Crystal Structure and Thermal Solid-State Reactivity of ...znaturforsch.com/ab/v59b/s59b1229.pdftent with the interatomic distances N2-C1 116.5 pm and C1-N3 130.7, corresponding to

Crystal Structure and Thermal Solid-State Reactivity of AmmoniumCyanoureate NH4[H2NC(=O)NCN]

Bettina V. Lotsch and Wolfgang Schnick

Department Chemie und Biochemie, Lehrstuhl fur Anorganische Festkorperchemie,Ludwig-Maximilians-Universitat Munchen, Butenandtstraße 5-13, D-81377 Munchen, Germany

Reprint requests to Prof. Dr. W. Schnick. Fax +49-89-2180-77440.E-mail: [email protected]

Z. Naturforsch. 59b, 1229 – 1240 (2004); received July 30, 2004

Dedicated to Professor Hubert Schmidbaur on the occasion of his 70th birthday

The ammonium salt of cyanourea NH4[H2NC(=O)NCN] has been synthesised via an acid-baseroute from the parent acid and characterized by single-crystal and powder X-ray diffraction, NMR andvibrational spectroscopy, mass spectrometry as well as thermal analysis. The molecular salt (P21/c,a = 388.95(8), b = 1121.0(2), c = 1096.4(2) pm, β = 92.57(3)◦, V = 477.5(2)·106 pm3, Z = 4,T = 140 K) which may formally be derived from the related compound ammonium dicyanamideNH4[N(CN)2] by addition of one molecule water, consists of isolated ammonium and cyanoure-ate ions which are assembled via H· · ·N and H· · ·O hydrogen bonds, forming a three-dimensionalarrangement. At elevated temperatures, ammonium cyanoureate undergoes a complex transforma-tion affording the formation of urea and cyanoguanylurea H2NC(=O)NHC(NH2)=NCN or the iso-meric ammeline (C3N3)(NH2)2(OH) as the main products, depending on the reaction conditions. Thetransformation is accompanied by consecutive reactions such as proton transfer and the dis- and re-assembly of molecular fragments, yielding a macroscopic segregation of the reaction products. Theconversion represents yet another example of a complex reaction proceeding in the solid-state.

Key words: Solid-State Reaction, Crystal Structure, Cyanourea, Thermal Reactivity

Introduction

The detailed investigation of reaction mechanismshas largely been the domain of solution chemistry,owing partly to a large number of degrees of free-dom and the resulting direct and facile reaction path-ways, and to the well-established routine techniquessuch as solution NMR available for studying reactionsin situ. As a consequence, a number of reactions ofmajor importance, such as the historic conversion ofammonium cyanate into urea, have been investigatedthoroughly in the liquid phase, yet only very little at-tention has been directed towards the solid-phase be-haviour of the reactants [1 – 3]. Substantial progressconcerning solid-state reactivity has been achieved bythe systematic investigation of photoreactions such as[2+2] monomolecular and bimolecular photodimer-izations, which often exhibit single-crystal-to-single-crystal character [4]. However, satisfactory insight intothe mechanisms of thermally induced solid-state reac-tions is still rare.

0932–0776 / 04 / 1100–1229 $ 06.00 c© 2004 Verlag der Zeitschrift fur Naturforschung, Tubingen · http://znaturforsch.com

In order to close this gap, we have started a system-atic investigation of molecular solids that may be envi-sioned empirically as possible candidates for exhibit-ing solid-state reactivity due to their molecular andcrystal structures. On this background, we have beenfocussing particularly on dicyanamide salts, whosesolid-phase reactivity is largely driven by the respec-tive counter ions and which, therefore, provide a ver-satile and tuneable system for the study of solid-statereactions [5 – 8].

In this context, ammonium dicyanamide (2), whosereactivity may be understood by a simple isolobal anal-ogy with Wohler’s classic synthesis of urea, has turnedout to be a particularly suitable model system in termsof temperature, kinetics, purity, and completeness ofthe transformation [9, 10]. When heated to tempera-tures above 353 K, the ionic ammonium dicyanamideNH4[N(CN)2] is converted into its molecular iso-mer dicyandiamide (H2N)2C=NCN in a topochemicalsolid-state reaction [10 – 12]. Due to the fundamentalrole of the ammonium ion and the nitrile-group of the

1230 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

Scheme 1. Molecular formulas of several compounds whichare chemically or structurally related to the cyanourea struc-ture: (1) ammonium cyanoureate, (2) ammonium dicyan-amide, (3) cyanourea, (4) cyanoguanylurea, (5) cyanobi-uret, (6) ammeline, (7) 1,3-dicarbamoylguanidine, (8) biuret,(9) ammelide, (10) guanidinium cyanoureate, (11) cyanocar-bamoylguanidine, (12) [1-cyanoguanyl-3-guanyl]urea.

dicyanamide anion in this transformation, comparablesystems with similar reactive groups were devised inorder to address the question whether or not a coreframework of molecular and structural presuppositionsmay be established, which underlies and directs the ob-served type of solid-state reactivity.

Among the potential candidates, ammonium cyan-oureate (1) was found to be quite closely related toammonium dicyanamide (2), from which it can for-mally be derived by addition of H2O to one of thenitrile-groups, thereby maintaining the ammonium ionand one nitrile-group as the reactive centres. Althoughthe free acid cyanourea (N-cyanocarbamimidic acid)(3) has been mentioned as early as 1870 in the litera-ture [13] and its molecular structure and reactivity has

Table 1. Crystal data and structure refinement ofNH4[H2NC(=O)NCN].

