Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journal homepage: www.elsevier.com/locate/jpba

Characterization of polymorphic ampicillin forms

C. Baraldi a , A. Tinti b , S. Ottani c , M.C. Gamberini a,∗
a
Department of Life Sciences, University of Modena and Reggio Emilia, via Campi n.183, 41125 Modena, Italy
b
Department of Biomedical and Neuromotor Sciences, University of Bologna, via Belmeloro 8/2, 40126 Bologna, Italy
c
Istituto per la Sintesi Organica e la Fotoreattività, ISOF-CNR, via P. Gobetti 101, 40129 Bologna, Italy

a r t i c l e i n f o a b s t r a c t

Article history: In this work polymorphs of ␣-aminobenzylpenicillin (ampicillin), a ␤-lactamic antibiotic, were prepared
Received 19 May 2014 and investigated by several experimental and theoretical methods. Amorphous monohydrate and three
Received in revised form 12 August 2014 crystalline forms, the trihydrate, the crystal form I and the crystal form II, were investigated by FT-IR
Accepted 15 August 2014
and micro-Raman. Also data obtained by differential scanning calorimetry (DSC), thermogravimetric
Available online 26 August 2014
analysis (TGA), X-ray powder diffraction (XRPD) and hot-stage Raman spectroscopy are reported. Finally,
quantum mechanical calculations were performed by density functional theory (DFT) to assist the assign-
Keywords:
ment of spectroscopic experimental bands. For the first time, the ampicillin molecule in its zwitterionic
Ampicillin
FTIR/ATR spectroscopy
form was studied at the B3LYP/aug-cc-pVDZ level and the corresponding theoretical vibrational spectra
Raman microscopy were computed. In fact, ampicillin in the crystal is in zwitterionic form and concentrations of this same
DFT calculations form are quite relevant in solutions at physiological pH. Experimental and theoretical results allowed
X-ray powder diffraction identification of specific features for polymorph characterization. Bands typical of the different poly-
Thermal analysis morphs are identified both in IR and Raman spectra: in particular in the NH stretching region (IR), in
the amide I + ␦NH region (both techniques), in the 1520–1490 cm−1 region (IR), in the 1320–1300 cm−1
and 1280–1220 cm−1 (IR), in the 1200–1170 cm−1 (Raman), in the amide V region (IR), and, finally,
in the 715–640 cm−1 and 220–200 cm−1 (Raman). Interconversion among different polymorphs was
investigated by hot-stage Raman spectroscopy and thermal analysis, clarifying the complex pattern of
transformations undergone as a function of temperature and heating rate. In particular, DSC scans show
how the trihydrate crystals transform into anhydrous forms on heating. Finally, stability tests demon-
strated, after a two years period, that no transformation or degradation of the polymorphs occurred.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction solid solutions. Thus, the requirement that in pharmaceutical for-

Several active pharmaceutical ingredients (APIs) crystallize in
different forms (polymorphs) or with solvent molecules as inte-
gral part of their structures (pseudopolymorphs). Pharmaceutical
formulations of these compounds may prove challenging, since
differences in the solid state phase of the active ingredient may dra-
matically alter the final effects of the drug, with marked influence
on its bioavailability. Moreover, differences in crystal packing play a
major role on drug stability, especially as temperature and humid-
ity effects are concerned. Usually, crystalline forms are strongly
preferred for their stability and reproducibility; higher degree of
purification can be achieved, as compared to amorphous solids and
∗ Corresponding author. Tel.: +39 59 205 5157; fax: +39 59 205 5131.
E-mail addresses: mariacristina.gamberini@unimore.it, gamber@unimore.it
(M.C. Gamberini).
mulations of APIs only a single crystalline form be present, rises a
set of problems related to clinical protocols and legal and regula-
tory issues. The unpredictability (i.e. lack of obviousness) of crystal
structures and physical properties may rise legal issues and chal-
lenges in terms of obtaining and maintaining patent protection for
an API. Among the substances affected by this kind of problems,
an important case is provided by ampicillin, a molecule which has
obtained new attention in antibiotic therapeutic practices.
Ampicillin or ␣-aminobenzylpenicillin or 6-[d(-)␣-
aminophenylacetamido] penicillanic acid (Fig. 1) is a penicillin
class antibiotic with betalactamic structure. Literature reports
that ampicillin may exist in two anhydrous polymorphic forms (␥
form also named B or I; ␦ form also named II) and hydrate forms
(trihydrate; monohydrate or form A) [1–3]. The trihydrate and
anhydrous form I have been the most studied forms, while the
monohydrate and anhydrous form II did not receive analogous
attention. In particular, Shefter et al. [1] refer the existence of form

