There are a lot of phenomena at play here.
First of all, you already mentioned there's a factor of changing atmospheric composition. This one is easiest to understand. Consider a very simplistic model where your atmosphere have no turbulent mixing, there are just 2 type of gases and temperature of that gas is constant everywhere. Then you can model each gas with a simple model where gas density decays exponentially with height (but with different scale parameters) and these gases do not influence each other. Therefore at any given height this simplistic atmosphere can be modeled as a sum of densities of your two gases at this height
Now consider that one gas is 10x heavier (in terms of molecular weight) than other and it takes 90% of atmosphere at ground level. Let's play with our toy model in Python
import numpy as np
import matplotlib.pyplot as plt
h = np.linspace(0, 5, 100)
gas1 = 0.9 * np.exp(h * -10)
gas2 = 0.1 * np.exp(-h)
rho = gas1 + gas2
proportion = 100 * (gas1 / (gas1 + gas2))
plt.title("Proportion of gas 1 (%)")
You can see a clear knee in this model and change in atmospheric composition that starts at 90% heavy gas but quickly gets to 100% of light gas at high enough altitude. Once composition changes sufficiently, so does the slope of density chart. As you can see we did not needed anything here to get a knee except for 2 gases with sufficiently different density.
Now, the real atmosphere is of course a lot more complicated. One factor is a turbulent mixing of gases below certain height. This mixing makes atmosphere to act like it was a single gas with some "average" molecular weight rather than sum of non-interacting gases. In practice this means that "heavy" gas goes higher than it would in a simplistic model without turbulent mixing. But turbulent mixing is driven by heat (mostly heat of of Earth surface) and amount of available energy limit it in how high it can go. So at some altitude there's a transition between atmosphere dominated by turbulent mixing and atmosphere dominated by gas diffusion. This makes change in atmospheric composition and therefore "knee" more pronounced because turbulent mixing produces different mix at high altitude than diffusion.
Now, so far we were talking about atmosphere with constant temperature. This is unrealistic, temperature of real atmosphere changes considerably with a height. We get a lot of energy from Sun and some of it comes in form of UV and even X-ray light with highly energetic photons. Atmospheric gases absorb such photons well but this process goes only for highest layers of atmosphere because photons are absorbed before they can pass deep enough. A net result is a sharp and very strong increase of atmosphere temperature at high enough altitude. Furthermore, absorption of some of those energetic photons result in dissociation of molecules like O2 into simpler monoatomic gases. Both those processes (dissociation and high temperature) greatly reduce gas density and make it essentially "much lighter" for purposes of scale height. This means that even if you take pure noble gas atmosphere, rapid change in temperature will still produce two layers in it, "cold" layer of gas below some altitude and "hot" outer layer of gas that will behave differently. However it will be less pronounced than in earth-like atmosphere.
To sum things up:
- Atmosphere naturally separates in layers of different gases above certain altitude
- Lighter gases replace heavier ones because they have bigger scale height and their density decreases slower than density of heavier gases
- Solar light naturally produces outer layer of gas in atmosphere that behaves like it's very "light" and therefore have large scale height, very different from gas of lower levels
All these factors tend to produce "knee". However in case of Earth, I'd bet that main factor is the solar light.
You can still get an atmosphere without a knee, but you'll need to move your planet away from any sources of light and make sure that its atmosphere is composed from gases with similar density. A distant gas giant with pure hydrogen atmosphere maybe?