Pair your accounts.

Export articles to Mendeley

Get article recommendations from ACS based on references in your Mendeley library.

Pair your accounts.

Export articles to Mendeley

Get article recommendations from ACS based on references in your Mendeley library.

You’ve supercharged your research process with ACS and Mendeley!

STEP 1:
Click to create an ACS ID

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

Please note: If you switch to a different device, you may be asked to login again with only your ACS ID.

MENDELEY PAIRING EXPIRED
Your Mendeley pairing has expired. Please reconnect
ACS Publications. Most Trusted. Most Cited. Most Read
Highly Reliable Magnetic Memory-Based Physical Unclonable Functions
More

OR SEARCH CITATIONS

My Activity
Publications

Figure 1Loading Img
Download Hi-Res ImageDownload to MS-PowerPointCite This:ACS Nano 2024, 18, 20, 12853-12860
  • Open Access
Article

Highly Reliable Magnetic Memory-Based Physical Unclonable Functions
Click to copy article linkArticle link copied!

  • Jaimin Kang
    Jaimin Kang
    Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    More by Jaimin Kang
    Orcidhttps://orcid.org/0000-0003-0347-5100
  • Donghyeon Han
    Donghyeon Han
    Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    More by Donghyeon Han
  • Kyungchul Lee
    Kyungchul Lee
    Department of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea
    More by Kyungchul Lee
  • San Ko
    San Ko
    Department of Physics, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Korea
    More by San Ko
  • Daekyu Koh
    Daekyu Koh
    Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    More by Daekyu Koh
  • Chando Park
    Chando Park
    Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    More by Chando Park
  • Jaesoo Ahn
    Jaesoo Ahn
    Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    More by Jaesoo Ahn
  • Minrui Yu
    Minrui Yu
    Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    More by Minrui Yu
  • Mahendra Pakala
    Mahendra Pakala
    Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    More by Mahendra Pakala
  • Sujung Noh
    Sujung Noh
    R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    More by Sujung Noh
  • Hansaem Lee
    Hansaem Lee
    R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    More by Hansaem Lee
  • JoonHyun Kwon
    JoonHyun Kwon
    R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    More by JoonHyun Kwon
  • Kab-Jin Kim
    Kab-Jin Kim
    Department of Physics, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Korea
    More by Kab-Jin Kim
    Orcidhttps://orcid.org/0000-0002-8378-3746
  • Jongsun Park
    Jongsun Park
    Department of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea
    More by Jongsun Park
  • Soogil Lee*
    Soogil Lee
    Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    Department of Electronic Engineering, Gachon University, 1342 Seongnam-daero, Sujeong-gu, Seongnam 13120, Korea
    *Email: [email protected]
    More by Soogil Lee
  • Jisung Lee*
    Jisung Lee
    R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    *Email: [email protected]
    More by Jisung Lee
  • Byong-Guk Park*
    Byong-Guk Park
    Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    *Email: [email protected]
    More by Byong-Guk Park
    Orcidhttps://orcid.org/0000-0001-8813-7025
Open PDFSupporting Information (1)

ACS Nano

Cite this: ACS Nano 2024, 18, 20, 12853–12860
Click to copy citationCitation copied!
https://doi.org/10.1021/acsnano.4c00078
Published May 8, 2024

Copyright © 2024 The Authors. Published by American Chemical Society. This publication is licensed under

CC-BY-NC-ND 4.0 .

Abstract

Click to copy section linkSection link copied!
High Resolution Image
Download MS PowerPoint Slide

Magnetic random-access memory (MRAM), which stores information through control of the magnetization direction, offers promising features as a viable nonvolatile memory alternative, including high endurance and successful large-scale commercialization. Recently, MRAM applications have extended beyond traditional memories, finding utility in emerging computing architectures such as in-memory computing and probabilistic bits. In this work, we report highly reliable MRAM-based security devices, known as physical unclonable functions (PUFs), achieved by exploiting nanoscale perpendicular magnetic tunnel junctions (MTJs). By intentionally randomizing the magnetization direction of the antiferromagnetically coupled reference layer of the MTJs, we successfully create an MRAM-PUF. The proposed PUF shows ideal uniformity and uniqueness and, in particular, maintains performance over a wide temperature range from −40 to +150 °C. Moreover, rigorous testing with more than 1584 challenge–response pairs of 64 bits each confirms resilience against machine learning attacks. These results, combined with the merits of commercialized MRAM technology, would facilitate the implementation of MRAM-PUFs.

This publication is licensed under

CC-BY-NC-ND 4.0 .
  • cc licence
  • by licence
  • nc licence
  • nd licence
Copyright © 2024 The Authors. Published by American Chemical Society

Subjects

what are subjects

Keywords

what are keywords
Magnetic random-access memory (MRAM), which exploits the spin of electrons for information processing and storage, has attracted much attention due to its rapid operation, strong endurance, and low power consumption in comparison with other nonvolatile memories. (1−3) In particular, MRAM has been developed for various applications, including functioning as memory in automobile and space systems, (4−7) and this technology has been successfully commercialized through embedded memory implementations. (8−11) Beyond conventional nonvolatile memory applications, MRAMs are evolving to become a core element of non-von Neumann computing technology, (12−15) encompassing in-memory computing for neural networks and probabilistic computing. For example, in-memory computing has been realized with a crossbar array based on MRAM cells, achieving successful pattern recognition. (14) In addition, the utilization of stochastic magnetic tunnel junctions (MTJs) has enabled probabilistic-bit computing, effectively tackling challenges such as integer factorization. (15)
Among such unconventional functionalities offered by MRAM, one notable application is the physical unclonable function (PUF). PUFs leverage process variations occurring during the manufacturing process or the inherent stochasticity of a physical entity to create secure cryptographic keys, facilitating secure communication and/or authentication within electronic devices. (16−19) Various forms of stochastic variations have been utilized for the implementation of PUFs. (20−22) For instance, PUFs using Si-based semiconductor devices, such as the arbiter PUF, have been developed due to their compatibility with CMOS technology. (21) However, they require resilience improvements against side-channel analysis and environmental fluctuations. (23) Additionally, PUFs employing various materials, including metal-oxide memristors, graphene, polymers, perovskites, and organic crystals, are actively being explored. (24−28) They each offer distinct advantages as components of PUFs, such as a high degree of randomness with multiple entropy sources, concealability of stored information, or applicability to flexible substrates. However, the essential requirements for practical applications when using those materials, such as information retention under various environmental conditions and compatibility with CMOS technology, have yet to be fully realized. Therefore, it is important to develop a CMOS-compatible digital PUF that ensures reliability under various operation conditions.
Spintronic PUFs are being developed as an alternative to tackle these challenges and enable the development of on-chip PUFs integrated with CMOS technology. (29−36) A recent study demonstrated the secure access and data protection capabilities of CMOS-integrated on-chip spintronic PUFs within in-memory computing. (33) However, despite advances in spintronic or MRAM-based PUFs, some of the crucial PUF properties remain to be verified for practical applications. First, it is imperative to ensure reliability at various operating temperatures. For instance, PUFs should be capable of functioning in broader temperature ranges from −40 to 150 °C for automotive applications. (37) Second, it is essential to confirm the randomness of the entropy source to secure resilience against machine learning attacks. This concern has not yet been thoroughly explored in spintronic or MRAM-based PUFs, necessitating the development of highly reliable MRAM-based PUFs.
Here, we demonstrate highly reliable MRAM-PUFs by introducing a robust entropy source into 70 nm-sized MTJs. By arbitrarily manipulating the magnetization direction of synthetic antiferromagnetically (SAF)-coupled reference layers (RLs), we create an MRAM-PUF that generates random binary outputs with tunneling magnetoresistance (TMR) of ∼130%. It is demonstrated that our PUF exhibits ideal PUF properties. First, we achieve optimal values for the entropy, inter-Hamming distance (inter-HD), and correlation coefficient (CC). The randomness of the entropy source is verified through the National Institute of Standards and Technology (NIST) test. Second, these characteristics are maintained across a wide range of temperatures; identical PUF patterns with TMR ratios exceeding 100% are observed throughout a temperature range from −40 to 150 °C, thus satisfying industrial requirements. Lastly, we demonstrate the resilience of the proposed MRAM-PUF against machine learning attacks using simulations based on a generative adversarial network (GAN) model, where a large volume of challenge–response pairs (CRPs), consisting of 1584 CRPs with 64 bits each, are examined by magneto-optical Kerr effect (MOKE) spectroscopy. These results, combined with the merits available in the commercialized MRAM industry, highlight promising aspects of MRAM-PUF implementation.