http://dx.doi.org/10.1016/j.jpba.2014.08.021
0731-7085/© 2014 Elsevier B.V. All rights reserved.
330 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
Fig. 1. (a) Anhydrous ampicillin structure and (b) ampicillin zwitterionic form.
II and a crystallization method is reported [4], but IR characteri-
zation is inadequate. In the different commercial pharmaceutical
formulations like injectable or oral preparations, capsules or
tablets, the API is present as the anhydrous form or as sodium
salt or as the trihydrate form. Alburn et al. [5] report that the
anhydrous form I has the advantage of high storage stability
characteristics, related to its negligible water content. Moreover,
anhydrous ampicillin exhibits slower absorption in the gut and
prolonged blood levels with more effective action, relatable to its
lower solubility in water.
As far as ampicillin is concerned, a major problem is the poor
characterization of its polymorphs and the confusion in naming
such polymorphs. In fact several notation systems (letters – A, B,
C, Greek letters ␥, ␦ or Roman numerals – I, II, III, etc.) are used in
parallel, sometimes in inconsistent ways. Thus, it may happen that
the form defined as “I” by an author is not necessarily named in the
same way by another one. As a result, correlations between crystal
forms and experimental data reported in different studies, such as
vibrational spectra, may prove difficult and unreliable. The char-
acterization methods reported in the literature include FT-IR [6],
X-ray powder diffraction (XRPD) [1], single crystal x-ray diffraction
(SCXRD) [2], differential scanning calorimetry/thermogravimetric
analysis (DSC/TGA), diffuse reflectance infrared Fourier-transform
spectroscopy (DRIFT), THz time-domain spectroscopy (THz-TDS)
and NMR spectroscopy [7]. Table 1 summarizes the published stud-
ies concerning ampicillin forms and their nomenclature.
The study of ampicillin polymorphs, of their stability and
interconversion must face most of the above mentioned difficul-
ties in relating structural features to pharmaceutical formulations
and physical properties. SCXRD investigations report that, in the
crystals of the anhydrous form I and of the trihydrate, ampi-
cillin molecules are in zwitterionic form (Fig. 1b). In the form
I different molecules are held together by intermolecular H-
bonds [19]. In the trihydrate the crystallization water molecules
participate into a complex network of H-bonds connecting sev-
eral zwitterions [19]. In contrast, the crystal structure of the
anhydrous form II has not yet been reported and the exis-
tence of zwitterions in the solid state for this form is not
confirmed.
Literature methods report preparation of the monohydrate sol-
vate by heating in vacuo from room temperature up to 95 ◦ C, and
preparation of the anhydrous ␥ form by heating to 80–100 ◦ C [6].
In contrast, the less studied ␦ form has been crystallized from a
boiling xylene solution [4]. However, such methods lead to partial
decomposition of samples [1], as indicated also by the light yellow
color of the final specimens. Thus, in the present work, different
procedures have been used for the preparation, as reported in the
subsequent section.
The various polymorphic and solvate forms were crystallized
and characterized by different techniques. The molecular vibrations
of the different ampicillin forms were investigated by ATR/FT-
IR and micro-Raman. Quantum-mechanical calculations allowed
assignment of vibrational modes. The hot stage Raman microscopy
(HSRM) was able to follow the transition from trihydrate to amor-
phous monohydrate ampicillin. The same technique allowed, by
heating specimens in a sealed capillary, to control solid-solid con-
version from trihydrate to anhydrous forms. The results reported
in the present work allow a better identification and characteri-
zation of the different polymorphs and of their interconversion,
with a special focus on transformations undergone by the dif-
ferent forms and their range of stability. From this point of
view, they can also be useful in addressing the above reported
questions related to clinical protocols and legal and regulatory
issues.
2. Materials and methods
2.1. Compounds
The basic ampicillin was a commercial sample from
Sigma–Aldrich, Milano, Italy, corresponding to the trihydrate
form and was used in sample preparation as reported in the
following.
2.1.1. Preparation of polymorphs
Form I (). Commercial trihydrate ampicillin (0.5 g) was sus-
pended in water (0.8 ml) and heated to 90 ◦ C. In few minutes
crystals of the anhydrous form were precipitated [6]. The mix-
ture was filtered and placed in a dryer. The same anhydrous form
was also prepared by placing the trihydrate ampicillin in a sealed
capillary subjected to a slow heating (5–10 ◦ C/min) from 25 ◦ C to
150 ◦ C.
Form II (ı). Commercial trihydrate ampicillin (0.6 g) was dis-
solved in boiling xylene (6 ml, at 136 ◦ C) and then filtered under
vacuum. The precipitate was washed with ethyl acetate and diethyl
ether, dried in a oven at 50 ◦ C for 15 and then stored in a vac-
uum dryer at room temperature [4]. The final product was lightly
yellow. Alternatively, the same form was obtained by fast heating
(60 ◦ C/min) to 150 ◦ C in a sealed capillary. In this case, white crystals
were obtained.
Monohydrate. Commercial trihydrate ampicillin (0.6 g) was
heated in a oven from 30 to 95 ◦ C in 12 min and kept at this tem-
perature for 2 h [1,7]. The product obtained was lightly yellow.
Trihydrate. The trihydrate form was purified by recrystalliza-
tion from saturated aqueous solutions of the commercial powder
heated below 50 ◦ C [2]. The resulting precipitate was filtered by
sintered glass and the crystals were dried in vacuum at room
temperature.
 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340 331

Table 1
Existing studies and notation systems concerning ampicillin forms.

References and analytical Form I Form II Trihydrate Monohydrate

method

Ref. [1]: XRPD X X X X

“form 1” “form 2” “amorphous anhydrous
state”
Ref. [2]: single crystal X X
“anhydrous”
Ref. [3]: IR, XRPD X X
“anhydrous”
Ref. [4]: IR (nujol), XRPD X X X
“anhydrous, B, “new or ␦”
␥”
Ref. [5]: IR X X X
“anhydrous”
Ref. [6]: IR X X X
“other form”
Ref. [7]: DSC, TGA, XRPD, FTIR, X “anhydrous” X
NMR
Ref. [8]: DSC X
Ref. [9]: IR X X
“anhydrous”
Ref. [10]: IR X X
“B” “A”
Ref. [11]: XRPD, SEM X X
“amorphous state”
Ref. [12]: XRPD, FTIR, stability X
test
Ref. [13]: XRPD under heating X X X
in open/sealed holder, DSC “anhydrous” “new
under pressure anhydrous”
Ref. [14]: FTIR X X
“form II” “form I”
Ref. [15]: XRPD X X
“anhydrous”
Ref. [16]: XRPD, DSC, TGA X “anhydrous” X
“amorphous form”
Ref. [17]: THz–TDS X X
“anhydrous”
Ref. [18]: DRIFT X X
“anhydrous”
Ref. [19]: SCXRD X X
“anhydrous”

2.2. Methods Theoretical Raman and IR spectra were obtained by frequency

2.2.1. X-ray powder diffraction (XRPD)
Powder diffractograms were recorded by the X’PERT PRO celer-
ator Panalytical diffractometer (Panalytical, Almelo, Netherlands)
with the following experimental settings: Ni filtered Cu K␣ radi-
ation ( = 1.5418 Å); voltage 40 kV; tube current 40 mA; angular
speed 0.040◦ (2) min−1 ; divergence and anti-scatter slits 0.5◦ ;
beam mask 15 mm; angular range 4◦ <2 < 30◦ . The application soft-
ware version was X’Pert Date collector.
2.2.2. Computational methods (DFT calculations)
Quantum mechanical calculations were performed on the ampi-
cillin molecule by the Gaussian09 program [20] in the framework
of the density functional theory (DFT). The B3LYP functional was
employed, which includes the Becke’s three parameters (local,
nonlocal, Hartree–Fock) hybrid exchange functional [21] with the
Lee–Yang–Parr correlation functional [22]. Molecular orbitals were
obtained by a mixed basis set: the correlation-consistent polarized
augmented basis set, aug-cc-pVDZ [23], was used for all atoms,
whereas for sulphur the aug-cc-pV(D+d)Z was selected. Geome-
try optimization was carried on in redundant internal coordinates.
According to the implementation in Gaussian09, the convergence
criterion was met when maximum and root-mean square val-
ues of forces and next-step displacements were below predefined
thresholds.
calculations on the optimized geometries. Frequencies were com-
puted in the limit of the harmonic approximation, using the same
basis sets and method as in the geometry optimization steps. All
computed frequencies were positive, confirming that optimized
geometries correspond to minima on the Potential energy surface.
2.2.3. IR with attenuated total reflection (FT-IR/ATR)
The spectra were recorded by a VERTEX 70 (Bruker Optics,
Ettlingen, Germany) FT-IR spectrophotometer, equipped with a
deuterium triglycine sulphate (DTGS) detector. Setting parame-
ters: resolution 4 cm−1 ; apodization weak. The spectral range was
4000–600 cm−1 with 32 scans for each spectrum. The ATR spectra
were recorded using the Golden-Gate accessory.
2.2.4. Raman microscopy and hot stage Raman microscopy
(HSRM)
The Raman spectra were recorded with a confocal Labram
instrument (Horiba Jobin Yvon, Longjumeau Cedex, France)
equipped with a He–Ne laser at 632.8 nm and a CCD detector
(254 × 1024), cooled by the Peltier effect. The spectra were recorded
in backscattering after focalization in several positions within a
small area (ca.100 ␮m × 100 ␮m) of the sample. The maximum
laser power employed was 5 mW and the recording time for good
signal-to-noise ratio was 100–200 s. GRAMS/AI 7.02 was used for
the elaboration of the spectra.
332 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340

The hot-stage Raman analysis was carried out by direct coupling

of the Raman microscope with a hot stage.

