New light on an old story: Formation of melam during thermal condensation of melamine

Article abstract of DOI:10.1002/chem.200601291

We report on the existence and formation of the carbon nitride precursor melam (H₂N)₂(C₃N₃)NH-(C ₃N₃)(NH₂)₂, thereby clarifying one of the last unresolved issues posed by the complex thermal condensation of melamine C₃N₃(NH₂)₃. Experimental proof is put forward that melam is a direct condensation product of melamine, but can be detected only in small amounts under special reaction conditions owing to its rapid transformation into melem. The coexistence of melamine and melem during thermal condensation yields two adduct phases with distinct compositions [C₃N₃(NH₂)₃] ₂[C₆N₇-(NH₂)₃] and [C₃N₃(NH₂)₃][C₆N ₇-(NH₂)₃]₂. They may be considered as co-crystallizates of melamine and melem and can be isolated as intermediates between 590 and 650 K prior to the formation of single-phase melem C ₆N₇(NH₂)₃. Melam (C2/c, a = 1811.0(4), b = 1086.7(2), c = 1398.4(3) pm, β= 96.31(3)°, V= 2735.3(9) × 10⁶ pm³, T=130 K) adopts a ditriazinylamine-type structure with a twisted conformation about the bridging NH moiety and transforms into melem around 640 K. Two compounds deriving from melam have been synthesized by solution and solid-state reactions. The salt melamium diperchlorate C₆N₁₁H₁₁(ClO₄) ₂·2H₂O (C2/c, a = 1747.8(4), b = 1148.2(2), c = 993.6(2) pm, β= 118.79(3)°, V= 1747.4(6)×10⁶ pm ³, T=130K) crystallizes as a dihydrate and exhibits a doubly protonated, planar melam core. In the neutral complex Zn[C₆N ₁₁H₉]Cl₂ (P2₁/c, a = 743.00(15), b = 2233.2(5), c = 762.5(2) pm, β = 99.86(3)°, V= 1246.5(4) × 10⁶ pm³, T=200K), melam acts as a symmetrically bent bidentate ligand, which is coordinated to the Lewis acid Zn-site through two ring nitrogen atoms.

Full text of DOI:10.1002/chem.200601291

DOI: 10.1002/chem.200601291
New Light on an Old Story: Formation of Melam during Thermal
Condensation of Melamine
Bettina V. Lotsch and Wolfgang Schnick*[a]
Abstract: We report on the existence
and formation of the carbon nitride
co-crystallizates of melamine and
melem and can be isolated as inter-
mediates between 590 and 650 K prior
to the formation of single-phase melem
thesized by solution and solid-state re-
actions. The salt melamium diperchlo-
precursor melam (H2N)2
A
rate C6N11H11
1747.8(4),
A
G
U
the last unresolved issues posed by the
complex thermal condensation of mela-
C6N7
1811.0(4),
1398.4(3) pm,
(NH2)3.
Melam
b=1086.7(2),
b=96.31(3)8,
(C2/c,
a=
c=
V=
993.6(2) pm,
1747.4(6)”106 pm3, T=130 K) crystalli-
mine C3N3
A
zes as
doubly protonated, planar melam core.
In the neutral complex Zn[C6N11H9]Cl2
(P21/c, a=743.00(15), b=2233.2(5), c=
762.5(2) pm, b=99.86(3)8, V=
a dihydrate and exhibits a
is put forward that melam is a direct
condensation product of melamine, but
can be detected only in small amounts
under special reaction conditions owing
to its rapid transformation into melem.
The coexistence of melamine and
melem during thermal condensation
yields two adduct phases with distinct
2735.3(9)”106 pm3, T=130 K) adopts a
ditriazinylamine-type structure with a
twisted conformation about the bridg-
ing NH moiety and transforms into
melem around 640 K. Two compounds
deriving from melam have been syn-
AHCTREUNG
1246.5(4)”106 pm3, T=200 K), melam
acts as a symmetrically bent bidentate
ligand, which is coordinated to the
Lewis acid Zn-site through two ring ni-
trogen atoms.
Keywords: carbon nitrides · phase
transitions · solid-state structures ·
thermal decomposition
compositions
(NH2)3] and
(NH2)3]2. They may be considered as
[C3N3
A
ACHTREUNG
A
[C3N3
A
ACHTREUNG
ACHTREUNG
Introduction
products of melamine, melam [(H2N)2
G
melon “[C6N7(NH)(NH2)]n”, have not been ultimately clari-
A
ACHTREUNG
As part of the ongoing quest for graphitic carbon nitride
and its potentially ultrahard 3D analogue, the chemical
screening of suitable precursors has attracted significant at-
tention in the past few decades. Recently the focus has been
shifted to a class of historically important compounds assem-
bled around the common ancestor melamine (triamino-s-tri-
fied due to the lack of crystallographic data.[24,25–29] Owing to
their prototypic carbon(IV) nitride backbones, resolving the
identities of these potential CNx precursors may entail far-
reaching implications for the structure and formation of g-
C3N4, at the same time shedding light on the historically as
well as industrially important thermolysis reaction of mela-
mine.
Melam was first prepared in 1834 by Liebig on heating an
intimate mixture of potassium thiocyanate and an excess of
ammonium chloride at temperatures around 580–640 K.
From the fusion product a compound was isolated, after
treatment with potassium hydroxide solution, which was ar-
bitrarily given the name “melam” by Liebig.[1] Since the
19th century much effort has been devoted to the elucida-
tion of the composition and structure of the pyrolysis prod-
ucts of melamine, which, however, was hampered by con-
flicting statements with respect to their elemental composi-
tion and ways of formation.[24,25,30,31] Whereas the identity of
melamine was resolved by the elucidation of its crystal
azine) C3N3
G
modern carbon nitride chemistry:[2–21] While the crystal
structure of the heptazine derivative melem C6N7(NH2)3 has
E
been elucidated only recently,[22,23] the existence and, above
all, molecular structures of the conjectured condensation
[a] Dr. B. V. Lotsch, Prof. Dr. W. Schnick
Department Chemie und Biochemie, Ludwig-Maximilians-Universi-
tät, Butenandtstraße 5–13 (D), 81377 München (Germany)
Fax : (+49)89-2180-77440
E-mail: wolfgang.schnick@uni-muenchen.de
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
4956
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

FULL PAPER
structure in 1941,[32] chemical inertness and low solubility of
its condensation products has impeded thorough characteri-
zation ever since their early discovery.
native tautomeric form was introduced, accounting for reso-
nance within the two triazine rings (Scheme 2).[27,39,40]
Despite the above controversy, reports have been pub-
lished on the synthesis of melam and substituted melams
Melam has been considered the low-temperature de-am-
monation product of melamine, which is formed upon link-
ing two molecules of melamine with concomitant release of
one mole of ammonia. The resulting “ditriazinylamine”-type
structure model was set out as early as 1886 by Klason[33]
and accepted by later investigators.[24,25,34,35] Doubts about
the correctness of this proposed structure arose following
several unsuccessful attempts to obtain melam by pyrolyzing
melamine, the products of which were identified as melem
by IR and UV spectroscopy.[27–29,36,37] In contrast, May re-
ported on the isolation of melam, melem, and melon as the
pyrolysis products of melamine obtained at 633, 673, and
773 K, respectively, based on elemental analysis.[26] The two
alternative schemes proposed in the literature for the con-
densation process of melamine are outlined in Scheme 1.
Route a is based on the formation of melem as the first
condensation product, giving rise to a ratio of melamine to
liberated ammonia of 1:1. Melem then affords the polymeric
condensate melon, the identity of which has not been re-
solved as yet.[27,29,37] Melon may further lose ammonia to
form graphitic carbon nitride g-C3N4, which, however, has
not been proven until know and is subject of ongoing specu-
lation. In Route b, melamine progressively condenses to
graphitic carbon nitride via melam (melamine/ammonia=
2:1), melem, and melon (Scheme 1).[26,38,39] A detailed under-
standing of the pyrolysis of melamine, however, seems to be
pivotal to clarify the stability ranges of triazine and hepta-
zine systems with respect to the structure elucidation of g-
C3N4.[2–4,13,28,29]
Scheme 2. Two alternative tautomeric forms discussed for melam. The
conjugated form (right) exhibits a quinoid triazine ring system.
based on combinations of (alkyl-substituted) aminotri
with electrophilic triazine derivatives such as chlorotria-
zines.[40,41] The chloro derivatives of melam HN
(C3N3Cl2)2
and N(C3N3Cl2)3 were recently obtained by the reaction of
azines
AHCTREUNG
AHCTREUNG
melamine with cyanuric chloride and are being discussed as
potential intermediates in the formation of polymeric C3N4
compounds.[15,42] Presently, melam and its derivatives are
under consideration as flame-retardant formulations owing
to their exceptional thermal stability and decomposition
properties.[43–45] The synthesis of melam salts from melamine
in the presence of acids, among which is a zinc salt of as-
sumed formula Zn2+[C6N11H8]À , has been reported; howev-
2
U
er, no analytical data except for elemental analysis in the
latter case are available.[28,46] Costa and Jürgens have investi-
gated the thermal behavior of melamine salts, thereby
coming across melam hydrohalides and their ammonium
chloride adducts of the empirical compositions C6N11H9·HBr
and [C6N11H10]X·0.5NH4X (X=Cl, Br), whose structures
could, however, not be elucidated.[47,48]
While the originally proposed formula for melam as
sketched in Scheme 1 has long been acknowledged, an alter-
According to the literature data summarized above, the
knowledge on the formation of melam from melamine, as
well as its molecular structure,
is scarce and subject to much
controversy. In the following
we
will
comprehensively
sketch the thermal behavior of
melamine and present an alter-
native view on the formation
and existence of melam.
Results and Discussion
Thermal behavior of mela-
mine: A crucial factor when in-
vestigating the thermal behav-
ior of melamine is the choice
of the reaction conditions. Typ-
ically, such studies have been
conducted in closed systems, as
melamine exhibits
a strong
Scheme 1. Schematic representation of the two pathways proposed for the thermal decomposition of mela-
mine. Route a assumes direct formation of melem in the first condensation step, whereas according to Route b
melam is obtained prior to melem formation. Further loss of ammonia may yield polymers of melem
(“melon”) and, ultimately, graphitic carbon nitride.
tendency toward sublimation
above 570 K, which peaks
around 618 K. A lack of clarity,
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4957