Results and Discussion

Click to copy section linkSection link copied!
In order to construct the MRAM-PUF, we fabricated arrays of perpendicularly magnetized spin-transfer-torque MTJs (STT-MTJs) consisting of a CoFeB free layer (FL), an MgO tunnel barrier, and a CoFeB RL coupled to the SAF pinned layer (PL), as illustrated in Figure 1a. The diameter of the circular MTJs used here is 70 nm. First, we measured the current (I)–voltage (V) curve of a representative MTJ to examine its electrical properties. Figure 1b shows the result, which shows the typical bipolar switching behavior of an STT-MTJ, where magnetization switching occurs at a current of ±110 μA. From the I–V curve, we extracted the resistance (R)–I curve (Figure 1c), where a high resistance (RAP) of 7 kΩ and a low resistance (RP) of 3 kΩ are obtained at a current of 2 μA. The TMR ratio, defined by (RAP – RP)/RP × 100%, is calculated to be approximately 130%. Subsequently, we measured the resistance as a function of the out-of-plane magnetic field (Bz) to identify the magnetization switching of each magnetic layer. Figure 1d shows the magnetoresistance (MR) curve measured with a reading current of 2 μA, exhibiting two distinct resistance states equivalent to the RAP and RP values. Here, there are three magnetic fields at which the resistance changes; at Bz = 0.17 T, the magnetization of the CoFeB FL (red arrow) is switched, altering the magnetization alignment with the bottom CoFeB RL (blue arrow) from antiparallel to parallel. This results in a change from RAP to RP. At Bz = 0.38 T, the magnetizations of both the SAF-coupled PL (gray arrow) and the CoFeB RL are reversed simultaneously, switching back to RAP. When Bz exceeds 0.6 T, overcoming the RKKY coupling, all three magnetization directions of FL, RL, and PL are aligned in parallel, resulting in RP. Figure 1e shows the minor loop measured by sweeping Bz between ±0.27 T, where only the magnetization of the FL is changed, with that of the RL in the + z-direction retained, resulting in the two resistance states RAP and RP. These can serve as digital binary outputs “0” or “1” for the MRAM-PUF in our study.

Figure 1

Figure 1. Electrical and magnetic properties of the STT-MTJs. (a) Optical images of the STT MRAM (STT-MTJ) array (left and middle) and the device structure (right). The diameter of the circular pillar is 70 nm. STT switching characteristics: (b) current (I)–voltage (V) curve and (c) resistance (R)–I curve. (d) MR curves measured as a function of the out-of-plane magnetic field (Bz). Green (purple) symbols represent MR data obtained by sweeping the magnetic fields from +Bz (−Bz) to –Bz (+Bz). Red, blue, and gray arrows indicate the magnetization direction of the FL, RL, and PL, respectively. (e) A minor MR curve measured by sweeping Bz between ±0.27 T. Here, the magnetization direction of RL is fixed along the +z-direction.

High Resolution Image
Download MS PowerPoint Slide
To create an MRAM-PUF using the MTJ array, we first applied an oscillating magnetic field with a decreasing amplitude from 480 to 260 mT to randomize the magnetization directions of the SAF-coupled RLs (see Methods for details). Note that the range of the applied magnetic field was chosen to start near the coercivity of the RL and end at a value exceeding the coercivity of the FL. Consequently, the magnetization directions of the RLs were selectively randomized, while the magnetization directions of the FLs remained the same for all devices. Then, we measured the minor TMR curves of 64 individual MTJs. Figure 2a shows the distribution of the MR curves, which exhibit either clockwise (red) or counterclockwise (blue) polarity. Here, the clockwise and counterclockwise MR curves are evenly proportioned and are randomly distributed in space. This is attributed to the randomized magnetization direction of the RL. By reading the TMR value of each PUF unit device, we can construct MRAM-PUFs featuring digital binary outputs denoted as RP and RAP. This digital nature has the potential to eliminate the need for additional peripheral circuits such as an analog-to-digital converter, thus reducing the energy consumption and chip area. Figure 2b–d presents other MRAM-PUFs fabricated using an identical process. The MR curves are given in Supporting Information 1. Here, RP and RAP are denoted in red and blue, respectively. Note that the device shown in Figure 2b is identical to that in Figure 2a. The distinct patterns observed in each MRAM-PUF demonstrate their potential to serve as secure cryptographic keys in hardware security modules.

Figure 2

Figure 2. Construction and characteristics of MRAM-PUFs. (a) MR curves of 64 MTJs after the randomization process. Red (blue) represents the clockwise (counterclockwise) switching polarity. Distributions of the switching polarity in 64 bit MRAM-PUFs (b) 1, (c) 2, and (d) 3. (e) Entropy of 31 MRAM-PUFs. PMF as a function of the (f) normalized inter-HD and (g) CC. The red curves indicate Gaussian fittings.

