Nano and the oil industry

I went to an interesting lunchtime talk today by Sergio Kapusta, former chief scientist of Shell.  He gave a nice overview of the oil/gas industry and where nanoscience and nanotechnology fit in.   Clearly one of the main issues of interest is assessing (and eventually recovering) oil and gas trapped in porous rock, where the hydrocarbons can be trapped due to capillarity and the connectivity of the pores and cracks may be unknown.  Nanoparticles can be made with various chemical functionalizations (for example, dangling ligands known to be cleaved if the particle temperature exceeds some threshold) and then injected into a well; the particles can then be sought at another nearby well.  The particles act as "reporters".  The physics and chemistry of getting hydrocarbons out of these environments is all about the solid/liquid interface at the nanoscale.  More active sensor technologies for the aggressive, nasty down-hole environment are always of interest, too.

When asked about R&D spending in the oil industry, he pointed out something rather interesting:  R&D is actually cheap compared to the huge capital investments made by the major companies.  That means that it's relatively stable even in boom/bust cycles because it's only a minor perturbation on the flow of capital.  

Interesting numbers:  Total capital in hardware in the field for the petrochemical industry is on the order of $2T, built up over several decades.  Typical oil consumption worldwide is around 90M barrels equivalent per day (!).   If the supply ranges from 87-93M barrels per day, the price swings from $120 to $40/barrel, respectively.  Pretty wild.

Short term-ism and industrial research

I have written multiple times (here and here, for example) about my concern that the structure of financial incentives and corporate governance have basically killed much of the American corporate research enterprise.  Simply put:  corporate officers are very heavily rewarded based on very short term metrics (stock price, year-over-year change in rate of growth of profit).  When faced with whether to invest company resources in risky long-term research that may not pay off for years if ever, most companies opt out of that investment.  Companies that do make long-term investments in research are generally quasi-monopolies.  The definition of "research" has increasingly crept toward what used to be called "development"; the definition of "long term" has edged toward "one year horizon for a product"; and physical sciences and engineering research has massively eroded in favor of much less expensive (in infrastructure, at least) work on software and algorithms. 

I'm not alone in making these observations - Norm Augustine, former CEO of Lockheed Martin, basically says the same thing, for example.  Hillary Clinton has lately started talking about this issue.

Now, writing in The New Yorker this week, James Surowiecki claims that "short termism" is a myth.  Apparently companies love R&D and have been investing in it more heavily.  I think he's just incorrect, in part because I don't think he really appreciates the difference between research and development, and in part because I don't think he appreciates the sliding definitions of "research", "long term" and the difference between software development and physical sciences and engineering.  I'm not the only one who thinks his article has issues - see this article at Forbes.

No one disputes the long list of physical research enterprises that have been eliminated, gutted, strongly reduced, or refocused onto much shorter term projects.  A brief list includes IBM, Xerox, Bell Labs, Motorola, General Electric, Ford, General Motors, RCA, NEC, HP Labs, Seagate, 3M, Dupont, and others.  Even Microsoft has been cutting back.  No one disputes that corporate officers have often left these organizations with fat benefits packages after making long-term, irreversible reductions in research capacity (I'm looking at you, Carly Fiorina).   Perhaps "short termism" is too simple an explanation, but claiming that all is well in the world of industrial research just rings false.

News items: Feynman, superconductors, faculty shuffle

A few brief news items - our first week of classes this term is a busy time.

• Here is a video of Richard Feynman, explaining why he can't readily explain permanent magnets to the interviewer.   This gets right to the heart of why explaining science in a popular, accessible way can be very difficult.  Sure, he could come up with really stretched and tortured analogies, but truly getting at the deeper science behind the permanent magnets and their interactions would require laying a ton of groundwork, way more than what an average person would want to hear.
• Here is a freely available news article from Nature about superconductivity in H₂S at very high pressures.   I was going to write at some length about this but haven't found the time.  The short version:  There have been predictions for a long time that hydrogen, at very high pressures like in the interior of Jupiter, should be metallic and possibly a relatively high temperature superconductor.  There are later predictions that hydrogen-rich alloys and compounds could also superconduct at pretty high temperatures.  Now it seems that hydrogen sulfide does just this.  Crank up the pressure to 1.5 million atmospheres, and that stinky gas becomes what seems to be a relatively conventional (!) superconductor, with a transition temperature close to 200 K.  The temperature is comparatively high because of a combination of an effectively high speed of sound (the material gets pretty stiff at those pressures), a large density of electrons available to participate, and a strong coupling between the electrons and those vibrations (so that the vibrations can provide an effective attractive interaction between the electrons that leads to pairing).    The important thing about this work is that it shows that there is no obvious reason why superconductivity at or near room temperature should be ruled out.
• Congratulations to Prof. Laura Greene, incoming APS president, who has been named the new chief scientist of the National High Magnetic Field Lab.  
• Likewise, congratulations to Prof. Meigan Aronson, who has been named Texas A&M University's new Dean of Science.  

Anecdote 5: Becoming an experimentalist, and the Force-o-Matic

As an undergrad, I was a mechanical engineering major doing an engineering physics program from the engineering side.  When I was a sophomore, my lab partner in the engineering fluid mechanics course, Brian, was doing the same program, but from the physics side.  Rather than doing a pre-made lab, we chose to take the opportunity to do an experiment of our own devising.   We had a great plan.  We wanted to compare the drag forces on different shapes of boat hulls.  The course professor got us permission to go to a nearby research campus, where we would be able to take our homemade models and run them in their open water flow channel (like an infinity pool for engineering experiments) for about three hours one afternoon.  

The idea was simple:  The flowing water would tend to push the boat hull downstream due to drag.  We would attach a string to the hull, run the string over a pulley, and hang known masses on the end of the string, until the weight of the masses (transmitted via the string) pulled upstream to balance out the drag force - that way, when we had the right amount of weight on there, the boat hull would sit motionless in the flow channel.  By plotting the weight vs. the flow velocity, we'd be able to infer the dependence of the drag force on the flow speed, and we could compare different hull designs. 