W. Schnick and B. V. Lotsch
however, remains with regard to the processes taking place
in this temperature range, as it has not yet been specified
whether or to what extent melamine undergoes sublima-
tion,[29] melting,[27,36] or condensation reactions.
as shown in Figure 2. All efforts to index the powder diffrac-
tograms or grow single crystals were unsuccessful. Conse-
quently, the characterization of the two phases was conduct-
ed by spectroscopic methods.
We have performed temperature-programmed XRD
measurements (TPXRD) with heating rates of 1 KminÀ1 in
a temperature range between 298 and 933 K (Figure 1). It
Figure 2. X-ray powder patterns (CuKa1 radiation) of melamine, phase I,
phase II, and melem (bottom to top) recorded at room temperature. All
compounds exhibit well distinguishable powder patterns indicative of dis-
tinct crystallographic phases.
Figure 1. In situ temperature-programmed X-ray powder diffraction
measurements (MoKa1 radiation) of melamine recorded between room
temperature and 923 K. Four phase transitions are observed prior to the
formation of a graphitic carbon nitride material visible by a broad reflec-
tion at 2qꢀ12.18 (dꢀ336 pm).
Adduct phases: Mass spectrometry was carried out as a first
step to identify the intermediate LT and HT phases. For
both phases, only peaks at m/z 126 and 218 were detected,
which correspond exactly to the molecular mass peaks of
melamine and melem, respectively. At the same time, no
evidence was found for melam at m/z 235. According to
these results, which conform to the observations previously
made by Jürgens,[48] strong evidence is put forward for mela-
mine and melem instead of melam to actually constitute the
phases under consideration.
The composition of both phases was determined by ele-
mental analysis; the results are listed in Table 1. It is evident
that the composition of phase I indeed closely corresponds
to that of melam as stated by May and others.[26,38,39] The
composition of phase II, however, neither matches that of
melamine, nor that of melam or melem to a first approxima-
should be noted that heating a sample in an open quartz ca-
pillary affords conditions similar to those in closed systems
as has been observed previously.[20] We will therefore use
the term “semi-closed system” for this experimental set-up.
Under the slow heating conditions applied, at least four
transformations are observed: The reflections of melamine
disappear at 590 K without passing through an amorphous
state, with the new phase being stable up to approximately
630 K. At this temperature, new reflections gradually
appear, the phase change being defined rather poorly.
Melem is formed around 653 K and is stable up to 780 K, at
which temperature a single broad reflection characteristic of
so-called graphitic carbon nitride materials emerges, again
without passing through the melt.
When comparing with the above literature survey (cf.
Scheme 1), Route b conforms quite well to these observa-
tions, whereas the direct formation of melem from mela-
mine (Route a) without the formation of intermediates de-
tectable by in situ X-ray diffraction can be discarded under
these conditions. The observed low-temperature (LT) phase
transition may tentatively be associated with the proposed
formation of melam, which, however, does not account for
the second phase change (HT phase transition) observed at
630 K prior to the formation of melem.
Upon trying to isolate the above phases ex situ, a strong
dependence on both temperature and time is perceivable, as
the HT phase (phase II) is formed by isothermal treatment
from the low-temperature phase (phase I), the same apply-
ing to the formation of melem from the high-temperature
phase. After optimizing the heating conditions, both phases
could be isolated in a pure state without admixtures of mel-
amine, melem, or phases I/II detectable by X-ray diffraction
Table 1. Elemental analysis data observed and calculated for different
compounds and compositions.
Compound
phase I
Sum formula
N [wt%]
65.93
C [wt%]
30.56
H [wt%]
3.63
A
N
ACHTREUNG
phase II
A
N
ACHTREUNG
melamine
melam
ACHTREUNG
N
A
ACHTREUNG
AHCTREUNG
A
N
ACHTREUNG
melem
ACHTREUNG
A
N
ACHTREUNG
2:1 adduct[a]
1:2 adduct[b]
C
[C6N10H6]
U
G
ACHTREUNG
ACHTREUNG
A
N
ACHTREUNG
[a] melamine/melem=2:1. [b] melamine/melem=1:2.
4958
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

Graphitic Carbon Nitride
FULL PAPER
tion. Anticipating the results discussed below, the following
noticeable detail may be pointed out:
and 1700 cmÀ1 are largely identical, apart from pronounced
differences of the relative band intensities and splittings.
Whereas for phase II (top) a split band is seen at 1600–
1630 cmÀ1, this band is broadened and slightly weaker in the
spectrum of phase I (bottom). However, bands at 1585 and
1561 cmÀ1 observed for the latter, which are typical of tri-
In view of the fact that the elemental composition of
melam is in between those of melamine and melem, it may
be possible to obtain an identical composition by assembling
melamine and melem in different ratios into a co-crystal.
This has formally been done for two compositions according
to a 2:1 adduct of melamine and melem, as well as a 1:2
adduct, the theoretically expected compositions of which are
given in Table 1. Interestingly, the 2:1 adduct has exactly the
same composition as melam, and is thus in agreement with
the experimental data obtained for phase I. With similar
precision the compositions of phase II and the 1:2 adduct
can be considered identical. These observations thus nicely
explain the results from mass spectrometry, which yield only
the molecular mass peaks of melamine and melem.
We therefore emphasize that an ultimate proof of melam
formation solely based on elemental analysis data is intrinsi-
cally critical and requires corroboration by complementary
methods. To further substantiate this hypothesis, the results
from complementary analytical techniques will be discussed
in the following.
AHCTREUNG
AHCTREUNG
at 1650 cmÀ1, no absorption is found for both I and II higher
than about 1630 cmÀ1, which may be indicative of a signifi-
cantly altered hydrogen-bonding network. This is supported
by the fact that the sharp n(NH) stretching vibrations for
melamine at 3468 and 3419 cmÀ1 are replaced by a weak,
though sharp band centered at about 3464 cmÀ1 for I and II.
Phases I and II exhibit a strong split band around 800 cmÀ1
attributable to the sextant bend of both triazine and hepta-
zine rings, which is located intermediate between those
found in melamine and melem, respectively. Although no ul-
timate conclusions can be drawn from vibrational spectros-
copy, the following aspects may be pointed out:
For melam one would expect more pronounced absorp-
À1
À
tion in the n
A
The IR spectra of phases I and II are shown in Figure 3
(bottom), together with the spectra of melamine and melem
(top) for comparison. The principal absorption “envelopes”
of phases I and II in the characteristic ring-stretching and
NH2-bending region of strongest absorption between 1350
below), which is not observed for I and II. A simple super-
position of the spectra of melamine and melem can also be
discarded (see also XRD results, Figure 2), as too many
band shifts are observed for both phases relative to the
former. Nevertheless, a particularly significant similarity be-
tween the spectrum of phase II and that of melem is evi-
dent. In addition, the overall resemblance of the spectra for
I and II suggests that the two phases are related with respect
to their constituents.
NMR investigations have been carried out both in the
solid state and in solution. The 15N solid-state NMR experi-
ments greatly suffer from the long relaxation times of the
samples, rendering the signal-to-noise ratio rather low. The
13CCP-MAS solid-state NMR spectrum of phase I shows
two signal groups, of which one is located at d=155.3 ppm,
and a split signal has maxima at d=166 and 168 ppm
(Figure 4, top). A similar spectrum, yet with a slight high-
field shift of the signal group now centered around d=
166 ppm, is obtained for phase II (Figure 5, top). While the
resonance at d=155.3 ppm is unchanged, the intensity at
d=165.0 ppm is increased, and the signal portion around
d=167.4 ppm is reduced with respect to phase I. Both “pat-
terns” contain features reminiscent of melem (middle
curve),[22,48] whose peaks are located at d=155.1/156.0 ppm
and d=164.3–166.4 ppm, and melamine, which has two
close-by signals at d=167.5 and 169.2 ppm (bottom curve).
Hence, from the 13Cspectra of the two intermediates an in-
creasing similarity with melem becomes evident on going
from phase I to phase II.
The 15N spectrum of phase I exhibits peaks between d=
À198 and À210 ppm and two signals in the high-field region
(d=À274 ppm, doublet at d=À291/À294 ppm, Figure 4,
bottom). To assess the 15N proton environment, a CPPI ex-
periment has been carried out (see Figure S1 in the Support-
ing Information). In agreement with its chemical shift, the
Figure 3. Top: FTIR spectra of melamine (bottom curve) and melem (top
curve). Bottom: FTIR spectra of the LT phase I (bottom curve) and the
HT phase II (top curve), measured as KBr pellets.
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4959