2.2.5. Differential scanning calorimetry (DSC) and

thermogravimetric analysis (TGA)
The thermal behavior of samples was studied with a SETARAM-
LABSYS TG-DTA/DSC instrument (Setaram, Caluire, France) using
dry nitrogen as purge gas over the temperature range 35–220 ◦ C at
a scanning rate of 10 ◦ C/min. The instrument was calibrated with an
indium sample (99.99% purity), having a melting point of 156.6 ◦ C
and a melting enthalpy of 28.45 J/g. The SETSOFT 2000 software
was used to extrapolate onset value temperatures for each thermal
event.

3. Results and discussion

3.1. Characterization of polymorphic and solvate forms

3.1.1. X-ray powder diffraction (XRPD)

Results of X-ray diffraction investigations have been previously
reported for single crystals of the anhydrous form I [19] and of the
trihydrate form [19]. Atomic coordinates support the conclusion
that in both polymorphs molecules are in zwitterionic form and
the NH3 + and COO− groups participate to an infinite network
of H-bonds. However, it must be noted that the positions of H-
atoms were not refined, but rather obtained by Fourier difference
maps. Oscillations around these positions (<.01 nm), as induced
by thermal treatments, may change the zwitterionic character of
molecules in the crystals. Moreover, a single crystal structure of
form II has never been reported, therefore a definite experimen-
tal evidence of the existence of a zwitterion in the solid state for
this form has not been assessed. In subsequent sections data from
experimental and theoretical spectra are reported, providing some
useful insight on this question.
Another point of interest is the chain of structural changes
induced by thermal treatments on the trihydrate form. Actually,
it has been reported [19] that differences of the anhydrous and tri-
hydrate structures are too important to allow direct conversion in
the solid state between these polymorphs. However, depending on
experimental conditions, a different process of fusion and partial
recrystallization might be possible.
As a first step toward this type of investigations, in order to con-
firm the polymorphs identity, X-ray powder diffraction patterns
have been recorded (Fig. 4). Form I and trihydrate XRPD patterns
are in agreement with data published by Ivashkiv and James [2,3].
Polymorph I displays a X-ray pattern with characteristic peaks at
2 = 8.10 vs, 16.4 vs, 20.3m, 21.7s, 22.9s, 25.4m, while the trihy-
drate form shows peaks at 2 = 7.4m, 12.3 vs, 15.1m, 18.1 vs, 19.5
vs, 22.2s, 23.4s, 23.7s, 25.7m, 26.9 vs, 29.2s, 29.5m (00-029-1546
[2]). In contrast, the XRPD pattern of ampicillin form II shows char-
acteristic strong diffraction peaks at: 2 = 6.9 vs, 13.4m, 15.7 vs, 16.5
vs, 17.0s, 18.6s, 19.9m, 26.8mw.
The monohydrate form presents the diffraction spectrum char-
acteristic of an amorphous substance (Fig. 2).
3.1.2. Computational results
Vibrational spectra of the crystalline forms were approximated
by DFT quantum mechanical calculations on the isolated molecule
in vacuum. Atomic coordinates for the initial stage of geometry
optimization were taken from the Cambridge Structural Database
(CSD) [24]. Preliminary results showed the presence of small force
constants and low frequency vibrational modes. Thus, the tight
convergence criteria on an ultra-fine DFT integration grid was
used in the geometry optimization to ensure reliable results in the
subsequent frequency computation step. In fact, the zwitterionic
character of the anhydrous form I, as obtained from the CSD, was
Fig. 2. X-ray diffraction patterns of ampicillin forms: I, II, trihydrate and monohy-
drate forms (from the top).
retained throughout the whole geometry optimization. The final
optimized geometry of the isolated molecule is close, but not iden-
tical to the coordinates in the crystal structure. Probably, the most
significant difference is a slight increase of the curvature along the
molecular axis of the optimized geometry.
Comparisons between theoretical and experimental spectra
show differences attributable to several factors. First, experiments
are performed on the solid phase, while calculations refer to the
isolated molecule in vacuum. Thus, computed frequencies do not
account for effects stemming from intermolecular interactions in
the crystals. Additional differences may arise also from anharmonic
effects and from experimental peak broadening, that cannot be
satisfactorily reproduced in the computed spectra.
Anharmonic effects can be reduced by a scaling procedure [25].
Linear fitting of the theoretical frequencies to the experimental
ones was performed by the SPESCA program [26]. Scaled frequen-
cies were computed as scaled = b × calc + a. For the crystal form
I, the corresponding equations were: scaled = 0.9502 × calc + 12.46
for the IR spectrum and scaled = 0.9493 × calc + 10.72 for the Raman
spectrum. As expected for the used functional, values of b close
to 0.95 were obtained. Interpretation of the theoretical frequency
spectra was performed by the potential energy distribution (PED)
analysis of the fundamental vibrations modes. The procedure was
carried on by the program VEDA4 [26], allowing the identification
of the stretching, bending and torsion local modes for each com-
puted line. In the additional materials PED results are summarized
in the corresponding Tables, one for each frequency calculation.
Additional computations were performed, aimed to simulate
effects of intermolecular interactions on the vibrational spectra
due to crystal packing. The initial geometry was a molecular dimer
taken from the CSD structure of form I. In this model the two
molecules are bound together by two H-bonds, each one between
the NH3 + of a molecule and the COO− group of the other one.
However, for this larger system, a different basis set, 6-311++G(d,p)
 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
was used, allowing to keep computation time within reason-
able limits. Optimized geometries obtained by these computations
showed that the NH3 + group looses the hydrogen involved in the
H-bond toward the COO− group of the other molecule. Thus, the
H-bonds are retained but the zwitterionic character is lost. The cor-
responding IR spectrum shows two distinct features at 2870 and
3135 cm−1 (not scaled values) related to the O H stretching of the
atoms involved in the two intermolecular H-bonds. These features
are not detected in the IR spectra of the polymorphs. This result
is consistent with the presence of zwitterions in form I and in the
trihydrate, but it is of special relevance in the case of the anhydrous
form II, suggesting that zwitterions may be present also in its not
yet determined crystalline structure.
3.1.3. Infrared and Raman spectra
Fig. 3 shows the experimental FT-IR/ATR spectra of different
ampicillin forms recorded at room temperature. A detailed assign-
ment of the experimental bands of ampicillin is reported in Table 2,
according to the results of DFT frequency calculation, as usually
