On the Demystification of “HPbI3” and the Peculiarities of the Non-innocent Solvents H2O and DMF

Article abstract of DOI:10.1002/zaac.201800267

Detailed investigations by XRD reveal that the precursor “HPbI₃” that was obtained by reaction of aq. conc. hydroiodic acid HI and PbI₂ in DMF is (CH₃)₂NH₂PbI₃. (CH₃)₂NH₂⁺ is formed by solvent reaction as already described in the literature but not properly assigned. Attempts to synthesize HPbI₃ by gas phase reaction of PbI₂ with aq. conc. HI yielded light-yellow crystals of the oxonium salt H₁₈O₈Pb₃I₈ (Pbam, Z = 2, a = 10.075, b = 30.162, c = 4.5664 ?). The crystal structure of H₁₈O₈Pb₃I₈ consists of trimeric ribbons of edge-sharing PbI₆ octahedra. These ribbons [Pb₃I₈]^2– are separated by protonated fragments of crystalline ice [H₁₈O₈]²⁺ or (H₂O)₆(H₃O⁺)₂. H₁₈O₈Pb₃I₈ can also be precipitated from PbI₂ and aq. conc. HI. H₁₈O₈Pb₃I₈ is not stable at room temperature but transforms within a few days to light-yellow (H₃O)_2x(H₂O)_2–2xPb_1–xI₂ with x ≈ 0.23 (R3m, Z = 3, a = 4.5554, c = 29.524 ?). The crystal structure represents a CdCl₂-type layer structure with H₂O/H₃O⁺ in between. Charge compensation is achieved by Pb²⁺ vacancies. Via topotactic reaction (H₃O)_2x(H₂O)_2–2xPb_1–xI₂ releases H₂O/HI and forms crystals of the pristine PbI₂. All steps were characterized by P-XRD, IR/Raman spectra, and UV/Vis spectra. H₁₈O₈Pb₃I₈ acts as a precursor for the synthesis of MAPbI₃ because the reaction with gaseous CH₃NH₂ yields MAPbI₃, so it can mimic a composition “HPbI₃”.

Full text of DOI:10.1002/zaac.201800267

Journal of Inorganic and General Chemistry
DOI: 10.1002/zaac.201800267
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
On the Demystification of “HPbI3” and the Peculiarities of the
Non-innocent Solvents H2O and DMF
Michael Daub[a,b] and Harald Hillebrecht*[a,b]
Abstract. Detailed investigations by XRD reveal that the precursor stable at room temperature but transforms within a few days to light-
¯
“HPbI3” that was obtained by reaction of aq. conc. hydroiodic acid HI yellow (H3O)2x(H2O)2–2xPb1–xI2 with x ≈ 0.23 (R3m, Z = 3, a = 4.5554,
and PbI2 in DMF is (CH3)2NH2PbI3. (CH3)2NH2+ is formed by solvent c = 29.524 Å). The crystal structure represents a CdCl2-type layer
reaction as already described in the literature but not properly assigned.
Attempts to synthesize HPbI3 by gas phase reaction of PbI2 with aq.
structure with H2O/H3O+ in between. Charge compensation is achieved
by Pb2+ vacancies. Via topotactic reaction (H3O)2x(H2O)2–2xPb1–xI2 re-
conc. HI yielded light-yellow crystals of the oxonium salt H18O8Pb3I8 leases H2O/HI and forms crystals of the pristine PbI2. All steps were
(Pbam, Z = 2, a = 10.075, b = 30.162, c = 4.5664 Å). The crystal
structure of H18O8Pb3I8 consists of trimeric ribbons of edge-sharing
PbI6 octahedra. These ribbons [Pb3I8]2– are separated by protonated
characterized by P-XRD, IR/Raman spectra, and UV/Vis spectra.
H18O8Pb3I8 acts as a precursor for the synthesis of MAPbI3 because
the reaction with gaseous CH3NH2 yields MAPbI3, so it can mimic a
fragments of crystalline ice [H18O8]2+ or (H2O)6(H3O+)2. H18O8Pb3I8 composition “HPbI3”.
can also be precipitated from PbI2 and aq. conc. HI. H18O8Pb3I8 is not
powder X-ray diffraction (P-XRD), and X-ray photoelectron
Introduction
spectroscopy (XPS). From the observed weight loss (in the
range of 240 to 345 °C) of 23.7% they conclude the reaction
of HPbI3 into PbI2 and HI (theoretical weight loss: 21.7%) and
a PbI2:HI ratio of 1:1.1. The XPS measurements also showed
an PbI2:HI ratio of around 1:1. Furthermore, they measured a
P-XRD pattern. From comparison of some individual reflec-
tions they concluded a layer structure similar to PbI2 with an
increased distance between the layers because of the intercal-
ation of H+. According to the ratio Pb:I of 1:3 and the similar-
ity to KPbI3 and CsPbI3 they assumed a layer structure with
PbI6 octahedra connected by common edges and corners
(NH4CdCl3 type).[6] In the course of the reaction the color
changed from dark yellow (PbI2) to light yellow (“HPbI3”).
The material named as “HPbI3” showed an increased solubility
in DMF compared to PbI2 and an improvement of the film
quality and solar cell efficiency of 16% by converting the pre-
cursor to FAPbI3 (FA = formamidinium).
