Molecular structures of seven methylenedinitramines with plasticising abilities

Article abstract of DOI:10.1016/S0022-2860(02)00475-1

Crystal structures of seven methylenedinitramines with plasticising ability have been determined. These molecules, appropriate as energetic plasticisers of propellants, are used to give an environmental temperature independent combustion. The dinitramines are synthesized by condensation reactions of nitramines with formaldehyde in an acidic medium. This work gives their melting point and their melting enthalpy. Glass transition temperatures are also measured.

Full text of DOI:10.1016/S0022-2860(02)00475-1

Journal of Molecular Structure 644 (2003) 37–48
www.elsevier.com/locate/molstruc
Molecular structures of seven methylenedinitramines
with plasticising abilities
a
Denis Spitzer , Bernd Wanders , Michael Richard Schafer , Richard Welter
a
a
b,
*
¨
a
´ ´
French-German Institut of Saint Louis (ISL), 5, rue du General Cassagnou, 68301 Saint-Louis, France
b
´
Laboratoire DECMET UMR 7513, Universite Louis Pasteur de Strasbourg I, 4, rue Blaise Pascal, 67000 Strasbourg Cedex, France
Received 5 June 2002; revised 8 August 2002; accepted 20 August 2002
Abstract
Crystal structures of seven methylenedinitramines with plasticising ability have been determined. These molecules,
appropriate as energetic plasticisers of propellants, are used to give an environmental temperature independent combustion. The
dinitramines are synthesized by condensation reactions of nitramines with formaldehyde in an acidic medium. This work gives
their melting point and their melting enthalpy. Glass transition temperatures are also measured.
q 2002 Elsevier Science B.V. All rights reserved.
Keywords: Methylenedinitramines; Molecular structure; Plasticisers; Propellant; nitrocellulose
1. Introduction
surface. The final result is an uncontrolled combustion
leading to anomalous combustion pressures.
The present study concerns particularly barrelled
weapon propellants for small calibre (0.02 m),
medium calibre (0.09 m) and large calibre (0.12 m).
A typical propellant includes generally a polymeric
binder which is the structural base. It is generally
nitrocellulose [1]. Additionally to the stabilizer, which
prevents the nitrocellulose chemical degradation,
other high energetic charges are added to guarantee
the energetic level of the propellant, also it is common
to add a plasticiser which is of great importance.
Plasticisers had in the past the only function to
facilitate the propellant manufacturing. The best
known or the most employed plasticisers are of
course nitroglycerine (NGL or NG) and diethylene-
glycoldinitrate (DEGDN or DEGN). Nitroglycerine is
a high energetic nitrate ester [2,3] that is still
the most used component for the gelatinisation of
Propellants for barrelled weapons as solid proper-
gols are used in several temperature conditions. Thus,
a weapon can be fired after the propellant was
submitted to extreme temperatures like 240 or
þ50 8C. The main result is a loss in the shoot
precision due to the degradation of the material by its
cracking under low temperatures. The incidence is a
lack of the combustion surface control. The same
inconvenience can take place in the rocket or booster
launcher field by leading to a damage of the booster
matrix by its cracking. These defects then produce a
tremendous uncontrolled increasing of the burning
*
Corresponding author. Tel.: þ33-3-90-24-15-93; fax: þ33-3-
90-24-12-32.
E-mail address: welter@chimie.u-strasbg.fr (R. Welter).
0022-2860/03/$ - see front matter q 2002 Elsevier Science B.V. All rights reserved.
PII: S0022-2860(02)00475-1

38
D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
Fig. 1. Chemical structure of dinitramines.
nitrocellulose. About 70% of the propellant world
production is based on nitrocellulose/nitroglycerine
propellants. Used to establish chemical bonding
probably mostly done through solvatation process of
the nitrocellulose, nitroglycerine gelatinises the nitro-
cellulose to build mechanically stable propellants.
Since about 10 years, studies are conducted to
develop plasticisers which can induce a temperature
independent combustion of propellants. Thus,
Langlotz et al. [4] included in nitrocellulose and RDX
based propellants, a mixture of three dinitramines
namedDNDA57,andsynthesisedbyTartakovskiietal.

