Monday, June 22, 2009

Good question

When last seen Eli was providing an example of how Science lurches forward, a nice example, where a flawed experiment loosened a tidal wave of publicity, only to end in a quiet fizzle. Some of the bunnies asked good questions and the Rabett will here endeavor to provide some answers. The question was whether excited state NO2* could react with water to form OH radicals as first proposed about a year ago in Science, by Sinha's group. This was questioned (we are being nice) by Dwayne Heard and colleagues recently and there was a reply. Eli did not think much of the reply although he did not go into it at the time.

Chuck asked

Is zero zero, or is zero their detection limit, after background subtraction?
for4zim thought that Li, et als answer was fine and Ankh had a few
Isn't the problem here that their conclusions come from logical reasoning rather than from collecting relevant data?

They got a different result, using a different method and different materials. Well, okay.

So wouldn't they also run the experiment without using the concentrating lens, just to reduce that difference?

They used different source chemicals. I don't know if that's something their setup allows them to experiment with or not.
Carr, Heard and Blitz first calculated the amount of OH radical that would be formed if Li, Matthews and Sinha were correct. They then generated OH, by photolysis of acetone at 280 nm followed by its reaction with O2 to form OH

(CH3)2CO + hv (280) --> CH3CO +CH3

CH3CO + O2 --> OH + other products

They detected the OH from this process as shown by the red dots in the figure to the right, which establishes that they can detect the radical at the levels implied by Li, Matthews and Sinha. The acetone was used to calibrate the signal. The thin black line shows the expected OH signal if Li, Matthews and Sinha were correct.

Now let's look at the response from the Shinha group. There are some interesting ideas there that will help understand what happened, even beyond the Heard group's objection. We can summarize the response as follows
1. Focusing was necessary to produce enough excited NO2* the absorption coefficient of NO2 is small.

2. A different calibration photodissociation was used for OH

3. Carr, et al do not suggest a mechanism for production of the OH that Li, et al observe (Eli is going to start naming the groups by their leaders, Heard and Sinha respectively for convenience) and they also did not push the intensity high enough to observe multiphoton production of OH

4. The non-zero intercepts in the power law measurements are characteristic of what is found in the Sinha lab for other situations including their OH calibration by photodissociation of HNO3 and CH3OOH.

5. Other than the offset, the linear power dependence shows that the process was first order in the laser intensity. Therefore little to no doubly excited NO2 was produced

6. Introduction of H2 into the system did not result in production of OH from O(1D) + H2O, a well known reaction. This shows that no O(1D) [the first electronically excited state which will not collisionally or radiatively quench to the ground state by collision in almost all cases] was produced by multiphoton processes

7. Any ground state O(3P) would have to be improbably hot translationally to react with water vapor producing OH

8. Higher lying quartet states of NO2 are unlikely to react.

9. The action spectrum (you tune the laser, monitor the OH) matches the absorption spectrum of NO2 showing that the excitation was not multiphoton

10. Transfer of energy from NO2* to H2O followed by reaction of other NO2* with the vibrationally excited H2O would depend on the square of the NO2 concentration.
Of these, 6 is by far the strongest. However, let us turn to the others starting with 1, 5 and 9 which are related. The absorption cross-section of NO2 around 570 nm is ~ 10-19 cm-2 which is not huge, but not vanishingly small like the O2 absorption @ 760 nm ~4 x 10-24 cm-2. Sinha focused their laser, which depending on this or that made the diameter of the focal volume something smaller than 50 microns to be conservative. Thus the cross-section would be ~10-4 cm2.

They used 12-15 mJ @ 55o nm to excite the NO2, which can be converted into a number of photons using E = hv yielding ~3 x 1016 photons/pulse, the number divided by the cross-sectional area of the focus, F, is
F = 3 x 1016 photons/10-4 cm2 = 3 x 1020 photons/cm2 .
If the transition is not saturated the fraction of NO2 molecules excited is the product of F and the absorption cross section. If it is saturated we have to include stimulated and spontaneous emission and carefully account for transitions to other electronic states. Let us see what we get for Sinha's conditions assuming no saturation.
P = 3 x 1020 photons/cm2 x 10-19 cm-2 = 30
Hmm, obviously saturation is going to be an issue for the focused condition, and the relative number of NO2* to NO2 ground state molecules may become a serious concern. On the other hand, for unfocused conditions (assuming a beam area of 1 cm2 which is maybe an upper limit for the lasers used, 0.1 cm2 might be slightly better - personal communication E Rabett) the fraction of NO2* will be about 0.003 and saturation will not be an issue.

Points 5 and 9 are tightly coupled. For molecules and multiphoton excitation (MPE), one distinguishes between a simultaneous and step-wise aka sequential multiphoton excitation. In simultaneous MPE the molecule is never in an intermediate state and simultaneously absorbs or emits two or more photons. If such a transition is not saturated (and you can saturate it by focussing hard or using a room sized laser or using a femtosecond system) the result depends on the intensity of the laser to the power of the number of photons absorbed.

