match nowt pp. 83-119 4975 ON THE KINETIC DETECTION OF SHORT LIVED INTERMEDIATES IN COUPLED SETS OF FIRST ORDER INTERCONVERSTONS AND ON THE DETERMINATION OF THE REACTION CONSTANTS BELONGING TO THE RESULTING MECHANISTIC SCHEMES. J. Leitich Institut £Ur Strahlenchemie im Max-Planck-Institut flir Kohlenforschung D ~ 4330 MUlheim/Ruhr, Stiftstrasse 34 - 36 Abstract In systems of n species interconverting by first order or pseudo first order reactions, the mutual dependences, caused by the Principle of detailed balancing, among the phenomenological reaction constants kone » the ratios of kung, Of equal origin, Roney ia te) +) A) 7 (B)y mere called 2, are n and the ractos of Ronen of the type investigated and sone relationships given. then mechanistic schenes Involving comon reaction intermediates are introdvcod. To these mechanistic achenes belong the nachanistic reaction constants lagcy and the ratios of Knog, of equal origin, R, fnech: (A particular mechanistic scheme may require of the T to have the value = 1. This may be used to decide whether this scheme is acceptable for a= 84 = particular system whose k -values are known.) Criteria are then phen given by which one can tell whether or not all Rugcy of a mechanistic schene are calculable from the Ryygq (which are accessible experi- mentally)and this is the aim of this paper. If, ¢.g., the mechanistic schene, taken as a graph, represents a tree it is found that all Rech 2@ always calculable. If the graph contains cycles then the necessary and sufficent criterion for calculability is u' + y n(n ~ 1)/2 - e! where u! = number of independent T which are = 1 as required by the mechanistic scheme, y = number of independent Rj.o, made dependent by the mechanistic scheme in a way other than via T = 1, n = number of interconverting reactants, e' = number of the edges of the mechanistic graph. Since an easy procedure to enumerate u' is given and since y, as a rule, mostly = 0 (which however, can be checked by an other criterion, viz. (27)), this criterion should be easily applicable. If calculability has thus been shown, derivation of the equations which link the Roop, to the R is rather straight- ‘phen forward for simple cases, making use of the quasistationarity of the reaction intermediates included in the mechanistic scheme. I.) Phenomenological reaction schemes Assume a system of n reactants where each reactant converts into every other one by first order or pseudo first reactions, For n= 4, e.g., this system would be described by reaction scheme I. A—— ay Ks # GF- 85 - Let us call I the "phenomenological" reaction scheme (phenomen- ological since no assumptions regarding reaction intermediates are included) and k,g, Kpay etc. the phenomenological first order reaction constants Kjngq+'! No symmetries are assumed, i.e. the phen* Kphen @¥@ in principle all different. II be a shorthand notation 4 : pen’ for n reactants: GPgqi it 1s a complete graph because of our assunp- of I. III is the corresponding phenomenological reaction graph G. tion that every reactant should convert into every other one. II.) The effect of the Principle of detailed balancing. The number of all kp, in one n-reactant system = twice the number n phen = p. en of all edges in G p=n(n- 1) In every cycle any one k,y,,, 18 determined by all the other kj4, of that cycle by operation of the Principle of detailed balancing [21 cpp)!? ; e.g., for a three-membered cycle: cue so that a Kakaok3 generally, for a 1-membered cycle: ‘i (ta) IreSie Let us consider the P Kunen Of GBnen One after the other in a deli- ‘phen berate sequence. Let us then call a kyo, which is PDB ~ independent ‘rom ns re a anc er S from Khon Considered before a Ki, and the number of the ki, , of WB... a; thigh Gaen 4 th q= Btn y 2) (Gnuneration: In any tree containing n vertices (= a partial graph of Gegn) every edge (there are n- 1) will supply two kU,oq, all n nn = 3) the other edges which add up to Gl, ( there are MAS 3) + 1) close cycles and therefore supply one k3ngq each) - i Bong (R) Ato Of Fyngg Of emul origin (in