D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
39
Fig. 2. Synthesis of symmetrical dinitramines by the self-condensation of nitramines.
[5] in 1993. This mixture induces a temperature
independent combustion. Since this time, studies
made by the French German Institute of Saint-Louis
(ISL) have shown that this result can be obtained only
with a few set of well defined quantities of
each dinitramine in this ternary mixture. The
DNDA57 mixture is composed of three different
types of molecules. These molecules are the 2,4-
dinitro-2,4-diazapentane (or DNDA5), the 2,4-dinitro-
2,4-diazahexane (or DNDA6) and the 3,5-dinitro-3,5-
diazaheptane (or DNDA7). In this work, we have also
studied the crystal structures of analogous dinitra-
mines: the 5-methyl-2,4-dinitro-2,4-diazahexane
(or MDNH), the 2,6-dimethyl-3,5-dinitro-3,5-diaza-
heptane (or DMDNH), the 2,4-dinitro-2,4-diazahep-
tane (or DNH) and the 5,7-dinitro-5,7-diazaundecane
(DNU). The chemical structures of these seven
molecules are given in Fig. 1.
respectively with formaldehyde in an acidic medium
according to the reaction schemes given in Fig. 2. The
three unsymmetrical dinitramines DNDA6, MDNH
and DNH were obtained by a chromatographic
separation technique starting respectively from the
ternary mixtures DNDA57, DN57 and DN59, respect-
ively. The DNDA57 mixture is obtained by the
condensation of methylnitramine with ethylnitramine.
The DN57 mixture is obtained by the condensation of
methylnitramine with isopropylnitramine. The DN59
mixture is obtained by the condensation of methylni-
tramine with N-propylnitramine. All these conden-
sations are done with formaldehyde in an acidic
medium. The synthetic pathways to these ternary
mixtures are given in Fig. 3.
For the crystal structure analysis, each compound
has been crystallised from ethanol in order to obtain a
purity near to 100% (verified by HPLC).
2. Experimental
2.2. Thermal properties
Table 1 resumes the principal thermal parameters
measured of these seven dinitramine compounds [6].
The obtained values are characteristic for plasticiser
compounds. We have in fact compounds with relative
low melting points. In the case of the compounds
DNDA6 and DNH, the melting points are very low
and near the room temperature. In addition, their glass
2.1. Synthesis
All the described compounds have been synthesized
at ISL [6,7]. The symmetrical compounds DNDA5,
DNDA7, DMDNH and DNU were prepared by
self-condensation reactions of methylnitramine, ethyl
nitramine, isopropylnitramine and buthylnitramine,

40
D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
Fig. 3. Synthesis of ternary mixtures of dinitramines.
transition temperatures of about 250 8C are also very
low.
DNDA5 and MDNH. Thus, the occurrence of
amorphous mixtures over very large composition
ranges extending from 30 to 70 mol% of one
component has been shown. The stability in depen-
dence of time of such mixtures was also shown.
‘Amorphous mixtures’ means that in this case no
melting or crystallization signal was detected, and this
in a temperature range from 250 8C (corresponding
to the glass transition temperature) to þ100 8C.
Studies made by ISL in order to explain the
working mechanism of the DNDA57 plasticiser, were
based on the determination of the binary and ternary
phase diagrammes [8] and the measurement of the
combustion in a closed vessel [9]. They revealed
the existence of particular interactions between the
molecules DNDA5 and DNDA6 and between