Like many great ideas, this was wonderful right up until we actually tried to implement it in practice.  Because we were sophomores and didn't really have a good feel for the numbers, we hadn't estimated anything and tacitly assumed that our approach would work.  Instead, the drag forces on our beautiful homemade wood hulls were much smaller than we'd envisioned, so much so that just the horizontal component of the force from the sagging string itself was enough to hold the boats in place.  With only a couple of hours at our disposal, we had to face the fact that our whole measurement scheme was not going to work.

What did we do?  With improvisation that would have made McGyver proud, we used a protractor, chewing gum, and the spring from a broken ballpoint pen to create a much "softer" force measurement apparatus, dubbed the Force-o-Matic.  With the gum, we anchored one end of the stretched spring to the "origin" point of the protractor, with the other end attached to a pointer made out of the pen cap, oriented to point vertically relative to the water surface.  With fine thread instead of the heavier string, we connected the boat hull to the tip of the pointer, so that tension in the thread laterally deflected the extended spring by some angle.  We could then later calibrate the force required to produce a certain angular deflection.  We got usable data, an A on the project, and a real introduction, vividly memorable 25 years later, to real experimental work.

Drought balls and emergent properties

There has been a lot of interest online recently about the "drought balls" that the state of California is using to limit unwanted photochemistry and evaporation in its reservoirs.  These are hollow balls each about 10 cm in diameter, made from a polymer mixed with carbon black.  When dumped by the zillions into reservoirs, they don't just help conserve water:  They spontaneously become a teaching tool about condensed matter physics.

As you can see from the figure, the balls spontaneously assemble into "crystalline" domains.  The balls are spherically symmetric, and they experience a few interactions:  They are buoyant, so they float on the water surface; they are rigid objects, so they have what a physicist would call "hard-core, short-ranged repulsive interactions" and what a chemist would call "steric hindrance"; a regular person would say that you can't make two balls occupy the same place.  Because they float and distort the water surface, they also experience some amount of an effective attractive interaction.  They get agitated by the rippling of the water, but not too much.  Throw all those ingredients together, and amazing things happen:  The balls pack together in a very tight spatial arrangement.  The balls are spherically symmetric, and there's nothing about the surface of the water that picks out a particular direction.  Nonetheless, the balls "spontaneously break rotational symmetry in the plane" and pick out a directionality to their arrangement. There's nothing about the surface of the water that picks out a particular spatial scale or "origin", but the balls "spontaneously break continuous translational symmetry", picking out special evenly-spaced lattice sites.  Physicists would say they preserve discrete rotational and translational symmetries.  The balls in different regions of the surface were basically isolated to begin with, so they broke those symmetries differently, leading to a "polycrystalline" arrangement, with "grain boundaries".  As the water jostles the system, there is a competition between the tendency to order and the ability to rearrange, and the grains rearrange over time.  This arrangement of balls has rigidity and supports collective motions (basically the analog of sound) within the layer that are meaningless when talking about the individual balls.  We can even spot some density of "point defects", where a ball is missing, or an "extra" ball is sitting on top.

What this tells us is that there are certain universal, emergent properties of what we think of as solids that really do not depend on the underlying microscopic details.   This is a pretty deep idea - that there are collective organizing principles that give emergent universal behaviors, even from very simple and generic microscopic rules.  Knowing that the balls are made deep down from quarks and leptons does not tell you anything about these properties.

Anecdote 4: Sometimes advisers are right.

When I was a first-year grad student, I started working in my adviser's lab, learning how to do experiments at extremely low temperatures.   This involved working quite a bit with liquid helium, which boils at atmospheric pressure at only 4.2 degrees above absolute zero, and is stored in big, vacuum-jacketed thermos bottles called dewars (named after James Dewar).   We had to transfer liquid helium from storage dewars into our experimental systems, and very often we were interested in knowing how much helium was left in the bottom of a storage dewar.

The easiest way to do this was to use a "thumper" - a skinny (maybe 1/8" diameter) thin-walled stainless steel tube,  a few feet long, open at the bottom, and silver-soldered to a larger (say 1" diameter) brass cylinder at the top, with the cylinder closed off by a stretched piece of latex glove.   When the bottom of the tube was inserted into the dewar (like a dipstick) and lowered into the cold gas, the rubber membrane at the top of the thumper would spontaneously start to pulse (hence the name).   The frequency of the thumping would go from a couple of beats per second when the bottom was immersed in liquid helium to more of a buzz when the bottom was raised into vapor.  You can measure the depth of the liquid left in the dewar this way, and look up the relevant volume of liquid on a sticker chart on the side of the dewar.

The "thumping" pulses are called Taconis oscillations.  They are an example of "thermoacoustic" oscillations.  The physics involved is actually pretty neat, and I'll explain it at the end of this post, but that's not really the point of this story.  I found this thumping business to be really weird, and I wanted to know how it worked, so I walked across the hall from the lab and knocked on my adviser's door, hoping to ask him for a reference.  He was clearly busy (being department chair at the time didn't help), and when I asked him "How do Taconis oscillations happen?" he said, after a brief pause, "Well, they're driven by the temperature difference between the hot and cold ends of the tube, and they're a complicated nonlinear phenomenon." in a tone that I thought was dismissive.  Doug O. loves explaining things, so I figured either he was trying to get rid of me, or (much less likely) he didn't really know.

I decided I really wanted to know.  I went to the physics library upstairs in Varian Hall and started looking through books and chasing journal articles.  Remember, this was back in the wee early dawn of the web, so there was no such thing as google or wikipedia.  Anyway, I somehow found this paper and its sequels.  In there are a collection of coupled partial differential equations looking at the pressure and density of the fluid, the flow of heat along the tube, the temperature everywhere, etc., and guess what:  They are complicated, nonlinear, and have oscillating solutions.  Damn.  Doug O. wasn't blowing me off - he was completely right (and knew that a more involved explanation would have been a huge mess).  I quickly got used to this situation.

Epilogue:  So, what is going on in Taconis oscillations, really?  Well, suppose you assume that there is gas rushing into the open end of the tube and moving upward toward the closed end.  That gas is getting compressed, so it would tend to get warmer.  Moreover, if the temperature gradient along the tube is steep enough, the upper walls of the tube can be warmer than the incoming gas, which then warms further by taking heat from the tube walls.  Now that the pressure of the gas has built up near the closed end, there is a pressure gradient that pushes the gas back down the tube.  The now warmed gas cools as it expands, but again if the tube walls have a steep temperature gradient, the gas can dump heat into the tube walls nearer the bottom.  This is discussed in more detail here.  Turns out that you have basically an engine, driven by the flow of heat from the top to the bottom, that cyclically drives gas pulses.  The pulse amplitude ratchets up until the dissipation in the whole system equals the work done per cycle on the gas.  More interesting than that:  Like some engines, you can run this one backwards.  If you drive pressure pulses properly, you can use the gas to pump heat from the cold side to the hot side - this is the basis for the thermoacoustic refrigerator.