reported in literature [27].
However, in the range 4000–2600 cm−1 some significant dis-
crepancies are found. In this range, O H and N H stretching are
expected. The sharp and intense peak at 3332 cm−1 (form I) was
attributable to the N H stretching. For polymorph II, the pres-
ence of three sharp and characteristic peaks at 3419, 3334 and
3287 cm−1 can be observed. These bands are always related to
asymmetric and symmetric N H stretching, which for the pri-
mary amine group falls in the range 3400–3300 cm−1 and for the
secondary amide group in trans conformation falls between 3330
and 3100 cm−1 . The disappearance of the two bands at 3419 and
3287 cm−1 in the FT-IR spectrum of form I may be attributed to a
different H-bond scheme in the two forms.
In this spectral range hydrated forms exhibit the OH bond
stretching vibrations of water molecules. The trihydrate form spec-
trum shows characteristic peaks at 3501 and 3440 cm−1 related to
the crystallization water that establishes hydrogen bonds involving
N and O (amide), O (betalactamic ring), COO− , NH3 + [2]. More-
over, crystal structures of the anhydrous and trihydrate forms show
that ampicillin molecules are in zwitterionic form [19] thanks to the
pKa = 2.5 ( COOH) and pKa = 7.3 ( NH2 ) of the acid and basic groups
of the molecule. Thus, the NH3 + group interacts with the COO−
group of a second drug molecule. There is no evidence of a NH3 +
intramolecular H-bonding and the presence of a broad shoulder
at about 2873 cm−1 , corresponding to the NH3 + stretching, con-
firms the zwitterionic state. Furthermore, the trihydrate spectrum
shows two strong bands at 1604 and 1572 cm−1 related to ionized
COO− and NH3 + groups deformation, whose bands overlap in this
spectral region. In the other forms these bands are weaker, even if
observable as medium intensity bands at 1641 and 1581 cm−1 in
the spectrum of form I.
The large band at about 2900 cm−1 , together with those at
2742 and 2652 cm−1 (bands A, B and C) indicate the presence of a
strong asymmetric H-bond [28] in the trihydrate form. Also in the
monohydrate, the large bands at about 3300, 2900 and 2650 cm−1
indicate the presence of weaker H-bonds. Bands attributable to
H-bonds are less evident in the spectra of anhydrous forms. Nev-
ertheless, the IR spectrum of form I shows the presence of two
large and not very intense bands with maxima at about 2900 and
2600 cm−1 , while form II spectrum displays maxima at about 3280
and 2950 cm−1 , indicating for both polymorphs that the H-bonds
are weaker than those of the trihydrate form. Moreover, the respec-
tive H-bonds are weaker in form II than in form I. The absence
of the A, B and C bands typical of strong H-bonds, the presence
in the IR and Raman spectra of the as and s COO− bands (see
Table II), as well as the small difference in the melting temperature
333
(see Section 3.1.4.1), strongly suggests for polymorph II, as for form
I, the presence of the zwitterionic form.
The monohydrate form exhibits a general bands enlargement,
typical of a compound with a low degree of order (amorphous) and
also the OH (water) stretching bands at about 3501 and 3300 cm−1
are broader, thereby suggesting that water was adsorbed instead
of fixed in crystallization sites and involved in hydrogen-bonds of
variable geometry.
In the spectral region between 1800 and 1600 cm−1 the vibra-
tional modes due to the C O groups stretching occur with high
intensity. For the trihydrate form, the C O stretching band at
1769 cm−1 attributable to the carbonilic group of the lactamic ring
is strong. For the anhydrous forms it drops in intensity and shifts
slightly below at 1765 (form I) and 1763 cm−1 (form II); the mono-
hydrate form shows a medium and broader band at 1774 cm−1 .
The secondary amide C O stretching, associated with the N H
deformation mode (amide I), is at lower wavenumbers showing
intense peaks at 1692, 1705 with a shoulder at about 1666, 1686 and
1663 cm−1 for the four structures, respectively. Such differences, in
particular the red shift of this band for the two solvate forms, can be
related to the increasing involvement of the amide group in hydro-
gen bonds. In fact, associated linear amides show a C O stretching
band lying in the 1680–1630 cm−1 region, while non-bonded
amides absorptions are at 1700–1650 cm−1 [29]. Similarly, hydro-
gen bonded NH stretch is usually reported near 3300 cm−1 , whereas
non-bonded NH stretch falls at 3490–3400 cm−1 . Considering the
C O stretching at higher wavenumbers, it can be hypothesized,
relatively to form II, that H-bond involving it become weaker, as
underlined above.
In the region 1650–1500 cm−1 also the as COO− and ıNH3 +
bands of the zwitterionic form can be observed (form I). The cor-
responding s COO− can be observed both in the IR and Raman
spectra, with variable intensity, in the 1300 cm−1 region. The
medium-strong bands near 1120 cm−1 are assigned to the COO
group deformation.
In this region the solvate forms show the OH water deformation
band in the 1640–1620 cm−1 range. This is clearly observed in the
trihydrate form at 1621 cm−1 , disappearing in case of deuteration
(not shown here).
One of the most characteristic bands related to amide group is
the very strong “amide II” vibration near 1500 cm−1 that involves
both C N stretching and N H in-plane bending (amide II), while
the stretch open mode mixed with NH deformation (amide III)
absorbs more weakly in the IR spectrum in the 1330–1250 cm−1
region. Other important medium-to-strong modes concerning the
C N stretching and the N H deformation are reported close to
1180–1190 cm−1 .
A band associated to C C stretching and C H deformation of the
phenyl ring falls at 1581 cm−1 for the I form. The phenyl ring vibra-
tions exhibit significant bands for the different forms. Form I and
the trihydrate show intense bands due to C C stretching and C H
deformation of the phenyl ring falling at 1581, 1455, 1378 cm−1
and 1572, 1457, 1373 cm−1 , respectively; instead, form II and the
monohydrate form show a shoulder near 1550 cm−1 . Both exhibit a
medium intensity band at 1458 cm−1 . Moreover, other C H defor-
mations together with in-plane deformations of the ring give strong
peaks at 1305 and 1234 cm−1 (form I), 1336, 1318 and 1236 cm−1
(form II), 1334, 1307 and 1218 cm−1 (trihydrate) and 1307 and
1223 cm−1 (monohydrate). The ring breathing bands are situated
at about 640 cm−1 for all forms. The more significant phenyl ring
out-of-plane deformations fall near 870 cm−1 , 800 cm−1 and at 751,
760–745, 763–736, 758 cm−1 for the titled forms.
Fig. 4 shows the micro-Raman spectra of the four ampicillin
forms in the spectral range between 3500 and 100 cm−1 . As for
the FT-IR spectra, in the range 3500–2600 cm−1 the weak N H
stretching bands of the two anhydrous forms are comparatively
 334
Table 2
Experimentally observed FT-IR and Raman wavenumbers (cm−1 ) and their attribution for ampicillin forms.

