New insights for titanium(iv) speciation in acidic media based on UV-visible and31P NMR spectroscopies and molecular modeling

Article abstract of DOI:10.1039/d1ra04284j

Titanium chemistry in aqueous acidic media has been extensively investigated over the last decades. Hydrolyzed species such as Ti(OH)³⁺, TiO²⁺, Ti(OH)₂²⁺or Ti(OH)₃⁺have been identified and their equilibria have been studied in nitric and perchloric acid. A predominance of the divalent cations was found for low pH (i.e., pH <2). Nonetheless, recent literature reports the existence of small titanium oxo-clusters in aqueous acidic media for large titanium(iv) concentration (typically., >0.1 mol L^?1), as stable precursors for the formation of condensed titanium dioxide. The present paper reconsiders firstly previous knowledge about the speciation of titanium(iv) in non-complexing acidic media by giving evidence for the presence of polynuclear hydrolyzed species, even at very low Ti(iv) concentration (i.e., typically <0.1 mmol L^?1). UV-visible absorbance spectra recorded for diluted nitric acid solutions (a model of non-complexing acidic medium) containing titanium(iv) were compared to time-dependent density functional theory (TD-DFT) predicted excitation energies. Experimental and predicted maximal absorbance wavelengths showed significantly improved matches when polynuclear species were considered in TD-DFT calculation. Then, 0.1-12.7 mol L^?1phosphoric acid solutions containing titanium(iv) were studied by means of spectroscopic techniques (UV-visible, NMR) in order to identify qualitatively the presence of titanium(iv) complexes and to link this speciation to the acid concentration. Two different titanium(iv) orthophosphate complexes, potentially polynuclear, were detected, and the presence of free titanium(iv) is also expected for low phosphoric acid concentration (i.e., <0.1 mol L^?1). A general complexation scheme for a large range of H₃PO₄concentration was thus formulated.

Full text of DOI:10.1039/d1ra04284j

RSC Advances
View Article Online
View Journal | View Issue
PAPER
New insights for titanium(IV) speciation in acidic
3
1
media based on UV-visible and P NMR
Cite this: RSC Adv., 2021, 11, 27059
spectroscopies and molecular modeling†
ab
b
c
Lucas Mangold,
Hubert Halleux, Sebastien Leclerc,
´
d
e
and Alexandre Chagnes *a
Aurelien Moncomble,
Gerard Cote
´
´
Titanium chemistry in aqueous acidic media has been extensively investigated over the last decades.
3+
2+
2+
2
+
3
Hydrolyzed species such as Ti(OH) , TiO , Ti(OH)
or Ti(OH)
have been identified and their
equilibria have been studied in nitric and perchloric acid. A predominance of the divalent cations was
found for low pH (i.e., pH <2). Nonetheless, recent literature reports the existence of small titanium oxo-
ꢀ1
clusters in aqueous acidic media for large titanium(IV) concentration (typically., >0.1 mol L ), as stable
precursors for the formation of condensed titanium dioxide. The present paper reconsiders firstly
previous knowledge about the speciation of titanium(IV) in non-complexing acidic media by giving
evidence for the presence of polynuclear hydrolyzed species, even at very low Ti(IV) concentration (i.e.,
ꢀ1
typically <0.1 mmol L ). UV-visible absorbance spectra recorded for diluted nitric acid solutions (a
model of non-complexing acidic medium) containing titanium(IV) were compared to time-dependent
density functional theory (TD-DFT) predicted excitation energies. Experimental and predicted maximal
absorbance wavelengths showed significantly improved matches when polynuclear species were
ꢀ
1
considered in TD-DFT calculation. Then, 0.1–12.7 mol L
phosphoric acid solutions containing
titanium(IV) were studied by means of spectroscopic techniques (UV-visible, NMR) in order to identify
qualitatively the presence of titanium(IV) complexes and to link this speciation to the acid concentration.
Two different titanium(IV) orthophosphate complexes, potentially polynuclear, were detected, and the
Received 2nd June 2021
Accepted 30th July 2021
DOI: 10.1039/d1ra04284j
ꢀ1
presence of free titanium(IV) is also expected for low phosphoric acid concentration (i.e., <0.1 mol L ). A
rsc.li/rsc-advances
3 4
general complexation scheme for a large range of H PO concentration was thus formulated.
cadmium, iron or titanium. For this goal, the most appropriate
1
. Introduction
4
–10
technology at the industrial scale is solvent extraction.
This
The wet phosphoric process produces different grades of technology relies on the use of an extraction solvent, which can
phosphoric acid (known as “wet” phosphoric acid by opposition efficiently extract phosphoric acid from the leach solution.
to H
can be used in the agriculture sector (fertilizers), in food, or in against several impurities including titanium(IV).
3 4
PO
produced by pyrometallurgy) and phosphate salts that However, most common solvents exhibit a lack of selectivity
11,12
In order to
the pharmaceutical or electronic industries. This process is remove titanium from wet phosphoric acid, it is therefore
based on the leaching of the phosphate rock with concentrated mandatory to implement additional purication steps.
sulfuric acid and a series of purication steps to reach the purity
specication requested by the target application. The puri- titanium is challenging. It requires knowing nely the specia-
cation steps aim at removing inorganic anions such as uo- tion of titanium(IV) in concentrated phosphoric acid and its
3 4
The design of extraction solvents selective of H PO against
1
–3
1
2–14
14
rides, sulfates, chlorides and metal impurities including liquid–liquid extraction equilibria. UV-visible,
Raman and
15
16
NMR spectroscopies, liquid–liquid extraction, solubility
17,18
19
a
measurements
and potentiometric methods were used to
Universit ´e de Lorraine, CNRS, GeoRessources, F-54000, Nancy, France. E-mail:
alexandre.chagnes@univ-lorraine.fr
investigate the speciation of titanium(IV) in various media such
as nitric, perchloric, sulfuric, hydrochloric and phosphoric
acids. The divalent titanium(IV) aquocation was generally found
b
Prayon, Rue Joseph Wauters 144 a` , B-4480, Engis, Belgium
c
Universit ´e de Lorraine, CNRS, LEMTA, F-54000, Nancy, France
d
Univ. Lille, CNRS, UMR 8516, LASIRE-Laboratoire Avanc ´e de Spectroscopie pour Les
to be predominant in perchloric or nitric acid solutions in the
Interactions La R ´e activit ´e et L’Environnement, F-59000, Lille, France
13,16,18
absence of any complexing agents.
The exact structure of
e
PSL Research University, Chimie ParisTech, CNRS, Institut de Recherche de Chimie-
this divalent cation is still debated as it was initially identied
Paris (IRCP), F-75005, Paris, France
2
+
20
as a titanium oxocation (titanyl TiO ) whereas more recent
studies suggested an intramolecular rearrangement leading to
†
Electronic supplementary information (ESI) available. See DOI:
0.1039/d1ra04284j
1
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27059

