The HOOH–HOO complex. A matrix isolation study

The peroxy radical forms a cyclic complex with hydrogen peroxide, with both entities acting as hydrogen bond donors and as hydrogen bond acceptors.

The strength of the complex is very similar to that of the hydrogen peroxide dimer.


The peroxy radical participates in numerous oxidation reactions.1

Its interactions with stable molecules influences the stabilization of newly formed hot radicals and it may influence its reactivity.

We have used matrix isolation combined with infrared spectroscopy to study its interactions with small, stable molecules.2–6

In these experiments, we have observed weak bands, which we suspected to be due to a hydrogen peroxide peroxy radical complex, since hydrogen peroxide, which forms from the reaction between two peroxy radicals, is always present in our experiments.

The addition of extra hydrogen peroxide to matrices containing peroxy radicals confirmed our suspicions.

However, a more complete assignment of the spectrum of the complex required studies of the hydrogen peroxide dimer 7 and of the water hydrogen peroxide complex,8 since water is another byproduct of the peroxy radical synthesis.

Both hydrogen peroxide and the peroxy radical act as hydrogen bond donors2,9,10 and as hydrogen bond acceptors,2,11 and both form cyclic complexes whenever possible.

It therefore seems reasonable to expect their complex to have a cyclic structure, essentially the structure of the hydrogen peroxide dimer,7 minus one of the two nonbonded hydrogens.

Our spectra give clear evidence for this structure, we observe two strongly red shifted OH stretches and two strongly blue shifted HOO bends one from each complex component.

A very recent calculation appears to support this structure.12

During the study of the HOOH–HOO complex, we noted that the positions of some of its bands were sensitive to the concentration of dopants in the matrix, shifting several wavenumbers between different experiments.


The experimental procedure has been described in earlier publications.2,7,8

A brief description is given below.

Two separate deposition lines were used.

Their gas flows were regulated by Nupro needle valves.

In one volume argon (L′Air Liquide 99.9995%) was mixed with O2 (L’Air Liquide, 999 995%).

The Ar/O2 ratio was varied between 50/1 and 75/1.

This gas mixture was then swept over the urea–hydrogen peroxide adduct kept a fixed temperature as described by Pettersson et al.13

The temperature of the adduct was +16 °C in most experiments.

Deuterated hydrogen peroxide was prepared as described in .ref. 13

In a second volume argon was mixed with H2(AGA) or D2 (L’Air liquide).

The Ar/hydrogen mixture was passed through a microwave discharge.

The gas mixtures were deposited on a CsI-window, cooled to 17 K by a Leybold RDK 10–320 closed cycle cryocooler.

Peroxy radicals formed on the surface of the growing matrix by the addition of hydrogen (deuterium) atoms from the discharge to molecular oxygen.

The deposition time was 2 h and the deposition rate 10 mmol h−1.

Infrared spectra were recorded, at 17 K, between 185 and 4000 cm−l with a Bruker 113 v spectrometer.

A resolution of 0.5 cm−1 and an average of 512 scans was used in the mid-infrared while in the far infrared it was sufficient to use a resolution of 1 cm−1 and 128 scans.

An MCT detector was used in the mid-infrared and a liquid helium cooled Ge-bolometer in the far infrared.

After the initial spectra had been recorded the matrix was irradiated with 266 nm radiation from a quadrupled Continuum NY 20 C YAG laser, and new spectra were recorded.


In complexes of the type studied here, the intramolecular vibrations of the complex components retain their original character in the complex.

Therefore, the perturbed ith fundamental of A in a complex with B will be denoted as νi(A–B).

In order to indicate which hydrogens are engaged in the formation of the hydrogen bonds we write the complex with these hydrogens next to each other, for instance DOOH–HOO and OOH–HOOD denotes a complex where HOOD is H-bonded to HOO.

It is possible to control the HOOH concentration in the matrices by changing the temperature of the urea complex.

It is rather difficult to control the concentrations of HOOD and DOOD, since different preparations give different degrees of deuteration.

It is also hard to vary the peroxy radical concentration in a systematic way, since the efficiency of the microwave discharge, which produces the hydrogen atoms, varies between different experiments.