For a couple of years, hybrid organic-inorganic lead halide
perovskites APbX3, MAPbI3 (MA
= methylammonium,
CH3NH3) in particular, are of great interest. The application of
MAPbI3 as an excellent light harvester in solar cells[1] has in-
duced a strong boost on the research of such materials. In re-
cent years, the power conversion efficiency (PCE) of such per-
ovskite solar cells (PSCs) was enhanced to over 20%[2]
through optimization on the cell architecture and the composi-
tion of the absorber material. One great advantage of PSCs is
the easy processing that is done close to room temperature.
Solubility and reactivity of the components enable the use of
simple organic solvents (DMF, DMSO, …) and gas phase reac-
tions (HI, CH3NH3I, …) close to ambient conditions.
This intensive research activity has also led to the use of
different precursor materials, solvents, and reaction conditions
to form MAPbI3.[3] In 2015 Wang et al. claimed “HPbI3” as a
“new precursor compound for highly efficient solution pro-
cessed perovskite solar cells”.[4] Subsequently, several contin-
uing publications referred to “HPbI3” as a precursor and gave
more or less detailed information on the nature of “HPbI3”.[5]
Wang et al. made experiments with PbI2 and HI in DMF and
characterized the remaining solid by thermogravimetry (TG),
With direct reference to the work of Wang et al. the group
of Pang et al. studied the transformation of the “HPbI3” pre-
cursor with methylamine gas (CH3NH2) into MAPbI3.[7] They
characterized the “HPbI3” compound also by P-XRD and
claimed that “Pawley fitting shows an orthorhombic (space
group P63mc) structure…” In contrast to Wang et al. they as-
sumed from the P-XRD and high-resolution transmission elec-
tron microscopy (HR-TEM) data a structural model that is
based on the hexagonal perovskite that consist of 1D chains of
face sharing PbI6 octahedra with protons in between. As an
additional proof they showed that “HPbI3” could be transferred
into MAPbI3 by the reaction with methylamine gas, whereas
KPbI3 could not be transferred.
* Prof. Dr. H. Hillebrecht
E-Mail: harald.hillebrecht@ac.uni-freiburg.de
[a] Institut für Anorganische und Analytische Chemie
Albert-Ludwigs-Universität
Albertstraße 21
79104 Freiburg, Germany
[b] Freiburger Materialforschungszentrum FMF
Albert-Ludwigs-Universität,
Stefan-Meier-Straße 21
79104 Freiburg, Germany
Long et al.[8] obtained “HPbI3” by mixing PbI2 and excess
HI (probably conc. aq. HI) in DMF. They postulated a “solvate
HPbI3·xDMF” with x Յ 1. For both samples they described
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/zaac.201800267 or from the au-
thor.
Z. Anorg. Allg. Chem. 0000, ᭿,0–0
1
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
needle-shaped light-yellow single crystals. Single crystal in- eters for the hexagonal dimethylammonium (DMA) compound
vestigations revealed a hexagonal unit cell with very similar (CH3)2NH2PbI3 [a = 8.769(1), c = 8.188(2) Å].[13] The inten-
lattice parameters for HPbI3 (a = 8.7339, c = 8.1779 Å) and sities were similar to the published P-XRD and the calculated
the solvate HPbI3·xDMF (a = 8.7517, c = 8.1802 Å). The re- P-XRD pattern of DMAPbI3 fits perfect the published P-XRD
finement in space group P63mc (no. 186) revealed the posi- pattern of Wang et al.[4] and Pang et al.[7]
tions of Pb and I. The cations were not localized because of
Furthermore, it is a well-known phenomenon that DMF de-
the weak scattering power (H+) and severe orientation disorder composes into formic acid and DMA, especially in acids.[14]
(DMF). The similar volume of the unit cell of the two com- Eppel et al. already showed the synthesis of DMAPbI3 from
pounds was explained by the high mobility of H+. The subse- PbI2/HI/DMF.[15] They also showed that DMAPbI3 crystallize
quent treatment of HPbI3 with excess CH3NH2 yielded in the orthorhombic space group Cmc21 with a pseudo hexago-
MAPbI3 that was characterized by P-XRD and yielded a work- nal metric [a = 8.7560(8), b = 15.1670(1), c = 8.1830(4) Å;
ing device with a PCE of 16%.
b ≈ Ί
3a], that may explain the confusion by Pang et al.[7] Also
In a very recent paper Zhang et al. mentioned PbI2·xHI that the transformation of DMAPbI3 by methylamine gas into
was synthesized from PbI2/conc. HI/DMF.[9] They reported a MAPbI3 has recently be shown in the literature.[16] Addition-
P-XRD pattern for “HPbIx”, which is very similar to the ally, the higher solubility of “HPbI3” is obvious, as it is known
hexagonal pattern of Pang et al. and Long et al. assigned to that the solubility of PbI2 is increased in the presence of iod-
HPbI3. Additionally, they report on a P-XRD pattern for a ides.[17] The decomposition temperature of the TG measure-
phase “PbI2·xHI”, that showed no similarity to any known ment by Pang et al.[7] fits with the one reported from Eppel et
phase.
al.[15] for DMAPbI3. The weight loss for the formation of PbI2
The great significance of “HPbI3” as a precursor is demon- is calculated as 27.3%, i.e. slightly higher than observed by
strated by the fact that it is mentioned in several reviews[10] or Pang but within the experimental range.
used without further characterization.[11] Furthermore, the idea
In order to proof our guess we reexamined the “HPbI3”
to “recover” moisture-sensitive perovskites like MAPbI3 by powder by solving PbI2 and aq. HI in DMF. We have preci-
transformation to “HPbI3” might be an important advantage in pated yellow needles/rods by overlaying the solution with ethyl
processing of the PSC.