W. Schnick and B. V. Lotsch
Figure 4. 13C(top) and 15N (bottom) CP-MAS solid-state NMR spectra of
the LT phase I (top curve; recycle delay: 300 s; tc: 4 ms for both meas-
urements; spinning frequency 10 kHz) in comparison with melem
(middle curve; recycle delay: 160 (13C) and 30 s (15N); tc: 4 ms for both
measurements; spinning frequency: 7 kHz) and melamine (bottom curve;
recycle delay: 20 s; tc: 8 ms (13C) and 4 ms (15N); spinning frequencies:
6 kHz (13C) and 5.5 kHz (15N). The signal assignments for phase I were
additionally corroborated by means of a CPPI experiment (see text).
Spinning side bands are marked by asterisks.
Figure 5. 13C(top) and 15N (bottom) CP-MAS solid-state NMR spectra of
the HT phase II (top curve; recycle delay: 360 s; tc: 4 ms for both meas-
urements; spinning frequency 10 kHz) in comparison with melem
(middle curve) and melamine (bottom curve, for parameters cf. Figure 4).
Spinning side bands are marked by asterisks.
Although the spectra of phases I and II largely parallel
those of melamine and melem with respect to the signal po-
sitions, obvious differences are observed regarding the
number of signals, especially if the NH2 regions for phases I
and II are compared to those of melem and melamine, re-
spectively. This observation may reflect the varying ratios of
melamine and melem in the two adduct compounds, as well
as pronounced differences in terms of the molecular ar-
rangements in the solid state.
Notably, the lack of signals pertaining to NH groups is a
strong indication that none of the nitrogen atoms of the con-
stituents of phases I and II is protonated (or deprotonated),
and that no larger molecular aggregates condensed via NH
bridges are present. This, however, further substantiates the
view that both phases are built up by a co-crystallizate of
melamine and melem, forming an array glued together by
hydrogen-bonding forces rather than by acid–base or cross-
linking reactions. The noticeable differences between the
spectrum of phases I and II on the one hand, and that of a
simple superposition of the spectra of melamine and melem
on the other hand, suggest that no simple heterogeneous
mixture of melamine and melem crystals is present, and that
the molecular arrangement and intermolecular forces in the
adduct phases are significantly different from those domi-
nant in the crystal structures of the pure compounds.
signal group centered at d=À200 ppm can be assigned to
tertiary nitrogen atoms, whereas the signals below d=
À270 ppm were identified as NH2 groups owing to their
characteristic polarization inversion behavior.[21,22,48,49] The
nitrogen resonance expected for the central 15N nucleus of
the heptazine core is not observed in the experiment, which
may be a natural consequence of the low signal-to-noise
ratio.
The 15N spectrum of phase II exhibits both slightly differ-
ent signal positions and numbers, with an increase of signal
multiplicity in the chemical shift region characteristic of the
melem-type Ntert and NH2 groups (d=À198.5 to À204.4 ppm
and d=À269.2 to À295.9 ppm). Remarkably, the two up-
field NH2 signals in phase I, which may be attributable to
the triazine part of the compound, are collapsed to one
single resonance at d=À295.9 ppm in phase II. The signals
of the presumed triazine nitrogen atoms typically observed
around d=À210 ppm in melamine and phase I is obviously
shifted down-field in phase II (Figure 5, bottom). A reso-
nance attributable to the central nitrogen atom of the cya-
meluric nucleus can be weakly distinguished at d=
À233 ppm. The following aspects need to be emphasized:
4960
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

Graphitic Carbon Nitride
FULL PAPER
Figure 6 outlines the solution NMR spectra of adducts I
and II, as well as those of melamine and melem for compar-
ison. Most notably, this time the spectra of I and II appear
to be identical to a superposition of those of melamine and
These results corroborate the above line of argument that
phases I and II in fact consist of co-crystals formed by mela-
mine and melem in two different, yet well-defined ratios.
Although no evidence for melam was found by thermal
decomposition of melamine under the above conditions, the
existence of melam cannot be disputed based on these find-
ings alone. It may not even be legitimate to reject the possi-
bility that melam is indeed formed as a condensation prod-
uct of melamine if different reaction conditions apply. To
pursue this thread, the decomposition of melamine under at-
mospheric pressure was studied.
1
melem, the chemical shift of both H and 13Csignals being
largely identical.[50]
Melam: When melamine was pyrolyzed in a porcelain cruci-
ble only loosely covered with a lid and placed in a muffle
furnace at 613–633 K for approximately two days, tiny color-
less crystals formed on the surface of the grayish bulk mate-
rial. These crystals, which may easily be taken for melamine,
were identified as melam (1) by single-crystal X-ray analy-
sis.[51] The X-ray powder pattern of the bulk material resem-
bles that of a mixture of impure adduct phases I and II,
though of poor crystallinity. Melam is either formed as a
minor side product, or most likely further transformed into
melem with a high rate constant, such that it is essentially
invisible in the in situ X-ray powder patterns. In accord with
the latter conjecture, all melam crystals completely vanish
on prolonged heating at the same temperature.
Although based on this observation the existence of
melam has been unambiguously proven, we note that it is
detectable only if the heating and tempering conditions are
very carefully chosen. It appears that only a certain time-
window, as well as the condition of atmospheric pressure,
allow the observation of melam.
Melam crystallizes in the space group C2/c with 1.5 for-
mula units in the monoclinic unit cell, such that two crystal-
lographically independent melam molecules are present.
Relevant crystallographic data are given in Table 2. The mo-
lecular structure is characterized by two triazine units con-
densed through a bridging NH group (Figure 7). As melam
is non-protonated, both rings are essentially equivalent
which is in agreement with the special position occupied by
N15 and the generation of the second triazine unit by sym-
metry. Although the second melam molecule is fully con-
tained in the asymmetric unit, both triazine rings exhibit
similar features and are grouped around N4 in a “staggered”
arrangement. It should be noted at this stage that the hydro-
gen positions at the bridging nitrogen atoms have been
found explicitly from difference Fourier syntheses, therefore
supporting the tautomeric form without conjugation be-
tween the triazine ring systems (“ditriazinylamine”, see
Scheme 2, left).
Figure 6. 1H (top) and 13C(bottom) NMR spectra of adducts I and II
measured in [D6]DMSO solution. The respective spectra of melamine
(bottom) and melem (top) are shown for comparison.
Whereas for phase I, the intensity of the proton signals
belonging to melamine outweighs that of the melem protons
(ratio ꢀ2.3:1.0), the situation is opposite for phase II (ratio
ꢀ1.4:2.0). These findings must, however, be treated with
some caution, as in both cases the sample was not complete-
ly dissolved in DMSO, so that a preferential uptake of a
single component may not be excluded. Nevertheless, the
significant variation in the intensities between phase I and
II, together with the results from elemental analysis, suggest
that the latter effect may not be particularly strong and the
general trend can nevertheless be derived.
Chemical evidence for the disintegration of the adduct
phases in solution was obtained by boiling a suspension of I
and II, respectively, in water so as to “extract” melamine,
which is soluble in hot aqueous solution. In fact, melamine
was isolated from the filtrates, whereas the white residues of
adduct I and II exhibit identical IR spectra, which were
identified as melem–hydrate phases (see Figure S3 in the
Supporting Information).
Notably, the melam molecules are not planar but are
twisted about the central nitrogen atom, and display dihe-
dral angles about the bridging NH groups of about 118
(N15) and 148 (N4), respectively. Accordingly, the amino
groups attached to the triazine rings do not interfere, which
together with a more efficient hydrogen bonding network
may be the reason for the observed nonplanarity. The angle
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4961