Experiments with hydrogen through the discharge contain only HOO radicals and experiments with deuterium through the discharge contain DOO and only traces of HOO.

Bands are assigned to the hydrogen peroxide peroxy radical complex, if they appear only when hydrogen peroxide and peroxy radicals are simultaneously present in the matrix, if they have constant intensity ratios directly after deposition, independent of the concentrations of hydrogen peroxide and peroxy radicals, and if they decrease at the same rate when the matrix is irradiated.

It should be noted that the strongest bands which we assign to a hydrogen peroxide peroxy radical complex are too strong to be due to 1∶2 (2∶1) complex.

Two OH stretches at 3432 cm−1 and 3206 cm−1 have been observed as two weak bands in many previous experiments with peroxy radicals, see for instance Fig. 2 of .Ref. 2

When hydrogen peroxide is added to the matrix, they grow strongly as expected if they are due to a hydrogen peroxide peroxy radical complex (Fig. 1).

When DOO is used instead of HOO, the 3432 band shifts to 3429 cm−1, the 3206 cm−1 band disappears, and a new band appears in the OD stretching region at 2383 cm−1.

It therefore seems clear that the 3206 cm−1 band is due to the OH stretch of HOO complexed to hydrogen peroxide.

The complex shift, −207 cm−1, is slightly larger than the corresponding shift of the peroxy radical water complex, −177 cm−1.

Experiments with partially deuterated hydrogen peroxide show that the 3432 cm−1 band is due to HOOH.

The corresponding band of DOOD complexed to HOO is found at 2536 cm−1.

Also for the HOOH (DOOD) band the complex shift is relatively large, −156 cm−1 (−110.0 cm−1), giving clear evidence for a hydrogen bond from hydrogen peroxide to the peroxy radical.

At the same time as the 3432 cm−1 and 3206 cm−1 bands appear, bands are observed at 1428 cm−1, 1487 cm−1 and 1496 cm−1.

Deuteration experiments show that the 1428 cm−1 band is due to HOOH and the 1487 cm−1, 1497 cm−1 doublet is due to HOO.

Both bands are strongly blue shifted from the HOO bends of hydrogen peroxide and the peroxy radical as expected for a cyclic, hydrogen bonded complex.

The OO stretch of the complexed peroxy radical is observed at 1121 cm−1, very close to the corresponding band of the peroxy radical water complex (1120 cm−1, ref. 2).

The non hydrogen bonded OH stretch of hydrogen peroxide is observed at 3574 cm−1 and the corresponding HOO bend at 1284 cm−1.

A weak band at 868 cm−1 is assigned to the OO stretch of complexed HOOH.

The corresponding bands of HOOH–DOO have been observed in experiments where D2 was passed through the microwave discharge instead of H2.

Experiments with partially deuterated hydrogen peroxide have made it possible to observe some of the intramolecular fundamentals of the D2O2 and HDO2 complexes with HOO (Table 1) and DOO (Table 2).

The bands observed give clear evidence that both HOOD–DOO and DOOH–DOO exist.

In addition to the intramolecular fundamental bands, we observe a number of relatively weak bands which fulfill our criteria for assignment to the peroxy radical hydrogen peroxide complex.

The strongest of these, at 2926 cm−1, is probably due to the overtone of the HOO bend of complexed HOO.

The assignment of the mid-infrared spectrum is collected in Tables 1, 2 and 3.

We have observed a number of intermolecular bands of the hydrogen peroxide complex (Fig. 2).

The two bands with highest frequencies are observed in the region where the bolometer spectra and the MCT spectra overlap.

The results in this region are in very good agreement.

Compared to mid-infrared bands of matrix isolated molecules, bands in the far infrared tend to be weaker and broader.

One is closer to the phonon band in the far infrared, this makes relaxation a low order process and therefore decreases lifetimes of excited states.

Intermolecular vibrations, which are found in this region, involve the motions of the complex forming molecules as rigid entities which tends to make the coupling to the matrix stronger than in the mid-infrared where the vibrations tend to be localized in one of the complex forming molecules.

It is more difficult to obtain precisely matched sample and background spectra in the far infrared, where the wavelengths are comparable to for instance the matrix thickness.