ether. The measurement of a selected single crystal yielded
On the whole the situation for the precursor “HPbI3” is in- the known structure of DMAPbI3 (Tables S1–S3, Supporting
consistent according the published data and also to our experi- Information). The refinement of the room temperature data
ence with hybrid perovskites.[12] On the one hand there was was done assuming the hexagonal unit cell and space group
clear evidence, that a compound acting as “HPbI3” is an impor- P63/mmc, as we observed neither super structure reflections,
tant intermediate during the formation of FAPbI3 and MAPbI3. nor a splitting of the reflection profiles or a reduction of the
On the other hand the existence of HPbI3 as a well-defined Laue class. Positions of Pb and I were obtained from direct
crystalline compound seemed to be quite implausible, because methods, N and C were localized from difference Fourier syn-
then a “naked” proton has to act as a cation. One indication thesis. The organic cation is disordered. This model corre-
for this point is the lack of knowledge on the ionic radius of sponds to the data given by Mancini et al.[13] We have found
H+. Therefore, we have investigated the reactivity of the sys- no evidence for a symmetry reduction to the orthorhombic
tem HI/PbI2/DMF and tried to obtain HPbI3 from PbI2 and aq. non-centrosymmetric model in Cmc21 as given by Eppel et
conc. hydroiodic acid as it was assumed in the literature. The al.[15] and also no reason for the symmetry reduction to P63mc.
samples were characterized by XRD investigations of powders Figure 1 shows the excellent agreement of the experimental
and single crystals, vibrational spectroscopy, and UV/Vis spec-
troscopy. The results show that a compound “HPbI3” is not
formed but reversible reactions with the solvents DMF and
H2O, respectively, mimic its intermediate existence.
Results and Discussion
Formation of (CH3)2NH2PbI3 from HI/PbI2/H2O/DMF
The first part of our results deals with the role of DMF in
the system PbI2/HI/H2O. Pang et al.[7] published a hexagonal
unit cell and lattice parameters of a = 8.765(1) and c =
8.177(2) Å for “HPbI3” and very similar values for the DMF
adduct HPbI3·xDMF (a = 8.7517, c = 8.1802 Å). This was not
comprehensible because H+ and DMF should show signifi-
cantly different volumes. In the case of a 1D perovskite the
size of the chain of face-sharing PbI6 octahedra should be sim-
ilar and the lattice parameter a relates to the cation size. A
comparison to literature revealed very similar lattice param- (Mo-Kα1 radiation).
Figure 1. Experimental and calculated diffraction pattern DMAPbI3
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
2
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
P-XRD pattern to the theoretical pattern calculated from the octahedron around Pb1 is nearly regular, while the octahedron
sc-XRD data (see Supporting Information).
around Pb2 is distorted in a 1+2+2+1 pattern. This distortion
is typically for a lone pair activity of a ns2 cation like Pb2+.
“HPbI3”/H18O8Pb3I8 from aq. conc. Hydroiodic Acid and
PbI2
When it became clear that the reaction of PbI2/HI/H2O in
DMF yields no compound “HPbI3” we tried to synthesize
HPbI3 directly from conc. hydroiodic acid. In the
literature[7–9,11b,18] several groups have described the synthesis
of “HPbI3” by gas phase reaction of HI and PbI2 with the
evaporation of gaseous HI from conc. aq. hydroiodic acid.
When PbI2 was placed beside an open container with conc. aq.
HI (57%) in an exsiccator, we observed the crystal growth of
light-yellow rods in the powder sample of PbI2. Single crystal
investigations by XRD revealed an unknown orthorhombic
unit cell. The preliminary structure determination produced a
model with trimeric chains of PbI6 octahedra [Pb3I8]2–. The
residual electron density between these chains were interpreted
as water molecules that were interconnected by hydrogen
bonds. According to the requirement of charge neutrality a part
of the water molecules should be labelled as oxonium cations
H3O+. With this model in mind we added PbI2 directly to conc.
aq. hydroiodic acid and obtained well-developed light-yellow
single crystals that showed the same orthorhombic unit cell as
the crystals formed before by vapor transport. The refinement
resulted in a composition H18O8Pb3I8.
The crystal structure (Figure 2) displays stripes of edge-
sharing PbI6 octahedra running in direction of the short c axis.
The motif can be seen as the transition from the NH4CdCl3
type, where the chains of edge-sharing octahedra form dimers
and the pristine PbI2 with layers of edge-sharing octahedra.
With respect to the NH4CdCl3 type there is one additional row
of PbI6/3 octahedra (each additional iodine is connected to
three Pb atoms). Vice versa the layers in PbI2 are separated by
additional I– anions into stripes. The stripes are arranged in a
fishbone-like manner with the H3O+/H2O subunits in between.
Figure 3. Pb–I and H3O+/H2O subunits in H18O8Pb3I8.
The O–O distances between 2.49 and 2.75 Å (Figure 3) are
typical for water molecules, which are linked by hydrogen
bonds.[19] The H3O+/H2O subunits represent a cut-out of the
crystal structure of ice I (O–O: 2.762 Å).[20] The short distance
O3–O4 is probably caused by a preferred localization of the
additional H+. In general, the positions of the hydrogen atoms
could not be determined, but it is known from vibrational spec-
troscopy and MD simulations, that the O–O distances are
+
shortened in oxonium cations like the Zundel cation H5O2 or
+ [21]
the Eigen cation H9O4 .