thie 6280.0) = 4 Foye co (Ble x (R), ana (f) aro given, chon atso (B) as givens () is ime which 49 4m 8) ana (8 puctely dependent on (8), and (8). net us call 2 Byron plicitly independent from Ajo considered before a AY,., and the number of the RI, of Chen ry then r= p-n=n(n- 2) Because of (1), (1a), i.e. by operation of PDB, in every cycle any one Ronen 18 determined by all the other Ryo, of that cycle: e.g. from (1) bhen considered a Let us call a R,,, which is PDB-independent from ¥) before and from the Ring, Which depend on them (implicitly or by pop) a RY. and the nunber of the RUY., of Gag, 81 then ‘phen ‘phenner = ain = 1) _ se BBS y (3) Enumeration: Any tree containing n vertices nm nm Shen’ phen Will close a cycle and supply one nore AUN... Tt ie thus seen that ¢ = a partial graph of ) wit supply n= 2 RUA; any further edge to add up to 6 equals the number of edges in Gon minus one, and it is seen generally that with every edge in a G" (which may be any reaction graph, not only the complete graph G,,,) but for the First one one RY is intro~ duced. In other words: A one - to - one mapping can be made between the set of the edges diminished by one of a reaction graph and the set of its R™. (3a) From what was said up to now one obtains prqer-s= 2M) aiee «ay ives, the number of PDB-dependent k,j,,, equals the number of PDB- ‘phen a ‘phen? cycles which are not generated by cycles considered before) in G! dependent R' both equal the number of independent cycles (i.e. ‘phen which is denoted by c. For a convenient synopsis relating p,q,r and s see below. uu Easy graphical procedure to choose a set of Riven! cy au The Roten a¥€ chosen deliberately except for the following restric— tions: Every chosen Rij, is introduced into G,,, in fat print, B eg. 8 and (2. in Schene 1V. 10 Raygq on the same center viich i Ls ‘phen have in common one branch (Scheme V) generate a third Rynon com prising the other to branches, thereever only one more Bayon required to complete a cycle (e.g. (B) in scheme 1¥) thie Teng, 12 aso introduced in fat print before continuing to choose another= Bim nit y+ ALL Ryygg SOnerated and introduced that way are no longer eligible as 3%, A 3 Synopsis of p.qvr.s: d % Fr Goumber of it priate anlesed by poo os ae a6 "Made = Comment: If in a system of n interconverting reactants one has already determined the product ratios Rio, of 3 - 1 reactants, then of tne n - 2 nY,,,, of ene 38 xeactant J - 2 are determined by PDB th and thus predictable. Thus of the n*” (the last) reactant all product ratios are predictable. If, on the other hand, in that system all k, or Runen have been phen ‘phen determined, of these only n(n + 1)/2 - 1 or n(n - 1)/2 - 1, respec~ tively, will yield real information on the system, the residual a oF Ronen of PDB and/or the experimental accuracy merely informing about the validity al nin = 9/2 +1 Kone= is III.) The T n Let us consider a four-cycle within Conon! Prom (1a) we see: (6) m uvery four-cycle contains one and a second one which because of (6) equals the first one. Dependences between four-cycles: a.) If the T of four-cycle (ABCD) is given and that of the four- cycle (ABCE) hig pi-sB|enen the 7 of the 5, ks! ea Fa third four-eyele (arco) [= 32 = Sy 6 Fa Roa is determined by the other two= 390 = wk. kk. oe pt. AS - ae = ot @ 2 Bq 1% 6 6 Ka" wy b.) The same is found for the three four-cycles (ABCD), (ACBD) and (appc) + Reakey Roky 1 Kaisa 2 Bq Res Kg Ra Fs okeg oaks ree: xe - (8a) hence: E> Fy Kgiks ~. Kelis ie.) + 12 = Ty Thus if three four-cycles are coupled according to a.) or b.), then of the three T any one is determined by the other two according to (8) or (8a), respectively. Easy graphical procedure to detect dependences according to a.) or beds (9) coro Vg WA “Ny represented as:=n ad. b.) Three coupled cmos [| pd represented as x —s I : at ‘ Let us call a four-cycle and its T dependent if its T is determined dependence rule: by the T of four-cycles considered before and independent if this is not so. Let us denote a T of an independent four-cycle by T°. c.) A four-cycle will always be independent if it contains at least one edge not belonging to any four-cycle considered before. This is so because this edge contains a pair of ky... which cannot ‘phen be contained in any T of four-cycles considered before and this, as 1s easily seen from (7), will make that T, i.e, that four-cyle, independent. 