Assorted items

Time is getting short before our semester starts here, and there is much to be done, so I'll be brief:

• A library of nice online interactive physics applets/demos from the University of Colorado.  However, between the ongoing security pains of both Java and Flash, it's getting more and more difficult to have a resource like this that is of maximal use to students.  If someone finds a security hole in HTML5, then there will be no platform left for this kind of application, and that would be very disappointing. 
• A very impressive demonstration of the Magnus force.
• The new Fantastic Four movie has been reviewed very unfavorably by critics.  "Legendary physicist" [sic] Michio Kaku was tapped to do four featurettes about the science of the film.  Coincidence?  
• Finally, two months after publication in the UK, my book is available through Amazon in the US - see the advertisement in the upper right of this page.
• If you're in the US (not sure the stream will work elsewhere) and you missed it, here is the recent PBS documentary "The Bomb", about the events of 70 years ago.   Likewise, here is their other timely documentary, "Twisting the Dragon's Tail", about uranium.  If books are more your thing, you can't do much better than The Making of the Atomic Bomb.
• This is brilliant.

Sci-fi time: Do laser pistols make sense?

The laser pistol (or similar personal directed energy weapon) is a staple of science fiction, and you can imagine why:  No pesky ammunition to carry, possible some really cool light/sound effects, near-speed-of-light to target (though phasers and blasters and the like are often shown in TV and movies with laughably slow velocities, so the audience can see the beam or pulse propagate*).  Still, does a laser pistol make sense as a practical weapon?  This isn't "nano", but it does lead to some interesting science and engineering.

I'm a complete amateur at this topic, but it seems to me that for a weapon you'd care about two things:  Total energy transferred to the target in one shot, and the power density (energy per area per time).  The former is somewhat obvious - you need to dump energy into the target to break bonds, rip through armor or clothing, etc.  The latter is a bit more subtle, but it makes sense.  Spreading the energy transfer out over a long time or a big area is surely going to lessen the damage done.  Think about slapping a board vs. a short, sharp karate chop.

From the internet we can learn that a typical sidearm 9mm bullet has, in round numbers, a velocity of about 400 m/s and a kinetic energy of about 550 J.  If at bullet stops about 10 cm into a target, you can use freshman physics (assuming uniform deceleration) to find that the stopping time is half a millisecond, meaning that the average power is 1.1 megawatts (!).

So, for a laser pistol to be comparable, you'd want it to transfer about 550 J of energy, with an average power of 1.1 MW spread over a beam the size of a 9 mm bullet.  Energy-storage-wise, that's not crazy for a portable system - a 1 kg Li-ion battery when fully charged would contain enough energy for several hundred shots.  Batteries are not really equipped for MW power rates, though, so somehow the energy would probably have to be delivered to the beam-producing component by some exotic supercapacitor.  (Remember the whine as a camera flash charges up?  That's a capacitor charging to deliver a high-wattage pulse to the flash bulb.)  The numbers for portability there don't look so good there - megawatt power transfers would likely require many liters of volume (or interchangeable, many kg of mass).   Of course, you could start with a slower optical pulse and compress it - that's how facilities like the Texas Petawatt Laser work.  Fascinating science, and it does get you to ~ 100 J pulses that last ~ 100 femtoseconds (!!).  Still, that requires a room full of complicated equipment.  Not exactly portable.  Ahh well.  Interesting to learn about, anyway.

(*The beam weapons in sci-fi movies and TV are generally classic plot devices:  They move at whatever speed and have whatever properties are required to advance the story.  Phasers on Star Trek can disintegrate targets completely, yet somehow their effects stop at the floor, and don't liberate H-bomb quantities of energy.  The stories are still fun, though.)

What are Weyl fermions? Do they really move charge 1000x faster?

A new paper (published online in Science) with experimental work from Zahid Hasan's group at Princeton has made a splash this week.  In it, they argue that they see evidence in crystals of the compound TaAs for (quasi)particles that have the properties of so-called Weyl fermions.  What does this mean?

I've written before about quasiparticles.  The idea is, in a big system of interacting  degrees of freedom (electrons for example), you can ask, how would we best describe the low energy excitations of that system?  Often the simplest, most natural description of the excitations involves entities with well-defined quantum numbers like momentum, angular momentum, electric charge, magnetic moment, and even something analogous to mass.  These low energy excitations are quasiparticles - they're "quasi" because they don't exist outside of the material medium in which they're defined, but they're "particles" because they have all these descriptive parameters that we usually think of as properties of material particles.  In this situation, when we say that a quasiparticle has a certain mass, this is code for a discussion about how the energy of the excitation depends upon its momentum.  For a non-relativistic, classical particle like a baseball, the kinetic energy \(E = p^{2}/2m\), where \(p\) is the magnitude of the momentum.  So, if a quasiparticle has an energy roughly quadratic in its momentum, we can look at the number in front of the \(p^{2}\) and define it to be \(1/2m^{*}\), where \(m^{*}\) is an "effective mass".

In some materials under certain circumstances, you end up with quasiparticles with a kinetic energy that depends linearly on the momentum, \(E \sim p\).  This is reminiscent of the situation for light in the vacuum, where \(E = p c\), with \(c\) the speed of light.  A quasiparticle with this "linear dispersion" is said to act like it's "massless", in the same way that light has no mass yet still has energy and momentum.  This doesn't mean that something in the material is truly massless - it just means that those quasiparticles propagate at a fixed speed (given by the constant of proportionality between energy and momentum).  If the quasiparticle happens to have spin-1/2 (and therefore is a fermion),  then it would be a "massless fermion".  It turns out that graphene is an example of a material where, near certain energies, the quasiparticles act like this, and mathematically are well-described by a formulation dreamed up by Dirac and others - these are "massless Dirac fermions".