We expect an interaction of the
H3O+/H2O subunits to the [Pb3I8]2– chains via O–H···I
hydrogen bridges. The shortest O–I distances are around
3.75 Å that is comparable to the N···I distances in MAPbI3 and
other hybrid iodoplumbates.
The explicit characterization of H18O8Pb3I8 as the first prod-
uct from the gas phase reaction of aq. conc. hydroiodic acid
Figure 2. Crystal structure of H18O8Pb3I8, projection in direction
(001); black: Pb, turqois: I; red: O.
Pb–I distances reflect the usual findings for connected poly- with PbI2 demonstrates, that hydroiodic acid acts under this
hedra (Figure 3). The shortest distance is observed for the ter- conditions not as a simple source of HI leading to the forma-
minal distance Pb2–I4 (2.98 Å) and the longest distance to the tion of HPbI3, but H2O is also present in the gas phase. This
threefold connected iodine in trans position (Pb2–I1: 3.55 Å). is not a surprise, because concentrated hydroiodic acid is
Intermediate distances result for the other twofold (Pb2–I3: known as an azeotrop with a boiling point maximum of 127 °C
3.16 Å) and threefold connected iodine atoms (3.21 / 3.23 / under normal pressure. Furthermore, the formation of oxonium
3.34 Å). According to the position within the ribbon, the PbI6 cations H3O+ is in line with the property of conc. hydroiodic
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
3
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
acid as a very strong Brønsted acid. So species of the series
There are layers of edge-sharing PbI6 octahedra in a CdCl2-
(H2O)x(PbI2)y(HI)z are conceivable under the reported reac- type arrangement with partially occupied Pb2+ sites. The site
tions conditions.[22] H18O8Pb3I8 is the representative for x = 6, occupation factor is 78.4(5)%. Between the layers there are
y = 3, and z = 2. As Figure 2 demonstrates it can be expected H2O molecules that are connected via hydrogen bridges to a
that the values of x, y, and z depend on the pattern for the corrugated layer. According to charge neutrality there are ad-
arrangement of PbmIn motifs and H3O+/H2O subunits as well ditional protons. Assuming a ratio Pb:I of 1:3 (see below) the
as the demand for charge neutrality.
formula can be written as (H3O)2x(H2O)2–2xPb1–xI2 with x ≈
0.226(5) or (H3O)0.5(H2O)1.5Pb0.75 2 if the relation to
I
H18O8Pb3I8 is retained. The structure repesents an intercalation
of H3O+/H2O into PbI2 with a change of the anion stacking
from hcp to ccp. In comparison to PbI2 the distance between
the layer is enlarged by 40% or 2.84 Å per H3O+/H2O layer.
Because of the Pb2+ deficit and the corresponding H+ between
the Pb/I layers both layers are charged. The topology of the
H3O+/H2O layer is very similar to the situation in H18O8Pb3I8.
(H3O)2x(H2O)2–2xPb1–xI2 (x ≈ 0.23) / (H3O)0.5(H2O)1.5Pb0.75 2
I
Single crystals of H18O8Pb3I8 are not stable in air. The well-
developed rods precipitated from conc. hydroiodic acid trans-
form to smaller crystals of irregular shape. The P-XRD pattern
of the freshly prepared single crystals were in good agreement
to the calculated pattern (Figure S4, Supporting Information),
but powdered samples showed additional lines after a few
hours (Figure S6, Supporting Information). The irregular crys-
tals were characterized by sc-XRD and revealed a rhom-
bohedal unit cell with a = 4.56 and c = 29.53 Å. The structure
The transition from H18O8Pb3I8 to (H3O)0.5(H2O)1.5Pb0.75 2
I
leaves the composition nearly unchanged and requires only
small shift of the atoms. Furthermore, the volumes related to
one iodide atom are also very similar [H18O8Pb3I8: 86.73 Å3;
(H3O)0.5(H2O)1.5Pb0.75I2: 88.6 Å3], so it is comprehensible,
that the shape of the H18O8Pb3I8 crystals is maintained in the
course of the transformation.
When (H3O)0.5(H2O)1.5Pb0.75I2 was stored in air for several
days the light-yellow color changed to orange-yellow and the
P-XPD pattern showed the formation of PbI2 (Figure S7, Sup-
porting Information). A similar observation was made in the
course of the single crystal measurement. The single crystal of
(H3O)0.5(H2O)1.5Pb0.75I2 is stable for several hours but during
a long-term measurement that was done to compensate the
weak scattering power caused by the small crystal size, the
¯
solution in R3m yielded for the positions of Pb and I a CdCl2-
type structure (R1 = 0.071, wR2 = 0.166). Difference fourier
synthesis gave a residual electron density between the layers
of 6.2 e–·Å–3 on a site 2/3, 1/3, 0, that was interpreted as oxy-
gen and reduced the R values (R1 = 0.042, wR2 = 0.108). The
displacement parameter of Pb was significantly enlarged in
comparison to I (factor 2). The refinement with a free site
occupation factor for Pb resulted in reasonable values for the
displacement parameter and a further reduction of the R values
(R1 = 0.020, wR2 = 0.031). The interpretation of this model is
shown in Figure 4.
