Is science sometimes in danger of getting tunnel vision? Recently published ebook author, Ian Miller, looks at other possible theories arising from data that we think we understand. Can looking problems in a different light give scientists a different perspective?

Share this |

Share to Facebook Share to Twitter Share to Linked More...

Latest Posts

Archive for March, 2013
In a previous post, I argued that the issue of whether cyclopropane ring electrons are delocalizable was not exactly handled well. By itself, that may seem to be merely an irritant, but the question now is, how widespread is such an issue?
 
In a recent edition of Nature there was an article by Robert Antonucci, (Nature 495: 165 - 167) who argued that the scientific community was failing when trying to explain quasars. A quote: "In my opinion, the greatest limiting factor in understanding quasars is not a lack of intelligence, effort or creativity, nor is it a dearth of fantastic new facilities. It is a widespread lack of critical thought among many researchers. Theories are being published that have already been ruled out by observations. Observers cling to falsified theories when interpreting their data. Most of the AGN community is mesmerized by unphysical models that have no predictive power."
 
This is fairly stern stuff! It got worse. He accused scientists of continuing to use and refine overly simple versions of models that include disproved assumptions and which do not match observations without lots of special pleading.  Observers were not left out: “Some astronomers like to see what they believe." Even worse was to come. In 1984 a temperature was measured and found to be in accord with the disk accretion model. Later, an amateur found the calculation involved a factor of ten missing in Newton's gravitational constant! The correction, and the fact that the method now fails to account for observation, is hardly cited, while the original paper has 100 citations. He complained that this scientific community was producing fewer and fewer theoretical papers, but there is a burgeoning effort to find more examples, leading to statistical analyses leading to further problems, such as claiming causal links in plots of dependent variables. 
 
The question now is, is Chemistry in a better condition? I do not think so. How much original theory, as opposed to opaque computations, have you seen lately? My guess is, not many. How many of you think there is no further theory to find? I think the problem lies in reductionism. Everyone seems to believe that all chemistry is a consequence of the Schrodinger equation, but that cannot be solved. Therefore there is no point looking further. That, in my opinion, is simply false. I am not doubting that the Schrodinger equation is generally correct, but that does not mean that the only way to produce theoretical work is to solve it.
 
Final advice from Antonucci: "I urge my junior colleagues to spend 15 minutes every day thinking, palms down, eyes on the ceiling." Follow a Californian bumper sticker: "Don't just do something, sit there".
Posted by Ian Miller on Mar 24, 2013 10:45 PM GMT
Many/most scientists would probably say, no, you cannot; all you can do is to falsify a theory, while you believe a theory to be true because all evidence supports it. This raises the problem, what happens when the evidence that contradicts the theory are suppressed?
 
Thus further to my previous posts, there were further subtle cues as to why cyclopropane did not demonstrate conjugation. For example, if you have a sequence of olefins, pronounced conjugative effects are demonstrated. On the other hand, while a cyclopropane ring adjacent to either positive charge, or to potential positive charge such as with UV transitions, gives effects similar to conjugative units, add a second cyclopropane ring to the first, thus have two in a row, and the second one has no noticeable effect. On the other hand, if we put three cyclopropane rings around a positively charged centre, the effects are very close to being additive, which does not seem to happen with cross-conjugation.
 
Similarly, with the cyclopropylcarbinyl carbenium ion, you would expect the bond to the carbinyl centre to either make an angle of 120 degrees to the plane of the cyclopropyl ring (as required by the Walsh MO treatment) or approach closer to this angle as the ion forms, but it does not. Instead, the centre moves towards the cyclopropane ring, as if there were an attractive force pulling it. That, of course is exactly what should happen with my polarization field. While the fact that cyclopropane stabilized adjacent charge was taken as proof of conjugation, the associated minor details that contradicted that proposition were ignored.
 
An observation can be used to prove a scientific statement, provided you can write it in the form: “If, and only if, theory X is true, then you will observe Y”. The observation of Y proves theory X is true, as stated. Of course it may be incomplete, but it will be true as far as it goes. The problem is to justify the ”only if” part of the statement, because how can you know that there is not an alternative that has not been thought of yet?
 
The reason I have been writing these blogs on cyclopropane conjugation is not to justify my own youth. From a personal point of view, I could not care less whether anyone believes me, although I do feel that everyone should have the opportunity to consider the issue for themselves. If people want to believe the Earth is flat, well, I cannot do much about that. But people cannot form reasonable views on such matters if the “trivial details” that falsify a theory are suppressed. A review should be critical and complete, not merely fashionable. But suppose, you argue, the reviewer does not know about these details? That is why I think we need a new form of review, like the wiki, where everyone can contribute, and a number of moderators bring order to what is produced. . What do you think?
 