W. Schnick and B. V. Lotsch
Table 2. Selected crystallographic data for melam (1), melamium perchlorate hydrate (2), and the Zn–melam
complex (3).
the triazine rings (see below).
The amino groups also depart
from planarity in that they
slightly bend out of the triazine
ring plane. This may be indica-
tive of only moderate partici-
pation of the free electron pair
at the amino nitrogen atoms in
the ring p system, which is sup-
1
2
3
empirical formula
molar mass [gmolÀ1
crystal system
space group
temperature [K]
C6N11H9
235.24
monoclinic
C2/c (no. 15)
130
“
A
N
]
472.84
monoclinic
C2/c (no. 15)
130
diffractometer, monochroma-
tor
radiation, l [pm]
Z
Enraf–Nonius Kappa
CCD, graphite
MoKa, 71.073
12
STOE IPDS, graphite
Enraf–Nonius Kappa
CCD, graphite
MoKa, 71.073
4
MoKa, 71.073
4
À
ported by the fact that the C
Next bonds (133–135 pm) are
slightly elongated by analogy
with melamine.[32] Another fea-
ture is the obvious distortion
of the triazine rings, which are
neither completely flat nor reg-
ular hexagons, the latter being
characteristic of triazine rings:
While the N-C-N angles are
close to 1268, the C-N-C angles
vary between 1138 and 1148.
unit cell dimensions
a [pm]
b [pm]
1811.0(4)
1086.7(2)
1398.4(3)
96.31(3)
2735.3(9)
1.714
1747.8(4)
1148.2(2)
993.6(2)
118.79(3)
1747.4(6)
1.797
743.0(2)
2233.2(5)
762.5(2)
99.86(3)
1246.5(4)
1.980
c [pm]
b [8]
volume [106 pm3]
calculated density [g cmÀ3
crystal size [mm3]
]
0.20”0.12”0.04
0.127
0.23”0.14”0.12
0.453
0.16”0.12”0.03
2.409
absorption coefficient [mmÀ1
diffraction range [8]
index range
]
3.56ꢁqꢁ27.48
À23ꢁhꢁ23
À13ꢁkꢁ14
À18ꢁlꢁ18
285/0
3.66ꢁqꢁ27.49
À22ꢁhꢁ22
À14ꢁkꢁ13
À11ꢁlꢁ12
170/0
3.27ꢁqꢁ27.50
À9ꢁhꢁ9
À28ꢁkꢁ29
À9ꢁlꢁ9
217/0
À
parameters/restraints
total no. reflections
independent reflections
observed reflections
GooF on F2
The C N bond lengths within
5976
7067
5595
the rings do not vary to a large
extent, amounting to 134–
135 pm on average, whereas
3125 (Rint =0.0162)
2706
1.048
1907 (Rint =0.0359)
1620
1.058
2857 (Rint =0.0378)
2186
1.057
À
À
the bonds C4 N6, C1 N2, and
final R indices [I>2s(I)]
A
À
C7 N13 adjacent to the bridg-
R1
wR2
0.0350 (0.0414)
0.0926 (0.0970)
À0.244/0.205
0.0400 (0.0493)
0.1009 (0.1048)
À0.437/0.465
0.0306 (0.0505)
0.0697 (0.0754)
À0.491/0.345
ing NH group are systematical-
ly shorter (133 pm).
min./max. residual electron
density (e”10À6 pmÀ3
)
The molecules are connected
through medium strong and
weak hydrogen bonds with
donor–acceptor distances rang-
ing from 297 to 333 pm. No
[a] Idealized formula.
pronounced p-stacking is observed, reflected by a screwlike
arrangement of the molecules, which gives rise to a 3D net-
work of mutually tilted melam molecules (Figure 8).
The IR spectrum of melam is displayed in Figure 9. The
similarities to the spectrum of melamine (Figure 3, top) is
evident, which is particularly pronounced in the n(NH)
region above 3100 cmÀ1 as well as for the d
N
N
Figure 7. Representation of the molecular structure of melam. The two
crystallographically independent melam molecules are shown. Thermal
displacement ellipsoids (except for H) are drawn at the 50% probability
level.
C-N-C at the NH group is significantly widened (1298 (N4),
1328 (N15)) and strongly deviates from tetrahedral symme-
try, giving rise to a rather trigonal-planar arrangement. The
Figure 8. Unit cell of melam (view along [010]), showing the spatially ex-
tended, spiral-type stacking of the molecules. Thermal displacement ellip-
soids (except for H) are drawn at the 50% probability level.
À
À
C NH Cbond lengths are up to 4 pm longer than those in
4962
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

Graphitic Carbon Nitride
FULL PAPER
having a shoulder on the high-field side. Presumably, these
hardly separable resonances correspond to the three distin-
guishable carbon sites within a single triazine nucleus. The
theoretically possible number of nine carbon sites can clear-
ly not be resolved. The chemical shift values are almost
identical with those observed for melam hydrate (not
shown) and very similar to the signal positions in the ZnCl2–
melam complex (see Figure 14).
The 15N CP-MAS NMR spectrum of melam (Figure 10,
bottom) exhibits three separable tertiary nitrogen signals be-
tween d=À190.0 and À205.9 ppm, and two NH signals at
d=À252.0 and À254.8 ppm, the former one being slightly
less than half as intense as the latter. This finding corre-
sponds well to the crystal structure analysis, according to
which one would expect two NH groups with an intensity
ratio of 2:1 (one bridging nitrogen on a special position with
a site occupation factor of 0.5), which is almost exactly re-
produced in the experimental spectrum. Five different NH2
signals are visible at d=À282.8, À286.0, À288.9, À295.0, and
À300.4 ppm. As for the tertiary nitrogen positions, signal
overlap due to similar electronic and geometric environ-
ments must therefore be inferred, which reduces the theo-
retically expected numbers of independent 15N signals in
both cases.
The thermal behavior of melam as studied by TPXRD
measurements reveals the conversion of melam into melem
at 640 K without passing through the melt (see Figure S5 in
the Supporting Information), and thus confirms its direct re-
lation with this important carbon nitride precursor. The re-
sults presented above help to provide a new basis for the
controversial view on melam formation from melamine.
On pyrolyzing melamine in closed or semi-closed systems
as outlined above, melam is not observable as the first con-
densation product of the starting material, which, however,
may result from its high reactivity in the respective tempera-
ture range. Similarly, the direct formation of melem from
melamine without the formation of detectable intermediates
lacks experimental evidence.
Figure 9. FTIR spectrum of melam, recorded at room temperature be-
tween 4000 and 650 cmÀ1 in reflection geometry (neat solid).
N) absorption between 1640 and 1414 cmÀ1. Though having
similar intensities, the latter bands are well resolved and
sharp in contrast to those of melamine, and an additional
band at 1583 cmÀ1 is found for melam that is absent in both
melamine and melem. Similar to melamine, the characteris-
tic sextant ring bend is located at 807 cmÀ1. Whereas for
melamine and melem no pronounced absorption is found
À
below 1400 cmÀ1, two strong bands attributable to the n
(C
G
N) and d(NH) vibrations are found for melam at 1338 and
1250 cmÀ1.
13CCP-MAS NMR spectroscopy (Figure 10, top) reveals
two close-by signals at d=164.0 and 167.2 ppm, the latter
Van der Plaats and co-workers suggested that between
593 and 673 K melamine forms mixed crystals with its de-
composition product,[36] however, without further substanti-
ating this hypothesis. We are now in a position to confirm
this view, as two intermediate phases could be isolated that
are stoichiometrically built up from melamine and melem.
According to NMR investigations and dissolution/extraction
studies, the homogeneous phases comprise hydrogen-
bonded arrays of neutral melamine and melem molecules
rather than salts originating from chemical reactions of the
components (acid–base reactions).
Presumably, melem is gradually formed at elevated tem-
peratures and reintegrated into the starting material mela-
mine, upon which a crystallographically distinct phase is
formed. It may be argued that the onset of melamine subli-
mation greatly facilitates the condensation reaction to
melem, as melamine in the gas phase may be far more reac-
tive than the melamine residue. Note that under the reac-
tion conditions chosen, a large amount of melamine eludes
Figure 10. 13C(top) and 15N (bottom) CP-MAS solid-state NMR spectra
of melam (recycle delay: 16 s; tc: 4 ms for both measurements). The spin-
ning frequency was 10 kHz.
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4963