reflection profiles of (H3O)0.5(H2O)1.5Pb0.75I2 started to get
broader, showed diffuse scattering halos, and new reflections
occurred.[23] The indexing revealed the hexagonal primitive
unit cell with slightly enlarged c lattice parameter [a = 4.552(2)
and c = 7.105(6) Å] compared to the unit cell of the pristine
¯
PbI2. The refinement (P3m, CdI2-type) was biased by the
broad reflection profiles (R1 0.082, wR2 0.228),
=
=
but the results showed clearly, that a single crystal of
(H3O)0.5(H2O)1.5Pb0.75I2 transforms to a single crystal of PbI2
by a topotactic reaction. A comparison of the two orientation
matices shows, that both ab planes remain parallel. This at a
first view surprising result is in line with the observation, that
the ions I– as well as Pb2+(!) have a remarkably high mobility
in MAPbI3 and related compounds.[24]
Our investigations show, that PbI2, H18O8Pb3I8, and
(H3O)0.5(H2O)1.5Pb0.75I2 can be transformed into one another
according to the conditions of reaction (Figure 5). A gas phase
reaction transforms binary PbI2 into H18O8Pb3I8. The fact, that
powder samples of PbI2 result in small but well-developed sin-
gle crystals of H18O8Pb3I8 underline the importance of mobili-
zation of Pb2+ and I– by HI/H2O. In contrast to a first guess
and many statements in the literature conc. hydroiodic acid is
not only a resource for HI, but also for H2O. The compound,
that is formed at room temperature in a closed system (exsicc-
ator) contains HI and H2O in a ratio 1:4. This is close to the
molar ratio as is present in conc. hydroiodic acid with 57% HI
(1:5.5) and corresponds to the composition of the azeotrop.
Figure 4. Crystal structure of (H3O)2x(H2O)2–2xPb1–xI2 (left) with cor-
rugated layer of H3O+/H2O and layer of PbI6 octahedra; black: Pb,
turqois: I; red: O.
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
4
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
Figure 5. Scheme for the transformations PbI2 Ǟ H18O8Pb3I8 Ǟ (H3O)0.5(H2O)1.5Pb0.75I2 Ǟ PbI2 and the synthesis of MAPbI3 from
H18O8Pb3I8 / (H3O)0.5(H2O)1.5Pb0.75I2.
”HPbI3” is introduced to the literature as a precursor for in a change of color from orange-yellow to light-yellow, be-
MAPbI3 and FAPbI3. Our experiments have shown, that the cause the absorption edge shifts from 525 nm to 475 nm. Also
gas phase reaction with conc. hydroiodic acid yields not the reaction of (H3O)2x(H2O)2–2xPb1–xI2 back to PbI2 changes
HPbI3, but H18O8Pb3I8 instead. Nevertheless, a simple acid- the color, while the transformation from H18O8Pb3I8 lets the
base reaction with gaseous CH3NH2 should result in MAPbI3. color unchanged. The evaluation was done with the assump-
When a sample of H18O8Pb3I8 was placed beside a concen- tion of a direct band gap. There are other equations for indirect
trated aqueous solution of CH3NH2 we observed immediately bandgap semiconductors.[26] The bandgap increases from
the formation of black MAPbI3. The fast reaction results from 2.3 eV to about 2.6–2.7 eV.
the high vapor pressure of CH3NH2, that is at room tempera-
ture much higher than in the case of aq. conc. hydroiodic acid.
Subsequently, the sample became a yellow “jam”. Single crys-
tal XRD measurements revealed MA4PbI6·2H2O and
MA4PbI6. After annealing at 60–100 °C a black solid
was formed, that was identified as MAPbI3 with very few
PbI2.
Things might be different with the use of gaseous HI. Here,
it was shown, that surpassing of gaseous HI over MAPbBr3 or
MAPbCl3 leads to the formation of MAPbI3, in the case of the
bromide even in a topotactic reaction peserving the shape of
the single crystals.[25]
IR and Raman Spectra
Vibrational spectra
of
PbI2,
H18O8Pb3I8,
and
(H3O)0.5(H2O)1.5Pb0.75I2 were recorded for further characteri-
zation (see Figure 7). The strong IR-active frequencies around
3480 cm–1 are due to O–H valence modes and only very weak
in the Raman spectrum. The sharp modes around 2800 cm–1
are probably due to the oxonium cation. The modes around
1600 cm–1 (IR and Raman) are bending modes of H3O+/H2O.
The modes around 500 cm–1 are due to rotational modes.[27]
Significant differences occur for the very strong Raman-active
Pb–I valence modes between 123 and 72 cm–1 according to the
different connectivity of PbI6 octahedra. The highest frequency
is observed for H18O8Pb3I8 as the only compound
UV/Vis Spectroscopy
Optical
spectra
of
PbI2,
H18O8Pb3I8,
and with terminal Pb–I bonds. The higher frequency of
(H3O)0.5(H2O)1.5Pb0.75I2 were recorded for further characteri- (H3O)0.5(H2O)1.5Pb0.75I2 in comparison to PbI2 can be ex-
zation and are shown in Figure 6. The spectra confirm the vis- plaind by the missing Pb2+ because some of the I– are only
ual impression, that the reaction of PbI2 to H18O8Pb3I8 results connected to two octahedra instead of three.
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
5
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
Figure 7. Vibrational spectra of PbI2, H18O8Pb3I8, and
(H3O)0.5(H2O)1.5Pb0.75I2.
Figure 6.
UV/Vis
spectra
of
PbI2,
H18O8Pb3I8,
and
(H3O)0.5(H2O)1.5Pb0.75I2 measured in diffuse reflection (top) and cor-
responding Tauc plots (bottom, see text).
less, the occurrence of “HPbI3” can be mimicked by reversible
addition and release of H2O, but according to the moisture
Conclusions
The development of suitable precursors for the synthesis of sensitivity of MAPbI3 and related perovskites, this should be
hybrid organic inorganic perovskites (HOIPs) is a central ques- obeyed in the course of processing PSCs and also the interpret-
tion for the fabrication of PSCs. We have shown that claimed ation of experimental results.