High Resolution Image
Download MS PowerPoint Slide
To evaluate the characteristics of the PUF, we analyzed 31 MRAM-PUFs, each containing 64 MTJs, focusing on three key PUF metrics: the entropy (E), inter-HD, and CC. First, we calculated the E value to quantify the degree of uniformity using the following equation
E=−[plog2⁡p+(1−p)log2(1−p)]
(1)
Here, p is the probability of having “0” or “1” in the CRP. Therefore, when a PUF has an equal number of bits “0” and “1” (p = 0.5), the E value becomes 1, ensuring maximum randomness. Figure 2e shows the calculated E values for 31 MRAM-PUFs. The mean value is found to be 0.980, close to the ideal value of 1. This confirms the uniform distribution of bits “0” and “1” within our PUFs. Then, we calculated inter-HD and CC to examine differences and linear correlations between the created PUFs. Inter-HD refers to the number of nonidentical bits when comparing two different PUFs of the same bit length. Therefore, inter-HD is a measure of how different two PUFs are, representing the uniqueness of a PUF. CC indicates the strength of correlation between two statistical quantities, which is obtained by calculating the Pearson’s CC between two PUFs. (38) Figure 2f,g shows the probability mass function (PMF) versus the normalized inter-HD and CC curves, where the solid lines represent the Gaussian fittings. The mean values of the inter-HD and CC are extracted to be 0.502 ± 0.001 and 0.001 ± 0.003, respectively. This indicates that the proposed MRAM-PUFs generate unique CRPs with no intercorrelations.
We next demonstrated the robustness of the MRAM-PUFs at various operating temperatures. For the industrial application of PUFs, it is crucial to ensure consistent output responses across different operation conditions, especially in the presence of temperature fluctuations. To confirm the robustness of the entropy source in our MRAM-PUFs, we conducted MR measurements of an MRAM-PUF with 64 unit devices at various temperatures. Figure 3a presents representative MR curves measured at different temperatures ranging from −40 to 150 °C. Although the TMR ratio and coercivity field slightly decrease with an increase in the temperature, the switching polarities of the MR curves remain consistent across all measured temperatures. By analyzing all MR curves (Supporting Information 2), we calculated the mean TMR as a function of the temperature, as shown in Figure 3b. We find out that the TMR ratio remains above 100% even at 150 °C. This indicates that the MRAM-PUF can reliably generate binary output even at high temperatures. Subsequently, we examined whether the generated CRPs change with the temperature. Figure 3c shows the MR curves of 64 PUF unit devices at different temperatures of −40, 25, and 150 °C. Here, the colors of the MR curve and the background represent the switching polarity of the MR curve: red (blue) for clockwise (counterclockwise) polarity. We observe identical switching polarity distributions in all 64 PUF unit devices at the three temperatures, confirming that the bit streams produced by the MRAM-PUF remain the same regardless of the operating temperature. These results demonstrate the reliable performance of MRAM-PUF under temperature fluctuations. We believe that this reliability results from the fact that we used the RL as an entropy source, which is thicker and has a larger coercivity (or exchange bias) compared to that of the FL. This might reduce sensitivity to external stimuli, including temperature, applied magnetic fields, or currents. This also implies that our MRAM-PUF is robust against the process and voltage variations, which potentially causes reliability issues in MRAMs.

Figure 3

Figure 3. Temperature dependence of the MRAM-PUF. (a) Representative MR curves measured at different temperatures. (b) Temperature dependence of the TMR ratio. The error bar indicates the standard deviation. (c) MRAM-PUF consisting of 64 PUF unit devices at different temperatures of −40, 25, and 150 °C.

High Resolution Image
Download MS PowerPoint Slide
We finally examine the resilience against machine learning attacks. To this end, we employed a half MTJ structure consisting of only an antiferromagnetically coupled RL and utilized MOKE microscopy to measure their magnetization directions. We first fabricated samples containing only an RL with a Ta (2 nm)/Pt (5 nm)/Co (1.2 nm)/IrMn (4 nm)/Ta (3 nm) structure. Here, Co is exchange-biased with the antiferromagnet IrMn and thus acts as the RL. This approach allows for the generation of a large volume of CRPs while maintaining the same entropy source of the random magnetization direction of the RLs, as in the MTJ described above. After a randomization process at 220 °C for 5 μm × 5 μm squared samples (see details in the Methods Section), we observed the spatial distribution of the magnetic domains using MOKE. Figure 4a shows representative MOKE images in which the dark (light) gray dots indicate exchange-biased magnetization aligned in the + z (−z)-direction (Figure 4b). We obtained a total of 101,376 bits, equivalent to 1584 MRAM-PUFs of 64 bits each, where each magnetic domain represents either bit “0” or “1” (Supporting Information 3). We evaluated the PUF metrics of E, inter-HD, and CC for these bit streams (Supporting Information 4). These metrics were found to be close to ideal values, consistent with those obtained from the MTJ array shown in Figure 2. In addition, we conducted NIST statistical tests (39) using the generated bit streams (Supporting Information 5) and confirmed the success of all 12 NIST tests, thereby verifying the randomness of the entropy source in our MRAM-PUF.

Figure 4

Figure 4. Resilience against machine learning attacks. (a) Representative MOKE images of 5 μm × 5 μm square-patterned devices after the randomization process. (b) Hysteresis loops of single square samples. The red (blue) curve is obtained from the dark (light) gray area, indicating that exchange bias forms along the positive (negative) z-direction. (c) Schematic illustration of the GAN. The GAN consists of a generator DNN and a discriminator DNN. The generator DNN is trained to generate fake CRPs, while the discriminator DNN is trained to distinguish the fake CRPs from the real ones. Here, we use 1200 CRPs for training, with the remaining 300 CRPs used for a comparison with CRPs generated by the GAN model. PMF as a function of the (d) inter-HD and (e) CC between the real CRPs and the fake CRPs produced by the GAN model.

High Resolution Image
Download MS PowerPoint Slide
We then simulated the machine learning attack by employing a GAN, a deep learning approach for password guessing. (40) The GAN is trained through the adversarial competition of two deep neural networks (DNNs): a generator DNN and a discriminator DNN. As schematically shown in Figure 4c, the generator DNN attempts to generate fake CRPs using random noise as input, while the discriminator DNN tries to distinguish the fake CRPs from the real CRPs. During the training process of GAN, the generator DNN is trained to decrease the estimation probability of the discriminator DNN, while the discriminator DNN is trained to maximize its estimation probability. See the details of the training process using the GAN model in Supporting Information 6. Among the 1500 generated CRPs, we used 1200 CRPs to train the GAN to generate 300 fake PUFs and then compared these with the 300 remaining real CRPs. Subsequently, we calculated the inter-HD and CC to evaluate the difference and linear correlation between the fake and real 300 CRPs. Figure 4d,e correspondingly plots the PMF as a function of the inter-HD and CC, illustrating that the mean inter-HD and CC values are 0.502 and 0.003, respectively. This demonstrates that the GAN model cannot generate CRPs based on information from existing PUFs, thereby highlighting its resilience against machine learning attacks. Together with its performance at high temperatures, the results demonstrate that our MRAM-PUF exhibits superior reliability as compared to that of previously reported spintronic or MRAM-based PUFs (Supporting Information 7).