One final comment on this. One reason why everyone said cyclopropane conjugates was because they expected it to, because molecular orbital theory, mainly the CNDO/2 version popular at the time, and also a more sophisticated version of MO theory championed by John Pople, said it would. Remember, molecular orbital theory starts by assuming total electron delocalization, and special reasons are required to produce bond localization. As Aristotle would have said, to find delocalization when you assume it in the first place is not a great achievement. More on this issue later.
Posted by Ian Miller on Mar 18, 2013 1:50 AM GMT
How do you tell which of two theories is likely to be correct? The answer is that each gives a set of predictions, and you have to find an experiment where the two theories predict discernibly different effects. More formally, you cannot state that one theory applies and the other does not from data in the intersection of the two sets. Thus one could not decide whether cyclopropane conjugates with adjacent unsaturation from the fact that positive charge adjacent to a cyclopropane ring is stabilized, because both the electron delocalization theory and my polarization field theory gave essentially the same prediction that positive charge would be stabilized. Worse, calculations showed that to within the uncertainties inherent in each calculation, the two gave essentially the same degree of stabilization: a little over 100 kJ/mol. for the bare carbenium ion in a vacuum. On the other hand, the effects were qualitatively opposite for negative charge. As noted in an earlier post, a case could be made that the required destabilization occurred, and there was certainly no evidence of significant stabilization, but it was difficult to say this was definitive. Then I got lucky: key evidence was published.
 
One further piece of evidence sometimes quoted in favour of cyclopropane conjugating was that cyclopropane adjacent to a chromophore generally gave a bathochromic shift, and an enhanced extinction coefficient. Now, to absorb electromagnetic radiation, to reach the excited state, the system must undergo a change in electric moment, and the probability of a photon being absorbed is proportional to the change in electric moment. Thus something like benzene must have an instantaneous dipole moment in the excited state. The net effect is probably most easily seen using the canonical structure representation, even if it is not strictly accurate. The net result is that for most transitions, a positive charge can be adjacent to the cyclopropane ring in the excited state, hence the polarization field interpretation predicts a bathochromic shift and an enhanced electric moment, exactly the same as does the conjugation theory.
 
It was some time after this that for me a key observation was made: the change of electric moment was measured for the n → π* UV transition of formaldehyde. The important point was the change of electric moment was from oxygen to carbon, hence the same transition on a carbonyl adjacent to a cyclopropane ring would lead to a change of dipole moment with the negative end directed towards the cyclopropane ring. Now, that change of electric moment would interact with my proposed polarization field, which would lead to any strained compound giving a hypsochromic shift to that transition when compared with alkyl. This was important, because it was well-known that conjugative effects give a pronounced bathochromic shift to all such transitions. For example, the transition in acrolein has a bathochromic shift of approximately 25 nm from a saturated aldehyde. I used my pseudocharge to calculate the magnitude of the hypsochromic shift for some strained systems, and got the shift for cyclopropane to within a half a nanometer. (There was probably a certain amount of luck there because observation of these transitions gives broad signals, and picking the maximum is a little subjective.) Of course I could also calculate proportional shifts for π → π* transitions, which have bathochromic shifts. An interesting point here is that it was thought that a carbonyl adjacent to the bridgehead of bicyclobutyl had no n → π* transition. According to my calculation, it would have one but it would be buried underneath the π → π* transition, a consequence of the larger shifts due to the higher strain moving them in opposite directions and thus eliminating most of their separation.
 
So, a triumph? Well, actually, no. Two reviews on the issue of electron delocalization in cyclopropane came out around this period. The first (Bul. Soc. Chim. France  1967, 357-370.) simply stated that the hypsochromic shifts occurred, but they were unimportant! The second (Angew. Chem. Internat. Ed. (Engl.) 1979, 18, 809-886) got around this problem by simply ignoring it. It also ignored my work, and worse, it ignored all the references I had found to work that suggested there was no electron delocalization. That is not the science that I signed up for.
 
The problem with reviews is that once one is declared definitive, there is no place to debate a review. I later wrote a review that found over sixty different types of observation that falsified the delocalization theory but I could not get it published. Accordingly, I and the textbooks disagree on this matter.
 
Posted by Ian Miller on Mar 11, 2013 2:35 AM GMT
Since I found nothing during February relevant to my theory of planetary formation, I thought I should outline why I think we need an alternative. The following is a very condensed look at the giant planets, and my ebook (Planetary Formation and Biogenesis) has much more detail.
 