“HPbI3” does not have the expected composition, when it is
We have synthesized and characterized H18O8Pb3I8 as the
synthesized using DMF as a solvent. All the observed behavior first example for a well-defined compound of the system H2O/
in several publications that base on “HPbI3” as a precursor is HI/PbI2. The structure can be easily described by a combina-
explained by the fact that (CH3)2NH2PbI3 is formed as inter- tion of the structural features of the binary components PbI2
mediate compound. In a subsequent reaction the DMA+ cation and H2O (i.e. ice I), that are separated by “additional” HI into
+
is transformed to methylamonium CH3NH3 . Although this re- a fishbone-like arragment of stripes. At room temperature
action of DMF including the formation of HOIPs was already H18O8Pb3I8 transforms within a few hours without significant
described in the literature this was not taken into consideration change of the composition to (H3O)0.5(H2O)1.5Pb0.75I2. This
when “HPbI3” was introduced as an existing precursor. The layer structure is related to a CdCl2-type structure for the Pb–
reason is the missing substantial characterization of the inter- I substructure with vacant sites on the Pb2+ positions. The
mediate species in the course of the device manufacturing.
charge compensation takes place by intercalated layers of H3O/
In general, solvents used in processing PSCs are not as inno- H2O, that represent cutouts of the structure of ice with ad-
cent as sometimes expected. We have shown that in the pres- ditional protons. (H3O)0.5(H2O)1.5Pb0.75I2 releases in a topotac-
ence of H2O HI acts not as an active gaseous phase forming tic transformation HI/H2O and forms PbI2 closing the cycle
“HPbI3” as it is supposed in some publications. Instead, H3O+ that started with the uptake of HI/H2O from conc. aq. hy-
and higher aggregates (H3O+)m(H2O)n are formed. Neverthe- droiodic acid. This topotactic decomposition can be important
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
6
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
black (MAPbI3) and after a while it changed to a colorless-yellow
for the precipitation of MAPbI3 or related HOIPs in a preferred
mash-like solid. From this “mash” single crystals of MA4PbI6·2H2O
and MA4PbI6 were isolated. The “mash” turned back into black (indi-
cating the formation of MAPbI3) by heating on a hotplate at about 60–
100 °C. Very few PbI2 as by-product was detected by P-XRD.
orientation.
According to our results we expect the existence of variety
of compounds in the system H2O/HI/PbI2. The fishbone-like
pattern allows an expansion of the neutral parts according to
the formula (H3OI)2(PbI2)n(H2O)2n, where the PbI2/H2O layers
are separated by the additional I– anions and H3O+ cations,
respectively. H18O8Pb3I8 represents the example with n = 3.
For the subsequent transformation to the intercalation com-
pound (H3O)2x(H2O)2–2xPb1–xI2 the value of n relates to the
extent of the deficit x.
Taking the release of H2O at elevated temperatures into ac-
count representatives of the system H2O/HI/PbI2 like
H18O8Pb3I8 / (H3O)0.5(H2O)1.5Pb0.75I2 can serve as new pre-
cursors for the synthesis of HOIPs with “HPbI3” as a vitual
precursor. As a proof of principle we passed gaseous CH3NH2
over H18O8Pb3I8 / (H3O)0.5(H2O)1.5Pb0.75I2 and observed in-
stantly blackening of the light-yellow starting materials due to
the formation of MAPbI3. Gentle heating to 60–100 °C re-
sulted in crystalline MAPbI3 with small amounts of PbI2.
However, our investigations show the importance of a substan-
tial characterization of intermediate products for a comprehen-
sive unterstanding of the processes in the course of the fabrica-
tion of a PSC device.
P-XRD: The products were characterized by powder XRD (STOE
Stadi P, Mo-Kα1 radiation, Ge (111) monochromator, silicon microstrip
detector Mythen 1 K, Debye-Scherrer setup, transmission). Unit cell
refinement was done using the STOE WinXPOW software.
Single Crystal XRD: Measurements of single crystals were done
using a diffractometer equipped with an image plate detector (IPDS II,
Fa. STOE, Darmstadt, Germany)[28] or with a Bruker APEX II dif-
fractometer with a CCD detector and a microsource.[29] The structure
solution (direct methods) and refinements were done with the
SHELXL software package.[30] For DMAPbI3, we preferred a refine-
ment in the hexagonal space group P63/mmc. Refinements in the non-
centrosymmetric subgroup P63mc or the orthorhombic space group
Cmc21 yielded no improvement beyond the standard deviations and
the DMA cation remained disordered. Further details are listed in the
Supporting Information (Tables S1–S3).
Further details of the crystal structures investigations may be obtained
from the Fachinformationszentrum Karlsruhe, 76344 Eggenstein-
Leopoldshafen, Germany (Fax: +49-7247-808-666; E-Mail:
crysdata@fiz-karlsruhe.de, http://www.fiz-karlsruhe.de/request for de-
posited data.html) on quoting the depository numbers CSD-434387
(for H18O8Pb3I8) and CSD-434388 [for (H3O)0.5(H2O)1.5Pb0.75I3].
Experimental Section
IR and Raman Spectra: IR and Raman spectra were recorded with a
Bruker FRA 106/S module with a Nd:YAG laser (λ = 1064 nm).