Wait - it gets richer.  In materials with really strong spin-orbit coupling, you can have a situation where the massless, charged fermions have a spin that is really locked to the momentum of the quasiparticle.  That is, you can have a situation where the quasiparticles are either right-handed (picture the particle as a bullet, spinning clockwise about an axis along its direction of motion when viewed from behind) or left-handed.  If this does not happen only at particular momenta (or only at a material surface), but can happen over a more general energy and momentum range (and in the material bulk), these quasiparticles can be described in language formulated by Weyl, and are "Weyl fermions".   Thanks to their special properties, the Weyl particles are also immune to some back-scattering (the kind of thing that increases electrical resistance).  I'm being deliberately vague here rather than delving into the math.  If you are very motivated, this paper is a good pedagogical guide.

So, what did the authors actually do?  Primarily, they used a technique called angle-resolved photoemission spectroscopy (ARPES) to measure, in 3d and very precisely, the relationship between energy and momentum for the various quasiparticle excitations in really high quality crystals of TaAs.  They found all the signatures expected for Weyl fermion-like quasiparticles, which is pretty cool.

Will this lead to faster computers, with charge moving 1000x faster, as claimed in various mangled versions of the press release?  No.  I'm not even sure where the writers got that number, unless it's some statement about the mobility of charge carriers in TaAs relative to their mobility in silicon.    This system is a great example of how profound mathematical descriptions (formulated originally to deal with hypothetical "fundamental" high energy physics) can apply to emergent properties of many-body systems.  It's the kind of thing that makes you wonder how fundamental are some of the properties we see in particle physics.  Conceivably there could be some use of this material in some technology, but it is silly (and in my view unnecessary) to claim that it will speed up computers.

The nanoscale: The edge of emergence

One of the most profound concepts to come out of condensed matter physics is the idea of emergent properties - nontrivial properties of a system that are not trivially deducible from the microscopic aspects and interactions of the underlying degrees of freedom, and that become even better defined as the system size grows.  One example is the rigidity of solids:  A single carbon atom is not rigid; a small cluster of carbon atoms has a countable number of discrete vibrational modes; but a large number of carbon atoms coupled by sp³ bonds becomes a diamond, one of the hardest, most mechanically rigid solids there is, so stiff that compressive sound travels at 12 km/s, 35 times faster than in air.  Somehow, going from one atom to many, the concept of rigidity acquires meaning, and the speed of sound in diamond approaches a precise value.   

This idea, that something remarkable, exact, yet not at all obvious can emerge collectively and even generically, is why condensed matter physics is profound and not just "mopping up the details".  This is the heart of Bob Laughlin's first book, A Different Universe:  Reinventing Physics from the Bottom Down, and was articulated concisely by Laughlin and Pines in their "Theory of Everything" paper.  

I was recently rereading that book, and one chapter articulates Laughlin's basically dismissive take on nanoscience.  He refers to it as a "carnival of baubles" - his view is that otherwise smart people get sucked into playing around at the nanoscale because it's diverting and involves fun, cool toys (i.e., everything looks cool under an electron microscope!), instead of spending their time and effort actually trying to think about deep, fundamental questions.   Well, everyone is entitled to their opinion, but it won't surprise you that I disagree with much of that take.  Working at the nanoscale allows us to examine how emergence works in specific cases, sets the ground work for the materials and devices of future technologies (two topics I touch on in my book), and allows us to develop new probes and techniques precisely for asking (some subset of) deep questions.   Like being able to probe matter on ultrafast timescales, or over a huge temperature range, or in systems of unprecedented purity, pushing our control and manipulation of materials to the nano regime lets us ask new and different questions, and that's how we make progress and find surprises.  This isn't an infatuation with baubles (though everything does look cool under an electron microscope).  

A grand challenge for nano: Solar energy harvesting near the thermodynamic limit

As I'd mentioned earlier in the week, the US Office of Science and Technology Policy had issued a call for "Grand Challenges" for nanotechnology for the next decade, with a deadline of July 16, including guidelines about specific points that a response should address.  Here is my shot:  

Affordable solar energy collection/conversion that approaches the thermodynamic efficiency limit based on the temperature of the sun (efficiency approaching 85%).  

Physics, specifically the second law of thermodynamics, places very strict limits on how much useful energy we can extract from physical systems.  For example, if you have a big rock at temperature \(T_{\mathrm{hot}}\), and another otherwise identical big rock at temperature \(T_{\mathrm{cold}}\), you could let these two rocks just exchange energy, and they would eventually equilibrate to a temperature \(T_{0} = (T_{\mathrm{hot}}+T_{\mathrm{cold}})/2\), but we would not have gotten any useful energy out of the system.  From the standpoint of extracting useful energy, that process (just thermal conduction + equilibration) would have an efficiency of zero.  Instead, you could imagine running a heat engine:  You might warm gas in a cylinder using the hot rock, so that its pressure goes up and pushes a piston to turn a crank that you care about, and then cool the piston back to its initial condition (so that you can run this as a cycle) by letting the gas dump energy to the cold rock.  Carnot showed that the best you can do in terms of efficiency here is \( (1 - T_{\mathrm{cold}}/T_{\mathrm{hot}})\).  On a fundamental level, this is what limits the efficiency of car engines, gas turbines in power plants, etc.  If the "cold" side of your system is near room temperature (300 Kelvin), then the maximum efficiency permitted by physics is limited by how hot you can make the "hot" side.  

So, what about solar power?  The photosphere of the sun is pretty hot - around 5000 K.  We can get energy from the sun in the form of the photons it radiates.  Using 300 K for \(T_{\mathrm{cold}}\), that implies that the theoretical maximum efficiency for solar energy collection is over 90%.  How are we doing?  Rather badly.  The most efficient solar panels you can buy have efficiencies around 35%, and typical ones are more like 18%.  That means we are "throwing away" 60% - 80% of the energy that should be available for use.  Why is that?  This article (here is a non-paywall pdf) by Albert Polman and Harry Atwater has a very good discussion of the issues.   In brief:  There are many processes in conventional photovoltaics where energy is either not captured or is "lost" to heat and entropy generation.  However, manipulating materials down to the nm level offers possible avenues for avoiding these issues - controlling optical properties to enhance absorption; controlling the available paths for the energy (and charge carriers) so that energy is funneled where it can be harnessed.  On the scale of "grand challenges", this has a few virtues:  It's quantitative without being fantastical; there are actually ideas about how to proceed; it's a topical, important social and economic issue; and even intermediate progress would still be of great potential importance.  