Conclusions

Click to copy section linkSection link copied!
We successfully demonstrated highly reliable MRAM-PUFs created by utilizing nanoscale MTJ arrays, where the randomly distributed magnetization direction of the SAF-coupled RL serves as an entropy source. Our MRAM-PUF, which generates random binary outputs with a TMR ratio of ∼130%, exhibits ideal uniformity and uniqueness. The MRAM-PUF maintains consistent patterns across a wide range of temperatures, from −40 to 150 °C, meeting the rigorous requirements of automotive applications. Furthermore, we demonstrated the robustness of our MRAM-PUF against machine learning attacks through simulations based on a GAN model. These results, achieved using commercially available MRAM technology, highlight the promising potential of the implementation of MRAM-PUFs.

Methods

Click to copy section linkSection link copied!

Sample Preparation

70 nm-sized MTJ arrays were fabricated using standard STT-MRAM technology at the Maydan Technology Center of Applied Materials. The stacks consist of Co- and Ni-based synthetic antiferromagnet layers, a SAF-coupled CoFeB RL, and a CoFeB-based FL. After the deposition, MTJ patterning was performed using 193 nm dry lithography and advanced etching tools to fabricate MTJ arrays with a diameter of 70 nm and a pitch of 200 nm. These samples were grown and fabricated on a 12 in. wafer, subsequently diced into 26 mm × 10.5 mm dies for experiments. Each piece contains 10,240 MTJs, corresponding to 160 PUFs consisting of 64 MTJs. Exchange-biased layers of a Ta (2 nm)/Pt (5 nm)/Co (1.5 nm)/IrMn (4 nm)/Ta (3 nm) structure were deposited on Si/SiOx (200 nm) substrates using an ultrahigh-vacuum magnetron sputtering system. After the deposition, samples were annealed at 250 °C with an out-of-plane magnetic field to develop a perpendicular exchange bias. They were then patterned into 5 μm × 5 μm squares by photolithography and Ar ion etching for the optical measurements.

Randomization Process

To create a random orientation of SAF-coupled RLs in the MTJ samples, an oscillating magnetic field with a decreasing amplitude was applied along the z-direction from −480 to +260 mT in 0.4 mT steps at room temperature. With the Pt/Co/IrMn exchange-biased samples, an oscillating magnetic field with a decreasing amplitude was applied along the z-direction from −11 to zero in 0.055 mT steps at 240 °C. The samples were then cooled to room temperature without the application of an external magnetic field.

Electrical Measurements

The TMR was measured with a constant reading current of 2 μA at various temperatures from −40 to 150 °C while sweeping the out-of-plane magnetic fields.

Machine Learning Attacks

The GAN used here has two DNNs. The generator DNN consists of an input layer with 50 neurons, followed by three hidden layers with 256, 512, and 1024 neurons, respectively, and an output layer containing 64 neurons. The discriminator DNN is composed of an input layer with 64 neurons followed by two hidden layers with 256 and 512 neurons, respectively, and an output layer with one neuron. The output layer of the generator DNN serves as the input layer of the discriminator DNN. We utilized the hyperbolic tangent activation function for both DNNs. Training was done by back-propagation conducted with 400 epochs.

Supporting Information

Click to copy section linkSection link copied!

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.4c00078.

  • MR curves of MRAM-PUFs; MR measurements of an MRAM-PUF at various temperatures; generation of CRPs through optical measurements; evaluation of PUF performances of the bits generated by optical measurements; NIST randomness test results; GAN model; and performance comparisons of spintronics- or MRAM-based PUFs (PDF)

Terms & Conditions

Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.

Author Information

Click to copy section linkSection link copied!
  • Corresponding Authors
    • Soogil Lee - Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea;  Department of Electronic Engineering, Gachon University, 1342 Seongnam-daero, Sujeong-gu, Seongnam 13120, Korea;  Email: [email protected]
    • Jisung Lee - R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea;  Email: [email protected]
    • Byong-Guk Park - Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea;  Orcidhttps://orcid.org/0000-0001-8813-7025;  Email: [email protected]
  • Authors
    • Jaimin Kang - Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea;  Orcidhttps://orcid.org/0000-0003-0347-5100
    • Donghyeon Han - Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    • Kyungchul Lee - Department of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea
    • San Ko - Department of Physics, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Korea
    • Daekyu Koh - Department of Material Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-road, Yuseong-gu, Daejeon 34141, Korea
    • Chando Park - Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    • Jaesoo Ahn - Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    • Minrui Yu - Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    • Mahendra Pakala - Applied Materials, Inc., 3050 Bowers Avenue, Santa Clara, California 95054, United States
    • Sujung Noh - R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    • Hansaem Lee - R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    • JoonHyun Kwon - R&D Division, Hyundai Motor Company, 150 Hyundaiyeonguso-ro, Namyang-eup, Hwaseong 18280, Korea
    • Kab-Jin Kim - Department of Physics, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Korea;  Orcidhttps://orcid.org/0000-0002-8378-3746
    • Jongsun Park - Department of Electrical Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea
  • Author Contributions

    The study was performed under the supervision of B.-G.P. J.K. fabricated samples with the help of C.P., D.K., J.L., J.A., M.Y., and M.P. J.K. and S.L. conducted the electrical measurements with the help of J.L., S.N., H.L., J.K., D.H., and S.L. J.K. performed optical measurements with the help of S.K. and K.-J.K. K.L. and J.P. performed the machine learning attack. All authors discussed the result, and J.K., J.L., S.L., and B.-G.P. wrote the manuscript.

  • Notes
    The authors declare no competing financial interest.

Acknowledgments

Click to copy section linkSection link copied!

This work was supported by the National Research Foundation of Korea (NRF-2022M3I7A2079267 and RS-2023-00261042).

References

Click to copy section linkSection link copied!

This article references 40 other publications.