DMAPbI3: PbI2 (461 mg, 99%, Acros Organics) was dissolved in
0.7 mL DMF (99.8%, Roth). Subsequently, hydroiodic acid (0.2 mL,
57% aq. solution, Acros Organics) was added and the solution was
overlaid with ethyl ether. After a few minutes, light yellow crystals of
DMAPbI3 were formed.
UV/Vis Spectra: Diffuse reflectance spectra were recorded with a
Thermoscientific Evolution 600 equipped with an integration sphere.
The baseline was measured against SPECTRALON®. For the measure-
ment, a powdered sample was pressed into a pellet inside the sample
holder.
H18O8Pb3I8 / (H3O)2x(H2O)2–2xPb1–xI2: Initially, a mixture of both
compounds was synthesized by placing powdered PbI2 (1.0 g) in an
open glass vessel. This vessel with PbI2 was enclosed into a larger
container with an excess of conc. hydroiodic acid (57% aq. solution)
in a separate open vessel, so that the equilibrium gas phase (H2O/HI)
could slowly diffuse via the gas phase into the provided PbI2. After
1 week small yellow crystals were grown from PbI2.
Supporting Information (see footnote on the first page of this article):
SI contain additional data for sc-XRD measurement and experimental
and calculated P-XRD patterns.
Acknowledgements
H18O8Pb3I8: Pure H18O8Pb3I8 was obtained by cooling a hot solution
of PbI2 (3.3 g) in 5 mL hydroiodic acid (57% aq. solution) to room
temperature. The bright yellow crystals were isolated by suction fil-
tration and dried by pressing between filter papers. All characteriza-
tions of H18O8Pb3I8 were done instantly. After 3 d, when the powder
still looked unchanged, the measured P-XRD showed only reflections
of (H3O)2x(H2O)2–2xPb1–xI2 [or (H3O)0.5(H2O)1.5Pb0.75I2, respectively],
so all characterizations were repeated. One sample was washed with
ethyl ether and dried in vacuo. The responding P-XRD showed only
reflection of PbI2.
This work was supported by BMBF (Project “Meso-PIN”,
03 SF0484 D)
Keywords: HPbI3 precursor; Oxonium cation; Lead; Non-
innocent solvent; Hybrid perovskite; Single crystal
References
(H3O)2x(H2O)2–2xPb1–x 2
developed light-yellow platelets of (H3O)2x(H2O)2–2xPb1–x 2
(H3O)0.5(H2O)1.5Pb0.75 2 were obtained by dissolving the mixture of
I
or (H3O)0.5(H2O)1.5Pb0.75I2: Well-
[1] M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, H. J.
Snaith, Science 2012, 338, 643–647.
[2] National Renewable Energy Laboratory, “Best Research-
I
/
I
H18O8Pb3I8 / (H3O)2x(H2O)2–2xPb1–xI2 (see above) in acetone and slow
evaporation of the solvent.
Cell
Efficiencies”,
https://www.nrel.gov/pv/assets/images/
efficiency-chart.png.
[3] For example: a) T. Kollek, C. Fischer, I. Göttker-Schnetmann, S.
Polarz, Chem. Mater. 2016, 28, 4134–4138; b) Z. L. Zhang, B. Q.
Men, Y. F. Liu, H. P. Gao, Y. L. Mao, Nanoscale. Res. Lett. 2017,
12, 84.
MAPbI3 from H18O8Pb3I8 / (H3O)2x(H2O)2–2xPb1–xI2: A sample of
the initially synthesized mixture was placed in a glass next to CH3NH2
(40% aq. solution, Merck). The bright yellow solid turned instantly
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
7
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
[4] F. Wang, H. Yu, H. Xu, N. Zhao, Adv. Funct. Mater. 2015, 25, [18] a) G. E. Eperon, S. D. Stranks, C. Menelaou, M. B. Johnston,
L. M. Herz, H. J. Snaith, Energy Environ. Sci. 2014, 7, 982–988;
b) S. R. Raga, L. K. Ono, Y. Qi, J. Mater. Chem. A 2016, 4, 2494.
[19] T. Steiner, Angew. Chem. 2002, 114, 50–80; Angew. Chem. Int.
Ed. 2002, 41, 48–76.
1120–1126.
[5] For example: a) Z. Zhou, S. Pang, F. Ji, B. Zhang, G. Cui, Chem.
Commun. 2016, 52, 3828–3831; b) T. Zhang, N. Guo, G. Li, X.
Qian, Y. Zhao, Nano Energy 2016, 26, 50–56.
[20] a) A. Goto, T. Hondoh, S. Mae, J. Chem. Phys. 1990, 93, 1412–
1417; b) A. D. Fortes, I. G. Wood, D. Grigoriev, M. Alfredsson,
S. Kipfstuhl, K. S. Knight, R. I. Smith, J. Chem. Phys. 2004, 120,
11376–11379.
[21] a) I. Shin, M. Park, S. K. Min, E. C. Lee, S. B. Sih, K. S. Kim,
J. Chem. Phys. 2006, 125, 234305; b) M. Park, I. Shin, N. J.
Singh, K. S. Kim, J. Phys. Chem. A 2007, 111, 10692–10702;
c) W. Amit, G. Gallot, F. Hache, S. Bratos, J. C. Leicknam, R.
Vuilleumier, J. Chem. Phys. 2007, 126, 034511.