Empirical formula C2 H6 N4 OMolar mass [g·mol−1] 102.11Crystal system monoclinicSpace group P21/c (no. 14)T [K] 140Diffractometer, monochromator STOE STADI 4, graphiteRadiation [λ / pm] Mo-Kα (71.073)Z 4a [pm] 388.95(8)b [pm] 1121.0(2)c [pm] 1096.4(2)β [◦] 92.57(3)V [106·pm3] 477.5(2)Calculated density [g·cm−3] 1.420Crystal size [mm3] 0.51×0.29×0.18Absorption coefficient [mm−1] 0.115Scan type ωF(000) 216Diffraction range [◦] 2.60 ≤ θ ≤ 28.02Index range −5 ≤ h ≤ 5,

−14 ≤ k ≤ 14,−14 ≤ l ≤ 14

Total no. reflections 6513Independent reflections 1158Observed reflections 980 (Rint = 0.0360)Refined parameters / constraints 89 / 0Corrections Lorentz, absorption,

polarisation, extinctionGOF on F2 1.073Final R indices |I > 2σ(I)| R1 = 0.0297

wR2 = 0.0769R Indices (all data) R1 = 0.0397

wR2 = 0.0829 withw = [σ2(Fo

2)+(0.0409P)2

+0.1113P]−1

where P = (Fo2 +2Fc

2)/3Min./max. residual electron −0.193 / 0.203density [e·A−3] −0.193 / 0.203

been characterized [14 – 17], crystal structure informa-tion is only available for the silver and, though in-complete, for the potassium salt of cyanourea [18, 19].Studies on the behaviour of cyanourea in aqueous andorganic solution reveal the susceptibility of cyanoureato intermolecular addition and consecutive decompo-sition reactions. In neutral and organic solution, differ-ent degrees of cyanoguanylurea (4) and cyanobiuret (5)are formed [20], whereas ammeline (6) (2,4-amino-6-hydroxy-1,3,5-triazine) and 1,3-dicarbamoylguanidine(7) are the main products in alkaline and acidic solu-tion, respectively (Scheme 1) [21, 22].

On the background of our ongoing studies of themechanistic aspects of solid-state reactions we are in-terested in the structure-reactivity relations of ammo-

B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN] 1231

Fig. 1. Crystal structure of NH4[H2NC(=O)NCN], viewalong [100]. Ellipsoids are drawn at the 70% probabilitylevel.

nium cyanoureate in the solid phase. In this contribu-tion we report on the synthesis, characterization and X-ray crystal structure of ammonium cyanoureate as wellas its solid-state reactivity as compared to ammoniumdicyanamide.

Results

Crystal structure

Ammonium cyanoureate crystallizes in the spacegroup P21/c (no. 14) with four formula units in themonoclinic unit cell. The anions are stacked in paral-lel chains along [001], forming a framework in whichpairs of ammonium ions are embedded (Fig. 1).

As observed in the silver salt of cyanourea [18], themolecular structure of the essentially planar cyanoureaanions exhibits a cisoid arrangement of the nitrilegroup and the oxygen atom, rendering the amide grouptrans to the nitrile moiety. The central nitrogen atomcarries the negative charge without significant de-localisation into the nitrile group, which is consis-tent with the interatomic distances N2-C1 116.5 pmand C1-N3 130.7, corresponding to the localisationof the triple bond and, hence, to the molecular for-mula N≡C-N-C(=O)NH2 for the cyanoureate anion(Table 2).

The anions are hydrogen bonded via weak N2· · ·H6contacts (241.8 pm) between adjacent amide and nitrile

Table 2. Bond lengths (pm) and angles (◦) for NH4-[H2NC(=O)NCN].

O1-C2 125.7(2) N4-C2 133.8(2)N4-H5 90(2) N4-H6 89(2)N2-C1 116.5(2) C1-N3 130.7(2)N1-H1 95(2) N1-H2 92(2)N1-H3 91(2) N1-H4 91(2)N3-C2 137.5(2)

C2-N4-H5 121(1) C2-N4-H6 120(1)H5-N4-H6 118(2) N2-C1-N3 174.7(1)H1-N1-H2 109(2) H1-N1-H3 108(2)H2-N1-H3 109(2) H1-N1-H4 108(2)H2-N1-H4 110(2) H3-N1-H4 113(2)C1-N3-C2 115.61(9) O1-C2-N4 121.9(1)O1-C2-N3 123.90(9) N4-C2-N3 114.23(9)

Hydrogen bonding

N1-H1. . . N2 200(2) N1-H1-N2 167(2)N1-H2. . . O1 191(2) N1-H2-O1 163(2)N1-H3. . . N3 199(2) N1-H3-N3 170(2)N1-H4. . . O1 203(2) N1-H4-O1 157(2)

Fig. 2. Coordination sphere of the ammonium ion inNH4[H2NC(=O)NCN]. The hydrogen bonding network isindicated by dotted lines. Ellipsoids are drawn at the 70%probability level.

groups along [001], and via medium strong N2· · ·H5contacts (216.1 pm) along [001], [0 11] and [011].

The ammonium ion exhibits slight deviations fromthe ideal tetrahedral symmetry with H-N-H bond an-gles between 108 and 113◦ (Table 2). The cation iscoordinated by three pairs of anions, of which onepair is doubly coordinated via O1 and N2, forming aninner coordination sphere that resembles an irregularheptahedron. The six closest cation(N1)-anion contactsrange between 280 and 346 pm, involving O1 (3×), thecentral nitrogen N3 (2×) and the terminal nitrogen N2(1×). The cation-anion assembly is characterized byfour medium-strong hydrogen bonds in the range from191 to 203 pm, where two donors are oxygen and twoare nitrogen atoms (bridging and terminal) (Fig. 2).