  1. 1
    Wang, Z.; Wu, H.; Burr, G. W.; Hwang, C. S.; Wang, K. L.; Xia, Q.; Yang, J. J. Resistive Switching Materials for Information Processing. Nat. Rev. Mater. 2020, 5, 173– 195,  DOI: 10.1038/s41578-019-0159-3
    Google Scholar
    1
    Resistive switching materials for information processing
    Wang, Zhongrui; Wu, Huaqiang; Burr, Geoffrey W.; Hwang, Cheol Seong; Wang, Kang L.; Xia, Qiangfei; Yang, J. Joshua
    Nature Reviews Materials (2020), 5 (3), 173-195CODEN: NRMADL; ISSN:2058-8437. (Nature Research)
    A review. The rapid increase in information in the big-data era calls for changes to information-processing paradigms, which, in turn, demand new circuit-building blocks to overcome the decreasing cost-effectiveness of transistor scaling and the intrinsic inefficiency of using transistors in non-von Neumann computing architectures. Accordingly, resistive switching materials (RSMs) based on different phys. principles have emerged for memories that could enable energy-efficient and area-efficient in-memory computing. In this Review, we survey the four phys. mechanisms that lead to such resistive switching: redox reactions, phase transitions, spin-polarized tunnelling and ferroelec. polarization. We discuss how these mechanisms equip RSMs with desirable properties for representation capability, switching speed and energy, reliability and device d. These properties are the key enablers of processing-in-memory platforms, with applications ranging from neuromorphic computing and general-purpose memcomputing to cybersecurity. Finally, we examine the device requirements for such systems based on RSMs and provide suggestions to address challenges in materials engineering, device optimization, system integration and algorithm design.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXotFKnug%253D%253D&md5=735b7fa0a0779ecb7d9d3186c3712466
  2. 2
    Chih, Y.-D.; Shih, Y.-C.; Lee, C.-F.; Chang, Y.-A.; Lee, P.-H.; Lin, H.-J.; Chen, Y.-L.; Lo, C.-P.; Shih, M.-C.; Shen, K.-H.; A 22nm 32Mb Embedded STT-MRAM with 10ns Read Speed, 1M Cycle Write Endurance, 10 years Retention at 150°C and High Immunity to Magnetic Field Interference. In 2020; IEEE International Solid-State Circuits Conference (ISSCC); IEEE, 2020; pp 222– 224.
    Google Scholar
    There is no corresponding record for this reference.
  3. 3
    Bhatti, S.; Sbiaa, R.; Hirohata, A.; Ohno, H.; Fukami, S.; Piramanayagam, S. N. Spintronics Based Random Access Memory: a Review. Mater. Today 2017, 20, 530– 548,  DOI: 10.1016/j.mattod.2017.07.007
    Google Scholar
    There is no corresponding record for this reference.
  4. 4
    Gallagher, W.-J.; Chien, E.; Chiang, T.-W.; Huang, J.-C.; Shih, M.-C.; Wang, C. Y.; Weng, C.-H.; Chen, S.; Bair, C.; Lee, G.; 22nm STT-MRAM for Reflow and Automotive Uses with High Yield, Reliability, and Magnetic Immunity and with Performance and Shielding Options. In 2019; IEEE International Electron Devices Meeting (IEDM); IEEE, 2019; pp 2.7.1– 2.7.4.
    Google Scholar
    There is no corresponding record for this reference.
  5. 5
    Choe, J. Recent Technology Insights on STT-MRAM: Structure, Materials, and Process Integration. In 2023; IEEE International Memory Workshop (IMW); IEEE, 2023; pp 1– 4.
    Google Scholar
    There is no corresponding record for this reference.
  6. 6
    Lee, K.; Chao, R.; Yamane, K.; Naik, V. B.; Yang, H.; Kwon, J.; Chung, N. L.; Jang, S. H.; Behin-Aein, B.; Lim, J. H.; 22-nm FD-SOI Embedded MRAM Technology for Low-Power Automotive-Grade-l MCU Applications. In 2018; IEEE International Electron Devices Meeting (IEDM); IEEE, 2018; pp 27.1.1– 27.1.4.
    Google Scholar
    There is no corresponding record for this reference.
  7. 7
    Yang, T.-H.; Li, K.-X.; Chiang, Y.-N.; Linn, W.-Y.; Lin, H.-T.; Chang, M.-F. A 28nm 32Kb Embedded 2T2MTJ STT-MRAM Macro with 1.3 ns Read-Access Time for Fast and Reliable Read Applications. 2018; IEEE International Solid-State Circuits Conference (ISSCC); IEEE, 2018; pp 482– 484.
    Google Scholar
    There is no corresponding record for this reference.
  8. 8
    Lee, K.; Bak, J. H.; Kim, Y. J.; Kim, C. K.; Antonyan, A.; Chang, D. H.; Hwang, S. H.; Lee, G. W.; Ji, N. Y.; Kim, W. J.; 1Gbit High Density Embedded STT-MRAM in 28nm FDSOI Technology. In 2019; IEEE International Electron Devices Meeting (IEDM); IEEE, 2019; pp 2.2.1– 2.2.4.
    Google Scholar
    There is no corresponding record for this reference.
  9. 9
    Ikegawa, S.; Mancoff, F. B.; Aggarwal, S. Commercialization of MRAM - Historical and Future Perspective. In 2021; IEEE International Interconnect Technology Conference (IITC); IEEE, 2021; pp 1– 3.
    Google Scholar
    There is no corresponding record for this reference.
  10. 10
    Song, Y.; Lee, J. H.; Han, S. H.; Shin, H. C.; Lee, K. H.; Suh, K.; Jeong, D. E.; Koh, G. H.; Oh, S. C.; Park, J. H.; Demonstration of Highly Manufacturable STT-MRAM Embedded in 28nm Logic. In 2018; IEEE International Electron Devices Meeting (IEDM); IEEE, 2018; pp 18.12.11– 18.12.14.
    Google Scholar
    There is no corresponding record for this reference.
  11. 11
    Ikegawa, S.; Mancoff, F. B.; Janesky, J.; Aggarwal, S. Magnetoresistive Random Access Memory: Present and Future. IEEE Trans. Electron Devices 2020, 67, 1407– 1419,  DOI: 10.1109/TED.2020.2965403
    Google Scholar
    11
    Magnetoresistive random access memory: present and future
    Ikegawa, Sumio; Mancoff, Frederick B.; Janesky, Jason; Aggarwal, Sanjeev
    IEEE Transactions on Electron Devices (2020), 67 (4), 1407-1419CODEN: IETDAI; ISSN:1557-9646. (Institute of Electrical and Electronics Engineers)
    A review. Magnetoresistive random access memory (MRAM) is regarded as a reliable persistent memory technol. because of its long data retention and robust endurance. Initial MRAM products utilized toggle mode writing of a balanced synthetic antiferromagnet (SAF) free layer to overcome problems with half-selected bits that challenged traditional Stoner-Wohlfarth switching. With the development of spin transfer torque (STT) switching in perpendicular magnetic tunnel junctions, the capability for scaling MRAM products increased markedly, enabling a 1-Gb device in 2019. Ongoing research will allow scaling to even higher capacities. Compared to traditional memories, STT-MRAM can save power, improve performance, and enhance system data integrity, which supports the growing computing demands for everything from data centers to Internet of Things (IoT) devices. This article provides a review of the technol. that enabled present toggle and STT-MRAM products, future STT-MRAM products, and new MRAM technologies beyond STT.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhvVCgsbjO&md5=519031eae15190ce7ff1b89e1631508e