Active learning vs. lecturing: the annual hectoring

The new issue of Nature contains this article, discussing the active learning approach to teaching.  Actually, "discussing" is too neutral.  The title of the article and the subheadline flatly state that  lecture-based pedagogy is "wrong".  The article includes this quote about active learning: " 'At this point it is unethical to teach any other way' ", and presents that quote as a bold-face callout.   

The article says "Researchers often feel that a teacher's job is simply to communicate content: the factual knowledge covered in the course."  This is an assertion, and based on my experience, one that is incorrect.  If basic communication of facts is the job of teachers, we should just quit now, since books and more recently google have made us obsolete.  The whole point of in-person instruction is more than conveying a list of facts - in the case of physics, it's a matter of teaching people how to think about the world, how to think critically and translate concepts and ideas into the language of mathematics for the purposes of gaining predictive understanding and an appreciation for the beautiful way the universe works.

The article also implies that faculty are reluctant to migrate to active learning because it would require that we work harder (i.e., greater prep time, and therefore less time for research) to achieve its benefits.  I do think it was good for the author to raise the issue of incentives and rewards at the end:  If universities want to claim that they value teaching, they actually need to reward pedagogy.

By trying to cast active learning vs lecture-based pedagogy as a one-size-fits-all, good vs bad, modernists vs sticks-in-the-mud faceoff, the author does a disservice to the genuinely subtle questions at play here.  Yes, it looks like well-done active learning does enable large segments of the target audience (typically in intro courses) to retain concepts better.  Not all active learning approaches are implemented well, however; some lecturers can be outstanding, and the ones that engage the class in discussion and back-and-forth are blurring the line into active learning anyway; active learning definitely is a compromise in that the investment of personnel and time to achieve the benefits does mean leaving out some content; and different people learn best from different methods!  The author raises these issues, but the main thesis of the article is clear.
  
I want to raise a question that you will find in many physics departments around the US:  Who is the target audience in our classes, particularly beyond the large freshmen service teaching courses?   In a big intro class with 350 future engineers, or 400 pre-meds, maybe sacrificing some content for the sake of engaging a larger fraction of the students to better internalize and retain physical concepts is a good idea.   If we do this, however, in a way that bores or fails to challenge the top students, or leaves gaps in terms of content, is that a net good?  

My point:  Pedagogy is complicated, and in the sciences and engineering we are trying to do several competing tasks in parallel.  Oversimplifying to the level that "active learning = unalloyed good all the time; traditional lecture = unethical, abusive method clung to by lazy, hidebound, research-driven-teaching-averse faculty" is not helpful. 

Nano "Grand Challenges" for the next decade

Last month the White House Office of Science and Technology Policy issued a call for suggestions for "nanotechnology-inspired grand challenges".  The term "grand challenge" is popular, both within the federal agencies and among science/technology coordinating and policy-making groups.  When done well, a list of grand challenges can cleanly, clearly articulate big, overarching goals that a community has identified as aspirational milestones toward which to strive.  When done poorly, a list of grand challenges can read like naive fantasy, with the added issue that pointing this out can lead to being labeled "negative" or "lacking in vision".  To make up a poor example: "In the next ten years we should develop a computing device smaller than a grain of rice, yet more powerful than an IBM Blue Gene supercomputer, able to operate based on power acquired from the ambient environment, and costing less than $5 to manufacture."    Yeah, not too realistic.

It's worth thinking hard about these, though, and trying to contribute good ones.  The deadline for this call is this coming Thursday, so get yours in while you can.  I put one in that I will discuss later in the week.

What do IBM's 7 nm transistors mean?

Two days ago IBM and their consortium partners made a big splash by announcing that they had successfully manufactured prototype chips on the wafer scale (not just individual devices) with features compatible with the 7 nm "node".   What does that mean?  It's being reported that the transistors themselves are "7 nm wide".  Is that correct?

The term "node" is used by the semiconductor industry to characterize major targets in their manufacturing roadmap.  See here for more information that you could ever possibly want about this.  The phrase "xx nm node" means that the smallest spacing between repeated features on a chip along one direction is xx nm.  It does not actually mean that transistors are now xx nm by xx nm.  Right now, the absolute state of the art on the market from Intel are chips at the 14 nm node.  A cross-section of those taken by transmission electron microscope is shown to the right (image from TechInsights), and Intel has a nice video explaining the scheme here.   The transistors involve fin-shaped pieces of silicon - each fin is about 14 nm wide, 60-70 nm tall, and a hundred or more nm long.  One transistor unit in this design contains two fins each about 40 nm apart, as you can see in the image.  The gate electrode that cuts across the fins is actually about 40-50 nm in width.  I know this is tough to visualize - here is a 3-fin version, annotated from a still from Intel's video.  In these devices current flows along the long direction of the fin, and the gate can either let the current pass or not, depending on the voltage applied - that's how these things function as switches.

So:  The 7 nm node IBM chip is very impressive.  However, don't buy into the press release wholesale just yet - there is a lot of ground to cover before it becomes clear that these chips are really "manufacturable" in the sense commonly used by the semiconductor industry.  

I'll touch on two points here.  First, yield.  The standard architecture of high performance logic chips these days assumes that all the transistors work, and we are talking about chips that would contain several billion transistors each.  In terms of manufacturing and reliability, integrated semiconductor devices are freaking amazing and put mechanical devices to shame - billions of components connected in a complicated way, and barring disaster these things can all work for many years at a time without failure.  To be "manufacturable", chip makers need to have the yield of good chips (chips where all the components actually work) be high enough to cover the manufacturing costs at a reasonable price point.  That typically means yield rates over at least 30%.  Back in the '90s Intel was giving its employees keychains made from dead Pentium processors.  It's not at all clear that IBM can really make these chips with good yields.  Note that Intel recently delayed manufacturing of 10 nm node chips because of problems.

