1.3.4 · D2Materials & Atomic Structure

Visual walkthrough — Crystal lattice structure of silicon

1,941 words9 min readBack to topic

Step 1 — Draw the problem: one atom, four arrows

WHAT. Put one silicon atom at a point. Draw an arrow pointing out toward each of its 4 bonded neighbours. We will call these arrows bond vectors — an arrow is just a "here is a direction and how far" object, and a vector is the mathematical name for exactly that.

WHY. Before we can ask "what is the angle between two bonds?", we need bonds to be something we can point at and measure. Arrows from a shared centre are the simplest honest picture of four bonds sprouting from one atom.

PICTURE. Look at the red central dot in the figure. Four black arrows leave it. Notice they do not lie in a flat plane — two tilt up, two tilt down. That 3D spread is the whole story; a flat square would waste space.

Figure — Crystal lattice structure of silicon

Step 2 — Why a tetrahedron? Push four things apart on a sphere

WHAT. Imagine the four arrow-tips are beads that can slide anywhere on a sphere around the centre, but they repel each other. Let them settle. They stop when every pair is as far from every other pair as possible.

WHY. The four bonds are made of electron pairs, and electron pairs push each other away (this is the VSEPR idea from Valence electrons and the octet rule). "Spread 4 points as far apart as possible on a sphere" has one answer: the tetrahedron — a triangular pyramid. Not a square (that would leave the up–down directions empty), not a cross.

PICTURE. In the figure the four tips form a tent-shaped pyramid: every face is an identical triangle, and no tip is closer to another tip than any other pair. Perfect balance.

Figure — Crystal lattice structure of silicon

Step 3 — Give the corners coordinates

WHAT. Build a cube centred on the origin (the central atom) with corners at in each direction. Pick the four alternating corners that form our tetrahedron:

Reading the symbols right where they sit:

  • ::: an arrow to the first neighbour; its three numbers say "go along , along , along ".
  • The pattern: each vector has an even number of minus signs ( has 0, the others have 2). That is precisely what "alternating corners" means — skip every other corner so no two chosen corners share an edge.

WHY. Coordinates turn "the angle between two arrows" from a picture into numbers we can compute. We chose (not or anything) because the answer is an angle — scaling all arrows longer or shorter never changes the angle between them, so we pick the tidiest numbers.

PICTURE. The cube is drawn in black; the four chosen corners glow red. Trace them: pick a corner, jump to the corner diagonally across a face, and you land on the next red one. Never two reds sharing an edge.

Figure — Crystal lattice structure of silicon

Step 4 — The one tool we need: the dot product

WHAT. To measure the angle between two arrows we use the dot product. For two vectors and :

  • ::: multiply the two -parts together.
  • ::: add the product of the -parts.
  • ::: add the product of the -parts. One number falls out.

And this single number is tied to the angle by:

  • ::: the length of arrow (written with straight bars, like "size of").
  • ::: cosine of the angle between them — a dial that reads when the arrows point the same way, when they are perpendicular, and when they point opposite ways.

WHY this tool and not another? We are not asked for lengths or areas — we are asked for an angle between two directions. The dot product is the one operation whose entire job is to convert two direction-arrows into their angle. Its sign alone already tells us "same-ish way" (+) vs "opposite-ish way" (−), which is exactly the question a bond angle asks.

PICTURE. The figure shows the dial: two arrows swinging from side-by-side () through a right angle () to back-to-back (). Remember where lands — we are about to get a negative value, meaning "the bonds lean away from each other".

Figure — Crystal lattice structure of silicon

Step 5 — Compute the pieces

WHAT. Take two of our bond vectors, and . First the dot product:

  • ::: the -parts agree, contributing positively.
  • ::: the -parts disagree, pulling the total down.
  • ::: the -parts disagree too. Total .

Now the lengths. The length of an arrow is the 3D Pythagoras: .

  • Every square kills the minus sign, so both arrows have the same length — sensible, since all four bonds are identical.

WHY. These are the exact three numbers the formula demands. With them in hand, only is left unknown.

PICTURE. The figure isolates just these two red arrows and , labels their shared length , and shows they visibly lean apart — foreshadowing the negative cosine.

Figure — Crystal lattice structure of silicon

Step 6 — Solve for the angle

WHAT. Put the pieces into the angle formula and isolate :

  • The top () is the dot product from Step 5.
  • The bottom () is the two lengths multiplied.
  • Result: .

WHY. We rearranged by dividing both sides by the lengths, because we know everything except . The negative sign is the punchline: from Step 4's dial, a negative cosine means the angle is obtuse — bigger than . The bonds genuinely point away from each other, confirming "maximum spread".

PICTURE. The figure overlays the value on the cosine dial from Step 4: it sits just past the halfway point toward "back-to-back", so the angle is a bit past — a wide, obtuse opening.

Figure — Crystal lattice structure of silicon

Step 7 — Undo the cosine: arccos

WHAT. We have but we want itself. The tool that reverses cosine is arccos (also written ): it asks "which angle has this cosine?"

WHY arccos? Cosine takes an angle and gives a ratio; we have the ratio and need the angle back — that is exactly the job arccos exists for. (Compare: undoes to recover a slope's angle; here we undo to recover a spread's angle.) Because is negative, arccos lands in the obtuse range , matching our expectation.

Edge / sanity check — the three degenerate answers: to trust arccos, know what it returns at the extremes:

  • If the bonds were identical directions, (arrows on top of each other).
  • If perpendicular, (like a simple-cubic lattice — wrong for silicon).
  • If exactly opposite, (a straight line, only possible with 2 bonds).

Our sits between the and cases — closer to — giving . That is the widest four arrows can manage; only two arrows could reach the full .

PICTURE. The figure plots the cosine curve from to , marks the three checkpoints (), drops a red line from on the vertical axis across to the curve and down to on the angle axis — the "undoing" made visible.

Figure — Crystal lattice structure of silicon

The one-picture summary

WHAT. One frame collapses the whole journey: cube → 4 alternating corners → two red bond arrows → dot product → divide by lengths .

Figure — Crystal lattice structure of silicon
Recall Feynman: the whole walkthrough in plain words

I want the angle between two silicon bonds. First I draw one atom with four arrows poking out — its four bonds. Those arrows shove each other apart until they're as far as they can be, which makes a little pyramid shape called a tetrahedron. That shape is hard to do math on, so I hide it inside a cube: the four arrow-tips land on four alternating corners of the cube, corners like and — nice whole numbers. To get the angle between two arrows I use one gadget, the dot product: multiply the matching parts and add them up, which for these two gives . I also find each arrow's length, both . The angle rule says the dot product equals length-times-length-times-cosine, so cosine of the angle is over 's square roots, i.e. . Negative means the bonds lean away — an obtuse angle. Finally I press the "un-cosine" button, arccos, and out pops . That's it: no silicon chemistry, just four arrows told to spread out, and geometry handing back one famous angle.


Connections