D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
41
Table 1
Supplementary publication no. CCDC 186129
(DNDA7), 186130 (DNDA6), 186131 (DNDA5),
186132 (DNU), 186133 (DNH), 186134 (MDNH)
& 186326 (DMDNH). Copies of the data can be
obtained free of charge on application to CCDC, 12
Union Road, Cambridge CB2 1EZ, UK (fax: þ44-
1223-336-033; e-mail: deposit@ccdc.cam.ac.uk).
Thermal data of dinitramines
Compound Melting point Melting enthalpy Glass transition
(8C)
(J/g)
(8C)
DNDA5
DNDA6
DNDA7
MDNH
DMDNH
DNH
54.4
31.6
75
102.3
108.2
163.9
106.3
127
244.4
258.3
210.1
No
61.5
85.1
36
No
3. Discussion
110
257.6
No
DNU
72
125
The discussion is focused on to the thermal
parameters (melting temperature, melting enthalpy)
of the different methylenedinitramines, the bonding
length and the angles for each structure.
The origin of these interactions is actually studied
using the modulated DSC technique. Another way for
a better understanding of the crystallisation process of
these mixtures is to examine the intermolecular
interactions. Work concerning these aspects is
actually in progress, but remains confidential.
3.1. Evolution of the thermal parameters
The melting is discussed here because it is an
indication of the crystal cohesion. In Table 1, we
mentioned the melting point of the methylenedini-
tramines. This evolution is shown once again in Fig. 6.
It is known that the melting temperature increases
with increasing molecule weight. We noticed that
there are some exceptions to this rule. These
exceptions are the unsymmetrical linear dinitramines
labelled DND6, DNH, and probably MDNH. This
characteristic is particularly visible in the case of the
dinitramines DNDA6 and DNH which have low
melting temperatures. It can be explained by the fact
that in the case of these structures (DNDA6, DNH,
MDNH), there are very few hydrogen bonds leading
to a very low crystal cohesion. So, for DNDA6 we did
not detect any H-bond in the elementary cell, whereas
in the case of DNH and MDNH the H-type bonds are
rather weak.
2.3. X-ray structure analysis
The selected crystals have been mounted on a
Nonius Kappa-CCD area detector diffractometer
˚
(Mo Ka, l ¼ 0.71073 A). The conditions of data
collection (Denzo software [10,11]) and structure
refinements are gathered in Tables 2 and 3. The cell
parameters were determined from reflections taken
from one set of ten frames (1.08 steps in phi angle),
each at 20 s exposure. The structures were solved
using direct methods (SIR97 [12]) and refined against
F 2 using the SHELXL97 software [13]. No absorption
correction was carried out. All non-hydrogen atoms
were refined anisotropically. Hydrogen atoms were
generated according to stereo-chemistry and refined
using a riding model in SHELXL97.
Zeman et al. [15] studied the thermal parameters
of several other methylenenitramines including the
DNDA5 molecule. He proposed two relations
predicting respectively the melting temperature
(MpNA) and the melting enthalpy (DHNA) of the
methylenenitramines derived both respectively from
the melting temperature (MpHMP) and the melting
enthalpy (DHHMP) of the corresponding aliphatic
structural analogues (i.e. homomorphs). He
measured the parameters using DSC by applying
a heating rate of 20 8C/min. He found the following
relations for linear methylenenitramines:
The ORTEP plots [14] of the molecular structure
are drawn on Figs. 4 and 5. Selected geometric
parameters for the seven molecules are reported in
Table 4.
For all the studied compounds, no classical
hydrogen bonds have been detected. However, it is
worth to notice that only the compound DNDA5
has two unique molecules in the asymmetric unit.
Crystallographic data (excluding structure fac-
tors) for these seven structures have been deposited
in the Cambridge Crystallographic Data Centre as

