Using intrinsic properties of quantum dots to provide additional security when uniquely identifying devices

Abstract

Unique identification of optical devices is important for anti-counterfeiting. Physical unclonable functions (PUFs), which use random physical characteristics for authentication, are advantageous over existing optical solutions, such as holograms, due to the inherent asymmetry in their fabrication and reproduction complexity. However, whilst unique, PUFs are potentially vulnerable to replication and simulation. Here we introduce an additional benefit of a small modification to an established model of nanoparticle PUFs by using a second measurement parameter to verify their authenticity. A randomly deposited array of quantum dots is encapsulated in a transparent polymer, forming a tag. Photoluminescence is measured as a function of excitation power to assess uniqueness as well as the intrinsic nonlinear response of the quantum material. This captures a fingerprint, which is non-trivial to clone or simulate. To demonstrate this concept practically, we show that these tags can be read using an unmodified smartphone, with its built-in flash for excitation. This development over constellation-style optical PUFs paves the way for more secure, facile authentication of devices without requiring complex fabrication or characterisation techniques.

Figure 1

The concept of using the non-linear response of a quantum dot-based tag for authentication. (a) Illustrates a security tag’s surface, coated in quantum dots and divided into pixels for clarity. As the incident illumination increases (from left to right at 3 excitation powers, E1, E2, E3) the average amount of photoluminescence (shown in red) increases, but at different rates from region to region. T1, T2 and T3 show the switch-on intensities of different QD regions. (b) Sketches how the photoluminescence intensity from three example locations could vary differently with increasing excitation. (c)–(f) The top section of each panel shows photoluminescence from a thin film containing colloidal quantum dots at increasing excitation power, with pixel histograms underneath displaying the number of pixels recording each intensity value (0 to 65,355) at the different incident powers: 36.4 μW (c), 57.6 μW (d), 91.4 μW (e), and 145 μW (f). Scale bar represents 50 μm.

Figure 2

The non-linear optical response of a colloidal quantum dot film. (a) Log–log plot showing the emission intensity from a series of points chosen to represent 3 different region types on the sample: QDs on the Si substrate (labels Si1-3), QDs on the Al layer (Al1-3), and QD clusters on the surface (C1-3). (b)–(d) Show images created by taking the ratio of emission intensity maps recorded at adjacent excitation intensities. These are (b) 0.4/0.15 μW, (c) 3.6/1.4 μW and (d) 115/90 μW. Scale bar represents 50 μm. Z scale represents the relative difference in point-to-point variation on the sample surface.

Figure 3

How the emission from different spatial regions responds to changes in incident power. (a) shows the excitation intensity difference at which there is the greatest relative increase in emission intensity. (b) shows the ‘switch-on’ intensity, i.e. the excitation power at which the emission intensity exceeds the background level. The colour scale indicates the incident power (μW).

Figure 4

Measuring the non-linear response of a quantum dot-based tag using a smartphone’s camera and flash. (a) A smartphone image of a tag comprising a QR Code (black and white outer) surrounding a blank region in the centre that is coated with colloidal quantum dots (CQDs) in a polymer. (b) and (c) false-colour images showing the exponent extracted from a power fit on the emission-excitation curved measured across the area of a ‘fake’ tag (b) and a ‘real’ tag (c). The exponent in the centre of the QD region is lower degree than the surroundings. (d) For a series of ‘real’ (R1-5, containing CQDs in the centre region) and ‘fake’ (F1-3, no CQDs) tags, the difference in linearity is plotted, with whiskers on the boxes spanning two standard deviations.