IR anhydrous form I Raman IR anhydrous Raman IR trihydrate Raman IR Raman Assignment

ahydrous form form II anhydrous trihydrate monohydrate monohydrate
I form II

3501m 3501w,br O H (H2 O)

3440s O H (H2 O) intramolecularly bonded OH (dimeric)a
3419m 3423vvw N H
3332s 3335vvw 3334m 3370sh 3301s,br N H
O H (IR) intermolecularly bonded OH (polymeric)a
3170vw 3287s 3288vvw N H (amide)
3061w 3069w-3058w 3088w 3064w CH (aromatic)
[3000–2800] [3000–2800] [3000–2800] [3000–2800] CH (aliphatic)
2873w NH3 +

C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
1765vs 1766vvw 1763vs 1768m 1769vs 1777w 1774s 1770w,br C O (lactamic ring)
1692vs 1693m 1705s 1711m 1686vs 1691ms 1663vs,br 1693w,br amide I + ıNH
1641m 1646w 1654w,sh as COO− + amide I
1621m,sh 1624sh ıOH (crystal H2 O)a
1581m 1604m 1558sh 1606m 1604s 1607m 1606sh 1604m CC(ph) + ıCH(ph) + ıNH3 +
1584w 1588m 1572s 1595w 1586mw
1520vs,sh 1497vs 1515vs 1492vs 1515vs ıCH(ph) + amide II + ıNH3 + + as COO−
1496sh 1500vs
1455m 1457w 1458m 1453w 1457m 1467w 1458m 1458w,br CC(ph) + ıCH(ph) + ıCH3 /CH2
1437w
1390s 1396m 1397m 1401w 1383m 1386vw 1387m 1396vw,br ıCH3 + s COO−
1378vs 1362s 1368w,sh 1375w 1373s 1375vvw 1369m ıCH2 +
ıCH3 + CC(ph) + ıCH(ph) + ıop CH(CH3 ) + ıHCC(CH3 )
1324vvw 1336m 1337w 1334m 1330w,sh ıip CH (ph) + ıHCC(CH3 ) + CN (amide, lactam)
1305vs 1316mw 1318m 1310vw,sh 1307s 1312m 1307m,br s COO− + CC(ph) + ıip CH(ph) + CN (amide,
1299vw lactam) + ıHCC(CH3 ) + ıHNC(amine) + ıNH3 +
1278w,sh 1254w 1258w,sh 1252m 1257w 1261s 1267ms 1260vw,sh 1247mw,vbr ıHCN(lactam) + CN(lactam) + ıip CH(ph) + ıHNC(amide,
1250sh 1252w lactam) + ıNH3 +
1234w 1235m 1236m 1239m 1223m,br CC + ıHNC(NH3 + ) + ıip CH(ph) + ıHCN(lactam)
1214m 1219 w 1213w,sh 1210w 1218m 1220w ıHCS + CN(amine) + ˇHNH(amine) + ˇHCS
1181m 1186ms 1190w 1193ms 1194w 1198ms 1191sh 1188m,br ıHNC(lactam, amide) + CN(amide) + ıHCC + ıNH3 +
1179w 1171m 1178vw
1157vw- 1146vw 1157mw 1156w 1158ms 1154m 1166w 1157w 1158m,br ıip HCC(ph) + ıCH + CC
1152sh
1123m 1122vw 1126m 1127w 1116m 1122vvw 1126m 1125vw,br ıCH(CH3 ) + CC(CH3 ) + CN(lactam,
amide) + ıHCN(lactam)
1072m 1097m 1076w 1097w 1077m 1074w ıHCC(CH3 ) + CN(amine + lactam) + HCCS + CC(ph) + ıNH3 +
1049vw 1056w CN (amine) + ıNH3 +
1026m 1029ms 1026m 1031ms 1019m 1039s 1026m 1030m ıHNC (amine) + CC(ph) + CN(amide + lactam) + ıNH3 +
1023sh 1025vw
1004vw 992vw 1005vs 1006w 1005vvs 1000vw 1007vs 1005vw 1003vs ıip HCC(ph) + ıop CH(ph) + CC(ph) + ıHCCH(CH3 ) + ıNH3 +
996mw 975w 991vw 990w 993w
979vw
961vw 949vw 953w 954vw 960w 963vw 948w,br ıCNC(amide,
941vvw 957vw lactam) + CC(lactam) + ıop CH(ph) + ıHCCH(CH3 )
926w 931vvw 924w 928w 931w 931vvw 920vw NC(amine) + CC(CH3 , lactam) + ıCH3 + ıop CH(ph)
913vw 919vvw
889w 891m 887vw 891vw CC(CH3 ) + NC(amine) + ıop CH(ph) + ıCH3
873w 876mw 872w 876w 875w 882m 867vw 870vw CC(aliph) + NC(amine) + SC + ıop CH(ph) + COO−
847w 851mw 847vw 847w 852m 850vw 851w CC(aliph) + ıCCS + COO−
833s 833ms 843m 837m CC(aliph) + ıp OCOC + ıCCC (lactam chain) + ıop CH(ph)
820vvw
802m 804mw 804vw 805w 813w 808vw ıCNC(amine + amide) + ıop CH(ph) + ıip ph + NH3 +
778m 783ms 776mw 787w 786m 782w,br ıNCO(lactam) + CNCO(lactam) + ıip ph
Table 2 (Continued)

IR anhydrous form I Raman IR anhydrous Raman IR trihydrate Raman IR Raman Assignment

ahydrous form form II anhydrous trihydrate monohydrate monohydrate
I form II

751m 754vw 760s 762s 763w,br 758m HNCC lactam + ıop NH(“amide

C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
V”) + ıNCO + CNCO(lactam) + ıop ph
745m 749w 736m 741m HNCC (lat amide) + ıop NH (“amide V”) + ıop ph
728s 729w, sh 711mw 720m 727m ıCOO− + CC(aliph) + ıop ph
692s 697m 696vs 696s 694ms 697vs ıCCO + lactam + skeleton vibration
669 m 673w 672vw 666vw,br 660w,br 660w,br ıOCO− + Lactamic ring def.
663w
639s 640w 647w 651m 644m 652w ıip ph + HNCC(lat,
amide) + CC(aliph) + CNCO(lactam) + ıCOO− + SC(form
II)
616m 609w 618ms 623m 619m CC(aliph) + CNCO(lactam) + ıip ph + ıCNC (lactam,
603ms 595vw 609w,sh amide) + CS + ıCOO− (form I)
590s 585s 580ms 578 m,br ıip ph + CC(aliph) + ıCNC(lactam)
523m 535w 526ms 532 m,br ıop ph + CC +
ıCNCO(amide + lactam) + SC + ıCCS (ı lactam)
499vw 511m 513m SC + ıop ph + ıCNC (C5-N4-C3)
483vw- 481vw 499vw 484vvw ıHCC + ıCCN(lactam) + CNCO + ıip lactam
466vw SC + “skeletal deformation”
424w 424vw 447m ıCCS
389w 400w 389vw “C skeletal deformation”.
361m 375m 366w 361vw ıCCS
353sh 356w ıCCC +HCNH + ıNH3 + + HCCC(CH3 )
317w 322w 333 vw 336vw ˇCCC(CH3 ) + ıNH3 + ıCCN + HCCC(CH3 ) + ıop ph
320 vw
291vw 297w 296m CH3 + CNCO + ıCCS
271 vw 263 w
248m 228vw CH3 + ıCCC
228w
212m 202m 209vw ıCCN (amine) + HCCC + ˇCCC
175w 185w 173m 174vw ıCCC + ıCCS
160ms 168w 144w ıCCN (amine) + HCCC + ıCCC
148w 149m
123vw 116m-105s 124vw Lattice vibrations
111m
a
Ref. [30].