Table 2
Crystal data and details of the structure determination for DMDNH, MDNH, DNH and DNU (all crystals are colourless)
Crystal data
DMDNH
MDNH
DNH
DNU
Formula
C7H16N4O4
220.24
C5H12N4O4
192.19
C5H12N4O4
192.19
C9H20N4O4
248.29
Formula weight (g mol21
Crystal system
Space group
)
Monoclinic
P21 (No. 4)
7.898(5)
6.167(5)
10.933(5)
90
Orthorhombic
Pna21 (No. 33)
13.244(5)
9.470(5)
7.144(5)
90
Monoclinic
P21=c (No. 14)
5.985(5)
15.126(5)
10.521(5)
90
Monoclinic
C2=c (No. 15)
26.817(5)
9.413(5)
11.155(5)
90
˚
a (A)
˚
b (A)
˚
c (A)
a (8)
b (8)
g (8)
97.18(1)
90
90
108.845(5)
90
108.155(5)
90
90
3
˚
V (A )
528.3(6)
2
896.0(9)
4
901.4(9)
4
2675.7(19)
8
Z
Density (obs) (g cm23
)
1.323
1.396
–
–
Density (calc) (g cm23
)
1.384
1.425
1.416
1.233
m (Mo Ka) (mm21
F(000)
)
0.113
0.122
0.121
0.097
236
408
408
1072
Crystal size (mm)
0.08 £ 0.10 £ 0.13
0.09 £ 0.09 £ 0.11
0.10 £ 0.08 £ 0.12
0.15 £ 0.06 £ 0.09
Data collection
Temperature (K)
˚
Radiation (A)
183
183
183
183
Mo Ka 2 0.71069
2.6, 25.0
Mo Ka 2 0.71069
2.6, 25.0
Mo Ka 2 0.71069
2.5, 27.4
Mo Ka 2 0.71069
2.3, 27.5
u Min–Max (8)
Dataset ðh; k; lÞ
27/9; 26/7; 212/11
4060, 1777, 0.027
1632
213/17; 212/10; 28/9
6161, 1952, 0.036
1568
27/5; 218/19; 210/13
6517, 2045, 0.040
1353
234/33; 0/12; 0/14
2947, 2946, 0.034
1875
Tot., Uniq. Data, R(int)
Observed data ½I . 2sðIފ
Refinement
N reflections, N parameters
R, wR2, Goof
1777, 136
1952, 118
2045, 118
2946, 154
0.0357, 0.0933, 1.10
0.0395, 0.0924, 1.03
0.0555, 0.1671, 1.04
0.0539, 0.1628, 0.94
P ¼ (Fo2 þ 2Fc2)/3
w ¼ 1/[s 2(Fo2) þ (0.0464P)2 þ 0.0392P] w ¼ 1/[s 2(Fo2) þ (0.0503P)2] w ¼ 1/[s 2(Fo2) þ (0.0917P)2 w ¼ 1/[s 2(Fo2) þ (0.0846P)2]
þ 0.1179P ]
Max. and Av. shift/error
0.01, 0.00
0.01, 0.00
0.90(12)
0.00, 0.00
–
0.00, 0.00
–
Flack x 20.20(14)
Min., Max. Resd. Dens. (e/A ) 20.29, 0.22
3
˚
20.31, 0.19
20.29, 0.34
20.23, 0.23

Table 3
Crystal data and details of the structure determination for DNDA5, DNDA6 and DNDA7 (all crystals are colourless)
Crystal data
DNDA5
DNDA6
DNDA7
Formula
C3H8N4O4
164.13
C4H10N4O4
178.16
C5H12N4O4
192.19
Formula weight (g mol21
Crystal system
Space group
)
Orthorhombic
Fdd2 (No. 43)
20.815(5)
43.427(5)
6.236(5)
90
Monoclinic
P21=c (No. 14)
8.076(5)
10.827(5)
9.528(5)
90
Monoclinic
P21=c (No. 14)
8.531(5)
5.774(5)
18.500(5)
90
˚
a (A)
˚
b (A)
˚
c (A)
a(8)
b (8)
g (8)
90
101.27(1)
90
100.821(5)
90
90
3
˚
V (A )
5637(5)
32
817.0(8)
4
895.1(10)
4
Z
Density (obs) (g cm23
)
1.498
1.419
1.384
Density (calc) (g cm23
)
1.547
1.448
1.426
m (Mo Ka) (mm21
F(000)
)
0.141
0.128
0.122
2752
376
408
Crystal size [mm]
0.11 £ 0.11 £ 0.12
0.09 £ 0.09 £ 0.11
0.10 £ 0.08 £ 0.12
Data collection
Temperature (K)
˚
Radiation (A)
183
183
183
Mo Ka 2 0.71069
1.9, 27.5
Mo Ka 2 0.71069
2.9, 27.5
Mo Ka 2 0.71069
2.2, 40.3
u Min–Max (8)
Dataset ½h; k; lŠ
226/20; 256/52; 28/7
9714, 3123, 0.038
2207
210:10; 0:13; 0:12
1846, 1845, 0.020
1416
215/14; 28/10; 222/33
10250, 5559, 0.045
2592
Tot., Uniq. Data, R(int)
Observed data ½I . 2sðIފ
Refinement
Nreflections, Nparameters
3123, 199
1845, 110
5559, 118
R, wR2, Goof
0.0553, 0.1560, 1.03
w ¼ 1/[s 2(Fo2) þ (0.0957P )2]
0.00, 0.00
0.0499, 0.1481, 1.02
w ¼ 1/[s 2(Fo2) þ (0.0809P )2 þ 0.1530P]
0.0564, 0.1664, 1.00
w ¼ 1/[s 2(Fo2) þ (0.0620P)2 þ 0.0302P2]
P ¼ (Fo2 þ 2Fc2)/3
Max. and Av. Shift/Error
Flack x
0.00, 0.00
–
0.00, 0.00
–
0.10(19)
3
˚
Min, Max. Resd. Dens. (e/A )
20.31, 0.30
20.30, 0.44
20.46, 0.31