2 Tt can be shown that! 1.) A four-cyle is dependent if and only if it depends by couplings a.) and/or b.) and will be independent otherwise. (10) From (10) follows: Procedure to choose a set of T” ay tron the set of the 2(0) # one after the other ie choosen deli- herately and after each single choice those 7 which depend (from t already chosen or from T already found dependent) according to couplings a.) and/or b.) are detected (most easily graphically by means of (9)) and set aside, before continuing to choose. TSee “proofs and demonstrations"=a = 2.) the number t of the independent four-cycies or TY in Gh, equals to = a(n - 3) . 2 (2) ser (the total of ail four-cyles in Gh... = 3 (4) ) 3.) For any given set of TY it is always possible to map to each TY one RU, contained in it (every T is a ratio or a product of two Ronen’ 8° We say: it contains two Roo). 3) Because of (3a) this can also be stated in the form: In every set of independent four-cycles, to each independent four-cycle one edge contained in it can be mapped. Or: Every independent four-cycle is so because of c.) Furthermore one can see: If a subset of the T in GPoq, whose elements be denoted by T', is complete in the sense that every T dependent from two T* by couplings a.) or b.) is also a T', then the total number of the T'" (= 7" of this subset) must be inde- pendent from the mode of choice and therefore characteristic for n this subset. (Example: The 3 (3) T in GPne© Gonon ( a because of (6) > . A we note generally: Gay #11) amply a yr, say (7 (Q) = 3, We connections in Gay, from A to C, A to Dy B to C and B to D run. (17) Using this criterion we can in a straightforward manner find the set of ,T implied by a ch mech From this set of implied ,7 we can choose a set of implied ,1" by means of (11). AGE, 48 acceptable for a particular n-component system if all ,7° which it implies have been verified experimentally. Be? tees BOSD ‘The nunber of 1° implied by OR, = ul. Always utc t sae) 1 GP, 1s acceptable then also wee Fron (13) follows that for every ,T% implied by GRagy the RZ, contained in and mapped to that ,1" is made equal by G4, to the= 96 second R contained in that ,TY and is thus determined by the wu ‘phen* ‘phen’ via implied ,T° from the remaining n(n - 1)/2 - 1-u ‘phen second R, So of the n(n -1)/2- 18 u’ are made dependent ‘mech ai. There may be implications other than via ;7" by which cl makes ROO.,, dependent from others; the number of such extra impli- by ¢} ‘mech ‘phen cations = y. Let us call a RUU. | which is not made dependent by Gh, from Rou ‘phen imech Ronen considered before a Rong, and the number of ROAG, for a particular Gaon vi then according to what was said n(n = 1) ves-ut-y = BASYD - yey ni 19) n n Bavdtzaiog GRech *°" Chen! Tf a Ghnen (= Partial graph of GD,,,, for a particular n-reactant system) contains one ,7" which one would like to consider as implied rather than accidental, then it has to be replaced by a G4 ‘mech 1 compliance with (17). Of an infinite number of G4.., complying with (17) a simple one would be x. o~x Ox TE Ghigq contained two ,T (it can contain no more than two 7) which one would consider as implied then it can be shown that a.) there is then only one G4... complying with (17), viz. XI, b.) one may consider the two ,1" only as either both implied or botha accidental. (20) We note that in X all six edges of c4 have been replaced phen (Pour of them had to in order to comply with (17); the residual two could since they no longer represented the only connections bet- ween their respective reactants). 4 ‘phen note bY S5nen a 4 (such Cohen We dem ) share at least one edge we say they are linked to- If two G each containing at least one ,1" gether. If a third ,¢ shares at least one edge with at least ‘ Shen one of these two 64, we say all three are Linked together. 