The temperature behaviour of the lattice parame-ters and cell volume of ammonium cyanoureate is dis-

1232 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

Fig. 3. Temperature dependence of the lattice param-eters (left y-axis) and cell volume (right y-axis) inNH4[H2NC(=O)NCN] as determined by X-ray powderdiffraction between 150 and 293 K. The contraction of thec-axis with rising temperature indicates a negative thermalexpansion coefficient, giving rise to an anomalous tempera-ture behaviour.

played in Fig. 3. As evidenced by the negative slopeof the curve pertaining to the crystallographic c-axis,the latter exhibits a negative coefficient of expansion,resulting in an anomalous thermal behaviour in the in-vestigated temperature range. In contrast, the cell vol-ume increases with temperature. This phenomenon isparalleled by ammonium dicyanamide [11, 12], whichalso exhibits a contraction of the c-axis with risingtemperature. In spite of the formal similarity consid-ering the symmetry and temperature characteristics ofthe structural parameters, the inner coordination sphereand the arrangement of the ions in the unit cells of thetwo structures differ. The question at issue in case of asolid-state transformation is the nature of the productdetermining factor, which may either be the extendedsolid-state arrangement of the ions as determined bythe pre-organisation of the molecules in the unit celland packing effects, or the thermodynamic drivingforce generating the energetically most favoured prod-uct (i. e. molecular structure) irrespective of the crys-tallographic conditions (“solution-like” conditions), orthe co-action of both factors. This issue will be consid-ered in detail in the following chapters.

Thermal analysis

Since a comprehensive description of the solid-phase reactivity of ammonium dicyanamide had beengiven previously [12], a comparative investigation ofthe thermal behaviour of the title compound seemed tobe particularly intriguing. The first question that had tobe addressed was whether reactivity in the solid stateis likely to occur in this system, or if melting or com-plex decomposition are the prevailing temperature re-

Fig. 4. DSC heating and cooling curve (inset) ofNH4[H2NC(=O)NCN] recorded between RT and 800 K witha heating (cooling) rate of 0.5 K min−1.

sponses. As a primary step, large crystals of ammo-nium cyanoureate (1) were investigated under a mi-croscope during thermal treatment of the sample up to453 K. This procedure did not affect the bulk crystalshape, yet led to a continuous clouding of the initiallyclear crystal together with a keying of the surface andsimultaneous softening of its consistency, thereby in-dicating the conversion into polycrystalline material orpotential solid-state reactivity to occur.

DSC and DTA/TG measurements indicate signifi-cant differences in the thermal behaviour of ammo-nium cyanoureate depending on the heating and pres-sure conditions, thereby rendering the course of thethermal reactivity a complex interplay between the lat-ter variables. The DSC curve for a sample heated con-tinuously up to 773 K in a closed aluminium cru-cible (heating rate 0.5 K min−1) displays a weak andbroad exothermic signal at 355 K, followed by a sharpexothermic signal (onset 385 K), which is preceded insome cases by an endothermic deflection of the base-line. Between 390 and 463 K, a complex series of ther-mal events is observable (Fig. 4). The main productrecovered after the above thermal treatment is urea to-gether with ammeline (6) as proved by X-ray pow-der diffraction, a frequently encountered decomposi-tion product of C-N-O materials [20, 22 – 24].

It is known from the literature that ammeline (6) isformed to a small extent at temperatures above 498 Kas a minor decomposition product of urea besides themain products biuret (8), cyanuric acid, ammelide (9),and melamine [25]. At temperatures below 473 K,however, the formation of ammeline by urea decom-position is negligible, suggesting a different source ofammeline in the present case.

B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN] 1233

Fig. 5. Isothermal TG and DTA heating curves ofNH4[H2NC(=O)NCN] recorded at 358 K. The weight of thesample was 21.97 mg.

If the above heating experiment is conducted in anAl2O3 crucible open to the atmosphere (≤ 1 K min−1),the products differ substantially from those obtainedunder DSC conditions. No ammeline formation is ob-served, whereas urea is detected in all product mixturesobtained below ≈ 425 K (at higher temperatures subli-mation of urea occurs). Above 473 K the products aretransformed into amorphous graphitic C-N materials.

If the sample is heated to 358 K (1 K min−1) in un-sealed Al2O3 crucibles and annealed for 40 h, no ther-mal event is recorded (DTA) in spite of the ongoingtransformation of the sample as evidenced by X-raypowder diffraction. The powder pattern of the prod-uct phase recovered after 40 h is largely identical tothat obtained in the heating experiments. The TG curveshows a continuous mass loss, which decreases after5 h at 358 K where it amounts to ≈ 9.1%. Most likely,the observed mass loss is due to the formation of am-monia. When annealing further the curve flattens, yetthe mass loss continues beyond the end of the mea-surement after 40 h where it has added up to 16.8%(Fig. 5).

The number of products obtained by reaction inan open system is very sensitive to only slight varia-tions of the reaction conditions and may in the mostunfavourable case lead to a complex mixture of sidephases along with the main products. In the follow-ing, the thermal behaviour under optimised isothermalconditions (358 K, pressure equalisation), leading to aminimum of products, will be considered in detail.

High temperature X-ray diffraction

The in situ observation of the phase transforma-tion by means of variable temperature X-ray powder

Fig. 6. Variable temperature X-ray powder diffraction pat-terns of NH4[H2NC(=O)NCN] recorded at 358 K for 20 h.The measuring time for each diffractogram (7 – 15.5◦ 2θ )was 11 minutes and the heating rate up to the starting tem-perature 5 K min−1. The starting material and products areindicated at the right margin.

diffraction further approves the absence of melting un-der controlled temperature conditions.

