Electron band structure in germanium, my ass (2000) | Hacker News Reader

Electron band structure in germanium, my ass (2000)

pages.cs.wisc.edu

461 pointscrummy77 comments

As an odd coincidence, I did the same experiment on a shoestring budget with substandard equipment also. I too used a fancy computer algorithm to get a best fit. Except that I managed to get four significant decimal places in the result — an improvement over the (also outdated) textbook.

The author of the angry rant had a life-defining experience of overwhelming frustration.

The same scenario resulted in a positive life-defining experience for me

It’s funny how unpredictably things pan out even in identical circumstances…

But in some way we all end up in the same place :) (HN)

I really enjoyed the labs and it was seeing the predictions pan out in front of my eyes that actually got me interested in the first place. But you have to know what your getting into. The labs are extremely time consuming and required preparation.

Talking of a shoestring budget, sometimes it's a good thing. I actually learned more useful things about EE by fixing broken test gear than doing an undergraduate course. The course was so heavily theoretical that when it came to actually building prototypes, nothing worked and no one knew why, not even the lab techs. So I spent most of the time going round helping fixing people's limping final year projects.

Life is what you take out of it :)

Agree with this. I had to make my own equipment everywhere I went. The people I met later in my career who went to nice schools, with illustrious research experience had no idea what they were doing. Okay that's an exaggeration. But, when they entered industry they trusted every black box and rationalized every bizarre thing they saw as if their results depended entirely on their domain of interest.

Negative concentrations no problem! That must mean the machine is tracking the previous known value and this is the difference! Oh my oh my... The things I've seen the highly/expensively educated get paid lots of money to do, present, and be rewarded for in regulated industries....

Pretty funny.

Having taught low-temperature condensed matter labs, a big part of the grade is figuring out what went wrong, and either correcting for it, or at least acknowledging that it went wrong. The student needed to give more information about the experimental setup (what instruments did they use? four point or two point resistance? resistivity vs resistance? what is R_0?) and why they think the experiment didn't work. It looks to me like they had something miswired, so they only measured noise.

My understanding of undergrad lab work is that the point is not to get results, it's to learn how to do lab work.

Indeed, a scientist should have some exposure to experimentation. Experiments don't always work on the first try, and often require some knowledge and skill.

The ones who can't solder go into electrical engineering and sit at a computer terminal all day. (joking of course)

Usually on a Friday afternoon/early evening while other non-Science/Engineering types are out having something called a Social Life ...

Exactly my experience back in the days doing the mandatory "advanced experimental physics laboratory semester" where you had to do like 14 vastly different experiment of the caliber described in that post in the course of one semester on old equipment that would break during the experiment, with less than motivated PhD students or post grads as teachers. Of the 14 experiments only two worked and we got the expected results.

This experience drove me right into theoretical physics and writing computer simulations of electron dynamics and light-matter interactions in confined semiconductors (quantum dots, graphene and the like). That was fun.

Now I am working on medical device software development, as the other stuff does not pay the bills.

Yeah I don't quite understand how the physics curriculum is set. There is more and more physics discovered over time and yet the time you get to study remains the same. 4 years is not enough time to learn the math and physics you need to do anything real, not counting the other important stuff you need to do (like English comp and partying and getting your heart broken).

Heh this strikes me as funny. I remember studying hierarchical image classifiers... If you think physics changes fast you should check out CS sometime.

This strikes me as funny, because Lucas Kovar, the author, is now a software engineer at Google.

At my uni (and probably and most others) undergrad physics just ends around 1930s. Then for master's you get some of the mid-20 century (phase transitions and intro to qft). Then you specialize and learn the remaining 50 years

Physics course is set to do two things

(1) 'to teach you to think like a physicist' as my professor was fond of saying to us. No one is going to be able to learn all of the subjects knowledge in 4 years, let alone a complex and rigorous field like physics.

(2) to give you a core foundation of understanding on which all other physics is built upon.

By teaching the core subjects and lessons of physics, you can get a pretty good understanding of the world, how it works, and how to derive equations to explain it.

As for it not being able to learn enough to do anything real. I beg to differ, like all fields you tend to learn the most in 'the field'. Most physicists learn by doing just like any other field. Take for instance engineering, there are some exceptions but most of the engineers I have known in my life have really only become fully viewed as engineers once they had some years of experience under their belts... Same thing goes for comp sci, and same goes for physics.

