How Tiny Defects Hurt Quantum Computers

In the paper's words

Parasitic two‑level system (TLS) defects in the amorphous tunnel barriers of superconducting qubits are a major identified source of decoherence, yet device designers lack closed‑form estimates linking defect density and junction geometry to the probability that a device is degraded and to the resulting dephasing rate. This paper models the defect population of a Josephson junction as a random ensemble—a random dynamical system in which defect positions and frequencies are random variables and the qubit is a deterministic observer—and derives two complementary closed‑form results. A Poisson counting argument gives the probability that a junction hosts a strongly coupled defect: for an assumed density of 10²4 mathrmm^-3, a barrier volume of 4×10^-23 mathrmm³, and a fractional coupling window of 2×10^-3, the mean count is λ=0.08 and the single‑junction probability is P₁=0.0769; on a 100‑junction chip the expected count is $8$ and the affected‑chip probability is $0.99966$.

The abstract, verbatim, and its opening sentence is the claim. Every line below is quoted from the text.

Numbers it reports

1. Using the baseline geometry $$ V = 4\times10^{-23}\;\mathrm{m^{3}},\; f_{\Gamma}=2\times10^{-3}, $$ $$ \rho^{\mathrm{target}} = -\frac{\ln(0.99)}{(4\times10^{-23})(2\times10^{-3})} = \frac{0.0100503}{8\times10^{-26}} = 1.2563\times10^{23}\;\mathrm{m^{-3}} . $$ Thus the process must achieve a defect density an order of magnitude lower than the baseline $10^{24}\;\mathrm{m^{-3}}$ to meet the 1 % yield goal.

   from 5. Inverting the counting formula to set a target defect density

2. A further ten‑fold reduction drives the chip‑level probability to $$ P_{\mathrm{chip}}=1-e^{-0.08}=0.0769, $$ i.e. $\approx$ 7.7 % of chips would contain at least one strongly coupled defect.

   from 8. Practical Implications and Engineering Pathways

3. Require $P_{1}^{\mathrm{target}} = 0.01$ (i.e. at most 1 % of junctions affected).

   from 5. Inverting the counting formula to set a target defect density

4. Per‑defect factor: $$\left(\frac{g_{0}}{|\bar{\Delta}|}\right)^{2}= \left(\frac{1}{100}\right)^{2}=10^{-4}.$$ Per‑defect rate: $$\Gamma_{\varphi}^{(1)} = 10^{-4}\times10^{4}\ \mathrm{s^{-1}} = 1\ \mathrm{s^{-1}}.$$ Total dispersive rate: $$\Gamma_{\varphi}^{(\mathrm{disp})}= n\,\Gamma_{\varphi}^{(1)} = 5\ \mathrm{s^{-1}}.$$ Dephasing time: $$T_{\varphi}= \frac{1}{\Gamma_{\varphi}^{(\mathrm{disp})}} = 0.2\ \mathrm{s}.$$

   from 4.6 Dispersive dephasing rate

5. R6 (Sensitivities). $P_{1}$ is linear in $\rho$ and in $f_{\Gamma}$: a tenfold density reduction moves $P_{1}$ from $0.0769$ to $0.0080$ and $P_{\mathrm{chip}}$ from $0.99966$ to $0.5507$ (Section 4.8).

   from 5. Results

6. Because $\lambda$ is linear in $\rho$, $$ \lambda(\rho)=\rho\,V\,f_{\Gamma}\;\;\Longrightarrow\;\; \frac{\partial \lambda}{\partial \rho}=V\,f_{\Gamma}. $$ A ten‑fold reduction $\rho\to\rho/10$ yields $$ \lambda' = \frac{\lambda}{10}=0.008 . $$ The corresponding probability is $$ P_{1}' = 1-e^{-\lambda'} = 1-e^{-0.008}. $$ Using the series expansion $$ e^{-0.008}=1-0.008+0.000032-0.0000000853\approx0.992032, $$ $$ P_{1}' = 1-0.992032 = 0.007968\approx0.0080 . $$ Thus $P_{1}$ scales proportionally with $\rho$; a ten‑fold density reduction lowers the affected‑junction fraction from $7.69\%$ to $0.80\%$.