View Article Online
RSC Advances
Paper
2
+.15,21,22
the formation of titanium(IV) dihydroxide [Ti(OH)
2
]
a starting point the speciation of Ti(IV) in dilute nitric acid
ꢀ
1
According to the most recent literature, the hydrolysis equilibria solution (i.e., [HNO ] <2 mmol L HNO ) as a model of non-
3
3
of titanium(IV) in non-complexing acidic media can be written complexing acidic media. For this goal, UV-visible spectros-
as follows:
copy was used and completed with time-dependent density
function theory calculations (TD DFT). Then, the speciation of
Ti(IV) was investigated at its turn in a large range of concentra-
tion of phosphoric acid (i.e., in 0.1–12.7 mol L ) by P NMR
spectroscopy and UV-visible spectroscopy coupled with che-
mometric analyses.
+
+
] (aq.) + 2H
[
Ti(OH) ](aq.) + H
4
3
O # [Ti(OH)
3
2
O (KH,1) (1a)
ꢀ
1
31
+
+
2+
[
Ti(OH) ] (aq.) + H O # [Ti(OH) ] (aq.) + 2H O (KH,2) (1b)
3
3
2
2
2
+
+
3+
[
Ti(OH)
2
]
3 2
(aq.) + H O # [Ti(OH)] (aq.) + 2H O (KH,3) (1c)
The corresponding hydrolysis constants are reported in
Table 1.
In the presence of inorganic ligands in solution, titanium(IV)
can form various complexes as described in eqn (2) in the
particular case of acidic sulfate solutions:
2
. Experimental section and DFT
calculation
2.1 Materials and methods
4 4
Preparation of orthotitanic acid (H TiO ). All work was
carried under a fume hood. Orthotitanic acid (H
4
TiO
4
) used to
2
Ti(OH)2] (aq.) + nH SO
+
mꢀ2
2ꢀn(2ꢀm)
(aq.)
[
(aq.) # [Ti(OH) (H SO ) ]
(2)
m
4
2
m
4 n
where m ¼ 0 or 1 and n ¼ 1 or 2, and the corresponding equi- prepare a mother solution of various Ti(IV) was obtained as
librium constant is expressed as follows:
follows. A volume of 5 mL of titanium tetraisopropoxide (Sigma
ꢀ
ꢁ
Aldrich, purity ¼ 97%) was added dropwise to 25 mL of ethanol
2
ꢀnð2ꢀmÞ
ðTiðOHÞ ðH
ðTiðOHÞ2Þ
m
SO
i
4
Þ Þ
HmSO4
2
n
ꢂ
bn
¼
h
(3) (Carlo Erba, 96%) at 0 C under constant stirring. A white
2þ
n
SO4mꢀ2ꢁ
41
½H
m
precipitate of orthotitanic acid was formed immediately.
Aerwards, a mixture of 25 mL ethanol (Carlo Erba, 96%) and
ꢂ
3
mL of deionized water cooled at 0 C (resistance ¼ 18.2 U) was
Table 2 evidences signicant discrepancies between the
then added to the orthotitanic acid suspension. Aer 1 h of
agitation in an ice bath, the suspension was ltered under
vacuum by means of a fritted glass lter (cat. 4). The solid was
dried for 1 h under vacuum at room temperature (the solid was
regularly triturated during drying). A white powder (ꢃ3 g) of
titanium hydroxide was obtained.
m 4
H SO
formation constants b
n
in the presence of hydrogen sulfate and
21
28
sulfate ions. More recently, investigation of TiOSO
4
dissolution in
ꢀ
1
water and in 1.0 mol L HNO
3
by X-ray absorption spectroscopy and
small angle X-ray spectroscopy showed polycondensation phenomena
responsible for the possible presence of [Ti18
ꢀ
O27(SO
4
)
9.5
]
anions at
ꢀ
1
titanium(IV) concentrations ranging between 0.25 mol L
and
mol L . Likewise, other studies showed the presence of small
titanium clusters whose size changed depending on the experimental
Dissolution of titanium(IV) in nitric acid. All work was
carried under a fume hood. Approximately 0.3 g of the ortho-
ꢀ
1
1
ꢀ1
titanic acid was dispersed in 10 mL of 0.73 mol L nitric acid
2
9–32
conditions.
Such polytitanates species with bridging oxo groups
were also identied in anhydrous conditions in the case of titanium
alkoxide compounds such as titanium ethoxide.
cannot be excluded that polycondensation of titanium(IV) also occurs
in phosphoric acid. However, only few studies concern the investi-
gation of the speciation of titanium(IV) in phosphoric acid and the
conclusions of these studies are still subject of controversies.
To gain a better knowledge of titanium(IV) chemistry in
acidic aqueous media, we have decided to reinvestigate as
prepared by dilution of 65% (vol) nitric acid (analytical grade,
Merck) in deionized water. Aer 24 h under agitation, orthoti-
tanic acid was almost completely dissolved in nitric acid. In
3
3,34
Therefore, it
ꢀ1
order to complete the dissolution, 20 mL of 0.73 mol L nitric
acid was added again. Residual insoluble powder of orthotitanic
acid was removed by ltration under vacuum with a fritted glass
19,35,36

lter (cat. 4). The turbid solution was then ltered using an
Acrodisc Supor syringe lter containing a PES membrane
ꢂ
21
Table 1 Equilibrium constants as log K for Ti(IV) hydrolysis reactions at 25 C (reproduced with permission from Elsevier from Pichavant et al. ).
ND ¼ not determined in the original papers
16
23
24
25
17
18
26,27
Hydrolysis reactions
Liberti
Nabivanets
Lobanov
Nazarenko
Ziemniak
Schmidt
Janaf database
4
+
3+
+
Ti (aq.) + H
2
O # Ti(OH) (aq.) + H
ND
2.79
2.35
ꢀ0.16
ꢀ0.59
ꢀ0.92
1.05
0.097
ꢀ0.31
ꢀ0.64
ꢀ0.96
ND
ND
ND
ND
3.75
3.26
3+
2+
+
Ti(OH)2 (aq.) + H
2
O # Ti(OH)
O # Ti(OH)
O # Ti(OH)
2
(aq.) + H+ ꢀ1.8
(aq.) + H+ ꢀ2.4
+
2
+
Ti(OH)+
Ti(OH)
(aq.) +H
(aq.) + H
2
3
ꢀ2.1
ꢀ2.28
ꢀ1.85
ꢀ1.71
3
2
4
(aq.) + H
ꢀ2.1
ꢀ3.37
ꢀ1.9
ꢀ2.95
ꢀ3.07
27060 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
ꢂ
n
Table 2 Equilibrium constants for complexation of Ti(IV) (log b ) by hydrogen sulfate and sulfate at 25 C. Two constants are available for Szil a´ gyi
21
37
2 4 4 4
et al. depending on the nature of the added salt: (a) Na SO (b) (NH )SO (reproduced with permission from Elsevier from Pichavant et al. ). ND
¼
not determined in the original papers
38
39
40
Szil ´a gyi37
Nabivanets
Babko
Baillon
2
ꢀ
SO
SO
SO
4
4
4
SO
4
log b1
log b2
log b3
log b1
log b2
2.4
3.6
ND
2.15
ND
2.23
4.12
4.25
ND
0.53
1.08
ND
0.04
ꢀ0.04
0.85 (a)
0.68 (a)
ND
ND
ND
0.99 (b)
0.97 (b)
ND
ND
ND
4ꢀ
HSO
HSO
4
4
HSO
ND
(
0
porosity ¼ 0.2 mm, Pall) previously conditioned by using
In nitrate media, a good signal-to-noise ratio was obtained
ꢀ
1
.73 mol L nitric acid. The titanium concentration in the providing that the nitrate concentration was lower than 2 mmol
ꢀ
1