It's hard to explain because the field is viewed as an abstraction to those outside of it. The goal of understanding something new, discovering some new phenomenon, or better understanding an old one. But what that looks like in practice is years of working in a lab failing, trying things, taking results, tilting your head and going 'thats weird', and charting to colleagues about how to get something to work. Those who have done experimental physics, this is just experimentation and it makes sense, those who haven't, this probably sound simplistic and it is. Like your own fields, it's hard to really capture the depth of what experimentation looks like because even in physics it varies from sub field to sub field.

So it goes, sorry for the rambling from an experimental physicist

Thanks for the thoughtful response. I think maybe my objection to MY curriculum was being handed a lot of mathematical tools which I didn't need, and so it was impossible for me to "slot them in" to my toolbox. Most named solutions to differential equations were just like...why? (e.g. Bessel functions, or Green's function or...). I think it's a failure of pedagogy - I personally think you should give students the problem before giving them the solution. But the physics curriculum seems to want to give you solutions first and then tell you how you can use them. My brain doesn't work that way; tools fulfill a need. It would be like teaching a carpenter all the ins and outs of every tool in the shop, without ever building anything. How is the student supposed to organize that knowledge? Alphabetically?

I suspect that, just like how algorithms rarely pop up in software practice, so too do these kinds of tools pop up in physics practice, and when they do you probably get that same happy jolt of "Hey I finally get to apply this knowledge!" And that happens about once every 2 years.

Astronomy tends to be a way to get "newer" physics involved (at least in terms of discoveries), but it's also worth noting that formulations used usually evolve from what was originally published (Maxwell's equations are a well-known example), so you're getting more of a cleaned-up version, with insights based on newer discoveries.

> on old equipment that would break during the experiment

Isn’t that closer to the real conditions in a research lab?

I met some NASA scientists doing atmospheric sampling on a plane, and they had to accompany their equipment to manage any equipment glitches during the expensive sampling process (a custom modified 747 flying from Hawaii to New Zealand IIRC)

Can confirm, work in a top physics university in the lab. It is not a rare sight to see equipment from 10, 20, 30 years ago still in operation. Though it really depends on the equipment. Optics? They don't really go bad if you treat them right. Electronics? You'd be surprised as the quality of design and fab went into that plastic box that looks older than you.

Really physics discovery is partially limited by equipment, but in my time in the lab I have seen great physicists get remarkable results with equipment or setups that I personally thought was not up to the task.

As to college level experimental physics lab: the goal is not to get you to reproduce the nobel winning results, but to learn how to think like an experimental physicist. To hunt down issues, to calculate sources of error, to find out that some guy keeps running the microwave while you are taking sensitive measurements and that it actually impacted them and how.

It is the unimaginative or poorly taught that think the experimental physics lab in undergrad isn't important.

Yeah I've bought some old test equipment and it's a wonder than you just plug it in and it is still within specs. They knew how to build them back when HP was called HP and not agilent/keysight/whatever they rename themselves in next decade.

> 14 vastly different experiment of the caliber described in that post in the course of one semester

My one remaining did-not-graduate nightmare is that I never completed that course. And I was an experimentalist.

> This experience drove me right into theoretical physics

Had a summer student work on my nearly-intractable dissertation experiment. He immediately switched to theory.

It's not that bad of a course, it's just a struggle because it's different than all other physics courses you take.

In that class you aren't being graded on getting to the right answer. You are getting graded on how you approach getting the right answer, and no one expects you to get the right answer - in the ball park is generally the average for the course.

> one semester on old equipment that would break during the experiment

I saw that as a learning opportunity, teaches you which glues are cryo-proof, how to switch fuses on amplifiers (and other, more complicated electronics quick-fixes), and how important knowing people to borrow equipment from is.

Unless your lab is swimming in money, those are valuable skills for an experimentalist. And even if you swim in money, buying fancy new stuff has a minimum 8 week lead time while walking down the hall to the correct shelf to pilfer takes 5 minutes and gets you results before lunch.

> with less than motivated PhD students or post grads as teachers

This is an unfortunate truth everywhere I taught. Unfortunately, teaching the same thing 14 times a semester is no fun, and every single person who enjoys teaching (and is good at it) knows it, and instead teaches any available lecture, seminar, tutorial or exercise - anything, where you get to teach something new every week and that allows you to stay with the same students for longer.

Wow! I'm less than a year from applying to grad programs and all of my undergrad research experience is in very similar things. Theoretical is my jam. Did you do a PhD? If so do you now believe it was worth it?