When heated isothermally at 358 K for severalhours, the onset of the phase transformation is ob-served after 8 h (Fig. 6). As indicated by the compar-atively long coexistence of the starting material andproduct phases (> 2 h), the reaction kinetics of thephase transformation may qualitatively be assessed asrelatively slow. The transformation onset and rate in-crease exponentially by raising the transformation tem-perature, the crystallinity of the product however beingcorrelated inversely with the annealing temperature.When compared to ammonium dicyanamide, whoseoptimum transformation temperature is between 368and 373 K, the temperature characteristics for thetransformation process of the two compounds is quitesimilar, which may suggest the ammonium ion to playa decisive role for the solid-state reactivity in bothsalts.

Due to the moderate crystallinity of the productphase, indexing of the powder patterns was not suc-cessful. By comparison, a part of the reflection linescould again be attributed to urea as observed in theheating experiments in a closed vessel, yet all othercomponents could not be identified from the hetero-geneous product phase by X-ray diffraction methods.Therefore, the identification of the products was at-tempted by complementary analytical methods, whichwas however complicated by the poor solubility of alarge fraction of the product phase in water or organicsolvents apart from DMSO.

In accord with the above observations, a seriesof single-crystal X-ray diffraction measurements con-

1234 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

ducted isothermally at 348 – 351 K confirm the con-version of a single crystal of NH4[H2NC(=O)NCN]into polycrystalline material after approx. 14 days. Thetransformation is indicated by the complete disappear-ance of the reflections and the emergence of Debye-Scherrer rings. The refinement of the data sets revealsthe occupation factor of N1 (ammonium nitrogen) tobe smaller than one (0.89 < sof (N1) < 0.93), sug-gesting a slight tendency towards the release of am-monia without effecting a collapse of the structure. Inorder to fulfil the criterion of electro-neutrality, an in-termolecular proton transfer to the anion must occur,leading to a small yet undetectable fraction of the freeacid cyanourea, which may be incorporated statisti-cally into the parent structure. Allowing the site occu-pation factor for N1 to be smaller than 1 (≈ 0.944(6))for the data set collected at 140 K also slightly im-proved the refinement. This finding may further cor-roborate the occurrence of an energetically favouredstatic disorder due to facile proton exchange reactionsin the solid state even at low temperatures, which is ac-companied by the loss of small amounts of ammonia.

Vibrational spectroscopy

In accordance with the IR spectra of the transforma-tion product the presence of urea in the product mix-ture is corroborated, indicated by the intense absorp-tion lines around 1664, 1620 and 1450 cm−1 (Fig. 7,middle). Principally, the appearance of the productspectrum suggests the conservation of the majorityof molecular fragments as compared to ammoniumcyanoureate, since the presence and location of the ab-

Fig. 7. IR spectra of ammonium cyanoureate (top), the de-composition product obtained at 358 K (middle) and cy-anoguanylurea (bottom), recorded in a range from 4000 –400 cm−1 at room temperature.

sorption bands pertaining to the N-H, C=O and C≡Nmoieties are not significantly altered upon transforma-tion.

This is indicative of the core of the starting materialto stay intact and, in particular, a nucleophilic attackof ammonia at the nitrile carbon as observed in am-monium dicyanamide to be most unlikely due to thestill clearly visible nitrile band in the product spectrum.Owing to the product mixture and the appearance ofseveral minor bands and band shifts, the complexity ofthe spectrum is increased. Among the newly emergedbands are signals at 2199 / 2159 cm−1 and 1703, 1499,1346, 1136, 1079, 972, 739, 674, 635, and 458 cm −1.In order to establish the identity of the missing productcomponents, the IR data are not sufficient and need tobe complemented by other spectroscopic data.

NMR spectroscopy

1H and 13C spectra of a range of products that wereobtained under varying temperature conditions wererecorded in DMSO-d6 solution at ambient temperature.The 13C NMR spectrum of the product obtained af-ter optimisation of the reaction parameters, which con-sequently exhibits the highest degree of purity, showsthree signals at 160.2, 156.0 and 116.8 ppm (Fig. 8). Insome cases signals at 167.1, 158.6, and 124.3 ppm arealso present, whose intensities strongly depend on thethermal history of the sample.

When the spectral region between 150 and 170 ppmis visualised enlarged, the splitting of the signalat 160 ppm becomes evident. Comparison with the

Fig. 8. 13C NMR spectrum of the solid-state decompositionproduct of ammonium cyanoureate recorded in DMSO-d6solution at 293 K. The region between 150 and 163 ppm isshown enlarged in the inset.

B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN] 1235

Fig. 9. 1H NMR spectrum of the solid-state decompositionproduct of ammonium cyanoureate recorded in DMSO-d6solution at 293 K. The 1H spectrum of cyanoguanylurea isshown in the inset; the asterisk denotes an unknown impu-rity.

13C NMR spectrum of urea permits the assignmentof one of the signals at 160 ppm to urea. The signalaround 117 ppm is easily attributed to a nitrile car-bon atom, whereas chemical shifts of the signals at156, 158 and 160 ppm are characteristic of amide- oramino(imino)methyl-groups with sp2-hybridized car-bon atoms. The signals at 167 and 124 ppm areidentical with the 13C chemical shifts of ammoniumcyanoureate, yet are also observed when no ammo-nium ion is detectable by 14N / 15N NMR spectroscopy.The 1H spectrum of the product typically exhibits sixsignals at 5.40, 6.47, 7.13, 8.20, 8.71, 9.45 ppm and animpurity at 6.08 ppm, where the signal at 5.40 ppm canagain be attributed to urea (Fig. 9). In several spectra,the signal around 7 ppm is superposed by a broad fea-ture, indicating fast exchange of acidic N-H protons astypically observed for ammonium or guanidinium ions.Simultaneously, a medium intense signal at 5.10 ppmis present, which is close to the NH2 protons of am-monium cyanoureate. The integrated intensities of thefive major resonances, which appear in the region ofboth acidic and non-acidic NH and NH2 groups, areapproximately equal.