1f some ,G) hon AF€ Linked together we call these a completed set of Linked ,64,,,, 1f none of them is Linked to another ,64,,., not be- phen longing to that set. Let us call those 1 (n> 124) vertices A ‘phen mechanistic subsystem of the n-reactant system. Then we can genera- (reactants) which take part in a completed set of linked ,G.0. a lize what we had said before for 1 = 4 1 n The Conon (= @ partial graph of GP, anistic subsystem of 1 reactants out of the n has to be replaced by ) corresponding to a mech- a ch,y which has to be acceptable as judged by criteria (15) and 4 (07) and which will not (because the jS5 hon system are linked) be separable into independent sub-G,, partaking in the sub- ech* (21) Pen which do not take part ghen) 1 ren in any ;64yqn ax0 oft untouched by thie procedure. (Even inside a 1 Snech’ Those edges (and their two k......) in G' n - ) Of Conon MAY remain untouched.) This means that from the standpoint of the ,T" there is some of the edges (and their two kphen! no need to change the mode of connection and hence the two kien between a pair of reactants not belonging to a mechanistic subsystem = although of course one is free to do so if one has chemical reasons for it. A mechanistic subsystem of 1 reactants thus can be considered- 98 - completely by itself without considering the residual n - 1 reactants. Therefore, all what we are going to say in the following will be about mechanistic subsystems only; for continuity, instead of the index 1 the index n will be used further. (21a] For most conceivable combinations of ,T" there are also acceptable oe ‘Rech: This means that mostly the set of ,T" contains no redundancy as fas as information on the structure of 6” d, ever Mech 1 concerned, every u . - n ;T" supplying an independent information on Gia. nm yo Tf however for a certain set of ,T" there exists no acceptable Gna.y, than that or those ,T" must be accidental which one has to drop in order to obtain an acceptable Grou, provided by the following set of for the residual set of ,T". one example is j@ for n Here the ,T" to the right has to be taken as accidental. v.) the relation between the ratios of mechanistic reaction constants Riach and the ratios of phenomenological reaction constants Ronen: Tt can be shown” For a n-component system and any given G” mech’ the values of all " 30 Rid. are determined by, and only by, those of all AY... (22) From (22) follows: The number of the Ronoy the Agent cannot exceed that of stay (23)re From this and (16) and (19) follows uwity > s- -e (23a) : acho, makes at least as many RUU endent from. which says: a Glo, makes at least any Ronen dependent £ others as it has less edges than Gh... Since e' = n + x +c! ~ 1, where x = number of branching points x number of independent cycles in Goo, (analogous to ¢ for Spnen! we have wty 2 BOs x cre (23b) this with (12) gives wy op tex-ctet (230) uty 3 ec-cl-x (23a) Relations (23x) hold for all Goch: Now we can set out towards our goal and answer the question whether ‘phen (Which are accessible we can determine all RUQ., from the R’ experimentally). Necessary for all HWY, to be determined from the RIM (we say: for G2.) to be “calculable") would be: vz 8's since (23) holds, this reduces to ves! (24)= 100 = yuu au As the system of equations linking the Rose, to the Riga, is not separable into independent subsystems (which follows from (21) and (21a) since we are considering a mechanistic subsystem of reactants and since the Gj, of such a one is not separable into independent SUb-G,.goy) tt follows that (24) is also sufficient. Thus we can state: n (24) is necessary and sufficient for a Goon of a mechanistic subsystem to be “calulable". Comparison of (24) and (23) and keeping in mind (16) shows: only those G™, Mech ae "calculable" which for a given v (i.e., for a given u' 1 y) have the minimum number of edges. (24e) In analogy to (23a - a) condition (24) can be written: ulty= (24a) = +7 (24b) = tox-ch td (24) ercl-x (24a) n For a particular GPoop, enumerated. Unfortunately y is not readily evaluated and this all the above quantities except y are easily presents a problem, To circumvent the problem one has, however, the following possiblities:= lol - 1.) The necessary and sufficient condition (24x) contains the necessary condition n(n = 1) of we (25a) (25b - a) in analogy to (24b - a) 2.) Rule (not proven!) : (26) It appears that for a mechanistic subsystem (cf. (21), (21a)) always y = 0 except when s - u'>s' : then y = 5 - s' ~ ut (which is just condition (24) so that it means at the same time that a ot ech With y>0 is always calculable) 3.) Sufficient condition for calculability” 7 If a GO..y, contains only such edges whose cancellation would either interrupt the only connection between any pair of reactants or would imply a new ,T", and/or such edges between two X, whose elimination by merging the two X would imply a new ,T, then it is calculable. 4.) Tt can be shown?: a.) If GPa, is a tree it is always calculable. (28) b.) If GP mech $8 @ tree then always y = 0 (29) From (28) and (29) together with (24) follows further that for a tree= 102 - 1.) the REY, which is indicated by fat print in XII will always be even if the rest of ¢” Mech #8 not calculable and simply = calculable. =A. (30) ve s 2.) Since for a particular set of ,T" in general there is an infinite set of acceptable GP. Rech With only few of them having the minimum number of edges and thus being calculable (see (24e)) and since the ones with the minimum number will not always be those which are the most reasonable ones from a chemical point of view there may fre~ quently be a conflict between calculability and reasonableness of a Gags A remedy might then be to assume a relationship vhich uu Rech number of RY, and make the reasonable ?,,,, calculable. expresses one by others and thus to decrease the effective 3.) If the n-reactant system contains symmetries, i.e. if some and R, reactants have equal k, kite phen because they are optical antipodes or isotopic isomers then they also have equal kjoo, and Raech and there will be a symmetrical G”,.,, - To assess calculability- 103 - one then has to find the effective number of edges in GT... and ‘phen n . uuu wu Ghecn ( and Likewise the effective number of RUNY, and RUZ.) by eliminating those edges and R which are generated from others by a A symmetry operations, within the Glacy and the Grnaqs n 4.) Calculability of a c?oo, ‘phen If they don't then this GU... is falsified for that parti- does not mean that from real positive values of R there will always result real positive values of wo Rnech* cular reactant systen?. 5.) The concept may be extended to systems with reactions of higher order. One may keep the concentrations of reactants known and con- stant (or nearly constant) during experimental determination of the Kohen amd may thus use them as paraneters with which to define pseudo first order Kpnen a°4 Knech With all reaction constants being made pseudo first order in this from the higher order reaction constants. way any system is amenable to the present treatment. Moreover, the calculability situation is highly improved in such aystena: Go, become calculable even though they contain more uu o Rich than there are RUN + This is because the variation of the Rphen additional information; it is given as a function of the with the concentration of a reactant can be measured and yields Rnech Rnech by uuu Rphen respect to the reactant concentration. Thus the set of equations differentiating the equation expressing the by the with Re which to transform into the set of equations expressing the RWZ.,, by experimental values is increased. A general discussion of all possibilities which then exist would, however, blow up the scope of this paper and can be omitted since the reader probably will have no difficulties in working out the special cases he is interested- 104 3. A fine example is found in the work of Doering !41 ; ‘The (symmetrical) calculable G4,.,, on p.1178 leads for one particular R (ech t© the expression (in the authors" notation) + le, -@ 73 | 2 ‘x) - [Se FA ” @.