In total, yes it was. Working for some years in a field, having the time afforded to emerse yourself into a subject and deeply think about it, calculating yourself into frustrating dead ends and also into successes, writing papers, going to conferences presenting your results, have scientific exchanges with peers, write applications for research grants, sit in committees, work as a referee for scientific journals, teaching students, then writing a coherent and compelling thesis and finally defending it against & having a scientific discussion on eye-to-eye level with your supervisors will shape your character... a lot.

You might wanna skip that postdoc thing though if you know you don't wanna stay in academia (and trust me, you don't).

Original publication: Kovar, L. “Electron Band Structure In Germanium, My Ass.” Annals of Improbable Research, vol. 7, no. 3, May/June 2001, p. 4, https://web.archive.org/web/20010620051228/http://www.improb...

Here's a list of classics from Improbable Research

https://improbable.com/publications/classics/

I guess this must be from '99. Consequently the author went to get his Masters and PhD in CS.[0]

Having gone through a similar ordeal and frustration at the end of my uni career, in the end I went out with the highest admiration for the experimentalists and their tech staff.

[0]https://pages.cs.wisc.edu/~kovar/cv.html

This was one of the funniest things going around the Physics department when I was in school. I can assure you it is older than 2007. 1998 or ‘99 iirc.

I saw it in undergrad, 2001.

Earliest I've been able to trace is 2000. Anyone have an earlier copy?

https://web.archive.org/web/20001031193257/http://www.cs.wis...

I want the bibtex for that. ;)

Electron Band Structure in Germanium, My Ass - https://news.ycombinator.com/item?id=30690075 - March 2022 (1 comment)

Electron Band Structure in Germanium, My Ass (2001) - https://news.ycombinator.com/item?id=16360479 - Feb 2018 (38 comments)

Electron Band Structure In Germanium, My Ass - https://news.ycombinator.com/item?id=2513293 - May 2011 (97 comments)

Pretty funny, but what seemed disturbingly true:

> Going into physics was the biggest mistake of my life. I should've declared CS. I still wouldn't have any women, but at least I'd be rolling in cash.

I wish academia pays better.

It's not just the pay; the pressure as a graduate student and not-yet-tenured professor is immense. The darkly humorous "A (de)motivational letter", purportedly by one "Prof. Hardass Slavedriver", sums it up: https://lifesciencephdadventures.wordpress.com/2013/01/04/a-...

> The darkly humorous "A (de)motivational letter"

It's not just darkly humorous, it could almost be satire:

> It has to the the kind of chemistry that people read and then jizz in their pants, they’re so excited about it. That’s right: we only do jizz-worthy chemistry

I'll assume due to the nature of chemistry, the field doesn't suffer as much from non-reproducible data, but overall, this sentence alone might explain cases like the Francesca Gino fraud [0], or why the head of Stanford resigned over integrity issues [1]? Or the Schön scandal? [2]

I mean take the analogy seriously for a moment: What the author is asking for in the analogie's universe, is for one of his students to just invent Stable Diffusion so he can generate pornography for his colleagues to.. well?

But no word about integrity or honesty. Nothing. Just results.

Author took it as inspiration:

> The truth is that under all the profanities, and the slave-driver mentality of the professor I saw someone who truly wanted to develop their students.

PhD students of HN enlighten me, after reading this crap, how is this "slave-driver" not just incentivizing to cheat?

[0]https://www.theatlantic.com/science/archive/2023/08/gino-ari... [1]https://www.theguardian.com/us-news/2023/jul/19/stanford-pre... [2]https://en.wikipedia.org/wiki/Sch%C3%B6n_scandal

I think any system that rewards high performance also has the potential to reward (good enough) cheating. This letter can be motivational, and show the intense pressure to cheat in academia at the same time.

I'm not a PhD student, but I stopped after a masters exactly because I didn't want the pressure we're discussing.

Huh? Nothing matters but results anywhere. Why would it be different in academia, where funding is always tight and tenure-track jobs scarce and hotly contested?

Assuming this is real, it looks like he measured some high quality noise. No way the resistivity would jump 2x over 5 degrees and stay constant at constant temperature (the biggest suspect according to his own confessions).

When you do it right you get this sort of picture. https://pdfs.semanticscholar.org/3753/2b8a21825d66633f33684a...