[22] a) A. Flemström, T. K. Hirsch, L. Sehlstedt, S. Lidin, L. Ojamäe,
Solid State Sci. 2002, 4, 1017–1022; b) D. Katryniok, R. Kniep,
D. Mootz, Z. Anorg. Allg. Chem. 1980, 461, 96–100.
[23] The recrystallization from acetone yielded well-developed single
crystals that were used for the final structure refinement. Single
crystals recovered from the direct transformation of H18O8Pb3I8
produced identical results, but with higher R values.
[24] a) A. Senocrate, I. Moudrakovski, G. Y. Kim, T.-Y. Yang, G. Gre-
gori, M. Grätzel, J. Maier, Angew. Chem. Int. Ed. 2017, 56, 7755–
7759; b) T, M. Brenner, D. A. Egger, A. M. Rappe, L. Kronik, G.
Hodes, D. Cahe, J. Phys. Chem. Lett. 2015, 6, 4754–4757; c) A.
Walsh, D. O. Scanlon, S. Chen, X. G. Gong, S.-H. Wei, Angew.
Chem. Int. Ed. 2015, 54, 1791–1794.
[6] The terms “layer structure”, connection of PbI6 octahedra via
common edges and corners, and the assignment to the NH4CdCl3
type like KPbI3 are not consistent, because the NH4CdCl3 type is
a 1D structure with solely edge-sharing PbI6 octahedra.
[7] S. Pang, Y. Zhou, Z. Wang, M. Yang, A. R. Krause, Z. Zhou,
N. P. Padture, G. Cui, J. Am. Chem. Soc. 2016, 138, 750–753.
[8] M. Long, T. Zhang, Y. Chai, C.-F. Ng, T. C. W. Mak, J. Xu, K.
Yan, Nat. Commun. 2016, 7, 13503.
[9] a) Zhang, M. I. Dar, G. Li, F. Xu, N. Guo, M. Grätzel, Y. Zhao,
Sci. Adv. 2017, 3, e1700841; b) Y. Zhou, N. P. Padture, ACS En-
ergy Lett. 2017, 2, 2166–2176.
[10] J. Gong, P. Guo, S. E. Benjamin, P. G. Van Patten, R. D. Schaller,
T. Xu, J. Energy Chem. 2018, 27, 1017–1039.
[11] a) W. S. Yang, J. H. Noh, N. J. Jeon, Y. C. Kim, S. Ryu, J. Seo,
S. Seok, Science 2015, 348, 1234–1237; b) H.-H. Fang, F. Wang,
S. Adjokatse, N. Zhao, J. Evens, M. A. Loi, Light: Sci. Applica-
tions 2016, 5, e16056.
[12] a) M. Daub, H. Hillebrecht, Angew. Chem. 2015, 127, 11168–
11169; Angew. Chem. Int. Ed. 2015, 54, 11016–11017; b) M.
Daub, C. Haber, H. Hillebrecht, Eur. J. Inorg. Chem. 2017, 7,
1120–1127; c) M. Daub, H. Hillebrecht, Z. Kristallogr. 2018, 233,
555–564; d) M. Daub, H. Hillebrecht, Chem. Eur. J. 2018, 24,
9075–9084.
[25] K. Chen, X. Deng, R. Goddard, H. Tüysüz, Chem. Mater. 2016,
28, 5530–5537.
[26] a) J. Tauc, R. Grigorovici, A. Vancu, Phys. Status Solidi 1966,
15, 627–637; b) E. A. Davis, N. F. Mott, Philos. Mag. 1970, 22,
903–922.
[13] A. Mancini, P. Quadrelli, G. Amoroso, C. Milanese, M. Boiocchi,
A. Sironi, M. Patrini, G. Guizzetti, L. Malavasi, J. Solid State
Chem. 2016, 240, 55–60.
[14] For example: a) T. R. Amarante, P. Neves, A. A. Valente, F. A. A.
Paz, A. N. Fitch, M. Pillinger, I. S. Goncalves, Inorg. Chem. 2013,
52, 4618–4628; b) P. Thuery, J. Harrowfield, CrystEngComm
2015, 17, 4006–4018.
[15] S. Eppel, N. Fridman, G. Frey, Cryst. Growth Des. 2015, 15,
4363–4671.
[27] P. T. T. Wong, E. Whalley, J. Chem. Phys. 1976, 64, 2359–2366.
[28] Program XAREA, XRED, Stoe Cie., Darmstadt, Germany.
[29] T. Remove, SAINT, Data Reduction and Frame Integration Pro-
gram for the CCD Area-detector System; Bruker Analytical
X-ray Systems, Madison, WI, 2006.
[30] G. M. Sheldrick, SHELXTL, Crystallographic System, Bruker
AXS Analytical X-ray Instruments Inc.: Madison, WI, 1997–
2015.
[16] D. Ju, T. Zhao, D. Yangyang, G. Zhang, X. Hu, D. Cui, X. Tao,
J. Mater. Chem. A 2017, 5, 21919–21925.
[17] G. Jander, E. Blasius, Lehrbuch der Analytischen und Präpara-
tiven Anorganischen Chemie, 15th ed., S. Hirzel Verlag, Stuttgart,
2002.
Received: June 20, 2018
Published online: ᭿
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
8
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Journal of Inorganic and General Chemistry
ARTICLE
Zeitschrift für anorganische und allgemeine Chemie
M. Daub, H. Hillebrecht* ................................................. 1–9
On the Demystification of “HPbI3” and the Peculiarities of the
Non-innocent Solvents H2O and DMF
Z. Anorg. Allg. Chem. 0000, 0–0
www.zaac.wiley-vch.de
9
© 0000 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim