The Path of Oxygen from Water to Molecular Oxygen
Lawrence J. Heidt Massachusetts Institute of Technology Cambridge, Massachusetts
This article concerns mainly the path of the oxygen atom in water from its minus one and two states of oxidation in hydrogen peroxide (5, 31, 34-87, @) and water (13, 15-17, 19, 24, 25) respectively to its zero state in molecular oxygen (26). The study is part of a program of research concerning the mechanisms of the photochemical decomposition of water into gaseous hydrogen and oxygen by light absorbed by certain solute species (15,17,19, $1, 23-25, SO, S2), the production and stabilization of ozone (31) which is a part of the above problem as will be seen later from this article, and the charge and energy transfer processes (52) produced thermally or by light absorption which are of potential value in solar energy conversion and related systems. The latter include the studies of intra- and intermolecular electron transfer between donor and acceptor groups in substituted aromatics and aromatic complexes such as in substituted benzenes and benzene halogen complexes (32). I n order to emphasize the relationship of the study to the conversion of solar energy into chemical energy available in storage for man's use we present in a primitive way in Table 1some pertinent information regarding the in ziuo chlorophyll catalyzed natural photosynthetic process (41,46,47) and in Tables 2 and 3 and Figure 1 similar information regarding the in vitro cerium catalyzed process (15, 17, 19,81, 24, 27, SO). Table 1 and the pertinent reactions given in the text are not meant to imply that the steps from water to oxygen in photosynthesis are known while the reactions involving chlorophyll are not. The reverse is perhaps nearer the present situation (41, 46, 47). The natural photosynthetic system consists essentially of air, water, organic material, and very small amounts of certain metals known as the trace elements. Part one of the natural photosynthetic process consists of the charge and energy transfer processes associated with t,he absorption of sunlight by the organometallic substance chlorophyll and the subsequent release of the absorbed energy by the chlorophyll to produce the energy required for photosynthesis. This part may be represented by the reactions: Ground state chlorophyll
+ photon
Table 1. The Overall and Half Reactions for the in vivo Natural Photochemical Conversion of Water and Carbon Dioxide into Oxygen and Carbohydrate Material by Sunlight Absorbed by the Chlorophyll Catalyst in the Proper Wotery Environment
+ - -+ + + + + -
+ energy
(2)
The sunlight absorbed by the other colored substances in the system may not affect the overall process but t,his remains to be determined. There is no doubt, This is publication No. 83 of the M.I.T.Solar Energy Conversion Project. Financial support for this work hss come from the Godfrev Lowell Cabot Fund of the Massachusetts Institute
.....-
Water half reactions:
2H20
or 4H90
2H.02
followed by H20n and 2 e -
ZH+
0~ 4Ht
4Ht
On
(a)
(b)
+ 2e-
(c)
2H20
-
+ 4H+ + COZ
4e-
+ 4e-
4e-
2Ht
H202
Carbon dioxide half reaction:
(l/n)(C.HgO),
(dl
+ Hz0
Note: The rocess converts sunlight into chemical energy stored in the cfemicd bonds of the carbohydrate material and molecular oxygen. The sum of the water half reactions ( b ) , (el, and ( d ) is half reaction (a) and the sum of (e) and (d) 1s 2Hn01O2 2H20 45 kcal (heat). The highest possible quantum efficiencyfor the process in red limht of fifiOO A is~ 33% .---.. - - ~ - - if ~ the neroxide dismotation reaction does not occur and 25% if it doesbccur. The corresponding energy efficiencies are 78 and 66% as explained in the text.
+
+
~
Part two of the process, therefore, may be represented by the half reaction: COz
=
energy-rich chlorophyll* (1) Chlorophyll* = chlorophyll
however, that at least part of the sunlight must he absorbed by the chlorophyll in order to convert carbon dioxide into carbohydrate material (41). Part two of the natural photosynthetic process may be considered to consist of the charge and energy transfer processes associated with the conversion of the carbon dioxide into carbohydrate material (41, 46) through the agency of the photon-excited, energy-rich chlorophyll. The oxygen gas might also be produced from the carbon dioxide hut this is not the case since the oxygen has been found to come only from the water. This has been determined (47) by the use as a tracer of the heavy oxygen atom of mass 18.
+ 4H+ + 4e-
=
(l/n) (C.H$O).
+ H20
(3)
Part three of the process is associated with the prodnction from the water of the oxygen gas (26, 47) plus the electrons (or H atoms) (46) eventually transferred to the carhon dioxide as indicated by the above half reaction (3). Part three, therefore, may be represented by the half react,ion: 2HU = Oz
+ 4H+ + 4e-
(4)
Reactions (I), (2), (3), and (4) make it evident that Volume 43, Number 12, December 1966
/ 623
SUNLIGHT
Figure 1. Absorption spectra of the CellV1 and Cellll) perchlorate catalyrts in water. The short wavelength limit of visible wnlight~ir.bout 4 0 0 0 A ond of the sunlight reaching the earth's surface it is about 3 0 0 0 A or 3 3 0 0 0 cm-I. The short vertical liner are ot the wovelength ( 2 5 3 7 A) of the light employed for most of the quantum yield meosurementr. The seris solutions are amber in color 1241. The cereus d v t i a n r are mlorle$s 123). The strong mbsorption of visible and ultraviolet light by the CellV) discouraged the study of the light obrorbing intermediates produced by light in this system.
charge and energy transfer processes are involved. The overall reaction is presented in Table 1. The amount of chemical energy stored for man's use in the new chemical bonds produced when one mole of carbon dioxide and two moles of water are converted to carbohvdrate material and molecular is usually tagen to be 112 kcal (41) which is thLhkat of combustion of one C.H20 group in the carbohydrate glucose, CBHIIOC The equation for the combustion reaction is: O d d = CO&) 4- HxO(1) 112 kcal (5) ('/&%HISO,
+
+
Table 2. The Overall and Half Reactions for the in vitro Photochemical Conversion of W a t e r into Oxygen and Hydrogen b y Light Absorbed By the Cerum (IV) and (Ill) Catalyst in the Proper Watery Environment 4-
4e-
+ Ce(1II) + 2H20 + hv r-1
TC
~C~.(IV) 4Ce(IV)
-
4Cea(III)
+ Ox + 2H9 + 2H10 + 140 kcd stored
Water Half Reactions: 2Hz0
-+
or
4H.O
followed by
HlOn
and
2e4e-
--
O2
+ 4Ht + 4e-
2H102
0 3
2Hf
4e-
followed by
40H-
-- +-- +
+ 4Hf
Cerium Half Reactions:
1. 4e-
(b)
+ 2H+ + 2e-
+ 4HC + 4Hs0
or
(a)
+ 4H+ + 4e-
+ Hz03
(d)
2H2
(el
(f)
40H-
4Ho0
+ 4Ce(IV)
2. 4CelIII) - , ,
(c)
2Hz0
2Hs
+
(Q)
4Ce(III)
4CefIV) , ,
Note: The process converts light into chemical energy stored in the chemical bonds of the molecular oxygen and hydrogen. Here, as in the case of the natural photosynthetic process, the sum of the water half reactions ( b ) , (e), and (d) is half reaction (a) and the sum of (c) and ( d ) is 2H102+ Oz 21110 45 k c d (heat). The highest possible e6ciency for the in light of 2537 A on s. quentum yield basis is 25% and on an energy basis 30%.