I remember forwarding funny website links and word docs full of copy/pasted jokes around in emails back in the 90s and early 00s. This was a favorite of mine, having dropped out of an applied physics degree with the memory of the "final straw" optics lab where I lost the will to carry on. The equipment was awful, and I was paired with the clumsiest partner in the world. (During our 3rd year work experience semester, he was the only student who received an entirely negative feedback from his employer with the summary "unemployable" - I believe he caused a fire). Every time i stripped an optic fiber cable and mounted it in front of the laser, it seemed like the entire universe conspired to break the brittle glass core. When we did finally line everything up, happily taking measurements, I too ended up drawing a line through noise. In fact I haven't even read the piece yet, so it's been 15+ years since I've seen it, and I still remember laughing out loud at that line.

Not really related of the article, but a question I just realized I don’t know the answer to: where does the temperature of an atom “live”? Is it entirely in the electrons? Does the nucleus have any temperature?

https://physics.stackexchange.com/q/716282/43345 discovered by me some time ago. Long story short, nucleus cannot absorb low temperature thermal energy so it's virtually zero.

The "temperature of an atom" is something very different from the temperature of a material with billions of atoms. For the material, what matters most is usually vibrations of atoms, where the nucleus contributes the most to kinetic energy, because it is heavy, and the electronic configuration contributes to potential energy.

For an atom, you basically have to do quantum physics and talk about von Neuman entropy of it's quantum state. This kind of temperature is very removed from the everyday concepts of "heat" and "heat transfer".

Temperature is the motion of the whole atom. Individual atoms don't really have temperature, temperature is a bulk property, the average of motion of lots of atoms.

That's a good question. it depends how long you wait and how strongly coupled they are.

For electrons and nuclear motion they are often strongly coupled enough that they quickly reach equilibrium and thus are at the same temperature (e.g. in liquids or solids). But on short timescales they can absolutely be at distinct temperatures. In dilute gases they can also become somewhat decoupled (there can be a well defined electronic temperature and nuclear temperature).

One case whether this can take a much longer period of time is when dealing with nuclear spin -- this is much more weakly coupled to the electrons and so can take much longer to equilibriate.

Not a physicist, but I think temperature lives in motion of particles, let it be electrons, the nucleus, whole atoms, or otherwise. Because electrons shroud the nucleus and have much less mass, they are more prone with external interactions where they can change momentum by absorbing and releasing energy. The nucleus is not exempt from this, but takes a not more energy to change their momentum. That being said, "temperature" as we are accustomed to, is more like a statistical average of the total motion of all particles in the system.

>> Going into physics was the biggest mistake of my life. I should've declared CS. I still wouldn't have any women, but at least I'd be rolling in cash.

Hahaha, gold. I made that realization in the middle of my PhD and switched (40 years ago). Not regretting a moment. What broke me was the amount of politics you must engage in.

I was 3rd semester into theoretical math in 1999. Got a job on a side as a tech support. Then realised I am earning more in an entry level tech job than I can ever hope to earn in math. Did my 2 + 2 and went tech with no hesitation.

Applied math major here. It's practically CS anyway!

Well... I was in theoretical math (also not in US). We would be sometimes told by our profs that there is math and then there is this something that is being taught to everybody else, applied math included.

Applied math is essentially just learning the formulas to be able to solve real life problems. Not saying it is not useful. It is just something completely different from actually doing math research.

It is kind a like a difference between being a theoretical physicist and a mechanical engineer.

I feel this pain. On a lower level, my A Level physics practical exam was a life-changing experience that reduced me to tears because I could not assemble the experimental paraphernalia into a stable structure that would not fall over when I touched it.

What I'm taking away from the comment section is that if you have trouble replicating an experiment, you should take advantage of Cunningham's law: instead of having your pleads for help ignored by some highly demotivated PhD's posing as your thesis advisors, instead post a rant about your failure on HN and have those same advisors give you a snarky reply with how to do it correctly ( you even get references to correcting literature!)

I'm quite sure the author knows more than me, by far, so I'll err on the side of believing them, but I'm amazed that someone could read a physics textbook, but not be able to do a resistance measurement in a way they were happy with.

Anything above microohms would be trivial to measure compared to the skill needed to understand even the most basic physics textbook.

Of course, they didn't have 3D printers to make solderless mounts, and the equipment was much worse(I'd imagine they probably still had analog oscilloscopes lying around!).

But then again, I was once asked by a friend who was about to graduate to come help with a college project. I know absolutely nothing about any of the math involved, I was just there to make some sensor data go on a WiFi network. We got everything done in a day, after they were fighting for weeks.