335
336 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
Fig. 3. FTIR/ATR spectra of ampicillin forms: form I, form II, trihydrate, monohydrate (from the top).
different from the hydrated samples. Also in this case, the spec-
trum of form II shows two bands at 3423 and 3288 cm−1 (s N H
and N H of the amide group) while form I displays only one band
at 3335 cm−1 . These bands well correspond to those observed in the
IR spectra. The presence of these differences in the N H stretching
region, together with the X-ray diffraction and calorimetric results,
suggests a looser packing of crystal form II, with a possible dis-
tortion of the H-bond system, corresponding to these vibrational
features. Moreover, as regards hydrated forms, the bands due to
O H stretching vibrations are stronger in IR than in Raman spec-
tra. When carboxylic group forms hydrogen bonding, the result
is a broad band centered at 3100–2900 cm−1 , overlapped by the
much stronger C H stretching bands. This broad band is clearly
visible both in the IR and Raman spectra of the monohydrate
form.
Characteristic features are the C H stretching bands involv-
ing the benzene ring carbons, falling for the four forms at 3061,
3069–3058, 3088 and 3064 cm−1 , respectively.
Two C O stretching vibrations are expected, one in-phase
(symmetric stretching vibration) is Raman active and the other,
out-of-phase (antisymmetric stretching vibration), is IR active.
Medium-to-weak Raman bands are observed, that can be corre-
lated with these polar groups. The C O stretching band (lactamic
ring) for the various forms is located around 1770 cm−1 with a slight
shift to 1777 cm−1 for the trihydrate form, due to the carboxyl group
involvement in an intermolecular bonding.
The carbonyl stretching of the amide group shows a band at
about 1693 cm−1 for the anhydrous form I and the two hydrated
forms, while the peak is blue-shifted to 1711 cm−1 for the anhy-
drous form II, as noted in the infrared spectrum.
The NH3 + deformation modes are found as medium intensity
Raman bands observed at about 1605 cm−1 (all forms) and at 1584,
1588, 1595 and 1586 cm−1 for I, II, trihydrate and monohydrate
forms, respectively. Moreover, in the Raman spectra weak bands at
353 and 317 cm−1 for I form (similar values for the II form) can also
be assigned to the NH3 + deformation mode, while for solvates this
feature is observed at 333–320 and 336 cm−1 .
Unlike in the IR spectra, the amide II band is not normally
observed in Raman.
The total symmetric vibration (breathing) of the benzene ring
appears as a medium band at 640 and 651 cm−1 for the anhydrous
forms I and II, respectively, while for the trihydrate form it falls at
652 cm−1 and for the monohydrate at 660 cm−1 . Other significant
bands associated to the benzene ring are those at 1584, 1316, 1157
and 1005 cm−1 , referred to C C stretching vibrations and to the
C H bending deformation for form I. The other forms show slightly
different values at 1588, 1299, 1158, 1005 cm−1 (form II), 1595,
1312, 1166, 1007 (trihydrate) and 1586, 1158, 1003 (monohydrate).
In the lower wavenumber region of the Raman spectra, the C S
stretching vibration is observed between 650 and 610 cm−1 and
around 500 cm−1 . The variable wavenumbers and relative intensity
of these bands indicate different conformations of the pentatomic
ring, in particular as regards I and II forms. Furthermore, the C C S
bending and umbrella deformation modes appear as characteris-
tic bands in the Raman spectra. The first mode corresponds to a
medium band at 447 cm−1 for the trihydrate form, whereas the
other forms display very weak bands at lower values. Instead,
the umbrella mode shows medium bands at 361, 375, 366 and
361 cm−1 for the four forms, respectively.
In the range of the lattice vibrations, the bands are characteristic
and different for the two anhydrous forms. As regards the solvated
forms, the trihydrate spectrum displays strong and well resolved
bands, whereas the monohydrate shows broader bands with van-
ishing lattice vibrations. This behavior, as well as the broadening of
the main IR bands, suggests that water is not present as hydrated
molecules, but it is rather adsorbed.
In Table 2, for both IR and Raman spectra, the distinctive spec-
tral features for the various forms are printed in bold. Both IR and
Raman spectra display some peaks distinctive of the different poly-
morphs and are useful to differentiate them.
 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340 337

Fig. 4. Micro-Raman spectra of ampicillin forms: form I, form II, trihydrate, monohydrate (from the top).

3.1.4. Thermal analysis presence of strongly bonded crystallization water, in agreement

3.1.4.1. Differential scanning calorimetry (DSC) and thermogravimet-
ric analysis (TGA). Fig. 5 reports DSC and TGA profiles of ampicillin
forms. A temperature program from 30◦ to 225 ◦ C with heating rate
of 10 ◦ C/min was used. For all forms TGA data show a weight loss
onset just above 200 ◦ C.
The trihydrate form exhibits a large endothermic step at 131 ◦ C
(522.9 J/g) due to the dehydration and vaporization process. The
TGA confirms a significant weight loss before 150 ◦ C (close to 14%,
theoretical weight loss 13,4% [13]), a temperature indicative of the
with IR data on the presence of a strong H-bond. This process is
immediately followed by an exothermic peak, which, according to
the scheme reported in Fig. 6B, should be related to partial crys-
tallization of the anhydrous form I. Actually, after melting of the
trihydrate crystals, experimental conditions, such as slow heating
rate and presence of residual water molecules from the trihydrate,
favor the formation of anhydrous form I crystals. The formation of
amorphous monohydrate cannot be excluded too, depending on
the effect of the purge gas flow on the conditions in the DSC pan
338 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340

Fig. 5. DSC and TGA profiles (10 ◦ C/min) of ampicillin forms.