   from 2. Sensitivity of the single‑junction probability to defect density

How it is done

1. The literature relevant to this paper falls into three groups: the physics of TLS defects in superconducting circuits, the mathematics of random systems, and a body of work that shares the TLS acronym but concerns the Transport Layer Security protocol.

   from 2. Background and Related Work

2. Defects are modeled as point entities distributed independently and uniformly at random with number density $\rho$ inside a barrier of area $A$ and thickness $d$, so the barrier volume is $$V = A\,d.$$ Each defect has a transition frequency drawn uniformly from a band of width $\Delta_{0}$.

   from 3.2 Counting model

3. Inputs: $\rho=10^{24}\ \mathrm{m^{-3}}$, $V=4\times10^{-23}\ \mathrm{m^{3}}$, $f_{\Gamma}=2\times10^{-3}$.

   from 4.3 Single‑junction defect probability (spatio‑spectral model)

4. Inputs: $\bar{N}_{d}=100$, $B=19\ \mathrm{GHz}$, $2\delta=20\ \mathrm{MHz}=0.02\ \mathrm{GHz}$. $$\lambda = \frac{\bar{N}_{d}\,2\delta}{B}= \frac{100\times0.02}{19}=0.10526316.$$ Poisson approximation: $$P_{\ge1}=1-e^{-\lambda}=1-e^{-0.10526316}=0.09991221\approx0.0999.$$ Exact binomial check: $$P_{\ge1}=1-\left(1-\frac{2\delta}{B}\right)^{\bar{N}_{d}}=1-(0.99894737)^{100}=0.099962\ (\text{consistent}).$$ Mean defect spacing: $$\bar{s}= \frac{B}{\bar{N}_{d}} = \frac{19\ \mathrm{GHz}}{100}=1.9\times10^{8}\ \mathrm{Hz}=190\ \mathrm{MHz}.$$

   from 4.5 Spectral counting model and near‑resonance probability

What would falsify it

1. The results convert defect decoherence from a device‑by‑device pathology into an ensemble‑level design parameter with an explicit falsification register.

   from Abstract

2. A falsification register: the specific spectroscopy observations that would refute the counting model and the noise model, and a list of the assumptions on which each number rests (Sections 7 and 7).

   from 1. Introduction

3. This asymmetry is what makes the framework predictive: the fabrication process controls the ensemble parameters $(\rho,\Delta_{0},g_{0})$, and the observable decoherence is an ensemble average at fixed $\omega_{q}$.

   from 3.1 The random dynamical system formulation

Where the claim stops

1. The principal result is the conversion of two assumed densities and two assumed windows into device‑level probabilities and times: $P_{1}=0.0769$, $P_{\mathrm{chip}}=0.99966$, $T_{\varphi}=0.2\ \mathrm{s}$ (dispersive) and $800\ \mu\mathrm{s}$ (telegraph).

   from 7. Discussion

2. The value of these numbers is conditional on the assumptions, and the conditionality is the point of this section.

   from 7. Discussion

3. Equally, a measured density of strongly coupled defects inconsistent with $\rho V f_{\Gamma}$ by more than the Poisson fluctuation $\sqrt{N_{j}\lambda} \approx 2.8$ across a chip population would falsify the assumed $\rho = 10^{24}\ \mathrm{m^{-3}}$ or the uniform spectral distribution.

   from 7. Discussion

Where to read

1. Abstract191 words · 1 min
2. 1. Introduction615 words · 3 min
3. 2. Background and Related Work523 words · 2 min
4. 3. Methods837 words · 4 min
5. 4. Analysis571 words · 2 min
6. 5. Results287 words · 1 min
7. 6. Extended Derivations and Worked Examples724 words · 3 min
8. 7. Discussion571 words · 2 min
9. Appendix A. Divergence report20 words · 1 min
10. Appendix B. Claim attribution647 words · 3 min
11. 8. Practical Implications and Engineering Pathways549 words · 2 min
12. 10. Conclusion266 words · 1 min