Based on the spectroscopic information sketchedabove the molecular structure of the missing productcomponents may be derived:

Apart from urea, the major NMR signals in-dicate the formation of the isomeric urea deriva-tives cyanoguanylurea ([(cyanoamino)iminomethyl]-urea) (4, 4a) or [N-cyanocarbamoyl]-guanidine (11,

Scheme 2. Conformational isomers of cyanoguanylurea(4), (4a) and cyanocarbamoylguanidine (11) and (11a). In-tramolecular hydrogen bonding is indicated by dashed lines.

11a), which are sketched in Scheme 2. Exemplarily,two possible conformational or tautomeric isomers ofeach compound are shown, which are supposed tobe particularly favoured energetically if intramolec-ular hydrogen bonding is possible (“chelate effect”).Some spectra can only be interpreted conclusively byassuming the guanidinium salt of cyanourea (10) to bepresent in the product mixture along with the abovemain products. Thus, the presence of the 13C signalsat 158 (cation), 167 and 124 ppm (anion), as well asthe sharp 1H signal around 5 and the broad signal at7 ppm can be explained without invoking the presenceof small portions of unreacted starting material.

Mass spectrometry

The structural picture outlined above may be cor-roborated by mass spectrometry. Spectra obtainedby Direct Electron Ionisation (DEI+) clearly containpeaks at m/z 60 and 127, which can be assignedto the two main products urea and cyanoguanylurea(4) or cyanocarbamoylguanidine (11). The molecu-lar peaks are supplemented by peaks at 111 [M-NH2]+, 84 [M-HNCO]+, 44 [Murea-NH2]+ and 43[M-(H2N)2CNCN]+ and [Murea-NH3]+, belonging tomolecular fragments of the parent mass peaks. DEI+spectra of some decomposition products further con-tain weak peaks at m/z 169 and 152, which are consis-tent with the molecular structure of [1-cyanoguanyl-3-guanyl]urea (12). Although the formation of the lat-ter as a result of thermal decomposition reactions dur-ing sample evaporation cannot be excluded, the pres-ence of small amounts of 12 in the initial product mix-ture must be assumed and may be corroborated by theobservation of additional NMR signals in some prod-uct samples. Some spectra exhibit signals at m/z 126

1236 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

and 85, which are larger than the expected isotopepeaks of 127 and 84, respectively. These peaks maybe attributed to the trimerization of the intermediatelyformed cyanamide H2NCN to melamine and to the cor-responding decomposition products, or to the [M-H] +

peak of m/z 127. No mass spectroscopic evidence isfound for the formation of cyanobiuret (5), which wasclaimed to occur as a by-product in the decompositionof cyanourea in solution [20]. Mass spectra recordedin the FAB+ mode contain a large signal at m/z 60,whereas FAB− spectra show the presence of an anionwith m/z 84. These findings are fully consistent withthe assumption of guanidinium cyanoureate (m/z cation60, m/zanion 84) to be formed under certain conditionsas a minor conversion product.

As to the identity of the main transformationproduct, several reasons can be cited as evidencefor cyanoguanylurea (4) to represent the appropriatemolecular structure, yet the formation of small frac-tions of 11, depending on the temperature conditions,cannot be excluded:

Foremost, the decomposition of cyanourea or a mix-ture of the potassium salt of cyanourea with the par-ent free acid in aqueous solution was reported to entailthe formation of cyanoguanylurea (4, 4a) as the ma-jor product [20]. In order to compare the solid-statetransformation product with that obtained in aqueoussolution, a mixture of ammonium cyanoureate andcyanourea in a 3:1 ratio were reacted in water at 308 Kfor 7 – 10 days. Both the X-ray powder pattern as wellas the 1H and 13C NMR spectra of the above productcorrespond to those of the solid-state decompositionproduct, neglecting the reflections / signals due to ureapresent in the latter (Figs 8 and 9). Additionally, 15Nspectroscopic investigations of the sample synthesisedaccording to the above procedure give strong supportfor the validity of structure 4. In the proton coupled ordecoupled spectra, 15N signals are observed at −181.1(s, C ≡ N), −264.6 (s, NH), −278.0 (s, C=N), −291.2(t, NH2) and −295.3 ppm (t, NH2). For structure 11the 15N signal of the bridging -NH- unit would be ex-pected at δ < −300 ppm by analogy with the relatedcyanourea or cyanoureate structures.

Moreover, the band location of the cyano-groupsobserved in the IR spectra (2200 / 2159 cm−1)is very close to the nitrile signal in dicyandiamide(H2N)2C=N-C≡N (2205 / 2165 cm−1), whereas thatof cyanourea (H2N)(O=)C-HN-C≡N is positionedat 2254 cm−1, accompanied by a weak signal at2203 cm−1. The observed differences in the C≡N

stretching frequencies owing to the isolobal sub-stituents O and HN may further substantiate the valid-ity of the cyanoguanylurea structure.

In addition, the mass spectra strongly corroboratethe above conclusion by the presence of a fragmentpeak at m/z 84, which equals the molecular peak ofdicyandiamide. The facile fragmentation of the parentpeak at m/z 127 via a McLafferty type rearrangementinto HNCO (m/z 43) and dicyandiamide (m/z 84)is in coincidence with the experimental pattern. Theisomeric cyanocarbamoylguanidine structure (11, 11a)would generate fragments with m/z 42 and 85, the lat-ter of which however is not observed to a significantextent in the majority of the mass spectra.