Second, patterning technology.  All chips in recent years (including Intel's 14 nm and even their prototype 10 nm node products) are patterned using photolithography, based on a light source with a wavelength of 193 nm (!).  Manufacturers have relied on several bits of extreme cleverness to pattern features down to 1/20 of the free-space wavelength of the light, including immersion lithography, optical phase control, exotic photochemistry, and multiple patterning.  However, those can only carry you so far.  IBM and their partners have decided that now is finally the time to switch to a new light source, 13.5 nm wavelength, the so-called extreme ultraviolet.    This has been in the planning stages for years, with prototype EUV 300 mm wafer systems at Albany Nanofab and IMEC for about a decade.  However, changing to new wavelengths and therefore new processing chemistry and procedures is fraught with challenges.   I'm sure that IBM will get there, as will their competitors eventually, but it wouldn't shock me if we don't see actual manufacturing of 7 nm node chips for four or five more years at least. 

Ten years of blogging about CMP/nano

A couple of weeks ago this blog passed through its tenth anniversary (!).  That makes me about 70 in blog-years.  At the time, science blogging was going through a rapid expansion, and since then there has been a major die-off (we still miss you, Incoherent Ponderer) - people decide that they aren't reaching their desired audience, or don't have the time, or have run out of things to say, etc. 

A lot has happened in nanoscience and condensed matter physics in the last decade:  the rise of graphene (a late post on this), the recognition of topological insulators, the discovery of the iron pnictide superconductors, observations related to Majorana fermions, to name a few.  It's been fun to watch and discuss.  I've written quite a bit (but not so much recently) on physics as as undergrad and cultural attitudes, choosing a grad school, choosing/finding a postdoc position, giving talks in general, trying to get a faculty job, and other career-oriented topics, as well as science policy.  Over time, I've leaned more toward trying to explain CMP and nano concepts in more accessible ways, and at least identifying why this can sometimes be more difficult in our discipline than in high energy physics. 

Anyway, it's been a lot of fun, and I'm not going anywhere, thanks in large part to knowing that you are actually continuing to read this.  Thanks for your support.  If there are particular topics you'd like to see, please comment as always.

Can we use machine learning to solve really hard physics problems?

One of the most important achievements in 20th century physics and chemistry is density functional theory (DFT).  In 1964,  Walter Kohn and Pierre Hohenberg proved a rather amazing result:  If you want to know the ground state electronic properties of any condensed quantum electronic system (e.g., a solid, or a molecule), you can get all of that information (even in a complicated, interacting, many-body system!) just from knowing the ground state electronic density everywhere, \(n(\mathbf{r})\).  That is, any property you might care about (e.g., the ground state energy) can be computed from \(n(\mathbf{r})\) as a functional \(F[n(\mathbf{r})]\).  (A functional is a function of a function - in this case \(F\) depends in detail on the value of \(n\) everywhere.)   The Hohenberg-Kohn paper is the most-cited physics paper of all time, suggesting its importance.  So, truly figuring out the electronic structure of molecules and materials just becomes a matter of figuring out an extremely good approximation to \(n(\mathbf{r})\).

Moreover, Kohn and Lu Sham then went further, and found a practical calculational approach that lets you work with an effective system of equations to try to find \(n(\mathbf{r})\) and the ground state energy.  In this formulation, they write the total energy functional \(E[n(\mathbf{r}]\) as a sum of three pieces:  a kinetic energy term that may be written as a comparatively simple expression; a potential energy term that is easy to write and simply interpreted as the Coulomb repulsion; and the third bit, the "exchange-correlation functional", \(E_{\mathrm{xc}}[n(\mathbf{r})]\), which no one knows how to write down analytically.

You might think that not knowing how to write down an exact expression for \(E_{\mathrm{xc}}[n(\mathbf{r})]\) would be a huge issue.  However, people have come up with many different approximation methods, and DFT has been hugely useful in understanding the electronic properties of solids and molecules.

In recent years, though, some people have been wondering if it's possible to use "machine learning" - essentially having a computer derive an extremely good look-up table or interpolation - to approach an exact description of \(E_{\mathrm{xc}}[n(\mathbf{r})]\).  This is not a crazy idea at all, based on engineering history and dimensional analysis.   For example, actually writing down an analytical expression for the pressure drop of water flowing through a rough pipe is not generally possible.  However, dimensional analysis tells us that the pressure drop depends on just a couple of dimensionless ratios, and a zillion experiments can be run to map out a look-up table for what that un-writable multivariable function looks like.   Perhaps with computers that are becoming incredibly good at identifying patterns and underlying trends, we can do something similar with \(E_{\mathrm{xc}}[n(\mathbf{r})]\).  One of the main groups pursuing this kind of idea is that of Keiran Burke, and last week a preprint appeared on the arxiv from others arguing that machine learning can be useful in another many-body approach, dynamical mean field theory.  Who knows:  Maybe someday the same types of algorithms that guess songs for you on Pandora and book suggestions on Amazon will pave the way for real progress in "materials by design"!

How much information can you cram down an optical fiber?

A new cool result showed up in Science this week, implying that we may be able to increase the information-carrying capacity of fiber optics beyond what had been thought of as (material-dependent) fundamental limits.  To appreciate this, it's good to think a bit about the way optical fiber carries information right now, including the bits of this blog post to you.  (This sort of thing is discussed in the photonics chapter of my book, by the way.)

Information is passed through optical fibers in a way that isn't vastly different than AM radio.  A carrier frequency is chosen (corresponding to a free-space wavelength of light of around 1.55 microns, in the near-infrared) that just so happens to correspond to the frequency where the optical absorption of ultrapure SiO₂ glass is minimized.   Light at that frequency is generated by a diode laser, and the intensity of that light is modulated at high speed (say 10 GHz or 40 GHz), to encode the 1s and 0s of digital information.  If you look at the power vs. frequency for the modulated signal, you get something like what is shown in the figure - the central carrier frequency, with sidebands offset by the modulation frequency.   The faster the modulation, the farther apart the sidebands.   In current practice, a number of carrier frequencies (colors) are used, all close to the minimum in the fiber absorption, and the carriers are offset enough that the sidebands from modulation don't run into each other.  Since the glass is very nearly a linear medium, we can generally use superposition nicely and have those different colors all in there without them affecting each other (much).