ltrate was determined by using an Agilent Microwave-Plasma
L
as the nitrate ion exhibits a broad absorption peak at
Atomic Emission Spectrometer (see below). Titanium concen- 200 nm with a molar absorption coefficient approximately equal
ꢀ
1
5
ꢀ1
ꢀ1
tration was found to be equal to 2.3 g L
.
to 10 cm L mol . Therefore, the titanium(IV) solutions were
] ratio was controlled by adding the
Dissolution of titanium(IV) in phosphoric acid. All work was diluted and the [Ti]/[NO
3
carried under a fume hood. Approximately 2.7 g of orthotitanic requested amount of nitric acid in the solutions. The reference
acid was added under agitation to 30 mL of 85% (weight) solutions were prepared by using the same protocol. The blank
ꢀ
1
phosphoric acid (Sigma-Aldrich). Aer one night of agitation, was 0.73 mol L nitric acid, which was ltered by using a 0.2
a small amount of deionized water was added to decrease the mm syringe lter with PES membrane prior to UV-vis analyses.
3
1
31
viscosity. The suspension was ltered under vacuum with
P NMR spectroscopy. P NMR spectra were recorded at
ꢂ
fritted glass lter (Cat. 4) previously conditioned with deionized 161.975 MHz on a Bruker AVANCE III 400 spectrometer at 25 C
water. Titanium concentration in the ltrate was determined by equipped with a 5 mm BBFO broad-band probe. Samples were
using the Agilent Microwave-Plasma Atomic Emission Spec- introduced in a 5 mm NMR tube with a coaxial insert containing
trometer (the ltrate was still slightly turbid). The titanium(IV)
D
2
O as a lock solvent. The D
2
O is separated from the samples in
ꢀ1
concentration was found to be equal to 8.0 g L
Potentiometric titrations. The nitric acid concentration in are referred to external H
the diluted nitric acid stock solution was determined by acquired with 128 scans, 64 k points, a repetition delay of 2 s
.
order to avoid affecting the pH of the solution. Chemical shis
3
1
3
4
PO (0 ppm). P NMR data were
ꢀ
1
potentiometry with 0.1 mol L sodium hydroxide (VWR, AVS and a spectral width of 16 164 Hz.
Titrinorm, ꢄ0.2%). The pH-meter used for the potentiometric
titration was a pHenomenal pH 1100 L (VWR) equipped with
2.2 Computational methods
a pHenomenal pH electrode. The pH electrode was calibrated by
using two buffer solutions at pH ¼ 4 and pH ¼ 7 (VWR, AVS
Chemometric analysis. The R2019b version of Matlab was
Titrinorm). The concentrations of phosphoric acid in the stock used for chemometric analyses of the spectral dataset. This anal-
solution and in the titanium solutions were also determined by ysis relies on the determination of the Malinowski factor indicator
ꢀ
1
42–44
potentiometry with a fresh solution of 0.1 mol L sodium function criterion.
It was used to determine the number of
hydroxide (Merck, EMPROVE Expert ꢄ2.5%). The same pH- chromophores detectable within a given range of concentrations
45
meter and the same electrode were used but three buffer solu- and wavelengths as reported by Meloun et al. This criterion relies
tions at pH ¼ 4, pH ¼ 7 and pH ¼ 9 (VWR, AVS Titrinorm), on a component analysis calculated from the covariance matrix
T
respectively, were used to calibrate the pH electrode.
SS . The eigenvalue vector of this covariance matrix gives infor-
Elemental analyses. Titanium concentrations in the samples mation about the components associated to independent signals
were determined by using an Agilent Microwave-Plasma Atomic in the UV-visible spectra. The Malinowski factor indicator function
Emission Spectrometer (wavelength ¼ 323, 452 nm). The mother IND criterion is calculated by using eqn (4) for different k-values,
standard solution containing 1000 ppm of titanium in 2% (vol) which correspond to the number of chromophores in solution.
nitric acid (Sigma Aldrich, ICP Standard) was diluted in 5% (vol) The number of chromophores is equal to the value of k corre-
nitric acid to prepare ve standards containing 1 to 10 ppm tita- sponding to the minimum value of IND.
nium. In order to minimize possible matrix effects, two different
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
q
P
ꢂ
series of standards were prepared, without or with H PO
3
4
.
g
j
ðnðm ꢀ kÞÞ
j¼kþ1
UV-visible spectroscopy. An Agilent Cary 60 UV-visible spec-
trophotometer was used to record spectra between 190 nm and
INDðkÞ ¼
(4)
2
ðq ꢀ kÞ
400 nm with a step of 0.5 nm. A 2 mm light path cuvette (Agi-
j
where g , n and m denote the j-th eigenvalue of the covariance
lent) was used to analyze the solutions containing less than
ꢀ1
matrix, the number of spectra used in the dataset and the
number of wavelengths, respectively. The q parameter is equal
to the highest value between n and m.
To better identify the number of chromophores in solution
from the spectral data, a common data analysis procedure was
1
mmol L titanium. A 0.1 mm light path demountable cuvette
(
Hellma Analytics) was used to analyze the solutions containing
ꢀ
1
titanium at concentration greater than 1 mmol L
.
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27061

View Article Online
RSC Advances
Paper
used. Numerical noise (jitter) was added to the recorded spectra
in order to erase the contribution of the minor components and
to distribute more uniformly noise in the considered wave-
length range. Although jittering is usually used to reduce over-
plotting and better identify clusters in large datasets, it was
found efficient to identify the number of chromophores.
3
. Results and discussion
3.1 Titanium(IV) speciation in dilute nitric acid (non-
complexing media)
46
3.1.1. UV-visible spectroscopy. UV-visible spectra of
ꢀ1
ꢀ1
0.05 mmol L titanium(IV) in 0.73–2.2 mmol L nitric acid
DFT calculations. Time independent and time dependent were recorded in order to investigate the speciation of titaniu-
density functional theory (DFT and TD-DFT) calculations were m(IV) [Fig. 1a]. It is expected that no complexation of titaniu-
47
performed using the Gaussian package (version 16). The struc- m(IV) occurs in such nitric acid solutions as nitrate ions exhibit
tures considered in this work were optimized geometrically with only a weak tendency to form nitrato complexes with most
4
8,49
65–68
the PBE0 global hybrid functional
and the 6-311+G(d,p) basis metal ions.
For instance, the formation constants of mono
50–52
+
set. The solvation of the rst coordination sphere was manganese and cadmium nitrate complexes (MNO3 with M ¼
described explicitly whereas the outer solvation sphere was Mn or Cd) at 298.6 K are equal to 0.17 and 0.54, respectively.
described by using the PCM model.
67
5
3,54
Frequencies were calcu- Likewise, the formation constants of niobium(III), thorium(IV)
lated on the optimized geometries. Excitation energies were then and uranium(VI) nitrate complexes at 298 K are equal to 0.64 at
ꢀ
1
ꢀ1
determined by TD-DFT with the same level of theory with a linear ionic strength I ¼ 2 mol L , 0.22 (I ¼ 1 mol L ) and 0.24 (I ¼
ꢀ1
65,66,68
response non equilibrated PCM solvation model with water. 1 mol L ), respectively.
In the present work, it can be
69
Unless otherwise stated, a total number of 80 states were expected that only free titanium(IV) exists in solution as nitrate
ꢀ
1
computed in order to get a representative range of the experi- concentration is low, i.e. [HNO ] ¼ 0.73–2.2 mmol L
.
3
mental wavelength where light absorption occurred. The same
type of calculation was performed on a larger system, i.e. a tita- region.
Fig. 1a shows the presence of an intense absorption in the UV
13,14,70
The strong band located between 180 and 220 nm
71
nium oxo-cluster denoted Ti18, and the procedure and parameters (region 1) can be attributed to nitrate ions, and the band which
basis set, number of computed states and solvation model) used appears in the range 220–350 nm (region 2) on Fig. 1b aer
for this system is detailed in the ESI.†
baseline correction may correspond to the presence of one or
(
The use of hybrid functionals as well as the implicit solvation several titanium(IV) complexe(s). In coherence with the above
model PCM was found to be efficient to obtain reliable esti- statement, it is interesting that the intensity of the band located in
mations of the wavelengths of maximal absorbance for metal region 1 strongly and almost linearly increases with the concen-
5
5
56
complexes, molecular systems in complex environments as tration of nitric acid whereas it does not depend on the presence or
57
well as polyoxometalates. This scheme and the pitfalls of the absence of titanium(IV). On the opposite, the intensity of the band
implemented methods have been well-known for a decade. In located in region 2 (Fig. 1b) is independent on the concentration of
58
the current state of the art, the computed excitation energies nitric acid within the experimental uncertainty, which conrms
ꢀ
1
may thus be considered reasonably reliable as illustrated by our the absence of nitrato complexes of Ti(IV) in 0.73–2.2 mmol L
59–61
44
previous works using the same methodology.
HNO in agreement with Szil ´a gyi et al. . The noise observed in
3
The orbital transitions associated with the main predicted Fig. 1b for the shortest wavelengths is due to the strong absor-
excitation energies were described by calculating natural tran- bance of nitrate ions around 200 nm.
62
sition orbitals (NTO).
The large width of the absorbance peak for wavelengths
The structures of the optimized geometries and the representa- between 230 and 275 nm suggests an asymmetry with respects
tion of the Natural Transition Orbitals (NTOs) were generated using to the region between 200 and 230 nm. This may result from the
63
64
Avogadro soware and POV-Ray as post-treatment (ray-tracer).
presence of several titanium(IV) complexes in solution or the
ꢀ
1
ꢀ1
Fig. 1 UV-vis Spectra (after baseline correction) of (a) 0.05 mmol L of titanium(IV) in 0.73–2.2 mmol L nitric acid (solid lines) and aqueous
ꢀ1
solutions of 0.73–2.2 mmol L nitric acid (dotted lines) recorded with a 2 mm light path cuvette, and (b) difference between the spectra of Ti(IV)–
HNO and the spectra of HNO
3
3
.
27062 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
2
m
Fig. 2 Schematic structures suggested for the free form of titanium [(TiO)
2
m
]
considering various sizes, i.e. monomeric (a, m ¼ 1), dimeric (b, m ¼
), trimeric (c, m ¼ 3). Possible intramolecular proton exchanges are represented by dashed double arrows.
presence of a unique titanium(IV) complex exhibiting several
major possible electronic transitions.
3.1.2. DFT calculations. To get information about the
chemical structure and the stoichiometry of the titanium(IV)
In the following part of the paper, titanium(IV) species are written species in non-complexing media, computed transition ener-
2
m
as [(TiO)
ytitanate species as it is not possible to exclude the presence of Functional Density Theory (TDDFT) calculations.
polynuclear species in solution. For m >1, as illustrated in Fig. 2, an
Fig. 3 gathers the different congurations reported in the
m
]
with m ¼ 1 for mononuclear species and m >1 for pol- gies of several complexes were obtained by Time-Dependent
2m
octahedric structure with bonding oxo groups is suggested. A similar literature of free titanium in acidic medium [(TiO)m] (geometry
proton exchange equilibrium as TiO # Ti(OH)
nuclear species might occur in the case of polynuclear species, TiO and Ti(OH)2 cations as well as Ti(OH)3 and Ti(OH) were
2+
2+
2
13,16,18,22
for mono- optimized by DFT calculations in the present work).
Both
2+
2+
+
3+
leading to the formation of hydroxo bridging groups (see Fig. 2).
considered in order to gure out the predominant species in
2+
2+
2+
Fig. 3 Schematic representation of the first hydratation shell and optimized geometries for (a) [Ti(OH)
2
]
(cis) (b) [Ti(OH)
2
]
(trans) (c) [TiO] (d)
+
3+
[
3
Ti(OH) ] (e) [Ti(OH)] . Cartesian coordinates are available as part of the ESI.†
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27063

View Article Online
RSC Advances
Paper
Table 3 Typical bond lengths (in angstroms) and angles (in degrees) of optimized geometries for structures in Fig. 3
2+
2+
2+
3+
+
TiO
Ti(OH)
2
(cis)
Ti(OH)
2
(trans)
Ti(OH)
Ti(OH)
3
Mean Ti–H
Axial Ti–H
2
O distance
O distance (TiO and
2.07
2.29
2.11
n.a.
2.05
n.a.
2.02
2.09
2.22
n.a.
2+
2
3+
Ti(OH)
)
2+
Ti]O, Ti–OH distance (TiO and
1.58
n.a.
n.a.
n.a.
1.68
n.a.
3+
Ti(OH)
)
ꢂ
ꢂ
ꢂ
ꢂ
d
146
173
180
128
Ti ꢀ O ꢀ H angle
Mean Ti–OH distance
n.a.
1.74
1.76
n.a.
1.79
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
d
82 , 97
75 , 105
89 , 91
80 , 100
75 , 105
O ꢀ Ti ꢀ O angle (minimal and
maximal values)
solution. The starting geometries were thus obtained by placing
The structures used in the TD-DFT calculations were local
2+
one to three hydroxyl groups (and oxo group for TiO ) coordinated potential energy minima, as the vibration modes were all
to titanium(IV) cation. The rest of the octohedra sites of the associated with positive eigenvalues aer geometry optimiza-
complexes were lled with water molecules (solvation shell). The tion. The octahedricity was conserved during the optimization
outer solvation shells were described by an implicit model of water even though an important distortion was observed for the
+
2+
molecules (PCM) in accordance with other DFT works, which Ti(OH)
3
and Ti(OH)
2
cis structures as shown in Table 3. The
demonstrated that such kind of models are relevant to describe resulting Ti–(OH
transition metal complexes in aqueous media.
2
), Ti](O) and Ti–(OH) bond distances of the
61,72
Fig. 4 Comparison between predicted excitation energies with their relative associated oscillator strength (absolute values available in ESI†) for
2+
2+
2+
+
3
(
a) [Ti(OH)
2
]
(cis) (b) [Ti(OH)
2
]
(trans) (c) [TiO] (d) [Ti(OH)
3
]
(e) [Ti(OH)] and the spectrum of free titanium (1b).
27064 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
2+

nal geometrically optimized structures reported in Fig. 3 are mononuclear titanium(IV) [Ti(OH)
2
]
in acidic nitrate solutions
gathered in Table 3.
cannot be supported by this study. In other words, the free tita-
The excitation energies of the optimized complexes were nium(IV) complex in non-complexing acidic solutions (perchloric
2+
calculated using TD-DFT (excited states are available in the or nitric acid) may not match with [Ti(OH) ] cation as it was
2
13,15,16,18
ESI†). To do so, the scheme described in the Computational previously stated in the literature.
Section was used to predict UV-visible spectra as its accuracy The formation of polytitanate species is suspected even in
generally allows to compute transitions within 0.25 eV, which is acidic solutions. The computational cost of the level of theory
sufficient in the case of the qualitative approach of this work. chosen to perform the calculation (6-311+G**/PBE0) was
The oscillator strengths associated with the permitted transi- a limiting factor to perform the geometry optimization on large
tions are shown in Fig. 4. This value represents the probability of oxoclusters. Thus, the excitation energies were obtained only for
a transition, and thus it might be compared to an absorbance the dimeric form of titanium aquo-complexes. Different struc-
measurement for the wavelength corresponding to the transition tures containing several hydroxyl groups coordinated to tita-
energy. UV-visible plots were obtained by convolving the transi- nium(IV) atoms were considered. Two bridging oxygen atoms
tions with a Gaussian function (assuming a standard deviation of were inserted between the titanium atoms and their three
0.4 eV) to account for the band shape. The excitation energies and different protonation states were chosen for calculations. The
associated oscillator strengths, as well as the reconstructed spectra of the dimeric structures showed a fair agreement with
spectra were compared to the experimental spectrum of free the experimental spectrum obtained in diluted nitric acid
titanium reported in the rst part of the discussion. For each (Fig. 5a). The best agreement between the experimental and
mononuclear complex considered in these calculations, no calculated UV-visible spectra was obtained for the chemical
agreement was observed between the most relevant transitions structure displayed in Fig. 5b. The 3D representation shows that
and the experimental location of the maximum of absorbance this optimized structure consists of two titanium(IV) atoms
(235 nm). The computed transitions are signicantly blue-shied bonded via a mixed m-(OOH) bridge. The rst coordination
with respect to the experimental data whatever the structure spheres of the two titanium(IV) cations were completed with
considered. Therefore, the presence of any of these divalent water molecule as for the monomeric complexes. However, the
5+
Fig. 5 (a) Predicted transitions and their associated oscillator strengths for a dimeric structure [Ti
spectrum of free titanium calculated in 1b. The convolution of excitation energies by gaussians (wavenumber standard deviation of 0.4 eV) is
plotted with a dotted line. (b) Optimized geometry (upper and lateral view) and schematic representation of a dimeric polytitanate [Ti (m-
2
(m-(OOH))] compared to the experimental
2
5
+
(
OOH))] .
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27065