Finally, the existence of cyanocarbamoylguanidine(11, 11a) has not yet been established, since no datawhatsoever are available on this compound in the lit-erature. Therefore, the lack of information on the hy-pothetical cyanocarbamoylguanidine strongly suggeststhe instability of this compound, presumably with re-spect to the cyclisation to ammeline (6), which thusadds to the correctness of the above line of argument.

Discussion

The previous section has given an outline of thetentative reaction participants and characteristics, yetno conclusive picture of the reaction mechanism hasbeen presented. This can however only be done specu-latively on the argumentative basis sketched above.

In order to illuminate the potential course of thetransformation, crystallographic details have to betaken into account. First of all, the product distribu-tion strongly suggests the reaction pathway to differsubstantially from that observed for ammonium di-cyanamide:

Whereas in the latter case, the nucleophilic attack ofthe in situ generated ammonia occurs at the nitrile car-bon atom of the dicyanamide anion, the conservationof the nitrile moiety in ammonium cyanoureate indi-cates the role of the ammonium group to be fundamen-tally different in the title compound. The evolution ofammonia appears to be more pronounced in the lattercase, since it is observable by a weak exothermic eventaround 355 K in the DSC measurements when the sam-ple is heated gradually. In contrast, the small mass lossdue to evolving ammonia is not detectable by DSC inammonium dicyanamide, suggesting ammonia not tobe retained in the transformation products of ammo-nium cyanoureate stoichiometrically as is the case for

B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN] 1237

Scheme 3. Two alternative mechanistic pathways of thesolid-state transformation of 1. The two-step mechanismsketched to the right is characterized by the nucleophilic at-tack of the intermediately formed cyanamide at the nitrilecarbon of a neighboured anion, resulting in cyanoguanylurea(4) and urea. The alternative pathway (left) shows the nu-cleophilic addition of ammonia to the cyanamide carbon C1,leading to urea and the guanidinium salt of cyanourea (10)without evolution of one mole ammonia.

Scheme 4. Tentative reaction mechanism leading to the iso-meric cyanocarbamoylguanidine (11) and urea by the actionof the amide nitrogen of 1 as nucleophile. Subsequent cycli-sation of cyanocarbamoylguanidine to ammeline (6) may befavoured under certain temperature conditions.

ammonium dicyanamide. As described previously, theoverall assembly of the ions shows a tendency towardsH· · ·O hydrogen bonding in the present case, which isin contrast to the molecular arrangement in ammoniumdicyanamide (H· · ·N bonding) and may therefore sig-nificantly alter the solid-state reactivity as compared tothe latter.

In ammonium dicyanamide, the trajectory of theprotons during the transformation is directed towardsthe terminal and, to a lesser extent, to the central nitro-gen atoms of the anion. Whereas the nucleophilic at-tack of the in situ generated ammonia can only proceedat one of the chemically equivalent electrophilic car-bon atoms in ammonium dicyanamide, leading finallyto the molecular compound dicyandiamide in any case,

Scheme 5. Mechanism of the assumed formation of [1-cyanoguanyl-3-guanyl]urea (12) from cyanoguanylurea (4)by nucleophilic addition of the latter to the intermediatelyformed cyanamide.

the potential attack of ammonia in the title compoundmay proceed at two different carbon atoms, leading ineach case to different products. The conservation ofthe nitrile group in one of the products suggests theattack of ammonia to take place at the carbamoyl C2atom, thereby inducing a breakage of the C2-N3 bondand, ultimately, the generation of urea and cyanamideH2NCN after proton exchange. The intermediatelyformed cyanamide is prone to consecutive reactions,such as the nucleophilic attack at a neighboured nitrilecarbon to form cyanoguanylurea (4) (Scheme 3, right).An alternative scenario resulting in a different productdistribution comes into play if ammonia is not evolvedfrom the sample, but adds to the cyanamide interme-diate to form the stable guanidinium cation, therebyblocking the nucleophilic attack of cyanamide at theparent cyanourea anion (Scheme 3, left). The spec-troscopic data suggest the coexistence of both reac-tion pathways at low temperatures, which are realisedby the different targets of the initial nucleophile am-monia. However, the significant mass loss observed inthe isothermal DTA / TG experiments (first step 9.1%,final mass loss 16.8%) suggests the former reactionpathway (Scheme 3, right) to be the dominant one.Also, the observed exothermic event at 355 K in theDSC heating curve can be correlated with the evolutionof ammonia, which at this stage does not induce signif-icant structural changes as evidenced by X-ray powderdiffraction. This may be assessed as further evidencefor the facile release of ammonia, necessarily resultingin follow-up proton-transfer reactions. In a consecu-tive reaction, the main product cyanoguanylurea mayfurther act as a nucleophile, attacking the intermedi-ate cyanamide to form [1-cyanoguanyl-3-guanyl]urea(12), which is detected as a side phase by mass spec-trometry (Scheme 5).

Since the reaction sequence leading to the observedproducts proceeds in the solid state and, hence, is dom-inated by little molecular freedom, the pre-orientationof the reactive centres is assumed to be such that thediffusion length necessary for the reaction to occur isminimised. Taking into account the thermal behaviour

1238 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

of the lattice parameters, a competing reaction path-way may become increasingly favoured. Since the an-ions are arranged in a head-to-tail manner along [001],the amide group of one anion is located close to the ni-trile moiety of a neighboured anion (N4· · ·C1 355 pmat 140 K) (Fig. 1). Owing to the anomalous contractionof the c-axis with rising temperature, the reactive cen-tres move closer and a nucleophilic attack of the amideN4 at C1 prior to or after breakage of the C2-N3 bondwould lead to the isomeric cyanocarbamoylguanidinestructure (11) as sketched in Scheme 4.