The original spectra contain interference fringes from different optical elements.

Ideally these should cancel in the ratio between sample and background spectra.

Unfortunately even minute changes in for instance the cryostat position are sufficient to obliterate the cancellation.

This makes it harder to observe weak bands.

Both hydrogen peroxide monomer and dimer have clear and strong far infrared bands, which partially obscure the bands of the peroxy radical hydrogen peroxide complex.

The assignment is significantly aided by the fact that the hydrogen peroxide peroxy radical complex decomposes very rapidly when irradiated with 266 nm radiation, significantly faster than the hydrogen peroxide dimer.

The hydrogen peroxide monomer is hardly touched during the time it takes to eliminate the peroxy radical complex.

The irradiation has the additional advantage that the base line is almost unchanged by the irradiation.

The assignment of the far infrared spectrum is collected in Table 4.


The appearance of two strongly red shifted OH stretches from the peroxy radical hydrogen peroxide complex clearly suggest that the complex is cyclic, with both components acting as hydrogen bond donors and as hydrogen bond acceptors (Fig. 3).

This conclusion is supported by the observation of two strongly blue shifted HOO bends, one from each component.

In fact the HOOH part of the mid-infrared spectrum of the HOOH peroxy radical complex is quite similar to the spectrum of HOOH in a dimer formed between HOOH and DOOD where there are no resonant interactions between the dimer components.7

The shift of the bound OH stretch of HOOH is 25 cm−1 more negative in the peroxy radical complex than in the DOOD complex, indicating that the peroxy radical is a better hydrogen bond acceptor than hydrogen peroxide.

For the HOO bends of HOOH, the shifts are larger in the HOOH–DOOD dimer than in the peroxy radical complex.

The high frequency bend is found at 1436 cm−1 in the HOOH–DOOD dimer and at 1428 cm−1 in the complex and the low frequency bend appears at 1291 cm−1 in the dimer and at 1283 cm−1 in the complex.

For both bends, the complex shifts are smaller for the peroxy radical complex than for the mixed dimer.

In order to understand this apparent discrepancy between the stretch shift and the bend shifts we note that the two OH stretches of hydrogen peroxide are almost independent, in HOOD the OH stretch almost coincides with the two OH stretches of HOOH.13

The two HOO bends are strongly coupled, in HOOD the HOO bend is found at 1343 cm−1, approximately midway between the two HOO bends of HOOH (1385 cm−1 and 1271 cm−1) and the DOO bend is found at 981 cm−1 close to the average of the two DOOD bends.

The coupling between the bends depends on the dihedral angle between the HOO groups which may differ between the complex and the dimer.

In addition the 1428 cm−1 band is likely to be lowered somewhat by an interaction with the HOO radical bend at 1487 cm−1 and 1496 cm−1.

Ab initio calculations on the hydrogen peroxide dimer suggest that we should expect three intermolecular fundamentals above 200 cm−1 for the HOOH–HOO complex, librations of the two complex components around their OO axis and the torsion of hydrogen peroxide.14

The same calculations indicate that we should observe two torsion bands for the HOOH monomer since the trans barrier is low, and indeed we observe two bands for monomer HOOH, close to the expected positions.14

It seems reasonable to assume that the trans barrier of hydrogen peroxide complexed with a peroxy radical is relatively low so that there are two close low lying torsion states, a symmetric and an antisymmetric combination of the complexes with the free hydrogen of hydrogen peroxide above and below the plane of the other atoms.

If the barrier is not too different for complexed HOOH compared to monomer HOOH, the next two torsion states are close to the maximum of the trans barrier and therefore strongly split.

We observe four far infrared bands for HOOH–HOO (Fig. 2).

The lowest band, at 227 cm−1, is close to a band at 219 cm−1 of the hydrogen peroxide dimer, which we believe is due to the libration of HOOH in the dimer.14

We suggest that the 227 cm−1 band of HOOH–HOO should be assigned to the libration of HOOH around its OO axis.

The band at 557 cm−1 appears to shift to slightly above 400 cm−1 when DOO is substituted for HOO.

The precise position depends on the isotopic composition of the complex partner.