  12. 12
    Ryu, J.; Lee, S.; Lee, K.-J.; Park, B.-G. Current-induced Spin-Orbit Torques for Spintronic Applications. Adv. Mater. 2020, 32, 1907148,  DOI: 10.1002/adma.201907148
    Google Scholar
    12
    Current-Induced Spin-Orbit Torques for Spintronic Applications
    Ryu, Jeongchun; Lee, Soogil; Lee, Kyung-Jin; Park, Byong-Guk
    Advanced Materials (Weinheim, Germany) (2020), 32 (35), 1907148CODEN: ADVMEW; ISSN:0935-9648. (Wiley-VCH Verlag GmbH & Co. KGaA)
    A review. Control of magnetization in magnetic nanostructures is essential for development of spintronic devices because it governs fundamental device characteristics such as energy consumption, areal d., and operation speed. In this respect, spin-orbit torque (SOT), which originates from the spin-orbit interaction, has been widely investigated due to its efficient manipulation of the magnetization using in-plane current. SOT spearheads novel spintronic applications including high-speed magnetic memories, reconfigurable logics, and neuromorphic computing. Herein, recent advances in SOT research, highlighting the considerable benefits and challenges of SOT-based spintronic devices, are reviewed. First, the materials and structural engineering that enhances SOT efficiency are discussed. Then major exptl. results for field-free SOT switching of perpendicular magnetization are summarized, which includes the introduction of an internal effective magnetic field and the generation of a distinct spin current with out-of-plane spin polarization. Finally, advanced SOT functionalities are presented, focusing on the demonstration of reconfigurable and complementary operation in spin logic devices.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXksVKnu78%253D&md5=d40da93efcbd2999eee15efeeab76fe2
  13. 13
    Baek, S.-h. C.; Park, K.-W.; Kil, D.-S.; Jang, Y.; Park, J.; Lee, K.-J.; Park, B.-G. Complementary Logic Operation Based on Electric-Field Controlled Spin-Orbit Torques. Nat. Electron. 2018, 1, 398– 403,  DOI: 10.1038/s41928-018-0099-8
    Google Scholar
    There is no corresponding record for this reference.
  14. 14
    Jung, S.; Lee, H.; Myung, S.; Kim, H.; Yoon, S. K.; Kwon, S.-W.; Ju, Y.; Kim, M.; Yi, W.; Han, S. A Crossbar Array of Magnetoresistive Memory Devices for In-Memory Computing. Nature 2022, 601, 211– 216,  DOI: 10.1038/s41586-021-04196-6
    Google Scholar
    14
    A crossbar array of magnetoresistive memory devices for in-memory computing
    Jung, Seungchul; Lee, Hyungwoo; Myung, Sungmeen; Kim, Hyunsoo; Yoon, Seung Keun; Kwon, Soon-Wan; Ju, Yongmin; Kim, Minje; Yi, Wooseok; Han, Shinhee; Kwon, Baeseong; Seo, Boyoung; Lee, Kilho; Koh, Gwan-Hyeob; Lee, Kangho; Song, Yoonjong; Choi, Changkyu; Ham, Donhee; Kim, Sang Joon
    Nature (London, United Kingdom) (2022), 601 (7892), 211-216CODEN: NATUAS; ISSN:1476-4687. (Nature Portfolio)
    Implementations of artificial neural networks that borrow analog techniques could potentially offer low-power alternatives to fully digital approaches1-3. One notable example is in-memory computing based on crossbar arrays of non-volatile memories4-7 that execute, in an analog manner, multiply-accumulate operations prevalent in artificial neural networks. Various non-volatile memories-including resistive memory8-13, phase-change memory14,15 and flash memory16-19-have been used for such approaches. However, it remains challenging to develop a crossbar array of spin-transfer-torque magnetoresistive random-access memory (MRAM)20-22, despite the technol.'s practical advantages such as endurance and large-scale commercialization5. The difficulty stems from the low resistance of MRAM, which would result in large power consumption in a conventional crossbar array that uses current summation for analog multiply-accumulate operations. Here we report a 64 x 64 crossbar array based on MRAM cells that overcomes the low-resistance issue with an architecture that uses resistance summation for analog multiply-accumulate operations. The array is integrated with readout electronics in 28-nm complementary metal-oxide-semiconductor technol. Using this array, a two-layer perceptron is implemented to classify 10,000 Modified National Institute of Stds. and Technol. digits with an accuracy of 93.23 per cent (software baseline: 95.24 per cent). In an emulation of a deeper, eight-layer Visual Geometry Group-8 neural network with measured errors, the classification accuracy improves to 98.86 per cent (software baseline: 99.28 per cent). We also use the array to implement a single layer in a ten-layer neural network to realize face detection with an accuracy of 93.4 per cent.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhtVOlu7s%253D&md5=f42e9499f047cd7e8a03de243e651ae6
  15. 15
    Borders, W. A.; Pervaiz, A. Z.; Fukami, S.; Camsari, K. Y.; Ohno, H.; Datta, S. Integer Factorization Using Stochastic Magnetic Tunnel Junctions. Nature 2019, 573, 390– 393,  DOI: 10.1038/s41586-019-1557-9
    Google Scholar
    15
    Integer factorization using stochastic magnetic tunnel junctions
    Borders, William A.; Pervaiz, Ahmed Z.; Fukami, Shunsuke; Camsari, Kerem Y.; Ohno, Hideo; Datta, Supriyo
    Nature (London, United Kingdom) (2019), 573 (7774), 390-393CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