Although this reaction pathway seems to be facilein terms of the structural arrangement, no evidence hasbeen found for the formation of 11 as discussed above.It may however not be excluded that under certain re-action conditions the formation of this isomer 11 isfavoured, presumably followed by rapid cyclisation toammeline, which is found as the major product in theDSC experiments (Scheme 4).

Conclusion

The novel compound ammonium cyanoureate hasbeen synthesised and characterized by single-crystalX-ray diffraction and its solid-state decomposition elu-cidated with respect to the principle intermediates andreaction products. The thermal behaviour was foundto be very sensitive to the temperature conditions andmay yield a complex mixture of products owing toseveral competing reaction pathways. Whereas in aclosed system the formation of urea and ammeline isfavoured beyond 410 K, under optimised conditionsat 358 K cyanoguanylurea has been identified as amain product besides urea. The detection of the guani-dinium salt of cyanourea in varying amounts depend-ing on the reaction conditions corroborates the for-mation of cyanamide as a reaction intermediate. Inspite of its complexity, the transformation proceeds inthe solid phase, resulting in a microcrystalline mix-ture of the reaction products. Thus, the thermal reac-tivity of ammonium cyanoureate fundamentally differsfrom the related compound ammonium dicyanamidewhich may tentatively be correlated with the differinghydrogen bonding network, resulting in a more facilerelease of ammonia, and the altered reactivity of theanion due to the presence of an electrophilic amidecarbon. The above example further demonstrates theinterplay between the chemical functionality and thereaction conditions, leading to a variety of possiblereaction pathways different from those in solution,

which are not significantly limited by the rigid crystallattice.

Experimental Section

Synthesis

Ammonium cyanoureate was synthesised by neutralisa-tion of the free acid cyanourea with aqueous ammonia. Typ-ically, to 500 mg (5.9 mmol) cyanourea were added 1 –3 ml (13.3 mmol) conc. NH3 solution (Merck, 25%) and thewhite suspension dissolved in 12 ml water (pH = 9 – 10) andstirred for one hour. Afterwards the solution is filtered andleft for crystallisation by slow evaporation. Cyanourea wassynthesised by loading 8.0 g (50.6 mmol) urea-phosphate(Fluka, ≥ 98%) on a strongly acidic ion exchange resin(Merck, Ionenaustauscher I, H+-Form, Art. 4765), washingthe resin with 1.0 l water and pouring a solution of 1.0 g(11.2 mmol) sodium dicyanamide Na[N(CN)2] ] (Fluka, ≥96%, 0.5 M) onto the column. As a result, the dicyanamideis converted into cyanourea by acid hydrolysis without ionexchange taking place. A more convenient and direct syn-thesis for cyanourea, however, is the alkaline hydrolysis ofdicyandiamide in boiling 3N NaOH solution for 2 hours,which is described in detail elsewhere [24, 26]. 1H NMR(400.0 MHz, DMSO-d6): δ = 7.2 (s, 4H, NH4), 5.3 (s, 2H,NH2). – 13C{1H} NMR (100.5 MHz, DMSO-d6): δ = 123.5(C ≡N), 167.0 (C=O). MS (FAB−, 6 kV): m/z(%) = 84 (11)[H2NC(=O)NCN−].

Cyanoguanylurea was synthesised according to the pro-cedure described in [20] by constantly stirring 706 mg(6.9 mmol) of ammonium cyanoureate with 196 mg(2.3 mmol) of cyanourea in 15 ml water at 308 K. After 3 – 4days, the white product started to precipitate, yet the reac-tion was maintained for another 4 – 5 days and the insolubleproduct isolated by filtration. In order to increase the yield,100 mg cyanourea was added to the filtrate and the mixtureagain subjected the above reaction conditions.

The thermal conversion of ammonium cyanoureate waseither effected under “DSC conditions” as described in theThermal analysis section, or probed by heating the sample,which was placed in an unsealed Al2O3 crucible containedin a glass tube, with 1 K min−1 from 293 K to various tem-peratures between 400 and 463 K under pressure equalisa-tion (“DTA conditions”). Isothermal measurements were typ-ically conducted under the same conditions using target tem-peratures between 358 and 368 K. The annealing time rangedbetween 20 and 60 h.

X-ray diffraction

X-ray diffraction data of a NH4[H2NC(=O)NCN] sin-gle crystal were collected at 140 K and between 343 and351 K (high temperature measurements) on a four-circle

B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN] 1239

diffractometer (STOE Stadi 4) equipped with a 600 SeriesCryostream Cooler (Oxford Cryosystems), using graphitemonochromated Mo-Kα radiation (λ = 71.073 pm). Thestarting temperature for the high temperature measurementswas adjusted by applying heating rates of approx. 1 K min−1.The crystal structure was solved by direct methods em-ploying the program SHELXTS-97 [27] and refined on F2

using the full-matrix least-squares method implemented inSHELXTL-97 [28]. A semi-empirical absorption correctionbased on psi-scans was applied. The positions of all hydrogenatoms could be determined unequivocally from differenceFourier syntheses and all non-hydrogen atoms were refinedanisotropically. Details of the structure determination andcrystallographic data are summarised in Table 1. Intramolec-ular distances and angles are listed in Table 2. Further detailsof the crystal structure investigation are available from theFachinformationszentrum Karlsruhe, D-76344 Eggenstein-Leopoldshafen (Germany), on quoting the depository num-ber CSD-414279, the name of the authors, and citation of thepaper.