624
/
Journol o f Chemical Education
-
(H~o).c~oH.+~ + (H~O)&!~OH+S + H+ , .~ . . ~
-
(H20),HOCeOH~+'
(HIO),HOCeOH+'
+
(a)
+ H+ +
(6)
~ ( H x O ) ~ C ~ O H(H*O)~C~(OH)~C~(OH)P +~ 2 ~ (H2O).CeOH+'+ (H20)rCe(OH)2+2-
+
(H~0)sCe(OH)sCe(0H~)a+~ 3H90 2(H20)rCe(OH)2+'
~
(c) 0
(dl
(H.O)>HOCe(OH)sCe(OHW'
+ 3HzO ( e )
Note: The negative h droxide ions, OH-, link together the .ositivelv charged ce(1VYions. This is done by the hydroxide Ions shaiing 0% or more corners (two may make up a shared edge and three a shared face) of the octahedra of 0x3-gen atoms in the minus two state of oxidation surrounding each Ce(IV) ion in the water. Other details of the cerium catalyaed process in addition to the above reactions me given in the -4ppendix.
The energy required from sunlight to produce this stored chemical energy can he estimated in the folloying way (28, 29,SS) : One einstein of red light of G6OO A absorbed by chlorophyll to produce photosynthesis contains the energy: Nhc/A = 6 X 10" X 6.6 X = 18 X 10'' ergs/4.2 X
X 3 X 10'0/6600 X 10- ergs = 43 kcd
lo7 erg cal-'
I t follows that 112/43 = 2.83 or a minimum of three quanta of red light need to be absorbed by chlorophyll in the proper watery environment in order to convert one molecule of COz to carbohydrate material if the oxygen goes from water to molecular oxygen without the intervening formation of peroxide. If hydrogen peroxide is an intermediate, the intervening exothermic reaction: 2N101 = 21110
. 4e-
+
Table 3. Reactions Producing the Ceric Dimers Responsible for the Photochemical Conversion of W a t e r into Oxygen
+ Odg) + 45 kcal
+
(6)
wouldmake it necessary to employ (112 45)/43 = 3.7 Or a minimum of four quanta of red light. The maximum quantum efficiency for the natural photosynthetic process in red light is, therefore, 100/3 or 33% if there are no peroxide intermediates or 100/4 or 25% if there are peroxide intermediates. The cor-
responding energy efficiences are 100(112)/3(43) or 78y0 and 100(112)/4(43) or 66%. The experimental value for the energy efficiency is ahout 2% when based on all the energy in sunlight as it reaches the earth's crust falling on a natural photosynthetic system. A more detailed consideration of energy budgets in the natural photosynthetic process such as the inclusion of all the acutal reactions and their energies of activation would show that more than four quanta of red light are needed per molecule of oxygen evolved and COz reduced (dl). The path of rarbon from carbon dioxide to carbohydrate material has been partly revealed by employing radioactive carbon of mass 14 as a tracer. This has been done by work carried out largely under the supervision of 3Ielvin Calvin who has been awarded the Nobel Prize in Chemistry for this work. His accomplishments and working hypotheses have centered around the identification and relative amounts of the compounds of carbon produced and consumed in the reaction undcr a wide variety of conditions (46). Path of Oxygen
The pat,h of oxygen from water to molecular oxygen (47) has not been fully revealed by any of the methods employed so surcessfully by Calvin, et al., to identify and study the role of the isolatable carbon intermediates. One reason for this may he that the life times of the oxygen intermediates are for the most part too short to he isolated and identilied by such methods. These intermediates contain the oxygen atom in some negat,ive oxidation state between its minus two and zero states in water and molecular oxygen, respectively. They are stronger and often more rapid oxidizing agents than molecular oxygen when taking either one or two elertrons from organic material (14). Consequently it. is surprising that any molecular oxygen a t all is produced in the natural photosynthetic process where there exist,s relat,ively large concentrations of easily oxidizable organic species. Peroxide intermediates such as H202 and HOn, however, would produce gaseous oxygen not only by being oxidized by some other substance hut also by dismutation reactions (48) catalyzed by hoth organic and inorganic substances. Reactions of this kind in the natural photosynthetic process would decrease further the lifetimes of the oxygen intermediates and the chances of finding them by the techniques employed to determine the path of carbon. There is, however, some evidence (16) that the photochemical product,ion of molecular oxygen from water may not involve the formation of peroxides. This evidence comes from the ceric-water system where visible and ultraviolet light is absorbed by hoth dimers and monomers of hydrolyzed ceric ions but only the light absorbed by the dimers produces molecular oxygen (15). This may he due to the fact that the ceric dimers but not the monomers are capable of accepting two electrons from the water and thereby directly producing hydrated atomic oxygen rather than peroxide on the way to molecular oxygen. The pertinent reactions are presented in Tables 2 and 3. Information regarding the discovery of the in vitro cerium catalyzed process and further details of the pertinent chemical reactions are presented in the
Appendix. Also presented is information regarding the measurement of the net quantum yields which served as a basis for the discovery. The ceric catalyzed reaction was observed to stop as soon as the light was shut off. Thus the intermediates are short lived which led us to employ the methods provided by flash photolysis to try to detect and identify them and study their reactions. Unfortunately the ceric-water system does not favor such a study since any relatively weak light absor@ion by transients a t wavelengths shorter than 4500 A would be hidden by the strong absorption of this light by the ceric ions (24) as can be seen from Figure 1. Fortunately the colorless persulfate-water (IS, 16) and hydrogen peroxide-water systems (5, 39) do favor such a study at wavelengths longer than 2300 A where neither the reactants nor the final products strongly absorb light as can be seen from Figure 2. The principle light absorbing species in the persulfate system iz SzOs= in hoth acid and base (16). Light of 3200 A and shorter wavelength (IS), in particular of 2537 A, absorbed by the S208- in water produces oxygen gas with a quantum yield of about 0.6 mole of persulfate decomposed per mole of light quanta when the solution is at or above pH 5. At pH 3 or less the
Figure 2. Absorption rpaclro of the woter rolutianr of sodium persulfate 113, 16) ond hydrogen peroxide 15, 391 employed for Rash photolyris.