View Article Online
RSC Advances
Paper
shape of the band between 230 nm and 340 nm is poorly dimer structure, a difference of about 15 nm was found. In
reproduced. It is worth mentioning that a predicted transition comparison to the results obtained for monomeric structures,
associated with a non-negligible oscillator strength is present at this difference between the experimental and the calculated
275 nm as depicted in Fig. 5a. The width of the experimental wavelengths remains acceptable. Therefore, TD-DFT cannot be
peak between 250 nm and 300 nm might be explained by the used to conclude about the exact structure of titanium(IV) in
presence of electronic transitions at similar wavelength asso- nitric acid because of the existence of many structures whose
ciated with a higher oscillator strength.
calculated wavelengths match with experimental wavelengths.
Nevertheless, this study shows that the predicted position of
The main transitions located at 227 nm and 278 nm (the most
intense in their respective wavelength domains) were described with maximal absorbance shis toward higher wavelengths when
the natural transition orbitals (NTO) method (Fig. 6). The identi- the size of the polyoxometalate complex species increases. This
cation of the orbitals involved in the main transitions was found to tendency was conrmed for higher nuclearity as depicted in
be much easier with the use of the NTOs than with the standard Fig. 8, which shows the normalized convolution of the excita-
2
+
canonical orbitals (see ESI†). The rst transition (227 nm, 5.46 eV) tion energies by gaussians for TiO as well as for oxo-bridged
was decomposed into two contributions: (i) an electron transfer from polytitanate species with 2, 3 and 18 titanium(IV) atoms. The
the p-orbital corresponding to non-bonding electrons of oxygen cartesian coordinates of the optimized geometries of [Ti
2
(m-
)] as well as the procedure used to
one of titanium atom, and (ii) an electron transfer from p-orbitals of obtain the excited states for Ti18 are given in the ESI.†
the two bridging oxygens and of the oxygen atoms of coordinated The comparison with the experimental spectrum of titaniu-
4
+
6+
z 2 3 3
atoms from several coordinated water molecules to the d -orbital of (O) )] and [Ti (m-(O)
2
water molecules to the d-orbitals of titanium atoms. The second m(IV) in nitric acid reported in Fig. 1b shows that the maximum
excitation (278, 4.48 eV) was described by a similar transition, i.e. p- of absorbance at approximately 230 nm may be linked to the
orbitals of bridging oxygen atoms to d-orbitals of titanium atom. presence of a “small” polytitanate species (dimeric or trimeric
Both these descriptions support the ligand-to-metal charge transfer complex species). This Figure also shows that the width of the
character of the most intense transitions in the titanium(IV) complex. absorbance band in the 230–320 nm region might be due to the
The same type of calculation has been performed for two presence of other complex species of higher nuclearity such as
28
other titanium(IV) complexes whose chemical structure remains the oxo-clusters suggested by Kozma et al.
close to the previous one: the rst one in which a water mole- However, the comparison between the results obtained for the
cule in the solvation shell was replaced by an hydroxyl group [Ti (m-(OOH)) (OH)] and [Ti (m-(OOH))] (Fig. 5 and 7) and those
4
+
5+
2
2
5+
4+
(
Fig. 7a), and the second one corresponding to a trimer struc- obtained for the [Ti (m-(OOH))] and [Ti (m-(O) )] (Fig. 5 and 8)
2 2 2
ture with bridging oxo groups (Fig. 7b). The main excitation suggests that the presence of hydroxyl groups or water molecules
energy predicted for the trimer structure shows also a good coordinated to titanium(IV) and the protonation state of bridging
agreement with the experimental wavelength. As for the second oxo-groups may inuence the position of the absorbance band.
Fig. 6 Natural transition orbitals generated for the predicted transitions at (a) 278 nm and (b) 227 nm. The digits represent the contribution in
percent of the drawn NTO transitions.
27066 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
4+
2 3
Fig. 7 Predicted transitions and their associated oscillator strength for (a) a dimeric structure [Ti (m-(OOH)) (OH)] (b) a trimeric structure [Ti (m-
6+
(
O)
3
)] compared with the experimental spectrum of free titanium from Fig. 1b. The convolution of excitation energies by gaussians (wave-
number standard deviation of 0.4 eV) is plotted with a dotted line.
3
.2 Titanium(IV) speciation in phosphoric acid solutions
phosphoric acid concentration. In the region d ¼ ꢀ11 to
12.5 ppm, it is difficult to conclude if the two observed weak
ꢀ
When phosphoric acid is added to non-complexing solutions of
2
m
peaks correspond to (i) two different phosphorus complexes or (ii)
a unique titanium–phosphate complex containing two non-
equivalent phosphorus atoms characterized by two chemical
shis. Furthermore, the mean dynamic conformation could also
depend on experimental conditions such as pH and viscosity
correlated with phosphoric acid concentration. Previous works on
the synthesis of amorphous titanium(IV) phosphate reported two
ꢀ1 peaks at similar chemical shis obtained from the deconvolution
of MAS-NMR spectra of synthesized materials (d ¼ ꢀ5.0 to
titanium(IV), it is expected that the [(TiO)m] species identied
above (with m ¼ 1 for mononuclear species and m >1 for poly-
titanate species) will be converted into titanium-phosphate
complexes. The goal of this part is to investigate the forma-
tion of such complexes by using P NMR spectroscopy and
chemometric analyses of UV-visible spectra.
3
1
3
1
31
3.2.1.
P NMR. Fig. 9 shows P NMR spectra of 6.9–
ꢀ
1
12.7 mol L phosphoric acid solutions containing 17 mmol L
titanium(IV). The peaks located between d ¼ ꢀ6.1 and ꢀ6.9 ppm
may be attributed to a titanium–phosphate complex as the cor-
responding integration (aer weighting) depends on the
7
3
ꢀ
5.2 ppm and d ¼ ꢀ13.5 to ꢀ15.5 ppm). These peaks at d ¼ ꢀ5.0
to ꢀ5.2 ppm and d ¼ ꢀ13.5 to ꢀ15.5 ppm were attributed to the
2+
4+
6+
2 2 3 3
Fig. 8 Comparison of the normalized convolution of the excitation energies calculated by TD-DFT for TiO , [Ti (m-(O) )] , [Ti (m-(O) )] and
Ti18 with the experimental UV-vis spectrum of titanium(IV) in nitric acid from Fig. 1b.
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27067

View Article Online
RSC Advances
Paper
2
ꢀ
following types of phosphorous sites: P(OTi)
x
(OH)4ꢀx where x ¼ 1 orthophosphate groups (HPO
4
form) in a titanium(IV) phosphate
ꢀ
2ꢀ
or 2 (H PO or H PO environment) and P(OTi) OH (HPO
complex, possibly as bridging groups between two titanium(IV)
3
4
2
4
3
4
environment), respectively. Similar assignation, i.e. ꢀ4.5 to centers.
ꢀ
ꢀ
7.3 ppm attributed to H PO and ꢀ12.5 to ꢀ15.9 ppm attrib-
3.2.2. UV-visible spectroscopy and chemometric analysis.
2
4
ꢀ
74
75
ꢀ1
uted to HPO
4
were made by Trublet et al. and Zhang et al. for Fig. 10a shows the spectral dataset of 2.6–3.3 mmol L tita-
amorphous titanium(IV) phosphates. The MAS-NMR spectra of nium(IV) in 5.8–11.6 mol L phosphoric acid (aer baseline
ꢀ
1
76
a synthesized oxyuorinated titanium(IV) phosphate led also to correction). These spectra were used to determine the number
the conclusion that the peaks located at ꢀ7.0 ppm and ꢀ11.5 of chromophores detectable between 200 and 400 nm by using
42–45
could be attributed to orthophosphates groups linked to 1 or 2 the Malinowski factor indicator function IND criterion.
The
titanium(IV) centers, respectively. From these assignments, it can calculation of the IND criterion allows to identify the number of
be assumed that the peak observed in this study between ꢀ6.18 independent chromophores k. The orange curve (bottom) in
and ꢀ6.95 ppm may correspond to a complex in which the Fig. 10b shows the variation of this criterion as a function of the
ꢀ
orthophosphate group (in the form of H
PO
3 4
or H PO
2 4
) is linked hypothetical number of chromophores k without any external
to one titanium(IV) atom. For the second region, i.e. ꢀ11.3 to perturbation. This curve does not permit to determine the number
ꢀ
12.3 ppm, the peaks might be attributed to deprotonated of chromophores as it does not exhibit any minimum. Conversely,
3
1
ꢀ1
ꢀ1
ꢀ1
Fig. 9
(
P NMR spectra obtained for solutions with 17 mmol L of titanium and a phosphoric acid concentration of (a) 6.9 mol L (b) 9.8 mol L
ꢀ1
c) 12.7 mol L . The chemical shifts and integration of principal peaks are shown by black digits on top and red digits on the bottom of the spectra
respectively. The digits in dotted boxes are integration values ponderated by the concentration of phosphoric acid.
27068 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
Fig. 10 (a) Spectral dataset (after baseline correction) used for the determination of the number of chromophores (0.1 mm light path cuvette) (b)
Malinowski factor indicator function calculated for different dummy value of chromophores k and different numerical noise addition.
this curve reaches a plateau that can be explained by the presence experimental condition (n) (titanium(IV) and phosphoric acid
of non-uniformly distributed noise as well as by the presence of concentrations as reported for (n) ¼ 1 to 4 in Fig. 10). 3Tilm,1 and
species characterized by UV-visible signals of the same order of
magnitude as the noise. Aer perturbing the signal by adding the species Ti and Ti between 200 and 340 nm.
3
Tilm,2 denote the vectors of the molar extinction coefficients of
l l
m,1
m,2
more noise, the curves exhibit a minimum value for k ¼ 3, which
For the set of experimental condition n ¼ 1 ([Ti(IV)] ¼
ꢀ
1
ꢀ1
corresponds to the number of chromophores in solutions.
0.3 mmol L and [H PO ] ¼ 7.7 mol L ), Fig. 11 shows that the
3
4
l
m,2
The value of the Malinowski indicator function IND for the contribution of Ti
third chromophore without addition of noise is low in experimental conditions, the contribution of Ti in the expres-
was negligible. Therefore, under these
l
m,2
comparison to the value for the rst and second species. Thus, sion of A (1) can be neglected as well. The molar extinction coef-
l
m,2
even though this third species has to be taken into account to cient between 200 and 340 nm for Ti can thus be expressed
study the speciation of titanium(IV) in phosphoric acid, its as:
contribution in the experimental spectra of Fig. 10 is expected to
be negligible. Therefore, this third species will be neglected in
the next part of the present paper.
ꢃ
ꢄ
lm;1
lm;1
1
lm;2
Ti ꢁðnÞ
½Ti
ꢁ
ðnÞ
A
3Til
¼
A
ðnÞ
ꢀ
ð1Þ
(6)
m;2
½
½Ti ꢁð1Þ
ꢀ
1
Fig. 11 shows UV-visible spectra of 0.3 mmol L titanium(IV) in
ꢀ
1
2.0–7.7 mol L
3
H PO4. These spectra were used to nd more
l
m,2
To get information about the peak position Ti , it is therefore
necessary to evaluate the ratio of the concentrations of the rst
species Ti between the two different experimental conditions (n)
accurately the location of the absorbance peaks of the chromo-
phores. The rst peak with a maximum located at lm,1 ¼ 220 nm is
l
m,1
l
m,1
attributed to the presence of a species quoted Ti which might
31
also be at the origin of the P NMR peak observed between d ¼
ꢀ
6.1 and ꢀ6.9 ppm as both the UV-visible absorbance and the
31
ponderate integration values of the P NMR signal exhibit
a dependence versus the phosphoric acid concentration (Fig. 9).
A second band can be observed as a shoulder between 260
and 280 nm in the spectra of Fig. 11. In the following part of this
paper, the Beer–Lambert law has been used to nd the exact
position of this second peak (i.e., lm,2) attributed to the species
l
m,2
Ti
by considering, as stated above, that the contribution to
the total absorbance of third species detected by the Malinowski
indicator function IND was negligible.
The Beer–Lambert law was expressed by considering two
chromophores and a set of wavelengths comprised between 200
and 340 nm:
l
m,1
l
m,2
A
(n) ¼ 3Tilm,1[Ti
]
(n) + 3Tilm,2[Ti
]
(n)
(5)
where A (n) corresponds to the vector of the absorbance
measured between 200 and 340 nm for a given set of phosphoric acid after baseline correction (2 mm light path cuvette).
ꢀ1
ꢀ1
Fig. 11 UV-vis spectra of 0.3 mmol L Ti(IV) in 2.0–7.7 mol L
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27069