The standard theory for a giant is that solids come together by some unknown mechanism and form planetesimals, and these, through gravity, form larger bodies, and finally planets. It is usually assumed for giants that this takes less than a million years (My), then over a period of time that depends on the assumptions, these collide to form larger planets, until they reach about 10 times the mass of the Earth, then they start accreting gas as well. (Actually, they will accrete gas by the time they get to the size of Mars, but such early atmospheres contribute little to the mass. They then take about 10 – 15 My to get to a size where runaway gas accretion starts. So, what are the problems. I consider some of these to be:
(1) After 60 years, there is still no firm idea how the planetesimals form, therefore the distribution of them is simply an unverified assumption,
(2) Simulations agree that planetesimal collisions to reach Earth take about 30 My, so how does Neptune get there so fast, when matter density is much lower and velocities are much slower? Collision probability depends on the square of particle density, and initial particle density is proportional to r^-q, where q is usually taken as 1.5, although that too is an assumption, and it could be 2. If the average body contains n initial particles, particle density is now 1/n initial particle density.
(3) If material comes together by collision, to get things to go fast enough, relative velocities have to increase as particle size increases, so why do the bodies simply not smash to pieces, assuming they do form?
(4) The star LkCa 15 is approximately 2 My old, it is slightly smaller than the sun, and it apparently has a planet of nearly 6 times Jupiter's mass at about 16 A.U, or about three times as distant from the star as Jupiter.
 
In my opinion, (4) is critical. Accretion disks last between 1 – 10 My after primary accretion, so the LkCa 15 system is a very young one, so how did its gas giant get so big? Obviously, everything has to happen a lot faster than under standard theory. What are the possibilities? To start with, standard theory ignores chemistry, so what happens if we include it?
 
My concept is that the initial cores grow like snowballs. In the outer disk water and silicates condense to form amorphous particles that adsorb other gases (Icarus 63: 317-332) and retain them to past the melting point. As the particles fall inwards, the temperature rises, and at some point, occluded volatiles that have passed their melting point are emitted. If, however, the melting point is not reached, the volatile is retained, more or less as a solid, and fills the pores. Suppose two particles collide. If they are sufficiently below a melting point, they bounce off each other, but if the volatile can melt, the energy of collision is absorbed in melting it, in other words, kinetic energy is converted to heat and the collision is, for the moment, inelastic. Now, the liquid trapped in pores between the particles cannot escape, but it can merge, then when it cools, it solidifies, thus we have pressure-induced melt-welding of the particles. This is similar to how a snow-ball grows with pressure. If so, then we look at the ices, these are (separated into subsets that have similar melting properties) in order of decreasing melting points (temperatures in degrees K): {water (273)}, {methanol/ammonia/water eutectic and CO2 (164-195)}, {CH4 and Ar, (84-90)},  {CO and N2 (63-68)}, and {neon (25)}.
 
There are, therefore, zones where ice can accrete into larger bodies, which depend on the temperatures in the disk. The surfaces of the disks usually have temperature proportional to r^-0.75, but the interior should retain heat better. If we put the index = -0.825, and assume Jupiter is the optimal place for a water-based core, then we predict the solar system as Saturn (water/ammonia/methanol) at 7.8 – 9.6 A.U. (actual, 9.5); Uranus (methane/argon) at 20-21.7 A.U. (actual, 19.7); Neptune (CO and N2) at 28 – 31 A.U. (actual 30) and possibly a planet based on neon at about 95 A.U. The satellites are based on the same compositions so we predict the Jovian system to be based only on water (the rest having volatilized); the Saturnian system to have ammonia and methanol (which can undergo chemistry to produce nitrogen and methane, which explains why Titan has an atmosphere and Ganymede does not); the Uranian system to be the slowest starter, because methane and argon are relatively minor components, but which will grow faster than Neptune once total accretion gets under way because matter density is greater, while Neptune will initially grow faster than Uranus, because nitrogen and carbon monoxide are common, but slower when gravity becomes the driving force. As far as I know, this is the only theory that requires Neptune to be bigger than Uranus to start with, and always to be denser. It also predicts that planets grow proportional to their cross-sectional area, because they grow initially in a flow of ice particles, all of which are continually renewed by the stream of gas heading starwards. By not involving collisions between equally sized objects, the rate of formation increases dramatically. Note it also predicts no life under-ice at Europa because the Europan sea will be deficient in both nitrogen and carbon. There are thin atmospheres around the major Jovian moons (Thinner than the gases in a light bulb!) but on Europa, there appear to be no nitrogenous species to 7 orders of magnitude less than the major species. What do you think? 
Posted by Ian Miller on Mar 4, 2013 1:35 AM GMT