I think college just only teaches you the hard stuff, and then leaves you to figure out easy things like "Resistors have a power rating that isn't infinite" by yourself. Seems pretty reasonable given the fact college education people make 20 to 100 times more than me!

This gives me fond memories.

I was not too recently ranting to some poor captive audience about the things which should and should not be labs, and how the topics existed on a scale of "doing science for primitives and stepping back into your time machine having advanced science by millennia" and "I swear everything related to this might as well be somewhere between voodoo and 1920s psychology." My ire was directed at calorimetry. They saddle high school students with calorimetry experiments, when really, as a field of measurement, unless you've got some great equipment and an amazing understanding of the possible hidden inputs, the error bars on your standard calorimetry experiment could accommodate a battleship doing a handbrake turn. At a high school level, calorimetry should be restricted to samples of ammonium nitrate dissolving in water and perhaps thermite, with checkboxes to indicate "got hot" or "got cold."

This is a similarly fraught deal and a cruel joke to play on an undergrad. Poor quality crystals, no explanation of appropriate solder, temperature control is non-existent. It's crap like this that turns people off of science. I have a physics degree and spent a lot of time tutoring; my big takeaway is that there exists a cabal of lizard people attempting to retard the technological advancements of humanity through a variety of gambits such as "social media," but that their biggest payoff has to be "labs."

And then, as a kind of extra special grift, are the kits. These are end-of-budget-year purchases which collect dust until someone gets the bright idea to actually hurl one of these beasts at the students. We received some kit involving fiber optics that was beyond terrible. I ended up tearing it apart and, with a DAQ and an XY plotter, created the world's slowest scanner. One thousand dpi, which isn't bad for 1992. The kit itself had pre-built experiments which seemed designed for little but frustration.

Even chemistry labs have better yields than this sort of thing. It's disheartening and students take this more personally than you might imagine.

This was my experience when doing some of measurements back while getting higher ed.

Sometimes you just get random results that has no relation to what you expected to get based on theory (and not just theory, but common sense). They're not zero and not over the top, they're just useless.

Welcome to experimental science. In this case they actually knew what they were looking for. Once you actually start doing real science, the first difficulty is to understand if what you are observing is actually what to expect or not (and that is not always straight forward).

So what was the final grade - did whoever did the grading just look at the pictures?

did he quit and go into CS, is he on HN?

Did he make any other discoveries regarding the assiness of other fields of human endeavor as the years went by?

Stalking the LinkedIn, it seems he quit, switched to a PhD in CS and has been working at Google for the last 12 years.

Wisdom is taking your own advice. Good on them.

Also, there are many a decent xysicists in tech. Physics skews to supernerds like good CS folks, so there's quite a bit of problem-solving and subject matter crossover beyond the closest neighbor, EE/CS.

The Goog has pockets of okayness here and there, so maybe they've found a comfortable home after 12 years. :)

The CS conclusion was worthy though

The double irony would be for the author to have persevered in physics and now hold several patents over solid state batteries that are going to revolutionize society and fix climate change, while their CS major friends just got fired by some GAFAM for not getting enough minimum-pay workers to click on funny cat memes.

Even if true the CS students would still be ahead earnings wise. /source I went into physics/materials

Let's make it "triple irony", then, I suppose ?

> I should've declared CS. I still wouldn't have any women, but at least I'd be rolling in cash.

True dat.

> I understand this is the same process by which the top quark was discovered.

I can't stop laughing

There is a data point close to zero resistance, but sadly at 230K.

This way please Dr. Crummy

jokes apart, this is real, right?

(2001)

You sure it isn’t Schottky diodes formed by the metallic contacts that are interfering with the measurement of the bulk germanium properties?

I’d sputter deposit metallic contact regions and solder to those, then maybe compare two different thickness samples and look at the difference in resistivity vs. temperature, essentially de-embedding your fixture.

Came here to say this. You can’t solder leads on to semiconductors willy-nilly. The work functions are too different. So instead of an ohmic contact, you get a diode.

Ok, I'm glad someone posted something.

It's not my sub field, but the description of the setup seemed soo haphazard

Could they measure the resistance in two direction, then subtract one from another to isolate the symmetric part?

Possibly, but the OP is probably measuring the equivalent of back-to-back series diodes, which would be very symmetrical (if built on the same process). With soldering though, I’m sure they are not equivalent. As the other poster said, they need Ohmic contacts.

Why is this still hosted on .edu?

Reply on news.ycombinator.com