View Article Online
RSC Advances
Paper
where 1# m, 0# m #3, 0# n #3, 1# g #4m and 1# b #4m are
integers. As previously stated, polytitanate formation cannot be
excluded as the literature of titanium chemistry reports the
29
ability of titanium(IV) to polymerize.
l
m,2
l
m,1
The two chromophores Ti and Ti are still visible in Fig. 13
for phosphoric acid concentration ranging between 0.8 and
ꢀ
1
1
2
.4 mol L . For instance, two bands located at 220 nm and
60 nm can be clearly observed in the spectrum recorded for
ꢀ
1
[H
3
PO
4
] ¼ 1.4 mol L (curve (5) in Fig. 13). However, another peak
corresponding to another chromophore can be observed at 225 nm
ꢀ
1
when phosphoric concentration was lower than 0.5 mol L
curves (8) and (9) in Fig. 13). The absorbance at this wavelength
increases when phosphoric acid concentration is decreased from
.5 to 0.14 mol L . Such an observation suggests that another
chromophore, Ti , exists at low concentration of H
(
ꢀ
1
0
l
m,3
3 4
PO .
l
m,2
It is interesting to point out that the peak corresponding to this
Fig. 12 Molar extinction coefficient of Ti calculated for a ¼ 0.8, 0.9
l
m,3
third chromophore Ti in the spectra of titanium(IV) in phos-
phoric acid and the peak observed in the spectrum of titanium(IV)
in diluted nitric acid attributed to free titanium(IV) [Fig. 13, black
curve (10)] are located at similar wavelength and exhibit the same
shape. The shi in wavelength (225 nm against 240 nm) may be
and 1 by [eqn (9)].
and (1). This ratio may be expressed in eqn (7b) by using the
expression of the Beer–Lambert law at lm,1 ¼ 220 nm [eqn (7a)]:
lm;1
ðnÞ
lm;1
l
m;1
lm;1
l
l
m;2
l
m,2
A
¼ 3Ti
l
½Ti ꢁðnÞ þ 3Ti
½Ti ꢁðnÞ
(7a) attributed to the presence of the absorbance peak of Ti . Thus,
m;1
m;2
l
m,3
Ti can be identied to free titanium(IV).
lm;1
lm;1
ðnÞ
lm;2
ꢀ
1
The absorbance at 310 nm for 0.3 mmol L
Ti(IV) in
½
½
Ti ꢁðnÞ
A
A
1
3
3
2
1
½Ti ꢁðnÞ
ꢀ
1
¼
a
where ¼ 1 þ
(7b)
lm;1
Ti ꢁð1Þ
lm;1
ð1Þ
lm;1
1.4 mol L
3 4
H PO (Fig. 13, curve (5)) was very low because the
l
a
½Ti ꢁðnÞ
m,2
concentration of Ti
was low. A decrease of phosphoric acid
ꢀ1
concentration from 1.1 to 0.1 mol L is responsible for an
increase of the absorbance at 310 nm of 0.05 AU, while the
absorbance at 260 nm increases of 0.08 AU. The extinction coef-
l
m,2
The coefficient a corresponds to the contribution of Ti in
the absorbance at 220 nm under the experimental condition (n).
By combining eqn (7a) and (7b), the following expression of
the vector P, which is proportional to 3Tilm,2, can be deduced:

cient calculated with eqn (8) shows an absorbance at 260 nm
almost 5 times greater than the absorbance at 310 nm (Fig. 13).
The presence of a third species is therefore obviously responsible
for the increase of absorbance at 310 nm when phosphoric acid
concentration decreases. Along with the position of the absor-
bance peak, this second observation also supports the presence of
lm;1
A
ðnÞ
3
l
fP ¼ AðnÞ ꢀ a
A
ð1Þ
(8)
Ti m;2
lm;1
ð1Þ
A
2m
the free titanium(IV) [(TiO)m] in solution as Fig. 1b shows also
a signicant absorbance at this wavelength.
l
m,2
Fig. 12 shows the variation of P for the species Ti at three
hypothetic values of a as a function of the wavelength (experimental
In summary, the variation of the speciation of titanium(IV)
from non-complexing aqueous solution to concentrated
ꢀ
1
ꢀ1
condition (n) ¼ 4, i.e. 0.3 mmol L Ti(IV) and 2 mol L
3 4
H PO ; this
experimental condition was chosen because the spectra recorded
solution of H
3 4
PO can be explained by considering the two
under the experimental conditions corresponding to (n) ¼ 1 and (n)
following titanium orthophosphate complexes in addition to
¼
4 show a signicant difference between 260–280 nm).
l
m,3
2m
free titanium(IV), Ti , corresponding to [(TiO)m]
.
A maximum of extinction coefficient is clearly observed at
l
m,2
Ti as [(TiO)
2mꢀgm
g
) ]
2
70 nm and can be attributed to the second titanium(IV)
-
m
(H3ꢀmPO
4
l
m,2
complex Ti
concentration of Ti
present in solution. Fig. 11 shows that the
l
m,2
l
m,1
-Ti as [(TiO)
2mꢀgmꢀbn
b
) ]
decreases when phosphoric acid
m
(H3ꢀmPO
4
)
g
(H3ꢀnPO
4
concentration is increased (see region 270–320 nm for the
conditions (1) to (4) in Fig. 11).
These complexes are formed in phosphoric acid according to
the following reactions depending on phosphoric acid
concentration:
Finally, it can be expected that the following reaction takes
place in solution when increasing phosphoric acid concentra-
ꢀ1
tion from 2 to 7 mol L
:
2
mꢀgm
Þ ꢁ2mꢀgmꢀbnþnbH
þ
½
ðTiOÞ ðH3ꢀmPO
4
Þ ꢁ
þ bH
3
PO
4
#½ðTiOÞ ðH3ꢀmPO
4
ÞgðH3ꢀnPO
4
3
O
m
g
m
b
lm;1
(9)
lm;2
Ti
Ti
27070 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
2
m
2mꢀgm
þ
O
½
ðTiOÞ ꢁ þ gH
3
PO
4
#½ðTiOÞ ðH3ꢀmPO
4
Þ ꢁ
þ mgH
3
m
m
g
À
Á
(10)
(11)
ꢀ1
ðnon complexing mediaÞ moderate H
3
PO
4
concentration; . 1 mol L
2
mꢀgm
2mꢀgmꢀbn
þ
O
½
ðTiOÞ ðH3ꢀmPO
4
Þ ꢁ
þ bH
3
PO
4
#½ðTiOÞ ðH3ꢀmPO
4
Þ ðH3ꢀnPO
4
Þ ꢁ
þ nbH
3
m
g
m
g
b
À
Á
ꢀ1
high H PO
4
concentration; . 6 mol L
3
ꢀ1
ꢀ1
Fig. 13 UV-vis spectra of solutions with 0.3 mmol L Ti(IV) and a phosphoric acid concentration of 0.1–1.4 mol L after baseline correction
(
2 mm light path cuvette).
calculated UV-visible spectra. However, in spite of such a good
where 1# m, 0# m #3 and 1# g #4m are integers.
tting, the determination of the exact nature of the observed
In these equilibria, the size m of the complex is kept constant polytitanates (i.e., size and hydrolysis state) was not possible
for simplication, but it might evolve with the phosphoric acid due to the multiplicity and similarity of corresponding
concentration.
structures.
UV-visible spectroscopy coupled with chemometric analyses
and P NMR spectroscopy were also successfully used to
qualitatively investigate the speciation of titanium(IV) in a large
3
1
4
. Conclusion
range of phosphoric acid concentrations (i.e., 0.1 to
The use of UV-visible spectroscopy and molecular modeling
techniques was found to be relevant to reinvestigate the speci-
ation of titanium in non-complexing acidic solutions. Indeed,
mononuclear species such as titanyl ion or titanium dihydr-
oxide which were considered in previous studies to be
predominant in non complexing acidic media were found not to
be sufficient to correctly simulate UV-visible spectra by DFT
calculations. On the opposite, taking into account the existence
of small polytitanate clusters as dimeric or trimeric structures
lead to a satisfactory match between experimental and DFT
ꢀ1
1
1.6 mol L ). Thus, it is concluded that titanium(IV) is present
as mononuclear (m ¼ 1) or polynuclear (m >1) free cation(s),
2
m
ꢀ1
[
m
3 4
(TiO) ] , in diluted H PO solutions (typically <0.1 mol L )
and then progressively forms orthophosphate complexes as the
concentration of phosphoric acid is increased, namely
2
mꢀgm
[
[
>
(TiO) (H
PO ) ]
and
m
3ꢀm
4 g
2mꢀgmꢀbn
(TiO)
m
(H3ꢀmPO
4
)
g
(H3ꢀnPO
4
)
b
]
in moderate (typically
ꢀ1
ꢀ1
1 mol L ) and concentrated (typically >6 mol L ) H
3
PO
4
solutions, respectively.
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27071

View Article Online
RSC Advances
Paper
A further step in the study of the speciation of titanium(IV) in 11 U. K ¨o lle and P. K ¨o lle, Angew. Chem., Int. Ed., 2003, 42, 4540–
H PO solution might be to perform TD-DFT calculation as the 4542.
one reported above for diluted HNO3 solutions (non- 12 I. Cserveny ´a k, G. H. Kelsall and W. Wang, Electrochim. Acta,
complexing media). However, due to the large number of 1996, 41, 563–572.
possible structures (i.e., stoichiometry or polytitanate size) and 13 T. Sugimoto, X. Zhou and A. Muramatsu, J. Colloid Interface
3
4
necessary long computing time, such calculations were beyond
the scope of the present paper.
Sci., 2002, 252, 339–346.
14 I. Szil ´a gyi, E. K ¨o nigsberger and P. M. May, Dalton Trans.,
2
009, 7717–7724.
5 P. Comba and A. Merbach, Inorg. Chem., 1987, 26, 1315–
323.
6 A. Liberti, V. Chiantella and F. Corigliano, J. Inorg. Nucl.
Chem., 1963, 25, 415–427.
7 S. E. Ziemniak, M. E. Jones and K. E. S. Combs, J. Solution
Chem., 1993, 22, 601–623.
8 J. Schmidt and W. Vogelsberger, J. Solution Chem., 2009, 38,
1
1
1
1
1
2
Author contribution
1
This work was conceptualized by L. M., A. C., G. C. and H. H.
The access to the calculation resources was made possible by
A. M. The investigation process was performed by L. M., A. M.
and S. L. The original dra of the manuscript was written
by L. M., A. C. and G. C. All the authors contributed to edit and
review the nal version of the manuscript. G. C., H. H. and A. C.
were involved in the funding acquisition and supervision of the
project.
1267–1282.
9 V. I. Ivanenko, I. A. Udalova, E. P. Lokshin and V. I. Kravtsov,
Russ. J. Electrochem., 2001, 37, 530–535.
0 M. Pourbaix, Atlas of electrochemical equilibria in aqueous
solutions, National Association of Corrosion Engineers,
Houston, Texas, 1974.
Conflicts of interest
2
1 A. Pichavant, E. Provost, W. F u¨ rst and J.-F. Hochepied, J.
Chem. Thermodyn., 2019, 131, 184–191.
There are no conicts to declare.
2
2 J.-P. Jolivet and M. Henry, De la solution a` l’oxyde -
Condensation des cations en solution aqueuse. Chimie de
surface des oxyde, EDP Sciences, Paris, 2012.
Acknowledgements
M. Lucas Mangold acknowledges CIFRE program (ANRT) for
2
2
2
2
2
3 B. I. Nabivanets and V. V. Lukachina, Ukr. Khim. Zh., 1964,

nancial support. The authors acknowledge the Plateforme de
30, 1123.
RMN de l’Institut Jean Barriol, Universit ´e de Lorraine for the
NMR analysis. This work was granted access to the HPC
resources of CINES (Centre Informatique National de l’En-
seignement Superieur) and IDRIS (Institut du Developpement
et des Ressources en Informatique Scientique) under the
allocations A0070806933 and A0090806933 made by GENCI
4 F. I. Lobanov, V. M. Savostina, I. V. Serzhenko and
V. M. Peshkova, Russ. J. Inorg. Chem., 1969, 14, 562.
5 V. A. Nazarenko, V. P. Antonovich and E. M. Nevskaya, Russ.
J. Inorg. Chem., 1971, 16, 530–533.
6 M. W. Chase, J. L. Curnutt, H. Prophet, R. A. McDonald and
A. N. Syverud, J. Phys. Chem. Ref. Data, 1975, 4, 1–176.
7 A. Testino, I. R. Bellobono, V. Buscaglia, C. Canevali,
M. D'Arienzo, S. Polizzi, R. Scotti and F. Morazzoni, J. Am.
Chem. Soc., 2007, 129, 3564–3575.
(Grand Equipement National de Calcul Intensif).
References
28 K. Kozma, M. Wang, P. I. Molina, N. P. Martin, Z. Feng and
1
2
3
Council of Europe, European Pharmacopoeia, 10th edn,
Council of Europe, Strasbourg, 2019, p. 3553.
U.S. Pharmacopeial Convention, Food Chemicals Codex
M. Nyman, Dalton Trans., 2019, 48, 11086–11093.
29 G. Zhang, J. Hou, C.-H. Tung and Y. Wang, Inorg. Chem.,
2016, 55, 3212–3214.
30 G. Zhang, C. Liu, D.-L. Long, L. Cronin, C.-H. Tung and
Y. Wang, J. Am. Chem. Soc., 2016, 138, 11097–11100.
31 K. Manseki, K. Saka, M. Matsui, S. Vafaei and T. Sugiura,
CrystEngComm, 2017, 19, 5844–5848.
(
FCC), 11th edn, Rockville, MD, 2019, p. 927.
FAO-JEFCA Phosphoric acid monograph, http://www.fao.org/
leadmin/user_upload/jecfa_additives/docs/Monograph1/
Additive-312.pdf, (accessed 2021).