W. Schnick and B. V. Lotsch
the condensation reaction and resublimes at the cold end of
the ampoule. In fact, whereas for phase I the yield amounts
to 60–70%, it is only 30–35% for phase II. This finding,
however, strongly suggests that the transformation from
phase I into II is simply a consequence of melamine subli-
mation, which is thus removed from the solid. The forma-
tion of two adducts with distinct compositions may therefore
be driven by the particularly beneficial arrangements of the
components in the solid at the particular melamine/melem
ratios, together with the possibility for melamine to be re-
moved from the solid phase by sublimation. By this equilib-
rium between the solid and the gas phase, the transforma-
tion of adduct I into adduct II is feasible without major mo-
lecular rearrangements, and possibly even without further
condensation of melamine into melem.
From the above statements it may be inferred that the
adduct phases are only metastable at their formation tem-
peratures, which is evidenced by the strong time dependence
of their synthesis. Once the temperature sufficient for
melem formation is reached, the latter is continuously pro-
duced until all melamine present in the solid is consumed.
Thus, the adduct phases I and II are only intermediate
stages, stable for a limited time, on the way to the thermo-
dynamically stable product melem.
rapid transformation into melem is short-lived and—contra-
ry to the adduct phases—essentially undetectable on the
time scale of in situ TPXRD. The fact that melam was only
observed when working in open systems may be indicative
of a deceleration of its transformation kinetics with decreas-
ing pressure as outlined above. This slow-down may again
correlate with the reduced availability of ammonia in the
gas phase, a species which is known to enhance melem for-
mation.[22,48]
Melam compounds: To explore the largely unknown chemis-
try and structural diversity of the important CNx precursor
melam, the acid–base behavior as well as the suitability of
melam as a complex ligand has been studied. Reactions in
solution were carried out using melam hydrate as the start-
ing material, solid-state reactions were conducted based on
melamine and the respective salts.
The majority of the reactions carried out in acid solution
yielded amorphous or poorly crystalline products that could
not be fully characterized. In principle, the addition of weak
acids such as acetic or formic acid to aqueous suspensions of
melam hydrate only yield poorly crystalline hydrate phases,
whereas strong acids form the respective melamium salts, in-
dicating the weakly basic properties of melam by analogy to
melamine and melem.[52] On prolonged refluxing of the
product solutions, hydrolysis converts the melamium salts
into the respective ammelinium salts.[53]
An alternative approach to melamium salts is given by re-
acting melamine with acid salts, preferably ammonium salts,
in closed ampoules at temperatures between 623 and 723 K.
This strategy is particularly intriguing as it provides access
to melam on a preparative scale starting from the commer-
cially available compound melamine. Remarkably, the ten-
dency of melam formation from melamine is greatly en-
hanced in the presence of acid salts, which presumably facil-
itate the condensation of the triazine rings. If melamine is
heated with ammonium halides, polycrystalline melam hy-
drohalide–ammonium halide adducts are formed, from
which melam hydrate can easily be isolated on treatment
with bases. This observation may rationalize the early dis-
covery of melam by Liebig who unwittingly prepared melam
via its thiocyanate or hydrochloride salt, liberating the neu-
tral molecule by “purification” with potassium hydrox-
ide.[1,33] We note that not only Brønstedt acids are suitable
reactants but also Lewis acids such as ZnCl2, FeCl3, or even
CuCl2, which—via the respective cations—may bring the re-
active triazine sites into contact by coordinating to two ring
systems, thereby functioning as connectors (see Figure 13).
The crystal structures of two representative compounds of
melam, which were obtained by using alternative proce-
dures, will be highlighted in the following.
Therefore, thermal recrystallization of the adducts, which
are only available as polycrystalline powders, is not feasible,
as this would lead to the “extraction” of melamine from the
solid. Likewise, recrystallization in solution results in the
rupture of the hydrogen-bonding network and, ultimately,
phase separation into the components melamine and
melem.
According to our findings and as noted previously by Fin-
kelꢁshtein,[27] melam is detectable to only a small extent and
only under specifically tuned reaction conditions. This obser-
vation may be rationalized by invoking the formation of
melam as a minor by-product instead of a true intermediate
of the condensation cascade of melamine. Along these lines,
a scenario may be envisaged that involves the comparatively
slow, reversible formation of melam from melamine. In a
parallel reaction, the latter may transform into melem via
volatile species such as cyanamide or dicyandiamide, which
are generated rapidly by depolymerization of melamine
under increased pressure of ammonia. Melam may be trans-
formed back into melamine at elevated temperatures, which
then further reacts to melem with a significantly increased
reaction rate as compared to that of melam formation. Low-
ering the pressure of ammonia may enhance melam forma-
tion (by slowing down the re-transformation into melamine),
while depolymerization of melamine and, thus, generation
of melem, is concomitantly slowed down.
However, a more straightforward rationale is the rapid
consumption and further transformation of melam into
melem as the condensation reaction proceeds, which is in
accord with in situ X-ray studies of the thermal behavior of
melam (see Figure S5 in the Supporting Information). Ac-
cordingly, melam may be viewed as a highly reactive inter-
mediate of the thermolysis of melamine, which owing to its
Melamium perchlorate: Melamium perchlorate (2) was ob-
tained from reaction of melam hydrate with perchloric acid
in aqueous solution. Relevant crystallographic data are sum-
marized in Table 2.[51] The unit cell (space group C2/c) com-
prises four formula units, each consisting of a doubly pro-
4964
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

Graphitic Carbon Nitride
FULL PAPER
tonated melam molecule, two perchlorate anions, and ap-
proximately two molecules of water of crystallization, of
which one is disordered. Only one triazine ring and one per-
chlorate anion are contained in the asymmetric unit, there-
fore the cation comprises two identical triazine units and all
perchlorate anions are identical.
In contrast to structures containing neutral melam mole-
cules, the cationic melam core is planar within experimental
error (Figure 11). Also, the distortion of the triazine rings is
sumed crystal water O6. The melam molecules form planar
layers extending along b, the long molecular axes being par-
allel to the [10À1] direction (Figure 12). The perchlorate
ions are inserted between the melam molecules, linking the
latter by hydrogen contacts with donor–acceptor distances
between 288 pm (N6···O1) and 301 pm (N6···O3). Remarka-
bly, the stacking distance between adjacent planes is as
small as 325 pm.
Figure 12. Unit cell of melamium perchlorate hydrate, view along [010].
Thermal displacement ellipsoids (except for H) are drawn at the 50%
probability level.
Zn–melam complex Zn[C6N11H9]Cl2: The ZnCl2–melam
G
complex (3) was synthesized from melamine and ZnCl2 in a
solid-state reaction. The neutral adduct crystallizes without
crystal water in the monoclinic space group P21/c with four
formula units in the unit cell. Crystallographic data for Zn-
Figure 11. Representation of the molecular structure of melamium per-
chlorate hydrate, of which one half is contained in the asymmetric unit.
The oxygen positions O6 are only half occupied, and the proton positions
H1a and H3a are disordered over the melam backbone with SOFs being
close to 0.5. Thermal displacement ellipsoids (except for H) are drawn at
the 50% probability level.
A
complex, melam functions as a neutral “ligand” coordinated
to a ZnCl2 molecule through the ring nitrogen atoms N2
and N5, such that the Zn1 atom is located in a distorted tet-
rahedral environment comprising two N and two Cl atoms
(Figure 13). These results clearly disprove the assumption of
Gal’perin et al. that a chloride-free Zn–melam salt contain-
ing an anionic melam core is formed upon the reaction of
melamine with zinc chloride under these conditions.[46]
The steric requirements in the Lewis acid–base complex
forces the melam ligand to slightly bend out of the molecu-
lar plane, thereby introducing a symmetrical (C2v) distortion
about the bridging nitrogen atom. Although the distortion is
different from that found in pure melam, the flexibility of
the N-bridged triazine core again is obvious. The triazine
obvious, yet less pronounced than in melam, with the aver-
age angles C-N-C and N-C-N within the rings amounting to
À
1178 and 1248, respectively. The internal C N distances vary
À
between 132 and 136 pm, whereas the external C Nexo bond
lengths are shorter by about 2–3 pm (131/132 pm).
Notably, the two triazine rings do not possess a quinoid
À
structure with a bridging C1 N4=C1 unit, but a C1-NH4-C1
bridge with elongated bond lengths (138 pm). The presence
of two perchlorate ions per melam molecule suggests melam
to be present as a doubly protonated cation. However, four
proton sites attached to the ring nitrogen atoms are found
from difference Fourier synthesis, which have site occupa-
tion factors on free refinement between 0.5(1) (H3a) and
0.7(1) (H1a). Thus, disorder has to be inferred, entailing an
almost equal protonation of N1 and N3 such that in total
two protons are attached to the rings.
Another sign of static disorder is found at the crystal
water site O6. Instead of a single oxygen a split position is
found in the refinement with the centers of mass lying
119 pm apart, which is clearly too short for two adjacent
water molecules. The refined site occupation factor for O6 is
0.562(9), thereby indicating an approximately equal occupa-
tion of both split positions with the sum being equal to one.
It has to be noted, however, that in contrast to the situation
for O5 no hydrogen positions were found around the pre-
Figure 13. Asymmetric unit of the neutral zinc chloride - melam complex,
in which melam functions as a bidentate ligand. Thermal displacement el-
lipsoids (except for H) are drawn at the 50% probability level.
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4965