yield drops to values about one tenth as large. Alkaline solutions, therefore, would he expected to produce maximum yields of oxygen gas and presumably of the oxygen intermediates. The persulfate, moreover, is thermally more stable in base than in acid. The two-electron acceptor in the persulfate system is the persulfate ion, SzOs=which absorbs light to form HSOI- by the half reaction: 4e4H+ 2SzOs= = 4HS04-. This is the analog of the C02 half reaction (Table 1) and the ceric hal£ reaction (Table 2). The series of water half reactions are again those presented in Tahles 1 and 2. It is important to note, however, that no energy is stored by the overall reaction:
+
+
The oxygen intermediates in this system also are short lived since the overall reaction stops when the light is shut off. The principle light absorbing species in the hydrogen peroxide system is Hz02 below pH 11 (5) and HOzabove pH 13 (39) since the first ionization constant Volume 43, Number 12, December 1 9 6 6
/
625
(48) of H202ill wat,er at 25OC is about 10-12. Light of 3200 A a n i short,er wavelcngt,h (5),in particular of 3130 and 2.537 A, absorbed by either Hz02or H02- in water produces oxygen gas with a quantum yield of about 1.0 mole of the peroxide decomposed per mole of light quanta absorhed by the peroxide when the solution is either acidic or basic (5, 55, 49). Figure 2 also shows that light of 2650 1 and nearby longer wavelengths is much more strongly absorbed by the HOz- than by the Hz02or the SzOs- at the same concentration. The corresponding absorbance values, e, at 2650 A are 150, 10, and 10. These values of c limit the concentration of these species to not more than about 0.003, 0.05, and 0.05M respptively for 80% or less absorption of light at 26.50 A and longer wavelengths by the 1.25-cm depth of solution in the reaction vessel traversed by the flash light. The monitoriiig light enlployed to study the transient
i
ML
A 1
MONO
PMT
PART A Figure 3. Schematic representation of the flash photolyrh opparotus for observing light absorbing transients produced by photolysis 14.51. FL. flash lamp; FV, filter vesel; RV, reoction vessel; ML, monitoring light source; M O N O , monochromotor; PMT, photomultiplier tube; OSC, o~cilloscope. Photographs of Ports A and B are presented in Figures 4 and 10. The glass in the optical train uos of water white fused quartz.
species traveled, howcvcr, 20 cm instead of 1.25 cm through the solution in the reaction vessel; hence, the corresponding concentrations applicable to the monitoring light at 2600 A and longer wavelengths are only one sixteenth an large as for the flash light. The alternative is to limit the monitoring light to longer wavelengths not much shorter than 3150 A where c is 10 for the HOz- or to use smaller concentrations of the reactants. The two electron acceptor in the hydrogen peroxidewater system is the HzOzmolecule in acid or weak base and the peroxide ion, H02-, in base above pH 12. In acid the light absorbed by the peroxide produces water by the half reaction: 4e- 4H+ 2H202 = 2HzO. Here again, as in the persulfate-water system, the water half reactions are those postulated for the chlorophyll and cerium catalyzed reactions (Tables 1 and 2). In this, as in the persulfate system, no energy is stored by the overall reaction:
+
4e -
hv
rn
+ 2H202+ 2Hz0
=
41L0
k A
I
Figure 4. Pho+ogroph of Rorh unit m d reoction vessel, Part A d Figure 3. ML, monitoring light source; ID, iris diophrogm; RV-FV, reoction vessel-filter verse1 unit; PB, pressure b d a s t ; 1, lens; FL, flash lamp; FLT,fiorh lomp trigger.
Figure 4 presents a photograph of the apparatus comprising Part A . The flash lamps were of water white fused quartz; four of t,hem n w e placed at the corners of a square and coaxially alongside the fused quartz reaction vessel and its surrounding filter. Figure 5 presents the electrical circuitry for the flash unit designed to minimize inductance in order to reduce the rise and decay times of t,he light flash. Short straight copper bars were employed to carry the current between condensers and flash lamps. There was a common ground point to which both sides of the rapid discharge condensers were grounded r h e n not in use. The lamps were lit by the dischargc of up to 1800 Joules at 20 kv per lamp which tool; placed in less than 5 psec through oxygen gas initially at a pressure of about 30 mm Hg and 25% Figure 6 preseuts the effect of the initial pressure of oxygen in the lamps upon the electrical discharge. Figure 7 is an oscillogram of the rise aud decay of
+
+ 0~+ heat
(8)
but no sulfur intermediates are produced to complicate the reaction. Flash Photolysis Technique
The flash photolysis apparatus employed to study the transients is represented diagrammatically by Parts A and B of Figure 3.
Figure 5.
Dlagram of the flash lomp ond condenser bonk circuitry.
the flash produced by the discharge of 8 pF condensers at 12,000 v per lamp. The times of the rise and decay are seen to be about 1 and 5 psec, respectively. Figure 8 presents diagrammatically the reaction vessel, RV, and its surrounding filter vessel, FV. The filters were confined between the walls 0.5 em apart. All the flash light incident upon the reaction vessel had to pass through the filter which was either water, aqueous nickel sulfate (which transmitted light between 2300 and 3200 A), or dry carbon tetrachloride. These liquids limited the flash light incident upon the reaction vessel to wavelengths longer than 2000, 2300 and 2600 A, respectively. Figure 9 presents the absorption spectra of the dry
Figure 6. Fiorh lamp characteri~tics. The doto ore for oxygen gar. High-energy plasma conditjonr produce o continvovr spectrum of light between 2 0 0 0 and 8 0 0 0 A. Low-energy nonplasma conditions are less favorable for light production b y oxygen thon b y argon and the other inert gwes especially a t wavelengths shorter thon 3 0 0 0 A. The hold back pressure required to prevent spontaneous fioth discharge ir reen 10 increase almo3t linearly with increase in voltoge across the lamp over the range of our experiments. X Tesla-coil induced Rash dirchorge; 0 ,Teslo-coil induced glow discharge; 0 , spontaneous Rash dirchorge.