-G Lithr Ele ~ Ntha | upon inserion of the authors" experimental values an imaginary value for Fe) results. (The same conclusion, i.e. that this 64.4, is un- acceptable for the system studied, has been reached by the authors via a different route of reasoning.)- 105 = . wu VI.) Derivation of the expressions for the RY, 1.) Bquations which express the ,R¥MY, (RU of origin a) as a function of the RM, are obtained by starting from the x quasi~ stationarity conditions for the X and assuming the concentration of all reactants except A to be zero. The same as for A is then done for the other reactants. 2.) The system of v(= s') equations thus obtained expressing the ue ry i v Rohen bY the v RUG, is then transformed into the system of v equations expressing the v RYU. by the v RUM ‘mech ‘phen* VII.) Possible procedure, starting out from experimentally deter~ mined k, or R, ‘phen iphen* 1.) Split the information given by k, (n(n - 1) experimentally ‘phen determined values) into n(n ~ 2) Rg, and A Kynen Of UNeguAL Origin ‘phen 2.) Find all ,T; choose the ,T" from the ,T according to (11); decide which of the ,7" shall be considered as implied and which as acci- dental and at the same time choose that acceptable (in view to the aT) och Which is the most reasonable one chemically (see: Devel- oping GP. n mech ££ GPnens P+ 96 ££.) 3.) 0 every implied ,1% map one #YY.,, contained in it (see (5) (13)), complete this subset of u' Soha to give the complete set of nin -17/2-1 ROU, by choosing the residual UY, (which are the pu Shen) wsing (5). Check the experimental accuracy and/or106 = r a wa validity of PDB by calculating the depending RD, from the Rona, by means of PDB (1a) and comparing with the experimental values. 4.) Check calculability of the chosen G” Mech £OF every mechanistic subsystem (see (21) (21a)) according to p.% ff. If necessary, make it calculable (see 2.) p. 102), 5.) Derive the expressions which give the RoY,,, as functions of the uuu hen (ee VE) ow yuu 6.) Calculate the values of the RYU, from those of the RUM and check {f they are real and positive otherwise the chosen G®.o), would be falsified for that system. 7.) The sknech (ech Whose origin is a reactant J) are then also calculable from the n k of unequal origin (see 1.)), the Ron, phen phen and the RUY.,, by making use of the quasistationarity conditions for the x. 8.) Convert the 28%, ané agen into energy- and entropy~ infor~ nation. VIII.) Flow of information BCR 1) Kener 7 > —® Konen Of unequal origin: Energies and entropies of the reactants relative to each other and to the oo transition states.- 107 2) Bohen 2 wu 4 " Fonen 7 -¥et (a7 Re aa > energies and entropies neck * of the transition states corresponding to Gh, relative to each other IX.) Example n(n = 3) 1 dependent z Ronen? @xPerimental accuracy and/or validity of PDB Rohen uty 2 am= 3, z made dependent from other uu phen n ® : ‘mech. by 6) structure of GPooy- 108 - For this tree there must be 7 Rong, ( since v= s' =e! -1= 7) out of 14 RUM. ( so that there are 14 - 7 = 7 implied ,1" out of ‘phen 1 uu 9 tT"). Proceeding along VII 2.) 3.) these Rohen 2x chosen as, say: {p E E 2 %B, A, z A, ¢ A, B, vp, ou uuu 7 REL are to be calculated from these 7 RUM, | Three are obtained in a straightforward way (see (30)) = aed Ea] " »Io © x kop {B\ ee B For 12 Kip Bicer w( ne Kaas - Rgp a Big in= 109 - ar ¥ We note that these equations can be expressed in different ways uuu since one can replace the Roya, by other Ruicq using implicit or PDB - or ,T" - dependence. It is a question of convenience which of the possible expressions to use. We note further that from these equations one can obtain the equations for sub-trees of the tree above such as XIII, XIV, XV. D a D——X,--a | E -B E— XB F-—x;—C XIII xIVv xv- lo - Obtaining these equations will in general involve replacement of the RU ‘phen PY other R, ‘phen USing the three types of denpendences (in order to remove dropped reactants from the equations) and keeping in mind footnote 1 (p.94). The equations for system XIII are obtained in a straightforward way: Kip Kae Kia The expressions for (12 , 722 , -12 (using the expression in brackets hin “los Lac at the right) and ;2! can be used directly as is clearly seen 28 k (they were however derived independently also); for ;12 alter- 1D natively the left-hand expression may be used after cancellation of C and F. These expressions can be applied to particular systems XIII described in the literature to give an improved nonapproximative treatment instead of the approximate treatments given here.51(614 4 por the system studied by Bartlett!51, z-butene(2), E-butene(2) and the two stereoisoneric cyclobutane adducts of these to tetra- fluorosthylene, this makes a considerable difference: Te Yay, ~ values given in Schene 1 of Bartlett's papers 0,717 0,23; 2,25 0,5 are to be replaced by the values 0,149; 0,0495; 0,435; 0,106, respectively.-ul- Proofs and demonstrations: nm To see (10) and (12), let us consider all four-cyles in Gog successively in such a way that each four-cycle considered shares as many vertices and edges as possible with four-cycles considered before. Then we will find that each four-cycle will either depend by couplings a.) and/or b.) or will be independent according to c.); at the same time we will verify (12). n(n -3)/2, in this particular procedure, is thus both the total of’ all T mutually independent in respect to couplings a.) and b.) and the total of all T°, i.e. T mutually independent in respect to any coupling whatsoever. It will, of course, be possible to choose several different sets of TY and to choose them by procedures different from that above. But whatever the choice, both totals, viz., the number of all T mutually independent in respect to couplings a.) and b.) and the number of all TY, must remain the same, n(n - 3)/2. From this follows (10). To see (13), let us take from any complete set of TY (n(n - 3)/2 tT") a subset of 1 Combining (31) and (32) we get: The skroch whose origin is a reactant, J, not an X) and the ,R4,.,, together determine the k,y. phen? Tf ROW thE skeen are not given but only the jRU.cyr the information on the rates of flow from J is not given but only that on the distribution of Flow anong different directions. The sane information loss results if the k, ‘phen @¥€ not given but only the RB} nen: From the last three sentences follows (22). 5 (32) means that vice versa one can never draw conclusions from the Kphen 0 the ky, but only to the ,R,oo,+ This means for chemistry: From the kinetic experiment one can find out the differences in energy and entropy between transition states around one Xx, ( since these are given by the ,,R,o¢,)+ but never the energy and entropy (relative to the ones before) and the lifetime of that x, (since these would be given by k. ‘Kas- 14 = The reason for (27) can be seen in the following manner: According to (3a) to every edge in GN», but for the first one one ROU, an be mapped. The impossibility to cancel an edge there- fore means the impossibility to counteract a change of the value Re of the RUU., mapped to that edge (elimination of that edge means that it becomes 0 or oo) by changing the values of the other ROG. such that the values of the RUY,, renain unchanged. If this is true for all edges, ic. for all ,,, then in analogy to the argument for a tree (see below) follows that the RW, are deter- mined by the RUN (28) can be seen in the following manner: Since a change in the value of a AU, changes the distribution of flow of matter anong aif fer ent directions and since this distribution determines the UY, and since if there are no cycles in G,, y (which means that there is only one connection between any pair of reactants) that change cannot be compensated by changing the values of other RY, (except at the price of inducing more changes) it follows that a specific set of values of the RUZ, requires a specific value of that UY, and, since this holds for al1 RM, of the tree, it follows that it requires a specific set of values of all the nW¥,. This in o*her words means that the set of values of the RU, is determined by the set of values of the ‘phen R phen” and therefore, by the set of values of the R' (29) can be derived in the following manner: From (28) and (24c) we obtain for a tree: u' + y = t= x +1 (33). Because of (18) u' can maximally be =. the only Go, t0 imply ut = t is a star, te. a tree with x = 1. Inserting into (33) shows that for a star y = 0.