We therefore assign the 557 cm−1 band to the libration of complexed HOO around its OO bond.

This assignment implies that the 277 cm−1 band and the 477 cm−1 band are due to the HOOH torsion.

In HOOH monomer the corresponding bands are observed at 273.2 cm−1 and at 373.5 cm−1 respectively.

The shifts are surprisingly different, but we do not think this automatically shows that the assignment is wrong.

In the monomer calculation, the two lowest states were split by 10 cm−1 and were approximately 300 cm−1 below the trans barrier, the third lowest state was very close to the barrier and the next state approximately 100 cm−1 above the barrier.

If the barrier is lowered in the complex, the two upper states will be even further away from two interacting tunnel states than in the monomer and therefore even further apart than in the monomer.

If we accept the assignment of the intermolecular fundamentals of HOOH–HOO, the assignment of the corresponding fundamentals of the HOOH–DOO complex is straight forward.

The bands at 485 cm−1 and 272 cm−1 are assigned to the torsion of HOOH and the 220 cm−1 band to the libration of HOOH, the libration of DOO is observed at 415 cm−1.

We have made a number of experiments with partially deuterated hydrogen peroxide, both with HOO and DOO.

In these experiments we observe a number of far infrared bands whose photolysis rates clearly shows that they are due to hydrogen peroxide peroxy radical complexes.

We found it difficult to control the H/D ratio sufficiently well to make it possible to avoid the presence of significant concentrations of HOOD in these experiments.

This makes the assignment very difficult, since the mid-infrared spectra indicate that there are two different isomers of the HOOD peroxy radical complex.

One therefore expects to find 12 bands due to the DOOD and HOOD peroxy radical complexes in a given experiment.

We have tried to assign the observed bands guided by the results for HOOH and the assumption that the torsion frequencies are mainly determined by the mass of the free hydrogen of the complexed hydrogen peroxide.

The assignment is collected in Table 4.

Two cyclic structures are found to represent local minima on the potential energy surface of HOOH–HOO in ref. 12, both at the B3LYP and QCISD levels of theory.

The hydrogen peroxide forms a hydrogen bond to the end oxygen of the peroxy radical in both structures.

The most stable structure is a six-membered ring where the peroxy radical forms a hydrogen bond to the hydrogen peroxide oxygen, which binds the free hydrogen.

The second minimum is a five-membered ring with a hydrogen bond from the peroxy radical to the hydrogen peroxide oxygen, which is bound to the hydrogen of the hydrogen bond.

It is interesting to compare the experimental complex shifts of HOOH–HOO with the calculated shifts of ref. 12 (Table 5).

The B3LYP calculations give very large shifts for the OH stretches of hydrogen peroxide and the peroxy radical for the six-membered ring structure, 1.4 and 1.8 times larger than the observed shifts respectively.

For the five membered ring structure the OH shift of the hydrogen peroxide is 0.50 of the observed shift and the OH shift of the peroxy radical is 1.5 times the observed shift.

The calculated OO shifts of the peroxy radical are approximately 10 times larger than the observed shift for both structures.

We are therefore convinced that the B3LYP method gives a rather poor description of the complex.

The complex shifts of the QCISD calculations are in reasonable agreement with the observations, the OH shifts of the OH bonds involved in the hydrogen bonds are somewhat smaller the the observed shifts but the general pattern of the shifts agrees well with the observations.

It is more difficult to choose between the two structures.

The largest difference in the calculated shifts is obtained for the HOOH torsion, which is calculated to be blue shifted in the five membered ring complex and red shifted in the six-membered ring complex.

Unfortunately it is impossible to compare the calculated torsions with the experimental torsion since the low trans barrier of the complex allows rapid tunneling in the torsionally excited state.

This makes the harmonic oscillator approximation used in the calculations completely unrealistic.

The calculated bound OH shifts of the six-membered ring complex are both significantly closer to the observed shifts.

Also most other calculated shifts of the six-membered ring are closer to the experimentally observed shifts that those of the five membered ring (Table 5).

Experiences from many other systems suggests that with few exceptions, only the most stable complex structure is observed, this seems to be the case also for the HOOH–HOO complex.