4
5
6
M. C. Assuncao, G. Cote, M. Andre, H. Halleux and 32 H. Einaga, J. Chem. Soc., Dalton Trans., 1979, 1917–1919.
A. Chagnes, RSC Adv., 2017, 7, 6922–6930.
33 F. Babonneau, S. Doeuff, A. Leaustic, C. Sanchez, C. Cartier
and M. Verdaguer, Inorg. Chem., 1988, 27, 3166–3172.
34 M. Czakler, C. Artner and U. Schubert, New J. Chem., 2018,
42, 12098–12103.
35 S. E. Ziemniak and E. P. Opalka, Chem. Mater., 1993, 5, 690–
694.
I. M. Ahmed, R. N. Ammanoeil, E. A. Saad and J. A. Daoud,
Period. Polytech., Chem. Eng., 2019, 63, 122–129.
F. Ruiz, A. Marcilla, A. M. Ancheta and J. A. Caro, Solvent Extr.
Ion Exch., 1985, 3, 331–343.
7
8
T. A. Williams, US Pat., US4127640A, 1978.
A. Hannachi, D. Habaili, C. Chtara and A. Ratel, Sep. Purif. 36 Y. Komatsu and H. Einaga, J. Inorg. Nucl. Chem., 1981, 43,
Technol., 2007, 55, 212–216. 2449–2454.
R. Hamdi and A. Hannachi, J. Chem. Eng. Process Technol., 37 I. Szil ´a gyi, E. K ¨o nigsberger and P. M. May, Inorg. Chem.,
012, 3(2), DOI: 10.4172/2157-7048.1000134. 2009, 48, 2200–2204.
0 M. Alibrahim, Period. Polytech., Chem. Eng., 2007, 51, 39–42. 38 B. I. Nabivanets, Russ. J. Inorg. Chem., 1969, 7, 352.
9
2
1
27072 | RSC Adv., 2021, 11, 27059–27073
© 2021 The Author(s). Published by the Royal Society of Chemistry

View Article Online
Paper
RSC Advances
3
4
4
9 A. K. Babko, E. A. Mazurenko and B. I. Nabivanets, Russ. J. 54 J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005,
Inorg. Chem., 1969, 14, 1091. 105, 2999–3094.
0 F. Baillon, E. Provost and W. F u¨ rst, J. Mol. Liq., 2008, 143, 8– 55 C. Latouche, D. Skouteris, F. Palazzetti and V. Barone, J.
2. Chem. Theory Comput., 2015, 11, 3281–3289.
1 F. A. Deorsola and D. Vallauri, J. Mater. Sci., 2008, 43, 3274– 56 D. Jacquemin, B. Mennucci and C. Adamo, Phys. Chem.
1
3278.
Chem. Phys., 2011, 13, 16987–16998.
4
4
4
2 E. R. Malinowski, Anal. Chem., 1977, 49, 612–617.
3 S. Pack, J. Chemom., 1991, 5, 545.
57 D. Ravelli, D. Dondi, M. Fagnoni, A. Albini and A. Bagno, J.
Comput. Chem., 2011, 32, 2983–2987.
4 P. W. Yendle and H. J. H. MacFie, J. Chemom., 1989, 3, 589– 58 C. Adamo and D. Jacquemin, Chem. Soc. Rev., 2013, 42, 845–
00. 856.
5 M. Meloun, J. Capek, P. Miksık and R. G. Brereton, Anal. 59 G. J.-P. Deblonde, A. Moncomble, G. Cote, S. Belair and
Chim. Acta, 2000, 423, 51–68. A. Chagnes, RSC Adv., 2014, 5, 7619–7627.
6 J. M. Chambers, W. S. Cleveland, B. Kleiner and P. A. Tukey, 60 D. J. Thaviligadu, L. Labarri `e re, A. Moncomble and
6
ˇ
ˇ´
´
4
4
Graphical Methods for Data Analysis, Chapman and Hall/CRC,
018.
7 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
J.-P. Cornard, Spectrochim. Acta, Part A, 2020, 225, 117492.
61 L. Labarri `e re, A. Moncomble and J.-P. Cornard, RSC Adv.,
2020, 10, 35017–35030.
2
4
M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, 62 R. L. Martin, J. Chem. Phys., 2003, 118, 4775–4777.
G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, 63 M. D. Hanwell, D. E. Curtis, D. C. Lonie, T. Vandermeersch,
A. V. Marenich, J. Bloino, B. G. Janesko, R. Gomperts,
B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov,
E. Zurek and G. R. Hutchison, J. Cheminf., 2012, 4(17), DOI:
10.1186/1758-2946-4-17.
J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, 64 POV-Team, The persistence of vision raytracer Pty. Ltd. POV-
F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, Ray 3.7, 2013.
D. Ranasinghe, V. G. Zakrzewski, J. Gao, N. Rega, 65 P. D. Bernardo, P. Zanonato, L. Rao, A. Bismondo and
G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, F. Endrizzi, Dalton Trans., 2011, 40, 9101–9105.
R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, 66 L. Rao and G. Tian, Inorg. Chem., 2009, 48, 964–970.
O. Kitao, H. Nakai, T. Vreven, K. Throssell, 67 D. F. C. Morris, J. D. MacCarthy and R. J. Newton,
J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro,
M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, 68 L. Rao and G. Tian, J. Chem. Thermodyn., 2008, 40, 1001–
V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, 1006.
K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, 69 S. E. Rizk, M. I. Aly and J. A. Daoud, Sep. Sci. Technol., 2017,
J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, 52, 1206–1215.
R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, 70 K. Takamura and T. Matsumoto, Dalton Trans., 2020, 49,
Electrochim. Acta, 1978, 23, 1383–1386.
O. Farkas, J. B. Foresman and D. J. Fox, Gaussian 16
Revision C.01, 2016.
8 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,
690–696.
71 B. G. Ershov and N. M. Panich, Spectrochim. Acta, Part A,
2018, 188, 179–182.
4
4
5
5
5
5
1
996, 77, 3865–3868.
9 C. Adamo and V. Barone, J. Chem. Phys., 1999, 110, 6158–
170.
72 S. Boumendil, J.-P. Cornard, M. Sekkal-Rahal and
A. Moncomble, Chem. Phys. Lett., 2015, 636, 39–45.
73 M. V. Maslova, D. Rusanova, V. Naydenov, O. N. Antzutkin
and L. G. Gerasimova, J. Non-Cryst. Solids, 2012, 358, 2943–
2950.
6
0 R. Krishnan, J. S. Binkley, R. Seeger and J. A. Pople, J. Chem.
Phys., 1980, 72, 650–654.
1 A. D. McLean and G. S. Chandler, J. Chem. Phys., 1980, 72, 74 M. Trublet, M. V. Maslova, D. Rusanova and O. N. Antzutkin,
639–5648. Mater. Chem. Phys., 2016, 183, 467–475.
2 P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, 1973, 28, 75 W. Zhang, R. Koivula, E. Wiikinkoski, J. Xu, S. Hietala,
5
2
13–222.
3 M. Cossi, G. Scalmani, N. Rega and V. Barone, J. Chem. Phys.,
002, 117, 43–54.
J. Lehto and R. Harjula, ACS Sustainable Chem. Eng., 2017,
5, 3103–3114.
76 P. Bhanja, S. Chatterjee, A. K. Patra and A. Bhaumik, J.
Colloid Interface Sci., 2018, 511, 92–100.
2
©
2021 The Author(s). Published by the Royal Society of Chemistry
RSC Adv., 2021, 11, 27059–27073 | 27073