J. Chem. SOC., Faraday Trans. 1, 1982, 78, 771-784 Thermal Behaviour of 9-Cyanoanthracene Photodimer (9-CNAD) BY DONATO DONATI AND PIERO SARTI-FANTONI Centro di Studio del CNR sulla Chimica e la Struttura dei Composti Eterociclici e lor0 Applicazioni, c/o Istituto di Chimica Organica, Universita di Firenze, Via Gino Capponi 9, 50121 Firenze, Italy AND GIULIO G. T. GUARINI* Istituto di Chimica Fisica, Universita di Firenze, Via Gino Capponi 9, 50121 Firenze, Italy Received 25th March, 1981 The monomerization reaction of 9-cyanoanthracene photodimer (9-CNAD) in the solid state has been reinvestigated both in isothermal and in linearly increasing temperature experiments using thermal methods and optical microscopy. From both kinds of calorigrams the heat of monomerization has been deduced in agreement with previous findings; the heat of melting of the crystalline monomer (9-CNA) formed during the monomerization reaction has also been redetermined.With supporting evidence from optical microscopy studies, a three-stage mechanism is proposed for the monomerization reaction ; agreement with recent reports is found only for the induction period. Our isothermal experiments show that, in the temperature range of interest, the solid-state monomerization reaction of 9-CNAD requires times far longer than those recently reported by other authors. Additional evidence is presented supporting our previous interpretation of the nature of the initial exothermic peak typical of the monomerization thermal curves. Upon U.V. irradiation, anthracene and some anthracene derivatives give photodimers both in solution and in the solid state.l? As the photodimers are thermally reverted to the monomers, the importance of these compounds for the storage of solar energy has also been c ~ n s i d e r e d .~ ~ ~ In addition the thermal properties of some monomers and dimers have been inve~tigated.~ In particular it was found that the dimers studied were converted back to the corresponding monomers during melting point determinations; indeed Calas and Lalandel clearly pointed out that it is impossible to define exactly the melting point of the dimers owing to the simultaneous monomerization reaction. Then it appeared impossible to study the kinetics of the monomerization reaction by dynamic thermal methods, particularly for those dimers having the monomerization temperature above the monomer melting point, because of the simultaneity of the two thermal events.However, the literatures states that 9-cyano- 10-acetoxyanthracene dimer (9-CN- 10-ACAD) monomerizes at ca. 423 K, i.e. at a temperature lower than the melting point of the monomer (472-473 K). In addition 9-CNAD was reported to melt instantaneously with depolymerization near 478 K;' the same compound was also described as monomerizing under prolonged heating at temperatures much lower than the melting point of the monomer.' In a previous paper5 we investigated the thermal behaviour of anthracene and of some anthracene derivative photodimers in a reproducible way by using a Perkin-Elmer DSC-1 b differential scanning calorimeter.This study showed that 9-CNAD mono- merized before the melting point of the monomer if low scan speeds were used. It was thus possible to separate the exothermal monomerization reaction from the 77 1772 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER endothermic melting of the produced monomer, and we thus found it interesting to study the kinetics of monomerization of 9-CNAD as well as that of 9-CN- 1 0-ACAD, by using dynamic d.s.c. results.* Recently9 the thermal behaviour of 9-CNAD has been reconsidered by using both dynamic and isothermal data obtained by differential scanning calorimetry or by isothermal spectrophotometry ; in this work9 a different interpretation is reported for the initial peak typical of dynamic5? 8 y and isothermallo experiments performed by thermal methods; moreover in the same paper9 the reaction is reported as rather fast (ca.5 min for completion at 407 K by isothermal calorimetry) and relatively slow (ca. 240 min for 70% conversion at 400 K by isothermal spectrophotometry). Results differing from those reported in ref. (9) for the isothermal monomerization reaction by thermal methods have already been reported.1° From a kinetic point of view, zero order has recently been attributed to a reaction period defined as accelerat~ry;~~ l1 this appeared to deserve further consideration. According to the shape of the reported cal~rigrams,~~ l1 the attribution of zero order to the overall reaction in dynamic runsg is also questionable. In fact, a zero-order transformation results in thermal curves mainly parallel to the baseline in isothermal runs or increasing in dynamic experiments, the reaction rate being dependent only on temperature.The melting of 9-CNAD was reported as taking place at 480 K9 and later described by the same authors’l as occurring ‘ occasionally in crystalline samples following incomplete monomerization’; such melting of the dimer was never reported by us in previous s t ~ d i e s . ~ ~ 8 * lo The recently reported data prompted us to check our previous findings on the monomerization of 9-CNAD by renewed dynamic and isothermal experimentation by thermal methods. To improve the accuracy, the new determinations have been performed by a Mettler TA 2000 thermal analyser; this apparatus has a higher sensitivity than the Perkin-Elmer DSC-1 b.The thermal determinations were also supplemented by optical microscopic examinations of the progress of the reaction in 9-CNAD single crystals. EXPERIMENTAL 9-CNA prepared according to the literature12 was dissolved in benzene and irradiated with the glass filtered light of a 250 W G & C medium pressure mercury lamp to obtain the dimer. 9-CNAD so precipitated was filtered and recrystallized several times from THF. By this procedure crystals of different sizes may be grown. For the present work a batch of small crystals of ca. 40pm average size was mainly used. Large (millimetre-size) crystals for use with the optical microscope were grown by the same procedure just preventing a rapid evaporation of the solvent. An old batch of 9-CNAD crystals of ca.30 pm, prepared in 1970, was also used and found to give thermal curves indistinguishable from those obtained using freshly prepared material, thus showing that the dimer remains unaltered even after long storage at room temperature. Both isothermal and dynamic thermal curves were obtained by a Mettler TA 2000 apparatus. Because of the sublimation of 9-CNA, sealed aluminium pans both for the sample and for the reference were used ; excluding the experiments with millimetre-size single crystals, the samples were always gently pressed into the sample pans by means of the flat section of a glass rod of suitable diameter. Weighing was also performed before and after each experiment to check for eventual sample losses. In dynamic determinations the sample was heated using the fast heating rate (100 K min-l) of the apparatus up to 390 K.After a short period at this temperature (necessary to equilibrate the apparatus at the selected sensitivity) the run was started at the chosen scan speed. For the isothermal determinations the chosen temperature was reached again using the fast heating rate, the sensitivity of the apparatus was then increased up to the predetermined value, and the apparatus was allowed to equilibrate.D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 773 The whole procedure took ca. 3-5 min. Unless otherwise stated, the recorder paper was allowed to advance at 0.5 cm min-l. Area measurements for heat determinations were performed by polar planimetry and found to be in good agreement with the results of a Simpson integration procedure used mainly for kinetic purposes. The temperature scales of the plots referring to dynamic determinations have been corrected for At,,, according to the manufacturer's instructions.Corrections for the change in sensitivity of the thermocouples with temperature were disregarded for the monomerization reaction because in the temperature range of interest (403-443 K) the sensitivity does not change appreciably from the unit value. Conversely, this type of correction was applied to heat determinations concerning melting of the monomer formed. After each isotherm as well as after low-scan-speed dynamic experiments, the samples were heated dynamically up to 510 K to determine heats of melting. Occasionally, for the sake of comparison, the Perkin-Elmer DSC-lb was also used.Fluorescence properties of 9-CNAD crystals were checked by a Perkin-Elmer MPF 44A spectrophotofluorimeter, equipped with a solid-state device, using 365 nm exciting radiation. The fluorescence spectra show only two very-low-intensity peaks at 420 and 442 nm attributed" to very small amounts of monomer in the dimer matrix. RESULTS A. DYNAMIC DETERMINATIONS The use of the TA 2000 apparatus has confirmed our previous observations and in particular the dependence of the shape of the thermal curves on the scan speed (fig. 1). From the dynamic runs at low scan speeds (1 and 2 K min-l), in which the exothermal portion is well-separated from the endothermic melting of the formed monomer, values of the heats of monomerization (crystalline dimer + crystalline monomer) and of monomer melting were determined, being AH,,, = - 74.4 f 2.5 kJ (mol dimer)-l and AH,,,, = 26.9 f 0.2 kJ (mol monomer)-l, respectively.The superimposition of monomerization and melting is evident at intermediate scan speeds (8-24 K min-l), whereas at higher scan speeds (29 K min-l) only an exothermal event is recorded. In the latter case the heat evolved is due to the transformation crystalline dimer --* melt monomer and, within experimental error, is the algebraic sum of the above-reported heats (on a unit-weight base). Moreover the use of the more sensitive Mettler TA 2000 instrument allows a better definition of a series of peaks, beyond the first, which are superimposed (see fig. 1 for low-scan-speed curves) on the broad exotherm. These peaks have been seen previously8, lo in thermal experiments using the DSC-lb.The shapes and positions of these peaks are strictly reproducible using samples of the same batch under the same experimental conditions. The position of the whole exotherm and therefore the position of the first and subsequent peaks is shifted towards higher temperatures on increasing the scan speed. Thus the temperature of the first peak can be defined only if the corresponding scan speed and temperature at which the scan begins are specified. Similar behaviour has been described previously5 when the change in temperature range (AT) covering the whole decompo- sition was reported for 9-CNAD. The kinetic analysis of the dynamic thermal curves has also been performed for low-scan-speed runs by standard procedures,13 and an Arrhenius diagram in which the data are plotted considering the whole monomerization reaction as either a first-order or zero-order process is shown in fig.2. The first-order plot is exactly like the one already reported,8 being clearly formed by two rectilinear portions of different slope. The values deduced from the slopes for the activation energies differ slightly from those previously reported,8 being higher for the low-temperature portion (ca. 450 kJ mol-1 instead of ca. 360 kJ mol-l) and lower for the high-temperature portion (ca. 152 kJ mol-1 instead of ca. 176 kJ mol-l). Incidentally we note that in ref. (9) the774 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER FIG. 1 .-Dynamic thermal curves for the monomerization of crystalline 9-CNAD.For illustrative purposes the heights of the original calorigrams were multiplied by the factors 'f' listed below. (a) 5.61 mg, scan speed = 1 K min-l, f = 0.25; (b) 3.31 mg, scan speed = 2 K min-l, f = 0.5; (c) 5.17 mg, scan speed = 8 K rnin-I, f = 0.5; ( d ) 3.52 mg, scan speed = 24 K min-l, f = 0.7; (e) 4.68 mg, scan speed = 29 K min-l, f = 0.5. values for the activation energies quoted from our previous paper have been reversed : that referred as the low-temperature value is the high-temperature one and vice versa. Fig. 2 also shows that, as expected, a zero-order process is almost indistinguishable from a first-order one as long as the fraction decomposed is small (the low-temperature portion). As far as the disputed nature of the first peak is concerned,8v9p11 additional information has previously been obtained14 by d.s.c. on quickly cooling to room temperature a sample previously heated to just beyond the first peak.When the sample so treated was heated again under the same conditions, the typical first peak was absent from the broad exotherm corresponding to the monomerization (fig. 3). In slightly different experiments, in which the sample was quickly cooled to room temperatureD. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 775 0 -2 -Y E: - -4 -6 I I 2.3 2.4 103 K / T I 2.5 FIG. 2.-Arrhenius plot deduced from dynamic experiments at scan speeds of 1 and 2 K min-'. The rate constants were computed using first-order (0 for scan speed = 1 and 0 for scan speed = 2) and zero-order (0 for scan speed = 1 and for scan speed = 2) rate equations.just before the sharp rise of the first peak, the usual initial peak followed by the broad exotherm was obtained upon heating again under the same conditions. The same behaviour has now been found in isothermal experiments (vide infra). B. ISOTHERMAL EXPERIMENTS Typical monomerization isotherms, obtained using the TA 2000 apparatus and shown in fig. 4, indicate that, in agreement with previous reports,lO the beginning of the exotherm and therefore the position of the first peak is progressively shifted776 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER 410 400 430 460 TI K FIG. 3.-Dynamic thermal curve, obtained by d.s.c., showing the effect of interrupting the experiment near the end of the first peak (see text).50 100 tlrnin I 150 FIG. 4.-Typical isothermal calorigrams for the monomerization of 9-CNAD. For illustrative purposes the heights of the original curves were multiplied by the factors ‘f’ listed below. (1) 6.99 mg, T = 404 K, f = 0.5; (2) 5.16 mg, T = 408 K,f= 0.5; (3) 6.16 mg, T = 410 K,f= 0.4; (4) 5.88 mg, T = 415 K,f= 0.3. The first and subsequent peaks have been lettered progressively. towards the origin on increasing the temperature of the isothermal experiments. Again, besides the first peak, several other peaks are superimposed on the rising portion of the broad exotherm. Thus isothermal and dynamic monomerization calorigrams are strictly similar provided that suitable values for temperature and scan speed are chosen. After completion of every monomerization reaction the melting of the formed monomer was determined dynamically (scan speed = 4 K min-l).The heat determinations were also performed (by polar planimetry) both for monomerization and melting, giving mean values and standard deviations as follows: AH,,, = - 77.3 _+ 4.7 kJ (mol dimer)-l and AH,,,, = 26.7 f 0.6 kJ (mol monomer)-l.D. DONATI, P. SARTI-FANTONI A N D G. G. T. GUARINI 777 The values found for the heat of melting of the monomer formed assured us that at the temperatures investigated the monomerization is complete when the apparatus is allowed to reach the final, strictly horizontal base-line. The times required for the completion of the monomerization reaction are longer than those reported in ref. (9), but are in agreement with those expected on the basis of our previously reported dynamic runs at low scan ~peeds.~ We want to point out here that, in our melting determinations following the isothermal monomerizations and the dynamic runs at low scan speed, no peak at 480 K attributable to the melting of the dimer 9*11 has ever been observed.Also, in isothermal experiments the effect of interrupting runs at times just after and just before the appearance of the first peak was investigated. In fig. 5(a) and 50 100 t/min 150 FIG. 5.-Isothermal heat curves showing the effect of interrupting the experiments just beyond (a) and just before (b) the first peak. For illustrative purposes the heights of the original curves are multiplied by 0.7; the first and subsequent peaks are lettered progressively.The vertical broken lines indicate the times at which the temperature of the experiment was resumed after interruption and cooling to room temperature. (b) the corresponding isotherms are reported showing that the first peak is no longer present [fig. 5(a)] when the sample has been previously heated to just beyond the first peak, while the small subsequent peaks remain. Conversely, if the sample had been previously heated to just before the onset of the first peak, the peak in question appears [fig. 5(6)] and the whole exotherm shows the usual shape apparently unaltered even if characterized by a shorter induction period (vide infra). For samples of the same batch, the ratio of the area under the first peak to the area of the whole exotherm is nearly constant (ca.773, at least in the temperature range investigated. However, the shapes of the peaks are strongly dependent on crystal dimensions and crystallization procedures. Owing to the initial stabilization of the TA 2000 instrument at the predetermined temperatures (this problem of stabilization period also occurs for d.s.c.), there are uncertainties in the precise determination of the time origins; thus the kinetic analysis of the decay portion of the isothermal calorigrams has been performed as described previo~s1y.l~ In agreement with dynamic runs (the high-temperature portion), in the temperature range investigated (400-41 8 K) the decay portion of the monomerization reaction of 9-CNAD is well-fitted by a first-order law (fig. 6). An Arrhenius plot of 26 FAR 1778 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER i - 1 -s! c I C FIG.6.-First-order plot of the decay portion of an isothermal calorigram showing the excellent agreement. (h, is the distance of the curve from the base-line at the time considered and is proportional to the reaction rate.) -2.5 .y c rn -3.5 \. \ \ + \ \* \ \. \. \. * \ \ \ \ * \. * \. \. * ' \ \ \ '. \ \ 24 2: 5 103 KIT FIG. 7.-Arrhenius plot for the first-order decay period of the isothermal monomenzation of 9-CNAD.D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 779 the calculated first-order rate constants is shown in fig. 7 from which the following activation parameters have been deduced: E, = 134.3 kJ mol-l and A = 7.2 x 1015. We also attempted a kinetic analysis of the first peak by allowing the recorder paper to advance at a rate of 5 cm min-l.The result was that up to a’ = 0.5 (where a’ is the fractional decomposition referred to the first peak only) plots of In a’ against time were linear, thus giving evidence of the existence of an exponential rate law for the rising portion of the first peak. Even if, owing to the previously described stabilization uncertainties, the zero time is not precisely known, we have determined the time needed, at several constant temperatures, to reach the top of the first peak from the instant at which the digital display of the TA 2000 instrument first reached the temperature chosen for the isotherm. These time intervals, including the undetermined stabilization period and a small portion (< 5%) of the reaction, are nevertheless believed to be proportional io (and will be referred to as) induction periods.An Arrhenius plot of these induction periods is reported in fig. 8 showing a good alignment of the experimental points. From -1 h - c .- E -2 T, \ w c - - 3 2.4 2.5 1 0 3 KIT FIG. 8.-Arrhenius plot for the induction periods of the isothermal monomerization of 9-CNAD. the slope a value for the activation energy (E, = 143.5 kJ mol-l) has been deduced, in strict agreement with previously reported data.g Moreover, plots of the fraction decomposed (a) were computed from the isothermal calorigrams by a Simpson integration procedure. These plots are collected in fig. 9 for a number of isotherms. c. OPTICAL MICROSCOPY STUDIES We have already described8 optical microscopy investigations performed under progressive heating in a temperature range covering the first peak.Plate 1 shows a series of micrographs obtained in this way by a Leitz Panphot microscope equipped with a Nikon camera, which were the basis of our previous description.8 Our isothermal calorigrams [ref. (10) and this paper], apart from showing the same general shape as the dynamic thermal curves at low scan speeds, allowed a determination of 26-2780 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER I I 2 1 t l h FIG. 9 . - a against t plots for a number of isothermal monomerizations of 9-CNAD; the discontinuities evident in the initial portions of the curves correspond to the first peaks. 0, 8.22 mg, 402 K; A, 6.99 mg, 404 K; H, 5.16 mg, 408 K; 0, 5.76 mg, 412 K; A, 5.88 mg, 415 K. the time necessary to reach the end of the first peak at a given temperature. Thus we have now performed isothermally the same type of experiment by optical microscopy.In plate 2 a series of micrographs from an isothermal experiment at 408 f 1 K is shown. Plates 1 and 2 clearly demonstrate that at both increasing and constant temperatures the same types of phenomena occur, namely: (i) after a certain period black spots oriented along preferred directions appear suddenly [plates 1 (c) and 2 (6) and (c)] ; (ii) alternatively and contemporaneously but again suddenly the blackening spreads over relatively large zones of the crystal [plates l(6) and 2(d) and ( e ) ] ; (iii) the two previously described features then grow darker and spread fairly regularly until complete darkening of the crystal occurs.In isothermal optical determinations the time required to reach the complete darkening of the crystal is in reasonable agreement with the time taken (deduced from the calorigrams at the same temperature) to reach the end of the first peak. Crystals heated isothermally on the hot stage of the microscope up to complete darkening in transmitted light were collected and used to perform isothermal (408 K) determinations by the TA 2000 instrument in order to evaluate the degree of monomerization reached in optical microscopy experiments. Fig. 10(a) refers to the isothermal calorigram so obtained and is to be compared with fig. l0(6), in which the isothermal decomposition at the same temperature of approximately the same weight of crystals from the same batch as those used in microscopy is reported.The run in fig. lO(b) was stopped after the beginning of the decay period, the sample pan was allowed to cool to room temperature, and was then opened. The crystals so treated were collected and examined in the microscope in transmitted light between crossed nicol prisms. Plate 3 shows that, after the above treatment, the crystals possess small crystallites, extinguishing upon rotation, which protrude from the edges. These are thought to be crystals of 9-CNA. When a crystal has reached complete darkening in transmitted light between crossed nicol prisms, it remains dark even if one or both polarizers are eliminated. Switching to reflected light, the crystal appears yellow and strongly diffuses light from the apparently unaltered surfaces.J .Chem. SOC., Faraday Trans. 1 , Vol. 78, part 3 Plate 1 PLATE 1 .-Progressive darkening of a crystal of 9-CNAD observed between crossed nicol prisms while the temperature of the hot stage was slowly increased (ca. 0.3 K min-l) from the melting point of sebacic acid (404-406 K) to the melting point of 2-benzal-4-phenylpseudo-oxazolone-5 (410 K). The spread of the reaction over relatively large zones is evident in the low part of frame (b) while fast reaction along preferred directions is evident in frame (c). Magnification: x 34. D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI (Facing p . 780)J . Chem. SOC., Faraday Trans. 1, Vol. 78, part 3 Plate 2 PLATE 2.-Progressive darkening of a crystal of 9-CNAD observed between crossed nicol prisms in an isothermal experiment at 408 K.(a) t = 0; (b) t = 2; (c) t = 3; ( d ) t = 4; (e) t = 5 ; cf) t = 6 min. Spreading of the reaction along preferred transverse directions is evident in frames (b) and (c) while a spread over large areas is clear in frames ( d ) and (e). Magnification: x 43. D. DONATI. P. SARTI-FANTONI AND G. G. T. GUARINIJ . Chem. SOC., Faraday Trans. 1, Vol. 78, part 3 Plate 3 PLATE 3.-Transparent crossed-nicol-prism micrographs of small needle-like monomer crystals formed on the surfaces of 9-CNAD crystals partially decomposed inside the sample pans of the thermal apparatus. Extinction upon rotation by ca. 29' is evident. Magnification: x 62. D. DONATI, P. SARTI-FANTONI AND G. G.T. GUARINID. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI r- 12 24 36 tlmin L8 FIG. 10.-Isothermal (408 K) calorigrams of a sample of large crystals previously used in optical-microscopy determinations (a) and of approximately the same weight of unreacted crystals of the same batch (b). Curve (a) has been shifted to the right by an amount corresponding to the time needed to reach complete darkening of the crystal in optical-microscopy experiments at the same temperature (ca. 8 min). Base-lines were not drawn but a comparison with fig. 4, curve 2, shows that for curve (b) the run was interrupted after the beginning of the decay portion. DISCUSSION The aim of this work is to gain additional insight into, and to meet some of the discrepancies and different interpretations of, the thermal monomerization of 9-CNAD.5*8-1' A.HEAT DETERMINATIONS The heat data in this paper agree well with those recently redetermined by d.s.c.lo and reasonably well with results previously rep~rted.~? According to the present determinations, the ratio between the area of the melting peak and the area of the broad monomerization exotherm in the same experiment should be, on a weight basis, ca. 1 : 1.5 if no changes in sensitivity have been made between the two thermal events. Except for the top part of fig. 1 in ref. (9), in which the base-line cuts the endothermal melting peak, the same ratio can be deduced from the thermal curves in ref. (9) and (1 1)- The shape of the monomer melting peak also needs some discussion; indeed this has recently been reportedll as consisting of two peaks that the authors attribute to the subsequent melting of a crystalline and an amorphous phase of the 9-CNA formed in the monomerization.While we strongly doubt that an amorphous phase can exhibit a sharp melting point, particularly in a temperature range so near the melting point of the corresponding crystalline phase, the existence of a double maximum in the region of the melting peak, even if not as evident as that reported in ref. (1 l), has sometimes been observed in our d.s.c. experiments (but never in TA 2000 runs). In782 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER our case we attribute the splitting of the melting peak to the slightly retarded melting of crystalline 9-CNA sublimed on the inner part of the cover of the sample pan which, in d.s.c.but not with the TA 2000, may have a slightly lower temperature. Indeed, on carefully opening the d.s.c. sample pans after double-peak melting, a number of small needle-like crystals and/or small drop-like aggregates of crystals of 9-CNA have been observed, by microscopy, adhering to the inside of the aluminium pan covers. Conversely these features have not been observed on the inside of the TA 2000 pan covers. Indirectly a confirmation of this behaviour may be deduced from plate 3, in which needle-like crystals of 9-CNA are observed growing outwards from the partially decomposed parent 9-CNAD crystal. B. EVALUATION OF KINETIC PARAMETERS AND PROPOSED MECHANISM The kinetic analysis of the dynamic calorigrams performed as if the whole reaction were first or zero order (fig.2) gave us the same results as those previously reported8 for the first-order treatment, while excluding the possibility of an overall zero-order process. The existence of a first-order decay portion is confirmed by the kinetic analysis of the constant-temperature data. However, according to a better definition of the shape of the thermal curves (the number of peaks beyond the first one) the fitting of a first-order process to the low-temperature (high-activation-energy) portion of the dynamic thermal runs is probably an artifact and the corresponding activation energy, if meaningful, cannot be related to a single process. With the supporting evidence from optical microscopy, we believe that our isothermal and dynamic thermal determinations indicate the following sequence of events.(1) During the induction period monomer molecules begin to be formed (probably frdtn incipient dimer pairs) probably at defects randomly distributed within the parent dimer crystal and thus simulating a homogeneous distribution (yellowish appearance of the crystals). Very small submicroscopic nuclei of crystalline monomer are thus formed at the end of the induction period (assumed now not to include part of the first peak). (2) Three types of growth of nuclei are evidenced by microscopy: (i) very fast and along preferred directions; (ii) very fast and on particular planes; (iii) regular (i.e. not sudden or spasmodic) spread in all directions perpendicular to the line of observation. We assume that type (iii) is responsible for the general rising portion of the calorimetric (either isothermal or dynamic) exotherms, while the superimposed peaks (first and subsequent) are due to mechanisms of types (i) and (ii) which, because of their exothermicity and high rates, generate heat pulses.We tentatively attribute the fast stages (i) and (ii) to the progress of the reaction along dislocation cores and adjoining (elastically distorted) regions and to the spread of the reaction on slipped planes, respectively. However, if what we have named ‘directions’ are the traces of planes almost parallel to the direction of observation, growth mechanisms (i) and (ii) unify. The subsequent (and probably stress-assisted16) reaction of different planes characterized by progressively lower extension and/or reactivity can explain the presence of various peaks superimposed on the rising portion of the main broad exotherm.The fact that in isothermal and dynamic thermal experiments the shape and position of the peaks is strictly reproducible (if samples of the same batch are used and the experimental conditions replicated), as well as the observed fact that the shape of the peaks changes with crystal dimensions and crystallization procedures, are thought to be in agreement with our interpretation (uide infra under section C). (3) At the end of the acceleratory period, the situation is such that the original dimer crystal, even if preserving its primitive external shape, is so heavily internally cut byD. DONATI, P. SARTI-FANTONI AND G.G. T. GUARINI 783 planes of already formed crystalline monomer that it may be considered as consisting of small residual crystallites of dimer surrounded by Crystalline monomer. Thus we have the situation typically described1', for the first-order decay mechanism observed. When process (3) is also complete, the crystals usually preserve the shape of the original parent dimer crystals but must be considered as aggregates of extremely small particles of crystalline monomer. (The fact that decomposed crystals strongly diffuse light supports this interpretation.) The dimensions of monomer crystallites which constitute the aggregate simulating the original dimer crystal could not be determined by optical microscopy and are thought to be such as to simulate an amorphous phase when X-ray diffraction is used. Indeed the sharp melting observed on further heating the reaction product ensures that a crystalline monomer phase is formed during the reaction.C. NATURE OF THE FIRST (AND SUBSEQUENT) PEAKS The general shape of the thermal curves is the same in thermal experiments for both constant and linearly increasing temperatures (obviously curves of suitable temperatures and scan speeds must be compared). Thus, as expected, identical phenomena are taking place in the two cases. This is confirmed by microscopic examinations, at least for the initial portions of the monomerization reaction. Since our interpretation of the first peak in dynamic runs8 (exactly the same as is given here) was q~estioned,~~ l1 we wish to add some considerations to support our point of view.(1) If the first peak was due to crystallization of monomer formed during the induction period (< 5 7 3 , in our runs the absolute value of the heat of crystallization deduced from the first peak would be at least double the corresponding value for the heat of melting. (2) Again, assuming that the first and subsequent peaks are due to repeated crystallizations, it is not clear why the monomer formed in the decay period should not crystallize but rather form an amorphous phase, especially in the presence of previously formed crystalline monomer. The possibility that an amorphous phase is formed (in analogy to the case for 9-CN-10-ACAD8) is ruled out here by the melting experiments performed after each isothermal run. (3) The shape of the first and subsequent peaks changes with the crystal dimensions and with the crystallization procedures used to obtain the crystalline dimer.We believe that these changes in