W. Schnick and B. V. Lotsch
rings slightly deviate from planarity, with the angles C-N-C
and N-C-N systematically alternating between 1148/1158 and
À
1248–1268, respectively. Differences in the C N bond
lengths within the rings (132–138 pm) are more pronounced
than in melam and melamium perchlorate, but the varia-
tions in both rings appear to be systematic. This effect can
most probably be ascribed to a drain of electron density
from the ring to the Lewis acid Zn2+ site, which then is com-
pensated by a rearrangement of the charge distribution ac-
À
cording to a quinoid-type structure. The C Next distances
are on average about 2 pm (133 pm) shorter than the ring
À
À
C N bonds. The triazine rings are again bridged by a C1
À
N1H C4 unit with equally elongated bond lengths (138 pm)
and an angle at the imino nitrogen atom of 1348.
The complexes form undulated strands extending along
[010], with neighbored melam molecules being hydrogen-
bonded to each other by weak N···N contacts of 306 and
309 pm. The “layers” are spaced by approximately 362 pm
and stacked irregularly such that the triazine ring planes of
adjacent sheets exhibit varying offsets.
The 13Csolid-state CP-MAS NMR spectrum reveals three
resonances, the small chemical shift differences of which in-
dicating comparatively similar electronic environments of all
13Cnuclei. Nevertheless, the number of signals nicely corre-
lates with the three different carbon positions in each ring
and the symmetry of the complex. Conducting the CP meas-
urements with different contact times as shown in Figure 14
Figure 14. CP-MAS solid-state NMR spectra of the melam–ZnCl2 com-
plex. Top: 13Cspectra recorded with two different contact times to identi-
fy the proton environment of the different carbon sites (top curve: 2 ms,
bottom curve: 10 ms; recycle delay: 60 s). Bottom: 15N CP (top curve,
spinning frequency 10 kHz) and CPPI spectra (bottom curve, spinning
frequency 5.5 kHz); the signal assignments are indicated (recycle delay:
60 s; tc: 4 ms; inversion time ti: 160 ms).
(top) enables a differentiation between those (2
+ 2)
carbon atoms adjacent to amino groups (d=163.4/
165.7 ppm), which exhibit faster magnetization transfers
from the proton bath, and those two which are neighbored
to the bridging NH group (d=161.4 ppm) and exhibit
higher intensity at longer contact times (spectrum a;
Figure 14 top).
groups exhibit almost identical chemical shifts each, so that
the observation of two NH2 signals can be rationalized.
In the 15N CP-MAS NMR spectrum (Figure 14 (bottom),
top spectrum) six signal groups are found, of which the sig-
nals at d=À193.8 and À207 ppm are located in the chemical
shift range characteristic of the tertiary nitrogen atoms of
triazine rings. The signal at d=À54.0 ppm exhibits a chemi-
cal shift typical of NH groups, whereas the signals at d=
À280.7 and À291.6 ppm lie in the NH2 region. The signal at
d=À238.1 ppm is located intermediate between the Ntert and
NH region and can—apart from chemical considerations
based on the crystal structure—not be assigned unambigu-
ously without further data. Thus, a CPPI experiment has
been carried out with an inversion time typical for NH sig-
nals to pass zero intensity (160 ms).[22] As shown in Figure 14
(bottom), bottom spectrum, the magnetization of the NH
signal at d=À254 ppm is zero, whereas according to the ex-
pectations that of the NH2 groups is negative. The signal at
d=À238 ppm clearly has hardly lost intensity and can there-
fore be assigned to a tertiary nitrogen resonance.
Conclusion
In summary, we have comprehensively delineated the ther-
mal condensation of melamine, which was shown to yield
two crystallographically distinct phases composed of neutral
melamine and melem molecules prior to melem formation,
if the reaction is conducted under the pressure of ammonia.
At ambient pressure small amounts of melam single crystals
can be isolated, which may be considered either as a side
product or, more likely, as a short-lived, reactive intermedi-
ate which is rapidly transformed into melem by means of
further ammonia elimination. We hypothesize that owing to
the initial lack of analytical data apart from elemental anal-
ysis, the 2:1 co-crystallizate of melamine and melem may
have easily been mistaken for melam. Nevertheless, melam
can be produced on a preparative scale by reacting mela-
mine with acid salts and treating the resulting melamium
salts with aqueous bases, upon which melam hydrates are
formed. Melam crystallizes in a “ditriazinylamine”-type
structure, in which the two triazine rings are twisted with re-
spect to each other. Protonated melam cations, which are
The above results suggest that the coordination of melam
at two ring nitrogen sites to the Zn cation leads to a high-
field shift for the 15N nuclei (shielding). The remaining two
signals can then be attributed to ring nitrogen sites not in-
volved in the Zn coordination. Two of the four amino
4966
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968