Figure 8. Schematic representation of fused quartz reaction veael, and rvrrovnding filter vessel, FV.
RV,
center of the solution in the reactionvessel on a path perpendicular to the path of the photolyzing light flash. The monochromator was a Zeiss 31x1 12 instrument which separated the light into narrow bands of wavelength. The selected band mas made to fall on the photomultiplier tube, PAIT, whirh produced an electrical response proportional to the light intensity. This response was amplified and recorded by means of the other equipment shown and identified in Figure 10. The oscilloscope was triggered by light from the flash falling on the oscilloscope trigger phototube, OTPT. Figure 11 presents a representative oscillogram for a transient light absorbing intermediate. In this case the flashed solution mas 0.0005 n"lH?02 in 0.2 A t NaOH. The interval L on the oscillogram covers the time when there was interference from stray light from the flash and when the absorption of light. by the transient was reaching its maximum value. Stray light in the monitoring light beam produced the usual difficulties especially at wavelengths where
carbon tetrachloride filter and of the sodium hydroxide solutions. The latter behaved as inner filters when employed to make up the solutions which were flashed photolyzed in the reaction vessel. The special Hooker NaOH is seen to be optically cleaner than the Ralcer analyzed reagent grade material. Figure 10 presents a photgraph of the apparatus comprising Part B of Figure 3. The light falling on the entrance slit of the monochromator, MONO, came from the monitoring light source after traveling through the
61 Figure 7.
Intensity p r o f i l e d a representative light Rorh.
Figure 9. Absorption rpectro of light filter materials, 0. D. v g wavelength. 21 5M Hooker NaOH-10cm poth, 31 CCL1) 5M BA NoOH-lOcm 0.5 cm poth.
Volume 43, Number 12, December 7966
/
627
Figure 10. Photogroph of the equipment required to observe the transients produced by Rash photolysir. Port B of Figure 3. FLCBS, Rorh lomp condenser bonk charger; FLT, Rash lamp trigger; CFPS, cathode power supply; M O N O , follower power supply; PMPS, ph~tomuitipii~r monochromator; PMT, photomultiplier tube; POC, polamid camero; OTPT,orcillorcope trigger phototube; 0SC.orcillorcope.
the light. intensity of the beam was low or phototuhe response to the light at the selected wavelength was small although the intrinsic light intensity at this wavelength might have been relatively high. Figure 12 presents the response of the EAII 6256 photomultiplier tube, PAIT, as a function of the mavelength of the light falling on it from the different monitoring light sources. The response cnn be seen to belimited to light between 7000-2800 A or at best 2300 A. This information is needed in the calculations for normalizing the data obtained from the oseillograms in order to correct the apparent variations in absorbing power of the transients as a function of wavelength. The absorption spectrum of a transient was determined as folloxr-s. First one located the wavelength of maxiinum (peak) light absorption by t,he transient produred in the photolyzed solution during selected time intervals such as 100-1000 psec after the flash. Ncxt one measured the rates of decay of the transient absorpt,ion at these peaks and at neighboring n7avelengths in order to obtain data to construct plots like those presented in Figure 13. Figure 13 presents plots for the 4300 transient in flashed water solutions 0.003 M in H201and 0.003 M in NaOH. Each line begins 0.1 msec after the flash and is for t,he stated wavelength or, more properly, narrow wavelength band of the monitoring light. The straight lines were obtained by the met,hod of least squares.
Linei pf nearly the same slope cover the range 33005300 h and represent decays having nearly the same half-life; hence, the transient light absorptions are probably due to the same species. A(log OD)/& = A(1og c)/At, wherc the optical density, OD = clr, is for the transient concentration, c, over the light path, 1, and r is the absorbance value of the species at the selected wavelength, X. Absorption spectra are usually presented as plots of t or of log e versus X or cm-' after normalizing the apparent OD values over the range of wavelength of interest. The normalization inrludes corrections for differences in (1) the concentration, c, of the light absorbing species (2) the response of the instrumentation and photographic materials to light as a function of wavelength of the kind displayed in Figure 12, and (3) the light intensities of the monitoring light as a function
Figure 12. Monitoring lamp-photomultiplier tube chorocterlrtlcr for llght from xenon, rirconium, and tungsten lamps falling on the EM1 6256 phototube.
of the light source, wavelength, and time especially during the period of time covered by the oscillogram. The equipment did not permit us to record simultaneously differences in the light intensity of two or more distinguishably different xavelength bands or of light from the same source traveling two differentlight paths. Results and Discussion
d
Figure 11. Light absorption profile of the 4300 trmsient produced b y Rash photolyrit of an oikaline water rolution d hydmgen peroxide.
628 / Journal o f Chemical Education
Flash photolysis of the alkaline solutions of sodium persulfate and hydrogen peroxide produced two outstanding transient lighto absorption peaks, one at 4300A, the other a t 2600.4. Table 4 presents data for flashed solutions of persulfate and hydrogen peroxide at 0.002 M which was low enough to produce nearly uniform absorption of the flash light by the solution. I n 0.01 A[ KaOH the half-lives of bot,h peaks are seen to
be about a millisecond which is about the time it takes to blink an eye. In 7 M NaOH the half-lives of the 4300 and 2600 A peaks are seen to be about tenfold and over one million fold longer respectively than in 0.01 M NaOH. In 7 M NaOH the peak a t 2600 A has a halflife of about 40 min. which is long enough to obsenre the peak and follow its decay with the Model 11 Cary
Figure 13.