So, if you want to improve data carrying capacity (including signal-to-noise), what can you do?  You could imagine packing in as many channels as possible, modulated as fast as possible to avoid cross-channel interference, and cranking up the laser power so that the signal size is big.  One problem, though, is that while the ultrapure silica glass is really good stuff, it's not perfectly linear, and it has dispersion:  The propagation speed of different colors is slightly different, and it's affected by the intensity of the different colors.  This tends to limit the total amount of power you can put in without the signals degrading each other (that is, channel A effectively acts like a phase and amplitude noise source for channel B).  What the UCSD researchers have apparently figured out is, if you start with the different channels coherently synced, then the way the channels couple to each other is mathematically nicely determined, and can be de-convolved later on, essentially cutting down on the effective interference.  This could boost total information carrying capacity by quite a bit - very neat. 

What is quantum coherence?

Often when people write about the "weirdness" of quantum mechanics, they talk about the difference between the interesting, often counter-intuitive properties of matter at the microscopic level (single electrons or single atoms) and the response of matter at the macroscopic level.  That is, they point out how on the one hand we can have quantum interference physics where electrons (or atoms or small molecules) seem to act like waves that are, in some sense, in multiple places at once; but on the other hand we can't seem to make a baseball act like this, or have a cat act like it's in a superposition of being both alive and dead.  Somehow, as system size (whatever that means) increases, matter acts more like classical physics would suggest, and quantum effects (except in very particular situations) become negligibly small.  How does that work, exactly?   

Rather than comparing the properties of one atom vs. 10²⁵ atoms, we can gain some insights by thinking about one electron "by itself" vs. one electron in a more complicated environment.   We learn in high school chemistry that we need quantum mechanics to understand how electrons arrange themselves in single atoms. The 1s orbital of a hydrogen atom is a puffy spherical shape; the 2p orbitals look like two-lobed blobs that just touch at the position of the proton; the higher d and f orbitals look even more complicated.  Later on, if you actually take quantum mechanics, you learn that these shapes are basically standing waves - the spatial state of the electron is described by a (complex, in the sense of complex numbers) wavefunction \(\psi(\mathbf{r})\) that obeys the Schroedinger equation, and if you have the electron feeling the spherically symmetric \(1/r\) attractive potential from the proton, then there are certain discrete allowed shapes for \(\psi(\mathbf{r})\).  These funny shapes are the result of "self interference", in the same way that the allowed vibrational modes of a drumhead are the result of self-interfering (and thus standing) waves of the drumhead.

In quantum mechanics, we also learn that, if you were able to do some measurement that tries to locate the electron (e.g., you decide to shoot gamma rays at the atom to do some scattering experiment to deduce where the electron is), and you looked at a big ensemble of such identically prepared atoms, each measurement would give you a different result for the location.  However, if you asked, what is the probability of finding the electron in some small region around a location \(\mathbf{r}\), the answer is \(|\psi(\mathbf{r})|^2\).  The wavefunction gives you the complex amplitude for finding the particle in a location, and the probability of that outcome of a measurement is proportional to the magnitude squared of that amplitude.  The complex nature of the quantum amplitudes, combined with the idea that you have to square amplitudes to get probabilities, is where quantum interference effects originate.   

This is all well and good, but when you worry about the electrons flowing in your house wiring, or even your computer or mobile device, you basically never worry about these quantum interference effects.  Why not?

The answer is rooted in the idea of quantum coherence, in this case of the spatial state of the electron.  Think of the electron as a wave with some wavelength and some particular phase - some arrangement of peaks and troughs that passes through zero at spatially periodic locations (say at x = 0, 1, 2, 3.... nanometers in some coordinate system).   If an electron propagates along in vacuum, this just continues ad infinitum.

If an electron scatters off some static obstacle, that can reset where the zeros are (say, now at x = 0.2, 1.2, 2.2, .... nm after the scattering).  A given static obstacle would always shift those zeros the same way.   Interference between waves (summing the complex wave amplitudes and squaring to find the probabilities) with a well-defined phase difference is what gives the fringes seen in the famous two-slit experiment linked above.

If an electron scatters off some dynamic obstacle (this could be another electron, or some other degree of freedom whose state can be, in turn, altered by the electron), then the phase of the electron wave can be shifted in a more complicated way.  For example, maybe the scatterer ends up in state S1, and that corresponds to the electron wave having zeros at x=0.2, 1.2, 2.2, .....; maybe the scatterer ends up in state S2, and that goes with the electron wave having zeros at x=0.3, 1.3, 2.3, ....  If the electron loses energy to the scatterer, then the spacing between the zeros can change (x=0.2, 1.3, 2.4, ....).  If we don't keep track of the quantum state of the scatterer as well, and we only look at the electron, it looks like the electron's phase is no longer well-defined after the scattering event.  That means if we try to do an interference measurement with that electron, the interference effects are comparatively suppressed.

In your house wiring, there are many many allowed states for the conduction electrons that are close by in energy, and there are many many dynamical things (other electrons, lattice vibrations) that can scatter the electrons.  The consequence of this is that the phase of the electron's wavefunction only remains well defined for a really short time, like 10^-15 seconds.    Conversely, in a single hydrogen atom, the electron has no states available close in energy, and in the absence of some really invasive probe, doesn't have any dynamical things off which to scatter.

I'll try to write more about this soon, and may come back to make a figure or two to illustrate this post.

Brief news items

In the wake of travel, I wanted to point readers to a few things that might have been missed:

• Physics Today asks "Has science 'taken a turn towards darkness'?"  I tend to think that the physical sciences and engineering are inherently less problematic (because of the ability of others to try to reproduce results in a controlled environment) than biology/medicine (incredibly complex and therefore difficult or impractical to do controlled experimentation) or the social sciences.  
• Likewise, Physics Today's Steven Corneliussen also asks, "Could the evolution of theoretical physics harm public trust in science?"  This gets at the extremely worrying (to me) tendency of some high energy/cosmology theorists these days to decry that the inability to test their ideas is really not a big deal, and that we shouldn't be so hung up on the idea of falsifiability. 
• Ice spikes are cool.
• Anshul Kogar and Ethan Brown have started a new condensed matter blog!  The more the merrier, definitely.
• My book is available for download right now in kindle form, with hard copies available in the UK in a few days and in the US next month.

