylcyclopentane, presumably isopropyl chloride. These results are in agreement with the concluby acting as a trap for carbonium ions. In the cases of catalysts which could not alkylate benzene sion of Professor H. C. Brown that the catalytic (boron fluoride, stannic chloride and antimony ability of metal halides does not necessarily invokc trichloride) the co-catalyst (isopropyl chloride) complete ionization of a carbon-halogen bond. 1 3 f 1 J was only partially consumed. For catalysts capa- Instead, the complex between a metal halide and ble of alkylating benzene the co-catalyst (either alkyl halide can react with a nucleophile without the methyl or isopropyl halide) was completely con- intervention of actually free carbonium ions. This consequently results in a considerable gradasumed. tion of catalytic ability. Thus, to bring about the Discussion isomerization of methylcyclopentane we conclude The results suniiiiarized in Tablcs 1-111 indi- that a relatively free carbonium io11 is necessary. cate this sequence of catalytic ability for CHs CHJ Friedel-Crafts catalysts, i\.l2Br6 > GazBr6, Ga2C16 > FesCls, SbC16 > ZrC14,SnC14, BC&,BF3, SbC13. + R +AI2Br7- RH AI, Brj Xluniinum bromide is a more effective isoinerization catalyst than gallium bromide when isopropyl bromide is used as a co-catalyst while in the pres- On the other hand, catalyst systems which are inence of water aluniinum bromide is a more effectivc effective in the isomerization of inethylcyclopentane catalyst than gallium chloride. These results are still can alkylate benzene readily. Here a frec consistent with the observation that the reaction carbonium ion is not a requirement and alkylatioii of methyl broniide with toluene at 0" is more selec- need involve only an incipient carboniunl ion, for tive when catalyzed by gallium bromide13a than example, as in complex I.13.15In the case of the when catalyzed by aluminum bromide. 13b Nore- polymerization of styrene catalytic ability requires over, the rate of the alkylation reaction is rrAuch still less polarization of the carbon-halogen bond than in the alkylation of benzene. Thus, the poly(12) (a) A. Wohl and E:. Wertyporoch, Bel,., 64, 1357 (1931); ( h l merization of styrene can be catalyzed by various R . L. Burwell and S. Arc c r , THISJ O U R N A L , 6 ' , 032 (1942); ( c )
6"
G . F . Hennion a n d R . A . K u r z , i b i d . , 66, 1001 (1943). (13) (a) €1. C. Brown and C. R. Smoot, ibid.. 78, 6 2 5 5 (19.56); Ih) H. C.Bruwn and H. Jungk, ibid., 77, 5584 ( 1 9 5 5 ) ; (c) C. R . Smoot and H . C. B r o w n . i b z d . , 78, 0245, 0249 (19Zi5j; S. U . Choi ,nd H. C . Drown, ibtd., 81, 3315 (1959).
-
+
(14) Although the last system cannot alkylate an aromatic ring it does readily exchange deuterium atoms for aromatic hydrogen atoms [J. Kenner, 51. Polanyi a n d P. Szego, .\-atwe, 1 3 5 , 2G7 (1933)l. (15) €I. C.Druwn, II 11'. Peuriall, I,. 1'. E d d y , W. J . Wallncc. 31. Grayson and K . L X d m u , I d . En: Clierii , 46, 1402 (19.X:).