44
D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
Fig. 4. Molecular structure of DMDNH (a), MDNH (b), DNH (c) and DNU (d); ORTEP plot with atom numbering. Thermal ellipsoids are
drawn at 50% probability level.

D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
45
Fig. 5. Molecular structure of DNDA5 (a), DNDA6 (b) and DNDA7 (c); ORTEP plot with atom numbering. Thermal ellipsoids are drawn at
50% probability level.

46
D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
–
Melting enthalpy (in kJ mol21):
DHNA ¼ 2:5071DHHMP 2 3:3197;
correlation : 0:9980
–
Melting temperature (onset) (in 8C):
lnMpNA ¼ 0:7248lnMpHMP þ 2:2408;
correlation : 0:9490
These equations were used to verify the accuracy
of the so calculated thermal parameters with the
experimental values of the melting enthalpy and the
melting temperature of the DNDA5, DNDA6,
DNDA7, MDNH, DMDNH, DNH, and DNU
methylenedinitramines.
Table 5 shows the result of this comparison where
in the case of the symmetrical methylenedinitramines
(DNDA5, DNDA7) the accuracy of the melting
enthalpy is good once the information is obtained.
However, the relation established by Zeman is not
valid for unsymmetrical methylenedinitramines
(DNDA6, DNH).
Concerning the melting temperature, the relation-
ship given by Zeman cannot be used for the
compounds studied in the present work, excepted for
DNDA5.
3.2. Evolution of the bonding length
Analysing the two middle CN length, it is first
interesting to note that the symmetrical DNDA5
molecule has two central C–N covalent bonds with
˚
two different length (1.436 and 1.452 A). In the
case of the two unsymmetrical molecules (DNDA6,
DNH), these two length are different as expected
˚
(DNDA6: 1.447 and 1.453 A; DNH: 1.458 and
˚
1.443 A). For the unsymmetrical MDNH molecule,
this difference is not revealed, as the difference
between the two lengths is the measuring accuracy
of the experiment.
The NC bond outside the NCN dinitraza bridge
˚
has about the same value (range: 1.449–1.465 A),
except for the DMDNH molecule for which it is
˚
rather long (1.493 A).
The NO bond length, and the NN bond length are
˚
about the same for all the molecules (1.225–1.235 A)
˚
and (1.340–1.355 A), respectively.

D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
47
Fig. 6. The methylenedinitramine melting temperatures.
3.3. Evolution of the bonding angles
this angle is a little smaller (MDNH: 113.18;
DMDNH: 113.48; DNDA7: 113.58; DNH: 113.68).
This information is important because it shows that in
the case of DNDA5 and DNDA6, the dinitraza bridge
is more open so it might induce a higher hindrance
effect.
This parameter was studied on different locations
in each molecule and result in following remarks:
The dinitraza bridge angle (central bridge of the
molecule) NCN exhibits the highest value for the
molecules DNDA5 and DNDA6 (respectively 114
and 114.18) whereas in the case of the other molecules
The CNC angles are rather constant and the
evolution does not reveal any significant aspect
Table 5
Comparison between calculated and measured melting temperatures and enthalpies of the methylene dinitramines
Component Measured melting Calculated
Measured
Calculated
Homomorph
enthalpy (J/g)
melting
melting point melting
enthalpy (J/g)
(onset) (8C)
point (8C)
Name
Melting
Melting
enthalpy (J/g) point (8C)
DNDA5
DNDA6
DNDA7
MDNH
DMDNH
DNH
102.3
108.2
163.9
106.3
127
108.2
164.4
166.7
54.4
31.6
75
69.9
128.9
137.1
90.3
n-pentane
72.15
86.18
100.2
100.2
2130.15
295.5
n-hexane
n-heptane
290.15
2118.3
2102.9
290.15
225.6
61.5
85.1
36
2-methyl-n-hexane
116.1
137.1
237.5
2,6-dimethyl-n-heptane 128.26
110
166.7
n-heptane
86.18
DNU
125
72
n-undecane
156.31