Graphitic Carbon Nitride
FULL PAPER
down and stored in the refrigerator until colorless needles began to crys-
tallize. IR (KBr): n˜ =3608.2 (w), 3536.1 (w), 3419.6 (m), 3347.9 (m),
3256.9 (m), 2868.7 (w), 2748.7 (w), 1686.7 (m), 1639.8 (s), 1560.2 (m),
1525.5 (w), 1478.5 (w), 1444.9 (m), 1394.7 (m), 1328.5 (m), 1262.3 (m),
1088.6 (s), 1060.4 (s), 1018.1 (m), 993.1 (w), 931.5 (vw), 876.5 (vw), 800.7
(vw), 787.0 (m), 740.6 (vw), 627.3 (m), 583.5 (w), 512.2 (w) cmÀ1. MS
(70 eV, DEI+): m/z (%): 18 (100) [H2O], 126.1 (68) [C3N6H6]+, 235.1
(70) [C6N11H9]+; (6 kV, FAB+): m/z (%): 236.2 (5) [C6N11H10]+; elemen-
unstable in neutral aqueous solution owing to their acid
properties, adopt an essentially planar structure, whereas
the melam backbone in the neutral ZnCl2–melam complex
is symmetrically bent, thereby demonstrating the significant
steric flexibility of this potential “bidentate” complex ligand
and precursor for graphitic carbon nitride.
tal analysis calcd (%) for [C6N11H11][ClO4]2 ”2H2O (472.2): C15.26, N
G
32.63, H 3.18; found: C16.08, N 34.24, H 3.22.
Melam–zinc dichloride: Melamine (200 mg, 1.59”10À3 mol) was intimate-
ly mixed with ZnCl2 (108 mg, 7.95”10À3 mol) and quickly transferred
into a Duran ampoule (length 120 mm, 1ext 16 mm, 1int 13 mm), which
was sealed under dry argon. The ampoule was placed in a vertical tube
furnace and heated at 703 K for 12 h (heating rate 1 KminÀ1). IR (KBr):
n˜ =3454.8 (s), 3447.2 (s), 3429.3 (s), 3349.6 (s), 3179.1 (m), 1671.5 (s),
1645.7 (s), 1620.5 (s), 1568.3 (m), 1548.1 (s), 1474.0 (m), 1418.1 (s), 1358.1
(s), 1302.8 (m), 1155.3 (w), 1109.2 (w), 1089.3 (vw), 1016.0 (w), 799.8 (s),
769.1 (w), 738.8 (vw), 722.0 (w), 694.8 (w), 629.0 (m), 601.7 (w), 570.5
Experimental Section
Synthesis
Adducts I and II: The adduct phases I and II (see text) were synthesized
by heating melamine (200 mg, 1.59”10À3 mol, ꢂ99%, Fluka) in a sealed
Duran ampoule (length 120 mm, 1ext 16 mm, 1int 13 mm) placed in a
vertical tube furnace at 663 K with a heating rate of 1 KminÀ1. The tem-
pering time for adduct I was 2.5 h, for adduct II 14 h were appropriate
(for lower temperatures down to 630 K, the tempering time had to be
readjusted, that is prolonged depending on the exact temperature). Data
for adduct I: 1H NMR (400 MHz, [D6]DMSO, TMS): d=7.3 (s, 6H;
NH2), 6.0 (s, 12H; NH2) ppm; 13C{1H} NMR (400 MHz, [D6]DMSO,
TMS): d=167.0, 165.1, 155.3 ppm; MS (70 eV, DEI+): m/z (%): 126.1
(100) [C3N6H6]+, 218.1 (100) [C6N10H6]+ (relative intensities refer to suc-
cessive scans); data for adduct II: 1H NMR (400 MHz, [D6]DMSO): d=
7.3 (s, 12H; NH2), 6.0 (s, 6H; NH2) ppm; 13C{1H} NMR (400 MHz,
[D6]DMSO): d=167.5, 165.6, 155.8 ppm; MS (70 eV, DEI+): m/z (%):
126.1 (70) [C3N6H6]+, 218.1 (100) [C6N10H6]+. For further analytical data
see text.
(w), 518.6 (vw), 460.5 (w) cmÀ1 MS (DCI+): m/z (%): 127.2 (26)
.
[(C3N6H6)+H]+; (70 eV, DEI+): m/z (%): 36.0 (100) [HCl], 126.1 (24)
[C3N6H6]+, 235.1 (0.1) [C6N11H9]+; elemental analysis calcd (%) for Zn-
[C6N11H9]Cl2 (371.5): C19.39, N 41.57, H 2.42, Cl 19.09, Zn 17.60; found:
C18.82, N 40.52, H 2.50, Cl 19.98, Zn 18.50.
1
General methods: H and 13CNMR spectra were recorded as [D 6]DMSO
solutions on a JEOL Eclipse EX-400 instrument, the chemical shifts
being referenced with respect to TMS.
Mass spectra were obtained by using a Jeol MStation JMS-700 gas inlet
system. The substance was dissolved in a 3-nitrobenzyl alcohol matrix
(FAB+) on a target and ionized by bombardment with accelerated
(6 kV) Xe atoms. Direct insertion DEI+ (70 eV) mass spectra were ob-
tained by using a Jeol MStation JMS-700 gas inlet system. Electrospray
Ionisation (ESI) measurements were conducted on a Finnigan MAT 95Q
and a Finnigan MAT 90 mass spectrometer using an API-Interface II
with an ESI probe head and a capillary voltage of 2.5 kV.
Melam: A porcelain crucible of about 2 cm height and 1 cm diameter
was loaded with 2–10 g melamine, loosely covered with a lid and placed
in a muffle furnace. After heating at 613–633 K for approximately two
days, tiny colorless crystals had formed on the surface of the grayish bulk
material. Melam used for the solid-state NMR investigations and thermal
analysis was obtained from melam hydrate “C6N11H9·2H2O”, which was
placed in a muffle furnace and dehydrated at temperatures between 423
and 572 K for 1–2 days. The product obtained was identified as melam by
IR spectroscopy and X-ray powder diffraction (low crystallinity).
1H NMR (270 MHz, [D6]DMSO, TMS): d=9.2 (s, 1H; NH), 6.6 (s, 8H;
NH2) ppm; 13C{1H} NMR (270 MHz, [D6]DMSO, TMS): d=167.6,
164.6 ppm; IR (KBr): n˜ =3483.8 (vw), 3456.5 (vw), 3300.2 (w), 3165.4
(w), 1687 (vw), 1639.5 (m), 1610.0 (m), 1583.8 (s), 1545.8 (s), 1513.1 (s),
1450.8 (s), 1414.7 (vs), 1338.4 (vs), 1249.8 (s), 1174.9 (w), 1069.9 (vw),
1021.1 (vw), 972.2 (vw), 806.2 (vs), 782.1 (w), 748.0 (vw), 682.6 (m), 632.2
(s) cmÀ1; MS (2.5 kV, ESI+): m/z (%): 236.11 (69) [M+H]+; elemental
analysis calcd (%) for C6N11H9 (235.2): C30.63, N 65.53, H 3.83; found:
C30.67, N 64.25, H 3.60. For further analytical data see text.
Elemental analyses were performed by using a commercial C, H, N ele-
mental analyzer system Vario EL (Elementar Analysensysteme GmbH).
X-ray diffraction: Single crystals of melam (1) and the zinc chloride–
melam complex (3) were measured on an Enraf–Nonius Kappa CCD dif-
fractometer equipped with a rotating anode, and X-ray diffraction data
of melamium perchlorate hydrate (2) were recorded on a STOE IPDS
diffractometer using graphite-monochromated MoKa radiation (l=
71.073 pm). The crystal structures were solved by direct methods using
the program SHELXS-97[54] and refined on F2 by applying the full-matrix
least-squares method implemented in SHELXL-97.[55] For the ZnCl2–
melam complex a multi-scan absorption correction was carried out that is
routinely included in the software packages DENZO and SCALE-
PACK.[56] The positions of all hydrogen atoms were determined from dif-
ference Fourier syntheses and refined without constraints; all non-hydro-
gen atoms were refined anisotropically. Details of the crystal structure
determinations and refinements are summarized in Table 2.[51]
Melam hydrate: Melam hydrate was obtained from a melam–NH4Cl
adduct of approximate composition [C6N11H10]Cl·0.5NH4Cl. The adduct
was synthesized by heating melamine (364.4 mg, 2.89”10À3 mol, ꢂ99%,
Fluka) and NH4Cl (91.9 mg, 1.72”10À3 mol, ꢂ99.5%, Fluka) at 723 K for
12 h (heating rate 1 KminÀ1) in a glass ampoule with ground neck (length
200 mm, 1ext 25 mm, 1int 22 mm).[48] The obtained melam–NH4Cl adduct
(550 mg, 1.84”10À3 mol) was stirred in aqueous NH3 (60 mL, 25% (p.
a.), Merck) for 1 h to remove the ammonium chloride and subsequently
air-dried. According to elemental analysis and vibrational spectroscopy,
products with the approximate composition “C6N11H9 ”2H2O” were ob-
tained, though slightly varying water content was observed in different
samples. IR (KBr): n˜ =3463.0 (m), 3336.9 (m), 3184.0 (m), 1623.0 (s, b),
1548.8 (s), 1525.4 (s), 1427.9 (s), 1350.8 (s), 1262.4 (m), 1183.6 (w), 1087.1
(vw), 1031.9 (w), 810.1 (m), 790.1 (vw), 740.7 (vw), 623.1 (vw), 614.5 (w),
591.9 (vw), 507.5 (vw) cmÀ1; MS (2.5 kV, ESI+): m/z (%): 236.11 (67)
High-temperature in situ X-ray diffraction was carried out on a STOE
Stadi P powder diffractometer (Ge(111)-monochromated Mo-Ka1 radia-
G
tion, l=70.093 pm) with an integrated furnace using unsealed quartz ca-
pillaries (1 0.5 mm) as sample containers. Data were collected in a 2q
range of approximately 3–178 with an average scan collection time of 30
minutes. The samples were heated from 298 K to temperatures around
923 K in steps of 10 KminÀ1, using a heating rate of 20 KminÀ1 between
the scans.
Solid-state NMR spectroscopy: 15N and 13CCP-MAS solid-state NMR
spectra were recorded by using natural abundance samples at ambient
temperature on a conventional impulse spectrometer DSX 500 Avance
(Bruker) operating at 500 MHz. The samples were contained in 4-mm
ZrO2 rotors, which were mounted in a standard double-resonance MAS
probe head (Bruker). The 15N signals were referenced with respect to ni-
tromethane, the 13Csignals to TMS. A ramped cross-polarization se-
quence was employed to excite the 15N/13Cnuclei via the proton bath;
the power of the 1H radiation was linearly varied about 50%. The data
[(MÀ2H2O)
+
H]+; elemental analysis calcd (%) for C6N11H9·2H2O
(271.2): C26.57, N 56.83, H 4.80; found: C26.36, N 55.44, H 4.85.
Melamium perchlorate hydrate: Melam hydrate (200 mg, 7.38”10À3 mol)
was suspended in water (110 mL) and heated under reflux. HClO4 (3 mL,
60% (p. a.), Merck) was added dropwise until a clear solution was ob-
tained. The solution was refluxed for another five minutes, then cooled
Chem. Eur. J. 2007, 13, 4956 – 4968
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
4967

W. Schnick and B. V. Lotsch
collection of all experiments was performed by applying broadband
proton decoupling using a TPPM sequence.[57] CPPI experiments (cross-
polarization combined with polarization inversion) were performed to
probe the number of hydrogen atoms directly attached to the nitrogen
nuclei.[22,49] Typically, the contact times tc for both CP and CPPI experi-
ments were 2–10 ms, while the inversion time ti in the CPPI experiments
was set to 100–160 ms such that the polarization of NH groups was close
to zero.[22] The recycle delays were estimated from 1H T1 measurements.
Sample spinning frequencies of 10 kHz and 5 or 5.5 kHz for CP and
CPPI experiments, respectively, were employed.
[20] B. V. Lotsch, W. Schnick, Chem. Mater. 2005, 17, 3976.
[21] B. V. Lotsch, W. Schnick, Chem. Mater. 2006, 18, 1891.
[22] B. Jürgens, E. Irran, J. Senker, P. Kroll, H. Müller, W. Schnick, J.
Am. Chem. Soc. 2003, 125, 10288.
[23] A. Sattler, W. Schnick, Z. Anorg. Allg. Chem. 2006, 632, 238.
[24] E. C. Franklin, J. Am. Chem. Soc. 1922, 44, 486.
[25] C. E. Redemann, . H. J. Lucas, J. Am. Chem. Soc. 1940, 62, 842.
[26] H. May, J. Appl. Chem. 1959, 9, 340.
[27] A. I. Finkelꢁshtein, Russ. J. Gen. Chem. 1961, 31, 1046.
[28] N. K. Gavrilova, V. A. Gal’perin, A. I. Finkelꢁshtein, A. G. Koryakin,
Zh. Org. Khim. 1977, 13, 669.
[29] L. Costa, G. Camino, J. Therm. Anal. Calorim. 1988, 34, 423.
[30] C. Voelckel, Ann. Chim. Phys. 1844, 62, 90.
[31] W. L. Burdick, J. Am. Chem. Soc. 1925, 47, 1485.
[32] E. W. Hughes, J. Am. Chem. Soc. 1941, 63, 1737.
[33] P. Klason, J. Prakt. Chem. 1886, 33, 285.
[34] B. Bann, S. A. Miller, Chem. Rev. 1958, 58, 131.
[35] G. B. Seifer, Russ. J. Coord. Chem. 2002, 28, 301.
[36] G. van der Plaats, H. Soons, R. Snellings, Proc. Eur. Symp. Therm.
Anal. 2nd 1981, 215.
IR spectroscopy: FTIR measurements were carried out on a Bruker IFS
66v/S spectrometer. Spectra of the samples were recorded as KBr pellets
(1 mg sample, 500 mg KBr, hand press with press capacity 10 kN) in an
evacuated cell equipped with a DLATGS detector at ambient conditions
between 400 and 4000 cmÀ1
. IR spectra of the neat solids (650–
4000 cmÀ1) were recorded in reflection geometry using a Perkin Elmer
Spectrum BXII FT-IR spectrometer equipped with a DuraSamplIR II di-
amond ATR device (Smith Detection).
Thermal analysis: Thermoanalytical measurements between room tem-
perature and 873 K were conducted on a Mettler DSC25 with a heating
rate of 1 KminÀ1. The cold-welded, only moderately pressure-resistant
aluminum crucible used as sample container was placed in the calorime-
ter under an atmosphere of dry nitrogen.
[37] A. I. Finkelꢁshtein, Opt. Spectra 1959, 6, 33.
[38] M. Takimoto, Kogyo Kagaku Zasshi 1961, 64, 1452.
[39] M. Takimoto, Kogyo Kagaku Zasshi 1964, 85, 168.
[40] N. V. Spiridonova, A. I. Finkelꢁshtein, Khim. Geterotsikl. Soedin.
1966, 1, 126.
[41] W. J. Schnabel, R. Rätz, E. Kober, J. Org. Chem. 1962, 27, 2514.
[42] S. Tragl, K. Gibson, H.-J. Meyer, Z. Anorg. Allg. Chem. 2004, 630,
2373.
[43] T. Tojima, R. Saito (Mitsubishi Engineering-Plastic Corp.), JP-
2004217687 A2, 2004 [Chem. Abstr. 2004, 141, 157944].
[44] E. Schlosser, B. Nass, W. Wanzke, EP-1386942 A1, 2004 [Chem.
Abstr. 2004, 140, 147040].
[45] P. Flippo, J. Tijssen, C. K. Sham, A. E. H. De Keijzer, R. J. M. Hul-
skotte, US-2001008913 A1, 2001 [Chem. Abstr. 2004, 135, 93449].
[46] V. A. Gal’perin, A. I. Finkelꢁshtein, N. K. Gavrilova, Zh. Org. Khim.
1971, 7, 2431.
Acknowledgements
We thank W. Wünschheim for conducting the DSCmeasurements and
Dr. P. Mayer, Dr. O. Oeckler, and T. Miller for single-crystal X-ray data
collection. The authors gratefully acknowledge financial support that was
granted from the Deutsche Forschungsgemeinschaft (DFG) (project
SCHN 377/12–1), Fonds der Chemischen Industrie, the BMBF and the
Studienstiftung des Deutschen Volkes (scholarships for B. V. Lotsch).
[47] L. Costa, G. Camino, G. Martinasso, Polym. Prepr. Am. Chem. Soc.
Div. Polym. Chem. 1989, 30, 531.
[48] B. Jürgens, Ph. D. thesis, University of Munich (Germany), Shaker
Verlag: Aachen, 2004.
[49] C. Gervais, F. Babonneau, J. Maquet, C. Bonhomme, D. Massiot, E.
Framery, M. Vaultier, Magn. Reson. Chem. 1998, 36, 407.
[50] The slight low-field shift observed for the 1H and 13Cmelamine sig-
nals is presumably due to differences in the instrument calibration
[1] J. von Liebig, Ann. Pharm. 1834, 10, 1.
[2] E. Kroke, M. Schwarz, Coord. Chem. Rev. 2004, 248, 493.
[3] S. Matsumoto, E. Q. Xie, F. Izumi, Diamond Relat. Mater. 1999, 8,
1175.
[4] S. Muhl, J. M. Mendez, Diamond Relat. Mater. 1999, 8, 1809.
[5] C.-M. Sung, M. Sung, Mater. Chem. Phys. 1996, 43, 1.
[6] M. L. Cohen, Phys. Rev. B 1985, 32, 7988.
or concentration effects, since
a 2:1 mixture of melamine and
[7] M. L. Cohen, Science 1993, 261, 307.
melem yields an essentially identical spectrum to that of phase I
(see Supporting Information).
[8] A. Y. Liu, M. L. Cohen, Science 1989, 245, 841.
[9] A. Y. Liu, M. L. Cohen, Phys. Rev. B 1990, 41, 10727.
[10] C. Niu, Y. Z. Lu, C. M. Lieber, Science 1993, 261, 334.
[11] D. M. Teter, R. J. Hemley, Science 1996, 271, 53.
[12] E. Kroke, M. Schwarz, E. Horath-Bordon, P. Kroll, B. Noll, A. D.
Norman, New J. Chem. 2002, 26, 508.
[13] I. Alves, G. Demazeau, B. Tanguy, F. Weill, Solid State Commun.
1999, 109, 697.
[14] D. R. Miller, D. C. Swenson, E. G. Gillan, J. Am. Chem. Soc. 2004,
126, 5372.
[15] D. T. Vodak, K. Kim, L. Iordanidis, P. G. Rasmussen, A. J. Matzger,
O. M. Yaghi, Chem. Eur. J. 2003, 9, 4197.
[16] T. Komatsu, J. Mater. Chem. 2001, 11, 799.
[17] T. Komatsu, T. Nakamura, J. Mater. Chem. 2001, 11, 474.
[18] J. Kouvetakis, A. Bandari, M. Todd, B. Wilkens, Chem. Mater. 1994,
6, 811.
[51] CCDC-618430 (1), CCDC-618432 (2), and CCDC-618431 (3) con-
tain the supplementary crystallographic data for this paper. These
data can be obtained free of charge from the Cambridge Crystallo-
graphic Data Centre via www.ccdc.cam.ac.uk/data request/cif.
[52] A. Sattler, Diploma Thesis, University of Munich (Germany), 2005.
[53] B. V. Lotsch, W. Schnick, Z. Anorg. Allg. Chem. 2006, 632, 1457.
[54] G. M. Sheldrick, SHELXS97, Program for the Solution of Crystal
Structures, University of Gçttingen, Gçttingen (Germany), 1997.
[55] G. M. Sheldrick, SHELXL97, Program for the Refinement of Crys-
tal Structures, University of Gçttingen, Gçttingen (Germany), 1997.
[56] Z. Otwinowsky, W. Minor, Methods Enzymol. 1997, 276, 307.
[57] A. E. Bennett, C. Rienstra, M. Auger, K. V. Lakshmi, R. G. Griffin,
J. Chem. Phys. 1995, 103, 6951.
Received: September 7, 2006
Revised: February 12, 2007
Published online: April 5, 2007
[19] Z. H. Zhang, K. Leinenweber, M. Bauer, L. A. J. Garvie, P. F. Mc-
Millan, G. H. Wolf, J. Am. Chem. Soc. 2001, 123, 7788.
4968
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 4956 – 4968