High temperature in situ X-ray diffractometry was per-formed on a STOE Stadi P powder diffractometer (Ge(111)-monochromated Mo-Kα1 radiation, λ = 70.093 pm), withan integrated furnace using unsealed quartz capillaries. Thedata collection was restricted to a 2θ range of 7 – 15.5◦ anda single scan collection time of 11 minutes. The sample washeated up to the starting temperature (358 K) by applyinga heating rate of 5 K min−1 and subsequently tempered for20 h.

Vibrational spectroscopy

FTIR measurements were carried out on a Bruker IFS66v/S spectrometer scanning a range from 400 to 4000 cm−1.The samples were prepared as a KBr pellet (5 mg sample,500 mg KBr) under atmospheric conditions.

Thermal analysis

Thermoanalytical measurements were conducted betweenRT and 573 K on a Mettler DSC 25 applying a heating rate of0.5 or 1 K min−1. The alumina crucibles used as sample con-tainers were placed in the calorimeter under an atmosphereof dry nitrogen.

For combined DTA/TG measurements at 358 K using aheating ramp of 0.5 or 1 K min−1 a Setaram thermoanalyserTGA 92-2400 was available. The samples were placed in anunsealed Al2O3 crucible and the measurements conductedunder nitrogen.

General methods

1H and 13C NMR spectra were recorded as DMSO-d6 so-lutions on a JEOL Eclipse EX-400 instrument, the chemicalshifts being referenced with respect to TMS.

Mass spectra were obtained using a Jeol MStation JMS-700 gas inlet system. The substances were dissolved in a 3-nitrobenzyl alcohol matrix (FAB−, FAB+) on a target andionized by bombardment with accelerated (6 kV) Xe atoms.For direct insertion probes using DEI+ (70 eV), the samplewas suspended in MeOH and heated rapidly on a platinumfilament.

Acknowledgements

The authors thank Dipl.-Chem. J. Weigand for the dona-tion of cyanourea, A. Lieb and Dipl.-Chem. U. Baisch forthe single-crystal data collection, W. Wunschheim and Dipl.-Min. S. Schmid for conducting the DSC and DTA/TG mea-surements, Dr. G. Fischer and D. Ewald, as well as Priv.-Doz.Dr. K. Karaghiosoff for carrying out the mass spectromet-ric and NMR measurements, respectively. Financial supportfrom the Fonds der Chemischen Industrie (scholarship forB. V. Lotsch) and the Deutsche Forschungsgemeinschaft isgratefully acknowledged.

[1] J. Shorter, Chem. Soc. Rev. 7, 1 (1978).[2] J. D. Dunitz, K. D. M. Harris, R. L. Johnston, B. M.

Kariuki, E. J. MacLean, K. Psallidas, W. B. Schweizer,R. R. Tykwinski, J. Am. Chem. Soc. 120, 13274(1998).

[3] C. A. Tsipis, P. A. Karipidis, J. Am. Chem. Soc. 125,2307 (2003).

[4] F. Toda, Organic Solid-State Reactions, Kluwer Aca-demic Publishers, Dordrecht (2002).

[5] A. P. Purdy, E. House, C. F. George, Polyhedron 16,3671 (1997).

[6] E. Irran, B. Jurgens, W. Schnick, Chem. Eur. J. 7, 5372(2001).

[7] B. Jurgens, E. Irran, J. Schneider, W. Schnick, Inorg.Chem. 39, 665 (2000).

[8] B. Jurgens, E. Irran, W. Schnick, J. Solid State Chem.157, 241 (2001).

[9] W. Madelung, E. Kern, Liebigs Ann. Chem. 427, 1(1922).

[10] B. Jurgens, H. A. Hoppe, E. Irran, W. Schnick, Inorg.Chem. 41, 4849 (2002).

[11] B. V. Lotsch, J. Senker, W. Kockelmann, W. Schnick,J. Solid State Chem. 176, 180 (2003).

[12] B. V. Lotsch, J. Senker, W. Schnick, Inorg. Chem. 43,895 (2004).

[13] F. Hallwachs, Liebigs Ann. Chem. 153, 293 (1870).[14] Yu. I. Mushkin, A. I. Finkel’shtein, Russ. J. Gen. Chem.

33, 1883 (1963).[15] J. S. Blair, G. E. P. Smith, J. Am. Chem. Soc. 56, 907

(1934).

1240 B. V. Lotsch – W. Schnick · Crystal Structure and Thermal Reactivity of NH4[H2NC(=O)NCN]

[16] H. Beyer, H. Schilling, Z. Chem. 5, 182 (1965).[17] H. Beyer, H. Schilling, Chem. Ber. 99, 2110 (1966).[18] D. Britton, Acta Crystallogr. C43, 2442 (1987).[19] N. S. Magomedova, Z. V. Zvonkova, Zh. Strukt. Khim.

15, 165 (1974).[20] K. Iio, Bull. Chem. Soc. Jpn. 57, 625 (1984).[21] Yu. I. Mushkin, A. I. Finkel’shtein, Kinet. Katal. 7, 219

(1966).[22] Yu. I. Mushkin, A. I. Finkel’shtein, Zh. Org. Kh. 3, 507

(1967).[23] H. Beyer, Z. Chem. 6, 213 (1966).

[24] H. Bieling, P. Barth, H. Beyer, Z. Chem. 4, 146 (1964).[25] P. M. Schaber, J. Colson, S. Higgins, E. Dietz, D. Thie-

len, B. Anspach, J. Brauer, American Laboratory 31,13 (1999).

[26] H. Beyer, Z. Chem. 6, 213 (1966).[27] G. M. Sheldrick, SHELXS97, Program for the Solution

of Crystal Structures, University of Gottingen (1997).[28] G. M. Sheldrick, SHELXL97, Program for the Refine-

ment of Crystal Structures, University of Gottingen(1997).