Fig. 6. Crystal form transformations upon heat treatment: (A) Raman spectra of the trihydrate form at room temperature and during heating at 60 ◦ C, 75 ◦ C, 90 ◦ C and 100 ◦ C
(from the top). (B) Conversion of the trihydrate form to monohydrate and anhydrous forms I (slow heating) and II (fast heating).
 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
(see Fig. 6B). Unfortunately, uncertainties in the identification of
the baseline make quite difficult to analyze the DSC profile above
190 ◦ C. However, a comparison with the traces of the other poly-
morphs shows several similarities confirming the presence of the
anhydrous form I and possibly of the monohydrate in this tem-
perature range. A final endothermic peak is detected at 216 ◦ C
(44.2 J/g), in agreement with published data [7], caused by melting
with decomposition.
The thermograms of the monohydrate amorphous form, per-
formed with the same temperature program, show a large
endothermic peak corresponding to a water loss at 68 ◦ C (55.4 J/g),
indicative of weakly bound water molecules. The dehydration was
also confirmed by TGA data: weight loss of about 8% between 30
and 150 ◦ C, theoretical weight loss 4.5%. The discrepancy between
these values is attributable to adsorption water not stoichiometri-
cally bonded. Even anhydrous forms show 2.5% and 3.3% weight
loss, as reported by Grant et al. [10]. A second very broad and
weak endothermic peak is detected at 190 ◦ C (28.3 J/g). Finally,
the ampicillin monohydrate melts with decomposition at 209 ◦ C
(19.9 J/g).
The DSC profile of the anhydrous form I shows a broad and weak
exothermic peak near 206 ◦ C (18.5 J/g) [7], followed by endother-
mic melting at 218 ◦ C (72.7 J/g). The form II thermogram, not
previously reported in the literature, exhibits a sharp exothermic
event at about 200 ◦ C (1.9 J/g), immediately followed by melting
at 212 ◦ C (23.1 J/g). For both these forms the DSC patterns above
180 ◦ C display trends consistent with a loss of residual absorbed
water and a possible melting of less perfect crystals followed
by a crystallization process of residual amorphous material (the
exothermic peak) immediately before the final melting. However,
as reported in a previous section and in Fig. 6B, formation of form
II by fast heating is likely to produce less stable crystals as com-
pared to form I. Thus, the exothermic peak at 200 ◦ C might as
well correspond to a solid state conversion of form II into form
I.
As regards thermal analysis, the melting temperatures obtained
by DSC are: 216 ◦ C, 209 ◦ C, 218 ◦ C and 212 ◦ C for trihydrate, mono-
hydrate, forms I and II, respectively, with different enthalphy of
fusion (44.2 J/g, 19.9 J/g, 72.7 J/g, 23.1 J/g). However, the above men-
tioned difficulties in the unambiguous identification of the baseline
introduce important uncertainties in the measurement of these
variables. Thus, discrimination of the different polymorphs only
by thermal parameters is questionable and should be supported by
results from other techniques.
3.1.4.2. Hot-stage Raman spectroscopy (HSRM). Hot-stage mea-
surements were carried out with the same DSC temperature
program (10 ◦ C/min; 32–225 ◦ C temperature range). A specific
characterization of DSC events is possible by this technique.
On heating from 35 ◦ C to 90 ◦ C, the micro-Raman spectra of
the trihydrate form is completely preserved, whereas, above
90 ◦ C, the conversion of trihydrate into monohydrate form
is complete (Fig. 6A) and followed by melting. Thus, HSRM
shows that, even if spectral changes are detected, heating
below 90 ◦ C does not produce modifications in the crystal
structures.
In contrast, when specimens are slowly heated (5–10 ◦ C/min)
up to 150 ◦ C in sealed capillaries, anhydrous form I is produced
(Fig. 6B). On fast heating in sealed capillary (60 ◦ C/min until 150 ◦ C),
white crystals of form II are obtained (Fig. 6B). It is possible to
observe these events through the capillary, by focusing beyond the
glass wall.
Raman spectra show that, as temperature is increased, a notice-
able decrease in intensity is observed for bands at 1777 (C O of
the ␤-lactamic ring), 1267, and 1120 cm−1 (already decreasing at
60 ◦ C) related to the carboxylic group. Also the bands at 882 and
339
447 cm−1 , related to deformation of betalactamic ring, decrease.
The sharp band at 173 cm−1 due to the five member ring breath-
ing displays a sudden intensity decrease as the temperature is
increased. These results may well correspond to an increasing dis-
order of the crystal lattice above 70 ◦ C, especially in the structure of
the sensitive ␤-lactams. Such instability was already hypothesized
by Austin et al. [6].
3.2. Stability tests with time
In order to test the stability of ampicillin with time, the XRPD,
Raman and FT-IR spectra of raw polymorphs and solvates in powder
form were recorded with a time interval of 2 years. The spectra of
tablets containing 25% (w/w) ampicillin were also recorded shortly
after their manufacture and after 2 years. In both cases, and in order
to simulate a real life scenario, in the interim period the powders
were placed in a laboratory box for protection against the light
but neither the temperature nor the moisture were controlled. No
difference was observed between the two sets of data, thereby indi-
cating that no transformation or degradation of the active form after
a period of 2 years had occurred.
4. Conclusions
For the first time, the four ampicillin polymorphs have been
characterized by a large set of different techniques: FT-IR, Raman,
XRPD, hot stage Raman microscopy and thermal analysis have
been employed with the aim to investigate problems related to
the characterization and the stability of the crystal forms of these
polymorphs. In this respect particular attention has been paid to
variables affecting interconversion of polymorphs as induced by
different processing methods. Pharmaceutical formulations of APIs
require that only a single crystalline form be present. Thus, reliable
methods to characterize polymorphs are of great importance espe-
cially when clinical protocols and legal and regulatory issues are
concerned.
All the above mentioned techniques are relatively able to iden-
tify the different polymorphs: IR and Raman by some characteristic
bands, as well as XRPD by the assignment of specific diffrac-
tion peaks to different crystalline forms or by their absence as in
the amorphous. Thermal analysis can discriminate polymorphs by
their melting temperatures and enthalphy of fusion. Moreover, this
study has confirmed the efficiency of vibrational spectroscopies,
in particular Raman techniques, to distinguish, for an API, differ-
ent polymorphs and solvate forms and to monitor the transitions
among them as temperature is changed. For the first time, the
Raman spectra of the four solid ampicillin forms are reported and
the bands typical of the different polymorphs are identified both in
IR and Raman spectra. Also, for the first time, DFT quantum mechan-
ical calculations have been performed for the zwitterionic form
of ampicillin at the B3LYP/aug-cc-pVDZ level. Theoretical IR and