    Conventional computers operate deterministically using strings of zeros and ones called bits to represent information in binary code. Despite the evolution of conventional computers into sophisticated machines, there are many classes of problems that they cannot efficiently address, including inference, invertible logic, sampling and optimization, leading to considerable interest in alternative computing schemes. Quantum computing, which uses qubits to represent a superposition of 0 and 1, is expected to perform these tasks efficiently1-3. However, decoherence and the current requirement for cryogenic operation4, as well as the limited many-body interactions that can be implemented, pose considerable challenges. Probabilistic computing1,5-7 is another unconventional computation scheme that shares similar concepts with quantum computing but is not limited by the above challenges. The key role is played by a probabilistic bit (a p-bit)-a robust, classical entity fluctuating in time between 0 and 1, which interacts with other p-bits in the same system using principles inspired by neural networks8. Here we present a proof-of-concept expt. for probabilistic computing using spintronics technol., and demonstrate integer factorization, an illustrative example of the optimization class of problems addressed by adiabatic9 and gated2 quantum computing. Nanoscale magnetic tunnel junctions showing stochastic behavior are developed by modifying market-ready magnetoresistive random-access memory technol.10,11 and are used to implement three-terminal p-bits that operate at room temp. The p-bits are elec. connected to form a functional asynchronous network, to which a modified adiabatic quantum computing algorithm that implements three- and four-body interactions is applied. Factorization of integers up to 945 is demonstrated with this rudimentary asynchronous probabilistic computer using eight correlated p-bits, and the results show good agreement with theor. predictions, thus providing a potentially scalable hardware approach to the difficult problems of optimization and sampling.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhvVWit7%252FE&md5=665cb85a1bf3ab1148cbe6495c94ccaf
  16. 16
    Herder, C.; Yu, M.-D.; Koushanfar, F.; Devadas, S. Physical Unclonable Functions and Applications: A Tutorial. Proc. IEEE 2014, 102, 1126– 1141,  DOI: 10.1109/JPROC.2014.2320516
    Google Scholar
    There is no corresponding record for this reference.
  17. 17
    Gao, Y.; Al-Sarawi, S. F.; Abbott, D. Physical Unclonable Functions. Nat. Electron. 2020, 3, 81– 91,  DOI: 10.1038/s41928-020-0372-5
    Google Scholar
    There is no corresponding record for this reference.
  18. 18
    McGrath, T.; Bagci, I. E.; Wang, Z. M.; Roedig, U.; Young, R. J. A PUF Taxonomy. Appl. Phys. Rev. 2019, 6, 011303,  DOI: 10.1063/1.5079407
    Google Scholar
    18
    A PUF taxonomy
    McGrath, Thomas; Bagci, Ibrahim E.; Wang, Zhiming M.; Roedig, Utz; Young, Robert J.
    Applied Physics Reviews (2019), 6 (1), 011303/1-011303/25CODEN: APRPG5; ISSN:1931-9401. (American Institute of Physics)
    Authentication is an essential cryptog. primitive that confirms the identity of parties during communications. For security, it is important that these identities are complex, in order to make them difficult to clone or guess. In recent years, phys. unclonable functions (PUFs) have emerged, in which identities are embodied in structures, rather than stored in memory elements. PUFs provide "digital fingerprints," where information is usually read from the static entropy of a system, rather than having an identity artificially programmed in, preventing a malicious party from making a copy for nefarious use later on. Many concepts for the phys. source of the uniqueness of these PUFs have been developed for multiple different applications. While certain types of PUF have received a great deal of attention, other promising suggestions may be overlooked. To remedy this, we present a review that seeks to exhaustively catalog and provide a complete organisational scheme towards the suggested concepts for PUFs. Furthermore, by carefully considering the phys. mechanisms underpinning the operation of different PUFs, we are able to form relationships between PUF technologies that previously had not been linked and look toward novel forms of PUF using phys. principles that have yet to be exploited. (c) 2019 American Institute of Physics.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXjtVGrurc%253D&md5=215d6a7491cb431d49094c24fcb9bad9
  19. 19
    Suh, G. E.; Devadas, S. Physical Unclonable Functions for Device Authentication and Secret Key Generation. 2007; ACM/IEEE Dessign Automation Conference (DAC); IEEE, 2007; pp 9– 14.
    Google Scholar
    There is no corresponding record for this reference.
  20. 20
    Guajardo, J.; Kumar, S. S.; Schrijen, G.-J.; Tuyls, P. FPGA Intrinsic PUFs and Their Use for IP Protection. In 2007; International Workshop on Cryptographic Hardware and Embedded Systems; Springer, 2007; pp 63– 80.
    Google Scholar
    There is no corresponding record for this reference.
  21. 21
    Lee, J. W.; Lim, D.; Gassend, B.; Suh, G. E.; Van Dijk, M.; Devadas, S. A Technique to Build a Secret Key in Integrated Circuits for Identification and Authentication Applications. In 2004; Symposium on VLSI Circuits. Digest of Technical Papers; IEEE, 2004; pp 176– 179.
    Google Scholar
    There is no corresponding record for this reference.
  22. 22
    Zhang, J.-L.; Qu, G.; Lv, Y.-Q.; Zhou, Q. A Survey on Silicon PUFs and Recent Advances in Ring Oscillator PUFs. J. Comput. Sci. Technol. 2014, 29, 664– 678,  DOI: 10.1007/s11390-014-1458-1
    Google Scholar
    There is no corresponding record for this reference.
  23. 23
    Merli, D.; Schuster, D.; Stumpf, F.; Sigl, G. Side-Channel Analysis of PUFs and Fuzzy Extractors. 2011; International Conference on Trust and Trustworthy Computing (TRUST); Springer, 2011, pp 33– 47.
    Google Scholar
    There is no corresponding record for this reference.
  24. 24
    Gao, B.; Lin, B.; Pang, Y.; Xu, F.; Lu, Y.; Chiu, Y.-C.; Liu, Z.; Tang, J.; Chang, M.-F.; Qian, H. Concealable Physically Unclonable Function Chip with a Memristor Array. Sci. Adv. 2022, 8, eabn7753  DOI: 10.1126/sciadv.abn7753
    Google Scholar
    There is no corresponding record for this reference.
  25. 25
    Dodda, A.; Subbulakshmi Radhakrishnan, S.; Schranghamer, T. F.; Buzzell, D.; Sengupta, P.; Das, S. Graphene-Based Physically Unclonable Functions That Are Reconfigurable and Resilient to Machine Learning Attacks. Nat. Electron. 2021, 4, 364– 374,  DOI: 10.1038/s41928-021-00569-x
    Google Scholar
    25
    Graphene-based physically unclonable functions that are reconfigurable and resilient to machine learning attacks
    Dodda, Akhil; Subbulakshmi Radhakrishnan, Shiva; Schranghamer, Thomas F.; Buzzell, Drew; Sengupta, Parijat; Das, Saptarshi
    Nature Electronics (2021), 4 (5), 364-374CODEN: NEALB3; ISSN:2520-1131. (Nature Portfolio)