the 0- may be produced by the direct cleavage by light of the 0-0 bond in the ion H-0-0formed by the neutralization of the HzOsbut this direct cleavage is not supported by other evidence presented later in this article. In the case of the 2600 k peak the data in Figure 14 are for flashed solutions 7 M in NaOH and either 0.0005 M in hydrogen peroxide (dots) or 0.001 M in sodium persulfate (squares). The resulting absorption spectrum is similar to that of ozone gas, wet (8) or dry (lo), whose photolysis we had studied earlier. The spectrum of ozone in dilute acid as reported by H. Taube (43) (solid line) and M. Kilpatrick, et al. (40) (broken line) also has a peak at 2600 A and is of the same shape but slightly broader on the long wavelength side. This dierence led us to determine the spectrum of ozone in 7 M NaOH where it was found to coincide with the spectrum represented by the squares for the long lived 2600-A peak produced by flash photolysis. The possibility remains that the long lived 2600-A peak produced by flash photolysis is for the same species as that responsible for the broad peak in the neighborhood of 2400 A observed by G. Czapski and L. M. Dorfman (56) in the case of the pulse radiolyses of oxygenated aqueous neutral and alkaline water. The
Ratesof decay of thetronsient absorption a t and near 4 3 0 0 A.
spectrophotometer (20). Figure 14 presents the absorption spectra associated with the 4300 and 2600 A peaks (31). In the case of the 4300 A peak the flashed solution was initially 0.1 M in NaOH and 0.004 M in Hz02 (dots) or 0.001 M in NaSzOs (squares); the solid line is drawn through these data. The diffuse reflectance spectrum of solid sodium ozonide as obtained by Mcis represented by the circles. They Lachlan, et al. also obtained the dotted circles for the absorption spectrum of the species produced by the photolysis with light of 3650 A of a glass at - 173' containing a small amount of hydrogen peroxide in an equimolar mixture of about 8 M NaOH and 8M KOH. They attributed the spectrum to the ozonide ion Oa-. The near coioGdence of all these spectral data suggests that the 4300 A peak is indeed due to ozonide. The presence of the ozonide ion, 03-, as an intermediate in the flashed solutions suggested the equilibrium: 03-= 0% 0- in strong base or 03HzO = 0% +HO OH- in weak base. The 0- or hydroxyl radical HO in these equilibria suggested that
(a)
+
Table 4.
+
Half Lives of the 2600 and 4300 Absorption Peaks
0zone.O). tn voter 1
'
I
'
I
'
I
~
I
l
'
1 .
I
I
+
A
Transient
Figure 14.
Absorption spectra orsocioted with the transient peaks a t
i.
2 6 0 0 and 4 3 0 0 Also presented we the spectro of thespccier believed to b e rerponrible for there peokr 120, 311 bee text.)
Volume 43, Number 12, December 1966
/
629
peak at 2400 A, however, was very short lived compared to the life of the 2600 A peak observed by us for ozone and for our flashed solutions. They attributed their peak to 0%-. Additional evidence that the long lived 2600 A peak is due to a species such as ozone containing no unpaired electrons rather than to a species such as Oz- containing one or more unpaired electrons was provided by the fact that no electron spin paramagnetic resonance (epr) spectrum could be obtained for 7 M NaOH flashed peroxide solutions or for ozone solutions displaying concurrently the long lived 2600 A peak. The epr experiments covered the whole spectral range of the epr instrument under conditions where a spectrum would have been observed if there had been more than about mole of detectable spin in the 0.01 ml of solution in the epr cavity. If the lightabsorbing species had been 02-there would have been present in the cavity 10-8 mole of detectable spin based on the absorbance value of 1000 M-I cm-' at 2600 A obtained for 02-by Czapski and Dorfman.
Figure 15. Build-up and subsequent decay oftho 4300 A oronide obxlrplion peok produced in flashed baric~oivtionrofhydrogen peroxide at 27'C ond rotvrated with air mt one atrn ( 2 2 ) lree text.)
All the above data regarding the long lived 2600-A peak support the view that the species responsible for the peak is indeed ozone. Figure 15 presents the build-up (22) q d decay (22) of the transient absorption peak a t 4300 A in solutions saturated with air at 1 atm. The dots represent the results obtained on a flashed solution initially 0.002 M in H20% and 0.001 M in NaOH. The displayed build-up is for the period 5 M 0 rsec immediately following the onset of the flash. The linear part of the decay began 100 psec later. The apparent build-up may be due partly to incomplete recovery of the measuring system from light and electrical shock produced by the flash. The crosses in Figure I5 are for a flashed solution initially 0.0005 M in HzOzand 0.2 M in NaOH. The line depicting the decay covers the period beginning 0.3 msec after the flash and ending 37 msec later. The oscilloscope scanning speed in this case was too slow to reveal any build-up of the 4300 A absorption. 630
/
Journal o f Chemiml Education
peak in both the flashed The decay of the 4300 0.001 and 0.2 M NaOH solutions containing HzOe is represented by the equation:
This is the equation for a firsborder reaction of rate constant k where the haIf-life t.1, is given by the equation: t ~ , , = 0.7/k. It should be noted that Alog(03-) = A log(0D). Figure 16 presents the dependence of the rate constant, ko,-, upon the concentration of hydrogen peroxide. The experimental values are represented by x and @. The data are for Hz02solutions 0.1 M in NaOH which were a t 27°C and saturated with air. The concentration of the hydrogen peroxide is seen to have been varied one thousandfold from about 0.00001 to 0.01 M. The equation found to represent thevariation in koa- is:
where a, b, and d are constants under the prevailing conditions and [H202]represents the formal concentration of the H20z,i.e., the number n of gram formula weights of H202,namely, n(34 g), employed to make up a liter of solution. The solid line in the figure represents the equation and is seen to fit the data within the limit8 of error. Table 5 shows that the constant a increases about five hundredfold with increase in (NaOH) from 0.01 to 9.5 M. In addition a was found not to depend significantly upon the partial pressure of oxygen between 0 and 1 atm. The term b/ [HzOz] is negligible when [H202]is greater than about 0.001 M. The constant d can be seen from the table to decrease about fivefold with increase in NaOH from 0.01 to 9.5 M. In addition d was found to increase about threefold with increase in the partial pressure of oxygen from 0.2 to 1 atm. The relationship of the constants a, b, and c to the mechanism of the reaction will he presented later in this article. Figure 17 presents the build-up and decay of the 2600 A ozone peak and its relationship to the decay of the 4300 A ozonide peak (22). The data are for flashed solutions initially about 0.0001 M in NazS208 and 7 M in NaqH (Baker analyzed). The crosses represent the 4300 A decay beginning 2 msec after the flash and
I 2.5
5
10
l"lO., x
25 0.8
i10'
50
0
110
500
,000
le.l../,%r.ri-
Figure 16. Rote constants ot 27°C for oronide decoy over a one-thousand fold mnge in the concentmtion of hydrogen ~ e r o x i d e .
