Is it possible to detect a particle, without energy transfer?

All detection requires a signal, which in turn requires energy. You can't detect without taking energy from the signal!

We can use a qubit as a particle detector that doesn't change the particle's energy. This can be implemented as follows. We start out with a qubit initialized in the state $\left|0\right>$ and we apply the Hadamard gate $U$ that acts as follows:

$$ \begin{split} U\left|0\right> &= \frac{1}{\sqrt{2}}\left[\left|0\right\rangle + \left|1\right\rangle\right]\\ U\left|1\right> &= \frac{1}{\sqrt{2}}\left[\left|0\right\rangle - \left|1\right\rangle\right] \end{split} $$

Note that $U$ is its own inverse, so applying $U$ again will bring the qubit back to the state $\left|0\right\rangle$ we started out with. But now consider what happens if during the time the qubit spends being a superposition of $\left|0\right\rangle$ and $\left|1\right\rangle$ a particle collides with it, but such that no energy is exchanged. Then the qubit will get entangled with the particle, so the qubit-particle system will be in a state of the form:

$$\left|\psi\right\rangle = \frac{1}{\sqrt{2}}\left[\left|0\right\rangle \left|D_0\right\rangle + \left|1\right\rangle\left|D_1\right\rangle\right]$$

where the states $\left|D_{i}\right\rangle$ are the particle states after scattering off the qubit in state $\left|i\right\rangle$. You might think that because the qubit was not affected by the interaction at all, we cannot perform a measurement on the qubit's state to find out that it has interacted with a particle. But watch what happens if we apply the Hadamard gate again to the qubit:

$$U\left|\psi\right\rangle =\left|0\right\rangle\left|D^{+}\right\rangle+\left|1\right\rangle \left|D^{-}\right\rangle$$

where $D^{\pm} = \frac{1}{2}\left[\left|D_0\right\rangle\pm\left|D_1\right\rangle\right]$

So, had there been no interaction, the qubit would have returned to the initial state $\left|0\right\rangle$ but now we end up with an entangled state of the qubit and the particle such that there is now a finite probability of finding the qubit in the state $\left|1\right\rangle$, despite the fact that the collision with the particle happened in a purely elastic way such that it did not affect the physical state of the qubit in any way at the time of the collision. The probability of finding the qubit in the state $\left|1\right\rangle$ is $\frac{1}{2}\left[1-\operatorname{Re}\left\langle D_0\right|D_1\left.\right\rangle\right]$, so it depends on the overlap between the two particle states corresponding to scattering off the qubit in the two states of the superposition.

If the states $\left|D_i\right\rangle$ are orthogonal, then you have 50% probability of finding the qubit in the states $\left|0\right\rangle$ and $\left|1\right\rangle$; the density matrix after tracing out the particle state is $\frac{1}{2}\left[\left|0\right\rangle\langle 0| + \left|1\right\rangle\langle 1|\right]$.

As a rule in order to detect a set of excited quantum states you have to input energy necessary to excite the harmonic oscillator modes of those states. If you want to detect the Higgs particle, which occurs at $125$GeV energy you need to impart that energy into the vacuum by exciting virtual modes through an interaction with at least that much energy. There is also the further consequences of measurement fidelity, which then requires far more energy to make the signal robust. The Fermilab Tevatron did appear to detect the Higgs particles at the TeV energy level, but the statistics were not good enough to claim detection. Now leap to the Planck scale where to detect quantum gravity you have to excite modes that are $10^{19}$GeV in energy gaps, say from vacuum to the generation of a quanta of black hole. To do quantum gravity experiments it means one must have interaction energy on this scale. This makes quantum gravity very tough or impossible to do in the lab.

There are some other more subtle ways of detecting things. The Lamb shift is a low energy detection of QED processes that split energy levels in the atom. The Lamb shift is in the microwave domain that is far in the IR domain from the energy gap of the $^2S_{1/2}$ and $^2P_{1/2}$ energy levels. This still involves the detection of an excited state, but with far lower energy than the main energy-gap of the theory. In the case of quantum gravity these gaps are the formation of Hawking radiation that is far more IR than the Planck energy. For standard model physics this might occur in subtle corrections to the electronic states of atoms as the wave function has small “exponential tails” in the nucleus so there could be weak neutral currents that perturb electronic states. There were proposals along these lines back in the 1970s, but I am not sure what came out of that.

There are other subtle measurements, such as weak quantum measurements. In the end though one must has a detector that registers a voltage. Some measurements involve no voltage, such as using the polarization of the electric vector in photons to do a “non-measurement” of a hidden bomb. However, if a bomb is not present a homodyne detector registers a voltage. It is probably not physically possible to measure something without having some coupling that can transfer energy or momentum to a detector or needle state.