Molecular electronics: 40+ years

More than 40 years ago, this paper was published, articulating clearly from a physical chemistry point of view the possibility that it might be possible to make a nontrivial electronic device (a rectifier, or diode) out of a single small molecule (a "donor"-bridge-"acceptor" structure, analogous to a pn junction - see this figure, from that paper).  Since then, there has been a great deal of interest in "molecular electronics".  This week I am at this conference in Israel, celebrating both this anniversary and the 70th birthday of Mark Ratner, the tremendous theoretical physical chemist who coauthored that paper and has maintained an infectious level of enthusiasm about this and all related topics.

The progress of the field has been interesting.  In the late '90s through about 2002, there was enormous enthusiasm, with some practitioners making rather wild statements about where things were going.  It turned out that this hype was largely over-the-top - some early measurements proved to be very poorly reproducible and/or incorrectly interpreted; being able to synthesize 10²² identical "components" in a beaker is great, but if each one has to be bonded with atomic precision to get reproducible responses that's less awesome; getting molecular devices to have genuinely useful electronic properties was harder than it looked, with some fundamental limitations;  Hendrik Schoen was a fraud and his actions tainted the field; DARPA killed their Moletronics program, etc.    That's roughly when I entered the field.  Timing is everything.

Even with all these issues, these systems have proven to be a great proving ground for testing our understanding of a fair bit of physics and chemistry - how should we think about charge transport through small quantum systems?  How important are quantum effects, electron-electron interactions, electron-vibrational interactions?   How does dissipation really work at these scales?  Do we really understand how to compute molecular levels/gaps in free space and on surfaces with quantitative accuracy?  Can we properly treat open quantum systems, where particles and energy flow in and out?  What about time-dependent cases, relevant when experiments involve pump/probe optical approaches?  Even though we are (in my opinion) very unlikely to use single- or few-molecule devices in technologies, we are absolutely headed toward molecular-scale (countably few atom) silicon devices, and a lot of this physics is relevant there.  Similarly, the energetic and electronic structure issues involved are critically important to understanding catalysis, surface chemistry, organic photovoltaics, etc.

What does a molecule sound like?

We all learn in high school chemistry or earlier that atoms can bind together to form molecules, and like a "highly sophisticated interlocking brick system", those atoms like to bind in particular geometrical arrangements.  Later we learn that those bonds are dynamic things, with the atoms vibrating and wiggling like masses connected by springs, though here the (nonlinear) spring constants are set by the detailed quantum mechanical arrangement of electrons.  Like any connected set of masses and springs, or like a guitar string or tuning form, molecules have "normal modes" of vibration.  Because the vibrations involve the movement of charge, either altering how positive and negative charge are spatially separated (dipole active modes) or how the charge would be able to respond to an electric field (roughly speaking, Raman active modes), these vibrations can be excited by light.  This is the basis for the whole field of vibrational spectroscopy.  Each molecule has a particular, distinct set of vibrations, like a musical chord.

Because the atoms involved are quite light (one carbon atom has a mass of 2\(\times\)10^-26 kg) and the effective springs are rather stiff, the vibrations are typically at frequencies of around 10¹³ Hz and higher - that's 10 billion times higher than the frequency of a typical acoustic frequency (1 kHz).  Still, suppose we shifted the frequencies down to the acoustic range, using a conversion of 1 cm^-1 (a convenient unit of frequency for molecular spectroscopists) \(\rightarrow\) 1 Hz.  What would molecules sound like?  As an example, I looked at the (surface enhanced) Raman spectrum of a small molecule, pMA. The Raman spectrum is from this  paper (Fig. 3a), and I took the three most dominant vibrational modes, added the pitches with the appropriate amplitude, and this is the result (mp3 - embedding audio in blogger is annoying).

I thought I was being clever in doing this, only to realize that, as usual, someone else had this same idea, beat me to it, and implemented it in a very cool way.  You should really check that out.

Anecdote 3: The postdoc job talk and the Nobel laureate

Back when I was finishing up my doctoral work, I made a trip to New Jersey to interview in two places for possible postdoc positions.  As you might imagine, this was both exciting and nerve-wracking.  My first stop was Princeton, my old undergrad stomping grounds, where I was trying to compete for a prestigious named fellowship, and from there I was headed north to Bell Labs the following day. 

As I've mentioned previously, my graduate work was on the low temperature properties of glasses, which share certain universal properties (temperature dependences of the thermal conductivity, specific heat, speed of sound, and dielectric response, to name a few) that are very distinct from those of crystals.  These parameters were all described remarkably well by the "two-level system" (TLS) model (the original paper - sorry for the paywall that even my own university library won't cover) dreamed up in 1971 by Phil Anderson, Bert Halperin, and Chandra Varma.  Anderson, a Nobel laureate for his many contributions to condensed matter physics (including Anderson localization, the Anderson model, and the Anderson-Higgs mechanism) was widely known for his paean to condensed matter physics and for being a curmudgeon.  He was (and still is) at Princeton, and while he'd known my thesis adviser for years, I was still pretty nervous about presenting my thesis work (experiments that essentially poked at the residual inadequacies of the original TLS model trying to understand why it worked so darn well) to him.

My visit was the standard format - in addition to showing me around the lab and talking with me about what projects I'd likely be doing, my host (who would've been my postdoc boss if I'd ended up going there) had thoughtfully arranged a few 1-on-1 meetings for me with a couple of other postdocs and a couple of faculty members, including Anderson.  My meeting with Anderson was right before lunch, and after I got over my nerves we had what felt to me like a pretty good discussion, and he seemed interested in what I was going to present.  My talk was scheduled for 1:00pm, right after lunch, always a tricky time.  I was speaking in one of the small classrooms in the basement of Jadwin Hall (right next to the room where I'd had undergrad quantum seven years earlier).  I was all set to go, with my binder full of transparencies - this was in the awkward period when we used computers to print transparencies, but good laptops + projectors were rare.   Anderson came in and sat down pointedly in the second row.  By my third slide, he was sound asleep.  By my fifth slide, he was noticeably snoring, though that didn't last too long.  He did revive and ask me a solid question at the end of the talk, which had gone fine.  In hindsight, I realize that my work, while solid and interesting, was in an area pretty far from the trendiest topics of the day, and therefore it was going to be an uphill battle to capture enthusiasm.  At least I'd survived, and the talk the next day up at Murray Hill was better received.