Table 5. Influence of the Sodium Hydroxide Concentration Upon the Decay of the Ozonide a t 0.2 atm and 27'C. Concentrations Are Expressed in Moles per Liter and Time In Seconds
(NaOH) 9.5 0.0095
1O5(H20.)
10-aa
108b
lo4d
1 . 0-100/3* 4.0-125/6
0.012 5.5
0.17 220
1.4 6.0
1.0-10013 signifies that the formal HSO2concentration was lowest at 1.0 X 10-6 M , and highest at 100 X 10" M and that three different solutions were employed each havina . . - a different value for HnOl. The equation for the rate constant far the decay in terms of the parameters a, b, and d is given in the text.
followed for 28 msec. The circles represent the 2600 build-up also beginning 2 msec after the flash but followed for 48 instead of 28 msec. The dots represent the 260&.4 decay in 7 M NaOH; this decay was followed in a 10-cm cell in the Model 11 Gary spectrophotometer for about 70 min while the OD decreased 0.05 units with a half-life of 1400 see. The build-up of the ozone is seen to take place at least in part during the decay of the ozonide. I n addition it is seen that the decay of the ozone, like the decay of the ozonide, eventually follows a first-order reaction in any given experiment. The rate constant for this decay also depends upon the environmental concentrations especially upon the concentration of base as can be seen from Table 4 where t.1, = 0.7/k increases over one millionfold between 0.01 and 7 M NaOH. Figure 18 presents the data for the decay of the transient absorption a t 2600 A observed after the flash during thevery early stages of the reaction. The data are for a series of flash photolyses on a single initially deaerated solution 0.002 M in sodium persulfate and 0.03 M in NaOH. A negligible amount of molecular oxygen was initially in the solution but oxygen acrumulated as the number of flashes was increased. The numbers alongside the lines indicated
that the decay represented by line 1 was obtained after the first flash, 3 after the third flash, etc. The time between successive flashes was a t least 2 min. (which was the time required to charge up the flash condensers) consequently, there was no carryover of the transients from one flash to the next. Two effects stand out. First, the haIf-life of the transient absorption at 2600 A is seen to increase from very short values, even somewhat shorter than that reported for 02-,to values of that for ozone as the oxygen accumulated in the solution, i.e.. the half-lives cor-
Time
Figure 18. Rapid initial and rubsequent much slower decay of the transient obrorption at 2 6 0 0 A; (See text.)
0.2
OZONIDE DECAY
Figure 17. Ozonide 14300 obrorptionl decay, ozone 12800 A obrorption) build-up accompanying the 4 3 0 0 A decoy and the eventud decoy of the long lived ozone peok a t 2 6 0 0 A 122).
responding to the linear part of lines 1, 3, 7, 8, 16, and 21 are progressively longer, e. g., about 50 and 1000 psee for lines 3 and 21, respectively, so that progressively longer oscilloscope scanning speeds and time scales were required to follow the overall decay. Second, the initial optical density values became larger with successive flashings indicating greater yield per flash of the transients responsible for the light absorption a t 2600 A or the replacement of one transient by another of higher optical density a t this wavelength or both as the molecular oxygen accumulated in the solutions. Eventually thevery rapid decay represented by lines 1, 3, and 7 was found to be attributable to the ion radical 0%-which has an absorption peak at about 2400 A as already mentioned (36). The slower decay and larger values for the initial absorption represented by lines 8, 16, and 21 were attributable to ozone which absorbs over three times more light per mole than 02at 2600 A. The oscilloscope scanning speeds employed to measure the slower decay were too slow to reveal the very rapidly decaying species or the build-up of the slower decaying species. Figure 19 presents the data for the transient absorpVolume 43, Number 12, December 1966
/
631
tion observed in the red end of the spectrum in a flashed deaerated solution containing initially 0.001 M NaOH and 0.001 M sodium persulfate. The optical density values were normalized to 1 at 6200 A. The half-life of the peak which was located at about 6300 A was less than 1 msec. The spectrum and its decayoare similar to that of the peak observed at 6300-7300 A in pulse radiolysis studies by E. J. Hart and J. W. Boag (38) and attributed by them to the hydrated electron (e- aq.). Reactions Explaining the Observations
The decay of the ozonide and formation of part of the ozone are explained by the following reactions in which all of the species are hydrated:
then explained by the replacement of HO by 0- which reacts more rapidly than HO with O2 to produce Oawhich in turn reacts rapidly with Oz- to regenerate 03,HO1.-, and OH-. The appearance and decay of the 6300 peak attributable to e-aq. and of the other observed phenomena not involving the decay of the ozone in the hydrogen peroxide solutions are explained by the reactions:
0-
O3-
+ 0%+ HIO
kr
Ol
(12)
+ HOZ- + OH-
(13)
- '
+ k6(0.)] = 2k, when (On)is ~ufficientlysmall
b
=
(k-,)kdOdB/k*
d
=
k-i(Od/kn
so b/d
=
k4(O2),At 27'C, kl
k-1 = 3.7.
The reaction: 202-
+ H*O
-
=
2.8 X loa sec-I and
+ HOn- + OH-
OZ
appears to he negligible compared to the reaction ka given above. The decay of the ozone in strong base is explained by the reactions: Os
+ 20H-
-
+
+
0% 2 0 HzO 01+0--0~+0~01 0%--20a 0-
+
+
(15) (16)
i
Figure 19. Abrorptlon rpectr~mof the tronlient responsible for the 6300 ohsorption peak produced b y the Rash photolyris of deoeroted bas'r water solrlions of tho p e r d o e s ot 2 5 ' C 1381.
/
0% HO
HOz-
0-
+
+ H20 + e -
03-
+ OH-
0% 01-
o*-03-
+ + + + + Os- + HOH + 02-+ H02- + OH-
-
03- H 0 2 - - 0 2 OH0- HOz- OH02-
01-
Journal o f Chemical Education
The primary reaction in the photolysis of the hydrogen peroxide, is not taken to be the cleavage of the peroxide, 0-0, bond by light into 20H from *H202or into 0- and OH from *H02- because this would not produce the partial scrambling of the oxygen atoms in the resulting molecular oxygen as observed by M. Anbar (35) by the use of different oxygen isotopes. We propose that the partial scrambling takes place mainly by cleavage of either of the two 0-0 bonds in the reaction between * H ~ z -and HOz- to produce HzO,03-,and e- followed by the cleavage of either of the two 0-0 bonds in 03-to produce O2 and 0-. Radiolysis instead of photolysis of hydrogen peroxide in water was found by Anbar (35) to produce very little, if any, scrambling. This is explained if the OH produced by radiolysis largely from the water reacts with H202or HOI- to produce H 2 0 and H 0 2 or 02-. The H o p or Oz- in turn may react with itself to produce OZand HZOZor O r or HOZ- by an H atom or electron transfer.