- 1s - Upon transformation of the star into other trees by increasing x, every additional xX will reduce u' by 1, (34). This with (33) gives Q9). To see (34) we note that for a given n there are trees with max- mum x (viz.: x =n -2) which differ in the type of branching. For all these trees u’ = (n - 2) (n -3)/2 as is found by systematic enumeration using (11). For the tree with minimum x (the star) ul = t = n(n -3)/2. out = n(n = 3)/2 - (n= 2) (n= 3)/2en- 3. ax =1- (n= 2) - (n= 3), Since with every additional x in a tree at least one ,T" must be lost, it follows from 4 u' = - A x that with every additional x exactly one ,T ist lost. Acknowledgment. ‘The promotion and encouragement of this work by Prof. 0.£. Polansky are gratefully acknowledged.- 116 - Glossary = ratio of products A and B when formed from C; it : cn 2 Ronen are not generated by cycles considered before) ayets, on in nen? Pe 87 n e = analogous to ¢ in GPa, a, instead of Go o47 p.99 calculability, of a G,.,1 definition p. 99 contained (R. in 7); definition p. 92 ‘phen dependence, among four-cycles and 1; definition p.% e = number of edges in Goon: pe 95 Z = a mechanistic reaction graph for n reactants; p.%3 Ca hanist tion graph f tants; p. Phen = the phenomenological reaction graph for n reactants; Be 85 implicit dependence, among Ryo, OF Raggy? definition p. 86 = first order or pseudo first order reaction constant p. 85 for the reaction A» k, = a first order or pseudo first order reaction con- stant within a mechanistic reaction scheme, mech? P+ 93 associated with an edge in Groon? Kohen = a phenomenological first order or pseudo first order reaction constant; p. 85 @ Kynen Which is PDB-independent from kenon Kphen considered before; p. 86PDB Rnech Ronen Ronen uu ‘phen uuu ‘phen -u7- number of reactants of a system number of all k,.,,, in one n-reactant syste nm Principle of detailed balancing; p.85 nuaber of all Kj, in one neveactant system; p.85 wa a ratio of k, of equal origi hen = + analogous to R' in a mechanistic instead of the phenomenological reaction scheme; p.95 owe (8) . considered before; p.86 . : dered before and from the Rg, which depend on them (implicitly or PDB); p.86 which is not made dependent by Goon, considered before; p. 96 fron u n number of the R Of Co nen? ‘phen p.86 number of the RU, of p-86 2 phen a number of the RW, of GPoo,7 P-95 ‘mech? a ratio of the type (2) 7 (A); v.89 ‘c D oP an independent 1; definition p.91 a T whose value is = 1; p.92- 118 - a ,T which is independent from ,T considered be- fore; p. for "dependence", see p.92 n u number of TY in Giga? Ps 92 number of ,T" in a particular experimentally measured system; p. 93 number of ,T implied by 6”. p. 95 1 ‘mech number of RUA? for a particular G? ‘phen mech? P+ 96 intermediate reaction stage which represents a branching point in Grooy7 p- 93 number of X of a GP, 99 mech? P+ number of implications other than via ,T" by which n uu ¢ mech MKS Rope, dependent from others; p. 963.) 4.) 5.) 6.) - us - Bibliography For methods to determine the k.1, ments, see P.A. Matsen and J.L. Franklin, J. Amer. chem. Soc. 72, 3337 (1950), and reference 3.) , ftom kinetic experi- For a recent discussion of PDB, see J. Lee, Int. J. chem. Kin. 3, 491 (1971) and references given there A fine example for study is the (n = 4) - system of compounds (3), (4), (5), (6) in the paper by D. Hasselmann, Angew. Chem. 87, 252 (1975) W. von E. Doering, K. Sachdev, J. Amer, chem. Soc. 96, 1168 (1974) P.D. Bartlett, K. Hummel, S.P. Elliott, R.A. Minns J. Amer, chem. Soc. 94, 2898 (1972) R, Huisgen, G. Steiner, J. Amer. chem. Soc. 95, 5055 (1973)