peak shape are better explained by our interpretation of the peaks than by the hypothesis based on the crystallization of the monomer. (4) The crystallization hypothesis could not explain the fact that, for a given sample, the first peak does not appear again on reheating after quick cooling (see Results sections A and B), while the subsequent peaks remain unaltered. Instead we believe that when a certain family of planes has reacted it will not react again on repeated heating. (5) The absence of the first peak in powdered materialsll is easily accounted for by the present interpretation.In fact, owing to the reported preparation,ll the powdered material is apt to be very poorly crystalline or not crystalline at all. This prevents the fast spreading of the reaction on preferred planes. CONCLUSIONS Our thermal and microscopic results confirm that the overall monomerization reaction of 9-CNAD consists of three subsequent stages as recently s~ggested.~9 l1 As to the mechanism of the three stages, our suggestion, based on our data, differs from784 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER what has previously been indicated for both the second and third stages. This disagreement is due mainly to the impossibility that the dynamic thermal curves may be interpreted by an overall zero-order proce~s.~ However, in agreement with the above-mentioned r e p o r t ~ , ~ ~ l1 the first stage (induction period) of the reaction is attributed to the formation of submicroscopic monomer nuclei probably originating from incipient dimer pairs.Our microscopic results then show that the second (acceleratory) period of the reaction is of a complex nature and comprises different kinds of nucleus growth. Thus its interpretation in terms of one of the known kinetic laws is unsafe and further studies are needed for a definitive clarification. At the end of the second stage, the dimer crystal is so heavily intersected by crystalline monomer formed on preferential planes that the remaining portion of the reaction takes place by a first-order consumption of residual dimer entities surrounded by the crystalline monomer already formed.The first and subsequent exothermic peaks evidenced in both dynamic and isothermal experiments are due to fast spreading of the exothermal reaction on preferred planes or directions with formation of crystalline monomer. No endothermic peak at 480 K attributable to the melting of residual crystalline 9-CNAD was observed under our experimental conditions. On the grounds of the present and recently reported1* isothermal results, our previous report, based on d.s.c. dynamic experiments,s is confirmed. The product of the monomerization reaction is crystalline, i.e. the crystallization front is very near the reaction front, and no evidence of amorphous phases is found, since the reaction product (9-CNA) was always found to have a sharp melting point. Our findings show that thermal methods do give reliable results in the study of the monomerization of anthracene and anthracene derivative photodimers. We have also found that 9-CNAD crystals are stable at room temperature over a period of 10 years. This information may be of some interest in view of the possible utilization of 9-CNAD crystals for energy storage. R. Calas and R. Lalande, Bull. SOC. Chim. Fr., 1959, 763. R. Lalande and R. Calas, Bull. SOC. Chim. Fr., 1960, 144. W. R. Bergmark, G. Jones 11, T. E. Reinhardt and A. M. Halpern, J. Am. Chem. SOC., 1978,100,6665. T. Laird, Chem. Ind., 1978, 186. G. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1970, 6, 423. C. Dufraisse and J. Mathieu, Bull. SOC. Chim. Fr., 1947, 307. R. Calas and R. Lalande, Bull. SOC. Chim. Fr., 1952, 434. D. Donati, G. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1972, 17, 187. E. M. Ebeid, S. E. Morsi and J. 0. Williams, J. Chem. SOC., Faraday Trans. I , 1979, 75, 11 1 1. lo D. Donati, G. G. T. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1981, 65, 147. l1 E. M. Ebeid, S. E. Morsi and J. 0. Williams, J. Chem. SOC., Faraday Trans. I , 1980, 76, 2170. l2 L. F. Fieser and J. L. Hartwell, J. Am. Chem. SOC., 1938, 60, 2555. l3 A. K. Galwey and M. E. Brown, Thermochim. Acta, 1979,29, 129; see also G. G. T. Guarini and R. l4 D. Donati, G. G. T. Guarini and P. Sarti-Fantoni, unpublished results. l5 G. G. T. Guarini, R. Spinicci, F. M. Carlini and D. Donati, J. Thermal Anal., 1973, 5, 307. l6 A. K. Galwey and G. G. T. Guarini, J. Chem. SOC., Chem. Commun., 1978, 273. l7 D. A. Young, Decomposition of Solids (Pergamon Press, Oxford, 1966). M. E. Brown, D. Dollimore and A. K. Galwey, in Comprehensive Chemical Kinetics, ed. C . H. Bamford and C. F. H. Tipper (Elsevier, Amsterdam, 1980), vol. 22. Spinicci, J. Thermal Anal., 1972, 4, 435. (PAPER 1 /484)