(17)
plus the reactions given for the decay of the ozonide. The enormous inhibiting effect of the base, OH-, is
632
Os-
0 1
The relationship between the rate constants for these reactions and the constants a, b, and d of eqn (10) for k 01 1s.. a = 2[k,
*HOn-
0% 0-
*HO2C e-
0-
+ OH-
+ HOn--Oz-
-+ ++ + +
oa-
ekc
--
+ light *HOr- + HOz-
HOz-
Conclusions
The results support the hypothesis that the photochemical oxidation of water to molecular oxygen by twoelectron acceptor oxidizing agents produces peroxides and that the molecular oxygen produced from water in the natural photosynthetic process probably also is via peroxides and perhaps even via the very short lived species: HO, Hot, and 03-. The results also support the hypothesis that light absorbed by the peroxide H20z or HOZ- does not directly cleave the 0--0 bond but produces instead a relatively long lived electronically excited state, namely, *H202or *H02- which subsequently either returns to the ground state or reacts with itself or molecular oxygen to produce 03-. The results provide an explanation for the production of molecular oxygen from water by light absorbed by the dimerized hydrolyzed ceric ions but not by light absorbed by the monomers in acidic perchlorate solutions (15). Both these dimers and monomers have about the same absorption spectrum (64) in the visible
and near ultraviolet down to about 2000 A. This spectrum is characteristic of charge transfer producing Ce(II1) and OH from the attached Ce(IV) and OH-. The Ce(II1) and OH before they can separate apparently react with each other to reform Ce(IV) and OH- unless the OH can react with another species attached to it. Such a species is not present in the monomer but is present in the dimer as the other Ce(1V) ion which may react thermally with the newly formed OH as it does with H202to produce Ce(II1) and the H+. The OH+ hydrated intermediate OH+ or 0 in turn apparently reacts with the water to produce H202. The H202may react with itself or with Ce(IV) to vroduce molecular oxygen probably via the HOz .- . intermediate. I n ceric solutions containing NOa- or SOp=,the OH radical could react with these species to yield OH- and NO. or SO4-. The latter could be ~roducedalso by direct tranifer of an electron to ~ e from : ~ NOa- or SO4= when attached directly to the Cef4 ion. The NO3may eventually produce Oz. The SO4- instead of producing O2 probably reacts largely with the newly formed Ce(II1) to produce Ce(1V) and S o h since SOpstrongly quenches both the photochemical and thermal oxidation of water by Ce(1V).
+
Summary
Flash photolysis studies of the production of moleonlar oxygen by ultraviolet light absorbed by the peroxides H202,HOz- and S20sZ-have revealed the forma, tion of transient absorption spectra attributable to hydrates of the following intermediates: the electron, e-; ozonide, 03-; ozone, Oa; and the superoxide, HOz or its ion 02-. The hydroxyl radical HO and its ion 0- are also probably present. The flash and auxiliary apparatus and procedures are described. The outstanding light absorbing transients in alkaline solutions at room temperature are ozone and ozonide. I n 0.01M NaOH the half-lives of both these species are about a millisecond. I n 7M NaOH the ozonide and ozone live about ten and over one million times longer, respectively, the half-life of the ozone being about forty minutes. Reactions are given which explain the observations. The results cast light upon the path followed by the oxygen atom from water to molecular oxygen and provide a reasonable explanation as to why the in vitro oxidation of water to molecular oxygen is brought about by light absorbed by hydrolyzed ceric ion dimers but not by light of the same wave length absorbed by hydrolyzed ceric ion monomers in acid-cerium-perchloratewater systems. The cerium system is of special interest because the Ce(II1) and Ce(1V) in it act as catalysts in the photochemical decomposition of water into gaseous hydrogen and oxygen. The results provide also a basis for speculation regarding the intermediates in the in vivo oxidation of water to molecular oxygen by the natural photosynthetic system. Acknowledgments
Vincent R. Landi carriedout the flash photolysiswork. Barbara A. Wordell, now h'lrs. John L. Rabaglia, rendered technical assistance. The author held a
John Simon Gnggenheim Memorial Foundation Fellowship during part of the work. Appendix
The cerium catalyzed in vitro process for the photochemical storage of energy is reversible as is the natural photosynthetic process with regard to the material initial reactants and final products. Its discovery (15, 17, 19, 21, 25-25), as already noted, was the result largely of netsquantum yield measurements employing light of 2537 A for the reduction of ceric ions, to cerous ions, and for the oxidation of Ce(II1) to Ce(1V) in water solutions containing initially perchloric acid and largely either Ce(IV) or Ce(II1) perchlorates. The variations in the yields for the conversion of Ce(IV) to Ce(II1) produced by variations in the initial concentrations of Ce(IV) and Ce(II1) led to the conclusions that reduction of the Ce(1V) was initiated only by that part of the light absorhed by the hydrolyzed ceric ion dimers of all the light absorbed by all the ceric species and that oxidation of the Ce(II1) to Ce(IV) took place when part of the light was absorhed by the Ce(II1). The formulas of the dimers are given in Table 3 (15). The variations in the yields for the conversion of Ce(II1) to Ce(IV) produced by variations in the initial concentration of perchloric acid (21) led to the conclusion that the oxidation of the Ce(II1) occurred only when the light absorbed by the Ce(II1) brought about electron transfer from the Ce(II1) to the attached water or protonated water, H30+, and that hydrated hydrogen atoms, H, or hydrogen molecule ions, H+z, produced in this way eventually oxidized some of the Ce(II1) to Ce(1V). The H2+, therefore, had to be a strong enough oxidizing agent to oxidize the Ce(II1) and it had to be a weak acid to be produced from the H atoms and H+. I t was indeed shown (21) that the half reaction potential of the HZ+is large enough to bring ahout the oxidation of the Ce(II1) to Ce(IV) while the H,+ is reduced to Hz. It was also found that the Hz+ is a weak acid of about the same strength as HS04-. It follows that one Hz+ also is a strong reducing agent which can reduce Ce(1V) to Ce(II1) while the Hz+ is oxidized to 2H+ thereby decreasing the yield of H2. It follows that Hz+ can also oxidize and reduce itself to 2H+ and Hz thereby decreasing the yield of Hz. The net quantum yield measurements were made with monochromatic light because the efficiency of light in producing chemical reaction often varies with the wavelength of the light and with the fractions of the light absorhed by the different species in the system. I t is axiomatic, but sometimes overlooked, that light is most efficient in bringing about the measured reaction when it is a t a wavelength that is absorbed by the species most efficient in producing the reaction. The quantum yield work has been expedited by the development of intense discrete light sources, efficient ways of rendering light monochromatic, precise actinometers for the measurement of the light quanta incident upon the system, and analytical volumetrical methods employing colorimetric or potentiometric endpoints to determine the small concentration changes Volume 43, Number 12, December 1966