It is difficult to compare the calculated intermolecular vibrations with the experimentally observed far infrared bands, since no description of the calculated bands is given in the paper and the calculations are done only for the HOOH–HOO isotopologue.

It seems reasonable to assume that the highest calculated intermolecular fundamental at 647.6 cm−1(only the QCSID values for the six-membered ring complex are considered) corresponds to the observed HOO libration at 557 cm−1.

Both calculations and experiment agree that this is an intense band.

We have difficulties in understanding the nature of the second largest calculated intermolecular fundamental at 544.6 cm−1.

The librations of the two complex components involve only H-motions and are therefore expected to have high frequencies.

The remaining intermolecular fundamentals will involve oxygen atom motions and have much lower frequencies.

The hydrogen bond from the peroxy radical to hydrogen peroxide is calculated to be stronger than the hydrogen bond from hydrogen peroxide to the peroxy radical.

The hydrogen peroxide libration is therefore expected to be less than 0.7 of the peroxy radical libration, the calculated frequency is 100 cm−1 larger than one expects from this argument.

It is perhaps possible that the coupling between the hydrogen peroxide libration and the hydrogen peroxide torsion is very strong, giving one upper and one lower component both with significant torsion and libration contents.

The hydrogen peroxide peroxy radical complex is clearly more strongly bound than the hydrogen peroxide dimer.

The shift of the bound OH stretch of HOOH is −132 cm−1 in the dimer and −156 cm−1 in the peroxy radical complex.

The complex shift of the OH stretch of the peroxy radical is −207 cm−1 suggesting that the hydrogen bond from the peroxy radical to hydrogen peroxide is stronger than the hydrogen bond from hydrogen peroxide to the radical.

For the hydrogen peroxide dimer the dissociation energy (De) was calculated to be 6.3 kcal mol−1 on the MP2 level.7

We believe that the the dissociation energy of the peroxy radical hydrogen peroxide complex is of the order of 6.5 to 7 kcal mol−1, the calculations of ref. 12 gave 7.7 kcal mol−1 (De, corrected for BSSE).

The shifts of several wavenumbers of some complex bands between different experiments is very surprising (Fig. 4).

Even more surprising is that the hydrogen peroxide dimer bands, which have very stable positions in experiments where only hydrogen peroxide is present in the matrix7 show shifts in these experiments, these shifts correlate with the shifts of the hydrogen peroxide peroxy radical complex bands.

We have previously noted a phenomenon, which we believe is related to the one observed here.

In the hydrogen peroxide monomer spectrum, in matrices with a low content of polar impurities, the OH stretch and the antisymmetric HOO bend have two relatively sharp components.

When polar impurities are present, bands appear approximately midway between the two low concentration components.

The low concentration components weaken and ultimately vanish at high impurity concentrations.

We suggested in an earlier publication that polar impurities introduced electric fields in the matrix, which strongly perturbed the hydrogen peroxide inversion.8

The two components of the monomer bands are due to the inversion.13

Since the dipole moment of hydrogen peroxide changes its direction upon inversion, the presence of an electric field may stabilize one of the two torsion minima and destabilize the other.

A similar mechanism is possible for the hydrogen peroxide peroxy radical complex, which should have a large dipole moment whose direction is determined by the orientation of the non hydrogen bonded hydrogen of hydrogen peroxide.

The electric field from a peroxy radical within 10 Å from a complex may inhibit the tunneling of the non bonded hydrogen and perhaps also shift its equilibrium dihedral angle, which in turn will influence the fundamentals directly involved in the hydrogen bond system of the complex.

The equilibrium structure of the hydrogen peroxide dimer has Ci symmetry and consequently no dipole moment.

However, the C2 symmetric isomer, which has the two non hydrogen bonded hydrogens on the same side of the plane of the heavy atoms has a relatively large dipole moment and is expected from ab initio SCF calculations to be approximately 50 cm−1 higher in energy than the Ci isomer.

A strong electric field may stabilize the C2 isomer.

More likely is that a polar molecule not too far from one of the non-hydrogen-bonded hydrogens of the Ci dimer inhibits the inversion of the dimer.