Sept. 20, 1959 6-
6+
CH3 -Br - - - - AltBrs I
CH.4
0 I
-t HBr -t A12Br6
systems which are ineffective in the alkylation of benzene. We conclude that styrene is a better nucleophile than benzene, presumably because of the stability of 11. 6+
6-
CtjHjCH=CH:: f (CHj)2CH--Cl---EF3
+
I1
+
4837
EFFICIENCIES OF FRIEDEL-CRAFTS CATALYSTS
+
CtjHsCHCHsCH( CH3):: BF;ClI1 CsHaCH=CH:. +--f (CsHsCH-CH2)"-
The catalytic ability of these systems is of course determined by both the acidity of the metal halide and the ease of ionization of the alkyl chloride. Thus, gallium chloride-isopropyl chloride readily isomerizes methylcyclopentane whereas gallium chloride-methyl chloride is an ineffective catalyst. Here the difference in catalytic ability results from different stabilities of the incipient cations [ (CH& C H + more stable thau CH3+]. Differences in catalytic ability can also result from different stabilities of the incipient anions. For example, the system boron fluoride-isopropyl alcohol readily alkylates benzene whereas boron fluoride-isopropyl chloride does not. This difference between the alkylation ability of alcohols and chlorides is well known and has been explained as being due to the greater stability of BF30H- relative to BF3C1-.12C Interestingly, sec.-alkyl fluorides, or even prim.alkyl fluorides, readily alkylate benzene in the presence of boron fluoride.12bs16Here the added resistance to ionization of a carbon-fluorine bond is offset by the stability of BF4- relative to BF3Cl-. 12b ,I6
Experimental Preparation of Reaction Mixtures.-Samples were prepared in a Stock-type high vacuum apparatus.'? The vacuum line contained no greased stopcocks or picein seals and was thoroughly flamed before the preparation of each reaction mixture. Methylcyclopentane (Phillips research grade, 99.9% pure) was stored first over calcium hydride and then over a sodium mirror. The benzene used was a C.P. grade stored over a sodium mirror in the vacuum line. Freshly distilled styrene was introduced into the vacuum line when needed. Aluminum bromide and gallium chloride were prepared by tlie reaction of a n excess of the anhydrous hydrogen halide with the desired weight of aluminum or gallium. The metallic halides were prepared in bulbs with break-off tips which were sealed to the vacuum line, opened, and the metallic halide sublimed into an ampoule which was an integral part of the reaction vessel for the preparation of the metallic halide. The co-catalyst , methylcyclopentane, benzene and/or styrene were transferred into the ampoule, the ampoule sealed and placed in a thermostated water-bath for the desired reaction period. Distilled antimony pentachloride and stannic chloride were vacuum transferred into the vacuum line as were the boron halides. Sublimed ferric chloride was (16) G. Olbh, S. K u h n and S, Olah, J . Chem. Soc., 2174 (1957). (17) R . T. Sanderson, ' V a c u u m Manipulation of Volatile Com~ u u u d s , "Johu Wilcy and Sons, Inc.. N e w York, N . Y., 1948.
vacuum resublimed directly into an ampoule attached to the vacuum line. Zirconium tetrachloride and antimony trichloride were added t o ampoules in a nitrogen dry-box. The ampoules were attached t o a vacuum line by a greased joint and the volatile reactants introduced. Heterogeneous samples were shaken in a thermostated bath at 25'. At 80" the heterogeneous samples were not shaken. Analysis of Reaction Products.-The methylcyclopentane-cyclohexane ratio in the reaction products was determined by g.1.c. The reaction ampoules were opened at liquid nitrogen temperatures and connected to a simple apparatus whereby the contents of the ampoule could be distilled under vacuum. At the end of the distillation the reaction ampoule was warmed to 50" for a few minutes at 0.3 mm. pressure. The reaction products were collected a t liquid nitrogen temperature in a trap containing 2 ml. of frozen 10% aqueous sodium hydroxide. Air then was allowed to enter the trap, the trap removed from the vacuum apparatus and warmed t o room temperature. The stoppered tube was shaken t o ensure destruction of any hydrogen halide or metallic halide that might have been transferred or entrained in the distillation. Samples of 10 pl. were analyzed by g.1.c. for methylcyclopentane and cyclohexane. No other hydrocarbons were found in significant amounts. When an alkyl halide was used as a co-catalyst its concentration in the products was noted qualitatively by g.1.c. Samples which could possibly have contained methyl halides were never allowed to warm above 5'. Calibration of the analytical technique with prepared samples of methylcyclopentane and cyclohexane in an A-column18 of a PerkinElmer model l54E Vapor Fractometer using helium as a carrier gas indicated mole cyclohexane mole methylcyclopentane
"O'
area cyclohexane peak area methylcyclopentane peak
Experiments in which the methylcyclopentane-cyclohexane isomerization had gone t o completion left an oily residue after the vacuum distillation. The other experiments in Table I involving aluminum and gallium halides left a solid, colorless catalyst. From these experiments involving slightly moist aluminum bromide at 25" a total of 0.73 g. of catalyst-free oil was obtained by treatment of the residues with pentane and water. This represented 7 7 , by weight of the methylcyclopentane used. Distillation gave 0.33 g. of material boiling at 83-90' a t 9.2 mm. This material most likely is a mixture of compounds containing twelve carbon atoms derived from the addition of a carbonium ion containing six carbon atoms to a molecule of a cyclohexene or methylcyclopen tene.19
TABLE IV DISPROPORTIONATION OF SUBSTITUTED SILANES IN CYCLOHEXASE
Silane concn.,
SOLUTION'
Temp., Time, 'C. hr.
Silane M (CH&SiC2Hh 0.23 80 89 (CH&SiBr 0.73 80 20 (CH3)SSiH 1.06 40 10 ( C H ~ ) & C G H ~1 . 0 4 40 1 0.15 Maluminum bromide.
Silane, % disproportionated
7.1 2.0 4.6 0.07
Cyclohexane isomerized,
%