/
633
often produced by photolysis under adequately controlled conditions. Our first extremely intense discrete light source is described in (1) and (3). The lamp is especially useful for slit illumination and the irradiation of small volumes of material. Mercury vapor arc lamps of this kind are now commercially available. They are useful for the production of light at 2650, 2800, 3020, 3130, 3650, 4040, 4360, 5460, 5780, and 10140 especially $t 3020, 3130, 3650, 4040, and 4360 A but not at 2540 A. The spectral energy distribution of our lamp is given in (4). Other intense discrete light sources include sparks as well as arcs. A mercury arc source operating at atmospheric pressure is described in (2). A mercury arc lamp operating at low pressure is decribed in (11). Information regarding the performance of the lowpressure lamp is given in (18). Spark sources are described in (7). The light from these asparks suitable for photochemical work is at 1940 A from aluminum 1990, 2030, 2060, 2140, 2540, 2610, 2800, 3080, 3280, 3300, 3350, 4060, and 4800, especially at 2140 and 2800 A from zinc, and at 2270, 2290, 2980, 3130, 3260, 3470 3610, 4680, and 4800 4, especially at 2270 and 2980 from cadmium. Most of our quantum yield measurements a t present are carried out with monochromatic light of 2540 A. This light is obtained from a low-pressure mercury vapor arc lamp operating in an arrangement of apparatus $18) by means of which about half of the light of 2537 A can be used for the irradiation of either large or small volumes. The rendering of light monochromatic has been accomplished by the use of monochromators (4), focal isolation (7), and filters (11,18). Our fused quart.2 monochromator (4), useful down to 2600 A, had a 60' prism 12 cm on a side and 14 cm high and plano-convex lens about 15 cm in diameter each with a focal length of 35 cm (f 35/15) for red light. The instrument provided automatically for focusing and for keeping the prism in the center of the field and at the position of minimum deviation. The entrance slit was curved to make the exit beam rectangular for light of 3660 A and vicinal wavelengths. Our cry~tal~quartz monochromator (6), useful in air down to 1900 A, had a 60' Cornu prism 8.5 cm on a side and 6.5 cm high and plano-convex lens about 8 cm in diameter each with a focal length of 12 cm (j1.5). Focal isolation (7) is useful for the isolation of the groups of lines centered at 1240, 2100, 2540, 2800, and 3350 A from zinc, at J940 A from aluminum, and at 2280, 3000, and 3610 A from cadmium. Filters are useful when-employed with care for the isolation of light of 2537 A from the low-pressure mercury vapor arc lamp. A satisfactory filter combination (11, 18) for use at 0 to 100°C consists of a 5 cm or greater depth of water which absorbs the infraredlight and light of 1849 A and a 1 cm depth of dry chlorine at 1.5 atm which absorbs the light between 2600 and 4500
A.
Four actinometen have been employed. All are aqueous solution$. The vandate-tartrate system employed at 4360 A and longer wavelengths will be described soon in the literature. The other actinometers contain the photosensitive species:uranyl oxalate (9,IS,), 634
/
Journol of Chemical Education
U0z(C20r),2-2=, or ferric oxalate (%9), Fe(C20&-2n, or persulfate (13, 16), S208.= The overall photochemical reactions in these systems are respectively: ~
~
+ 0H20~= 2- H S 0 4 + 0 d 2
The gross quantum yields for these actinometers, namely, the moles of oxalate or persulfate destroyed by the light per einstein (mole of light quanta) absorbed by the system as a whole are all about 0.6 for light of 2540 A when the above stated light sensitive species absorb over 90% of the light. The ferric oxalate actinometer is usually considered to be the most light sensitive because the amount of Fe(I1) produced by the light can be determined very precisely by the use of the orthophenanthrolene (ferroin) indicator (29). The actinometers should be employed only after consulting plots of their absorption spectra which vary significantly and differently with changes in the composition and temperature of the system (25). The gross quantum yields also depend upon these variables as well as upon the wavelength. The dark reactions in the actinometers are negligible near room temperature but not necessarily above 50% Adequate mixing of the actinometer solutions is advisable during their irradiation; otherwise, the quantum yields may decrease significantly with the extent of reaction unless the attenuation of light by the actinometer is less than 80% over a light path of 1 cm or more and then corrections may be required for the light transmitted. The evaluation of both gross and net quantum yields for the system under investigation usually can be made only after aLso consulting plots of the absorption spectra of the system and of the individual species in the system as well as evaluating the relative concentrations of the light absorbing species (S5,28,33). Literature Cited (1) HEIDT,L. J. AND D-~NIELS,F., J . Am. Chem. SOC., 52, 2151-2152 (1930). (2) . . FORBES.G. S. AND HEIDT,L. J. J . Bm. Chem. Soc., 53, 434914350 (1931). (3) DANIELS,F., AND HEIDT,L. J., J . Am. Chem. Sac., 54, 2381-2384 (1932). (4) HEIDT,L. J. AND DANIELS,F., J. Am. Chem. Soc., 54, 2384-2391 (1932). (5) HEIDT,L. J., J. Am. Chem. Soc., 54, 2840-2843 (1932). Presented are data showing that the quantum yield is about unity for the photolysis by light of 3130 A of H201at 1.54.5 M in H.S04 at 0 . 0 4 . 1 M . The results also show that the H02- from H201 absorbs light of 3130 mare strongly than does H20*. (6) FORBES, G. S., XISTIAKOWSI