48
D. Spitzer et al. / Journal of Molecular Structure 644 (2003) 37–48
going from 121.18 for DNU to 125.88 for DMDNH.
The CNN angles are rather constant to (116.2–
119.28).
References
[1] J. Quinchon, J. Tranchant, Les nitrocelluloses et autres
` `
matieres premieres de base des poudres et propergols,
Lavoisier, Tec et Doc, 1984.
[2] T. Urbanski, Nitrate esters, Chemistry and Technology of
Explosives 4 (1984) 295.
4. Concluding remarks
[3] T. Urbanski, Production of nitroglycerine, Chemistry and
Technology of Explosives 2 (1984) 62.
The most interesting result from this present
study is the determination of rather unexpected
crystal structures in the case of the unsymmetrical
methylenedinitramines (DNDA5, DNDA6 and
DNH). This is especially the case of the DNDA5
molecule which has two unique molecules in the
asymmetric unit. For all the studied compounds, no
classical hydrogen bonds have been detected, in
some case no hydrogen bonds have been detected,
which probably could be the explanation of the
existence of amorphous mixtures over very large
composition ranges for DNDA5/DNDA6 and
DNDA5/MDNH, in a temperature range from
250 8C (corresponding to the glass transition
temperature) to þ100 8C, due to the presence of
particular intermolecular interactions and very low
crystallisation kinetics. The nature of these inter-
actions will be examined. Considering the thermal
results and the determined crystal structures,
molecular simulation to evaluate the interaction
between the plasticiser and the nitrocellulose in the
solid state is started. These calculations are
performed on the base of a collaboration between
ISL and the GMO group from IPCMS laboratory
(ULP, Strasbourg I).
[4] W. Langlotz, D. Diehl, I.C.T. Mueller, Propellant power for
barrelled weapons, PCT, WO 98/34891, 1998.
[5] V.A. Tartakovskii, A.S. Ermakov, V.A. Koroban, Preparation
of N,N0-dialkyl methylenebisnitramines, Russian Chemical
Bulletin 42 (1993) 11.
[6] D. Spitzer, H. Ritter, S. Braun, Etude du comportement
´ ´
´ `
thermique de dinitramines lineaires a proprietes plastifiantes,
Rapport ISL, S-R111/2001.
` ´
[7] D. Spitzer, B. Wanders, Synthese, caracterisation thermique et
´ ´ ´ `
etude des melanges de nouvelles methylene-dinitramines
´ ` ´ ´
lineaires a proprietes plastifiantes, Rapport ISL, S-R124/2001.
¨
[8] D. Spitzer, M.R. Schafer, F. Ciszek, A. Carriere, Diagrammes
´
binaires et ternaire du melange plastifiant DNDA57, Rapport
ISL, 2002.
¨
[9] D. Spitzer, B. Baschung, M.R. Schafer, M. Scharf, F. Ciszek,
´ ´
Etude du melange plastifiant DNDA57 utilise dans les poudres
propulsives TU, SPU 604/2002.
[10] KappaCCD Operation Manual, Nonius B.V., Delft, The
Nederlands, 1997.
[11] Z. Otwinowski, W. Minor, Methods in Enzymology 276
(1997) 307.
[12] M.C. Burla, M. Camalli, G. Cascarano, C. Giacovazzo, G.
Polidori, R. Spagna, D. Viterbo, SIR J. Appl. Cryst. 22 (1989)
389.
[13] G.M. Sheldrick, SHELXL97. Program for the refinement of
crystal structures, University of Gottingen, Germany, 1997.
[14] A.L. Spek, PLATON98, Utrecht University, The Netherland,
1998.
[15] S. Zeman, Thermochimica Acta, 302, 1997,11–16.