Raman spectra have been obtained and compared to the experi-
mental ones by the aid of the potential energy distribution (PED)
analysis. Several data reported in this work emphasize the impor-
tance of the zwitterionic form for ampicillin. Actually, solvated
zwitterionic ampicillin at physiological pH is present in significant
concentrations, increasing as the pH is decreased. In the crystals
of the trihydrate and anhydrous form I, ampicillin zwitterions are
stabilized by a complex network of H-bonds involving the NH3 +
group. In contrast, the theoretical DFT IR spectrum of an isolated
dimer shows that, in this case, ampicillin molecules are not in zwit-
terionic forms. Thus, the presence of zwitterions in the solid state
must be related to their stabilization by effect of crystal packing.
A comparison of vibrational spectra of the crystal forms I and II
supports this conclusion, suggesting that ampicillin zwitterions are
340 C. Baraldi et al. / Journal of Pharmaceutical and Biomedical Analysis 100 (2014) 329–340
present also in form II, whose structure, however, has not been
solved yet.
From the collected data, the monohydrate form appears to be
amorphous, though it has the appearance of crystals (50 × 100 ␮m
approximately); experimental results show that water is not
stoichiometrically bonded, but rather adsorbed in the solid. By
interpolating the analytical data here reported, it can be demon-
strated that the forms cited as monohydrate in certain articles and
as amorphous in other articles, coincide. As far as form I is con-
cerned, it should be noted that, generally, the literature cites an
“anhydrous form” that, by comparison with our data, almost always
matches the form I.
Finally, the stability of the polymorphs with time, after a two
years period, has been demonstrated.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jpba.2014.08.021.
References
[1] E. Shefter, H. Fung, O. Mok, Dehydration of crystalline theophylline monohy-
drate and ampicillin trihydrate, J. Pharm. Sci. 62 (1973) 791–794.
[2] M.N.G. James, D. Hall, Crystalline modifications of ampicillin I: the trihydrate,
Nature 220 (1968) 168–170.
[3] E. Ivashkiv, Ampicillin, in: K. Florey (Ed.), Analytical Profiles of Drug Substances,
vol. 2, Academic Press, New York, 1973, pp. 1–61.
[4] A.C. Parker, S.E. Staniforth, US Patent, 3,933,796 (1976).
[5] H.E. Alburn, W. Chester, N.H. Grant, process for the preparation of 6-(alpha-
aminoacylamino)penicillanic acids, US Patent Office 3,299,046 (1967).
[6] K.W.B. Austin, A.C. Marshall, H. Smith, Crystalline modifications of ampicillin,
Nature 208 (1965) 999–1000.
[7] H.G. Brittain, D.E. Bugay, S.J. Bogdanowich, J. DeVincentis, Spectral methods for
determination of water, Drug. Dev. Ind. Pharm. 14 (1988) 2029–2046.
[8] Y. Takahashi, K. Nakashima, H. Nakagawa, I. Sugimoto, Effects of grinding and
drying on the solid-state stability of ampicillin trihydrate, Chem. Pharm. Bull.
32 (1984) 4963–4970.
[9] N.H. Grant, H.E. Alburn, Peptide hydrates: influence of the hydration state on
the properties of ␣-aminobenzylpenicillin, Nature 207 (1965) 645–646.
[10] N.H. Grant, H.E. Alburn, US Patent Office 3,144,445 (1964).
[11] K. Moribe, A. Wongmekiat, Y. Hyakutake, Y. Tozuka, T. Oguchi, K. Yamamoto,
Influence of dehydration temperature on water vapor adsorption, dissolution
behavior and surface property of ampicillin, Int. J. Pharm. 288 (2005) 245–252.
[12] S. Nojavan, A. Ghassempour, Y. Bashour, M.K. Darbandi, S.H. Ahmadi, Deter-
mination of residual solvents and investigation of their effect on ampicillin
trihydrate crystal structure, J. Pharm. Biomed. Anal. 36 (2005) 983–988.
[13] J. Han, S. Gupte, R. Suryanarayanan, Applications of pressure differential
scanning calorimetry in the study of pharmaceutical hydrates. II. Ampicillin
trihydrate, Int. J. Pharm. 170 (1998) 63–72.
[14] H.H. Silvestri, D.A. Johnson, US Patent 3,941,773 (1976).
[15] N.V. Phadnis, R.K. Cavatur, R. Suryanarayanan, Identification of drugs in phar-
maceutical dosage forms by X-ray powder diffractometry, J. Pharm. Biomed.
Anal. 15 (1997) 929–943.
[16] H. Zhu, D.J.W. Grant, Influence of water activity in organic solvent + water mix-
tures on the nature of the crystallizing drug phase. 2. Ampicillin, Int. J. Pharm.
139 (1996) 33–43.
[17] H. Liu, Y. Chen, X. Zhang, Characterization of anhydrous and hydrate pharma-
ceutical materials with THz time-domain spectroscopy, J. Pharm. Sci. 96 (2007)
927–934.
[18] T. Kojima, Y. Yamauchi, S. Onoue, Y. Tsuda, Evaluation of hydrate formation of
a pharmaceutical solid by using diffuse reflectance infrared Fourier-transform
spectroscopy, J. Pharm. Biomed. Anal. 46 (2008) 788–791.
[19] M.O. Boles, R.J. Girven, The structures of ampicillin: a comparison of the anhy-
drate and trihydrate forms, Acta Crystallogr. B 32 (1976) 2279–2284.
[20] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,
X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M.
Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Nor-
mand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N.
Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo,
R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W.
Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador,
J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman, J.V. Ortiz, J.
Cioslowski, D.J. Fox, Gaussian 09, Revision A.1, Gaussian, Inc., Wallingford CT,
2009.
[21] A.D. Becke, Density-functional thermochemistry. III. The role of exact exchange,
J. Chem. Phys. 98 (1993) 5648–5652.
[22] C.T. Lee, W.T. Yang, R.G. Parr, Development of the Colle–Salvetti correlation-
energy formula into a functional of the electron density, Phys. Rev. B 37 (1988)
785–789.
[23] T.H. Dunning Jr., Gaussian basis sets for use in correlated molecular calcula-
tions. I. The atoms boron through neon and hydrogen, J. Chem. Phys. 90 (1989)
1007–1023.
[24] F.H. Allen, The Cambridge structural database: a quarter of a million crystal
structures and rising, Acta Crystallogr. B 58 (2002) 380–388.
[25] J.P. Merrick, D. Moran, L. Radom, An evaluation of harmonic vibrational fre-
quency scale factors, J. Phys. Chem. A 111 (2007) 11683–11700.
[26] M.H. Jamróz, J.C. Dobrowolski, R. Brzozowski, Vibrational modes of 2,6-,
2,7-,and 2,3-diisopropylnaphthalene. A DFT study, J. Mol. Struct. 787 (2006)
172–183.
[27] R. Mishra, A. Srivastava, A. Sharma, P. Tandon, C. Baraldi, M.C. Gamberini,
Structural electronic, thermodynamical and charge transfer properties of chlor-
amphenicol palmitate using vibrational spectroscopy and DFT calculations,
Spectrochim. Acta: A 101 (2013) 335–342.
[28] D. Hadzi, S. Bratos, in: P. Schuster, G. Zundel, C. Sandorfy (Eds.), The Hydrogen
Bond, Structure and Spectroscopy, vol. 2, North Holland, Amsterdam, 1976, pp.
565–611.
[29] N.B. Daimay Lin-Vien, W.G. Colthup, J. Fatelet, G. Grasselli, The Handbook of
Infrared and Raman Characteristic Frequencies of Organic Molecules, Academic
Press, Inc./Harcourt Brace Jovanovich Publishers, Boston/San Diego/New York,
1991.
[30] G. Socrates, Infrared Characteristic Group Frequency. Tables and Charts, 2nd
ed., Wiley, Chichester, 1997.