In sequential MPE, the molecule absorbs a photon and is excited to another electronic state in which it remains for some time longer than a few fs, and then absorbs another photon, is excited to a yet higher excited electronic state, and then, maybe another, etc. This is what happens when multiphoton excitation is used for ionization (MPI). Sequential MPE and MPI are expecially effective for molecules because the number of states available increases rapidly (almost exponentially) with energy, so there are always more states available the higher you go.

Because of this, the action spectrum that will be observed is always the action spectrum associated with the "hardest" step, usually the first, and Sinha's point 9 is simply contrary to experience. Such stepwise transitions have been seen by Romanini in cavity ringdown spectra of NO2 at much lower intensities
In addition, in coincidence with absorption by these near infrared transitions, an appreciable fluorescence signal was detected in the visible range. According to our interpretation, this fluorescence is from NO2 levels excited by two photons in a stepwise incoherent process, with a strongly allowed second step. Since the fluorescence spectrum has the same lineshapes as the CRDS absorption spectrum, it seems that the first transition step is the one limiting the overall two-step process.
Indeed, you can get very high up, even with much less energy then Sinha used, as shown by the fact that in 1983 emission was seen from the Schumann-Runge bands of O2 following focussed excitation of NO2 (open article), and there are lots of studies of multiphoton ionization following focussed laser excitation of NO2 (Ed Grant was hot on that in the 1980s).

In short, Sinha's conditions favored significant multiphoton excitation, it occured as sure as Eli likes carrots. While it is not clear what highly energetic fragments molecules were produced, it is clear that a large variety were available including electronically excited fragment molecules, ions, atoms, radicals (NO is an honorary free radical having one unpaired electron) and more.

That takes out the rest of the points with the exceptions of 2, 4 which is simply bad technique and 6 showing that there was little to no O(1D). Point 2 is not really important, both groups have previously demonstrated that their methods work for calibrating the amount of OH and other radicals. Both groups and their father, mother and grandparent groups have excellent reputations for such stuff.

Thanks for getting the bunny even more interested in this.

It will get prettied up tonight maybe even with a couple of more pictures.


jyyh said...

It almost looks like H* and O* don't like to mix, though in principle (chemstry-wise) there should be none of that, or so I see it. Are the respective orbitals of the wrong shape? N2 has a nice round orbital as well as O2 and H2, while the heterogenous NO is somewhat slanted towards O, but little less than in OH-(aq). How about using some other excitation frequencies? Or is this a manifestation of the strong nuclear force over possible chemical reactions?

Hank Roberts said...

I will be grateful when you can dumb this down enough for me to follow.
No hurry.

EliRabett said...

Simple version: Sinha et al took a howitzer to the NO2, creating just about every type of fragment you could imagine except O(1D). Their claim that they only absorbed one photon is risible. We were doing this sort of stuff 30 years ago, and understood what we were doing better (which is why, after a while we stopped doing it)

Interestingly, Eli came across a series of papers by Javan (invented the CO2 laser) and separately by MC Lin, where they passed a CW Ar ion laser through mixtures of NO2 and a lot else. Reading between the lines, they didn't get much reaction with the 514 nm line, but with the 488 nm line they did. Also they never published on NO2 + H2O, which, even at the time was an obvious thing to try. They did on NO2 + CO. FWIW

jyyh said...

inducing specific reactions in gaseous phase (or in droplets) via accurately targeted laser pulses... sounds like atmospheric engineering in this context. why meddle with sulfates and such if (only) this could be done... i don't know if i got this correctly.

Bryn said...

There are two sharply critical comments on the Science website linked to the Marcott et al paper.

Can anyone point me to substantive and helpful replies which deal with the issues they raise?


Anonymous said...

Bryn, the first comment is hilarious. Arno Arrak, not unknown as a pseudoskeptic, states Marcott et al used tree ring proxies. Fail 1. Second, he claims McIntyre found the uptick to be due to the redating. Fail 2 (see at the end). He even questions the greenhouse effect, based on a single paper and the *fake* temperature "pause" (people seem to ignore the deep ocean when it suits them).

Matthews' criticism are debunked at tamino's:

Many entries on the issues there, including the uptick (Matthews should ask himself why Tamino had no problem reproducing the results, whereas he could not).

Rather problematic in Matthews' comment is his referral to Marcott's thesis. Anyone working in science knows that results presented in a thesis are hardly ever referred to in publications, especially when it is the author of the thesis that also writes the paper. Second, in 2 years understanding may develop further. I know hardly a PhD student of mine who has had a thesis chapter published without some significant changes. Anyone with even the remotest scientific ability will see the thesis contains one approach and the paper multiple approaches to create the reconstruction. If Matthews did not check this, he's a bad scientist. If he *did* check this, but didn't notice, he's not qualified enough to comment. And if he *did* notice and still wrote his comment, he's deceptive.


EliRabett said...

Bryn, why don't you join the group at a recent post on Marcott etc. not to say that M. did not do a good job of answering your question.