    Abstr.: Graphene has a range of properties that makes it suitable for building devices for the Internet of Things. However, the deployment of such devices will also likely require the development of suitable graphene-based hardware security primitives. Here we report a phys. unclonable function (PUF) that exploits disorders in the carrier transport of graphene field-effect transistors. The Dirac voltage, Dirac conductance and carrier mobility values of a large population of graphene field-effect transistors follow Gaussian random distributions, which allow the devices to be used as a PUF. The resulting PUF is resilient to machine learning attacks based on predictive regression models and generative adversarial neural networks. The PUF is also reconfigurable without any phys. intervention and/or integration of addnl. hardware components due to the memristive properties of graphene. Furthermore, we show that the PUF can operate with ultralow power and is scalable, stable over time and reliable against variations in temp. and supply voltage.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhtVGqtrjJ&md5=ffb58cd03cc82a2b42ee745d857a70b3
  26. 26
    Kim, J. H.; Jeon, S.; In, J. H.; Nam, S.; Jin, H. M.; Han, K. H.; Yang, G. G.; Choi, H. J.; Kim, K. M.; Shin, J. Nanoscale Physical Unclonable Function Labels Based on Block Copolymer Self-Assembly. Nat. Electron. 2022, 5, 433– 442,  DOI: 10.1038/s41928-022-00788-w
    Google Scholar
    26
    Nanoscale physical unclonable function labels based on block co-polymer self-assembly
    Kim, Jang Hwan; Jeon, Suwan; In, Jae Hyun; Nam, Seonho; Jin, Hyeong Min; Han, Kyu Hyo; Yang, Geon Gug; Choi, Hee Jae; Kim, Kyung Min; Shin, Jonghwa; Son, Seung-Woo; Kwon, Seok Joon; Kim, Bong Hoon; Kim, Sang Ouk
    Nature Electronics (2022), 5 (7), 433-442CODEN: NEALB3; ISSN:2520-1131. (Nature Portfolio)
    Hardware-based cryptog. that exploits phys. unclonable functions is required for the secure identification and authentication of devices in the Internet of Things. However, phys. unclonable functions are typically based on anticounterfeit identifiers created from randomized microscale patterns or non-predictable fluctuations of elec. response in semiconductor devices, and the validation of an encrypted signature relies on a single-purpose method such as microscopy or elec. measurement. Here we report nanoscale phys. unclonable function labels that exploit non-deterministic mol. self-assembly. The labels are created from the multilayer superpositions of metallic nanopatterns replicated from self-assembled block co-polymer nanotemplates. Due to the nanoscale dimensions and diverse material options of the system, phys. unclonable functions are intrinsically difficult to replicate, robust for authentication and resistant to external disturbance. Multiple, independently operating keys-which use elec. resistance, optical dichroism or Raman signals-can be generated from a single phys. unclonable function, offering millisecond-level validation speeds. We also show that our phys. unclonable function labels can be used on a range of different surfaces including dollar bills, human hair and microscopic bacteria.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhvFGgt7bK&md5=e185c629460ee928fe6e0592cd9e3105
  27. 27
    John, R. A.; Shah, N.; Vishwanath, S. K.; Ng, S. E.; Febriansyah, B.; Jagadeeswararao, M.; Chang, C.-H.; Basu, A.; Mathews, N. Halide Perovskite Memristors as Flexible and Reconfigurable Physical Unclonable Functions. Nat. Commun. 2021, 12, 3681,  DOI: 10.1038/s41467-021-24057-0
    Google Scholar
    27
    Halide perovskite memristors as flexible and reconfigurable physical unclonable functions
    John, Rohit Abraham; Shah, Nimesh; Vishwanath, Sujaya Kumar; Ng, Si En; Febriansyah, Benny; Jagadeeswararao, Metikoti; Chang, Chip-Hong; Basu, Arindam; Mathews, Nripan
    Nature Communications (2021), 12 (1), 3681CODEN: NCAOBW; ISSN:2041-1723. (Nature Research)
    Phys. Unclonable Functions (PUFs) address the inherent limitations of conventional hardware security solns. in edge-computing devices. Despite impressive demonstrations with silicon circuits and crossbars of oxide memristors, realizing efficient roots of trust for resource-constrained hardware remains a significant challenge. Hybrid org. electronic materials with a rich reservoir of exotic switching physics offer an attractive, inexpensive alternative to design efficient cryptog. hardware, but have not been investigated till date. Here, we report a breakthrough security primitive exploiting the switching physics of one dimensional halide perovskite memristors as excellent sources of entropy for secure key generation and device authentication. Measurements of a prototypical 1 kb Pr pyridinium lead iodide (PrPyr[PbI3]) weak memristor PUF with a differential write-back strategy reveals near ideal uniformity, uniqueness and reliability without addnl. area and power overheads. Cycle-to-cycle write variability enables reconfigurability, while in-memory computing empowers a strong recurrent PUF construction to thwart machine learning attacks.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhsFKnsr%252FP&md5=bc01c08ac4b95f9e11066ee066d3e02c
  28. 28
    Im, H.; Yoon, J.; Choi, J.; Kim, J.; Baek, S.; Park, D. H.; Park, W.; Kim, S. Chaotic Organic Crystal Phosphorescent Patterns for Physical Unclonable Functions. Adv. Mater. 2021, 33, 2102542,  DOI: 10.1002/adma.202102542
    Google Scholar
    28
    Chaotic Organic Crystal Phosphorescent Patterns for Physical Unclonable Functions
    Im, Healin; Yoon, Jinsik; Choi, Jinho; Kim, Jinsang; Baek, Seungho; Park, Dong Hyuk; Park, Wook; Kim, Sunkook
    Advanced Materials (Weinheim, Germany) (2021), 33 (44), 2102542CODEN: ADVMEW; ISSN:0935-9648. (Wiley-VCH Verlag GmbH & Co. KGaA)
    Since the 4th Industrial Revolution, Internet of Things based environments have been widely used in various fields ranging from mobile to medical devices. Simultaneously, information leakage and hacking risks have also increased significantly, and secure authentication and security systems are constantly required. Phys. unclonable functions (PUF) are in the spotlight as an alternative. Chaotic phosphorescent patterns are developed based on an org. crystal and at. seed heterostructure for security labels with PUFs. Phosphorescent org. crystal patterns are formed on MoS2. They seem similar on a macroscopic scale, whereas each org. crystal exhibits highly disorder features on the microscopic scale. In image anal., an encoding capacity as a single PUF domain achieves more than 1017 on a MoS2 small fragment with lengths of 25μm. Therefore, security labels with phosphorescent PUFs can offer superior randomness and no-cloning codes, possibly becoming a promising security strategy for authentication processes.
    >> More from SciFinder ®
    https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXitVGnsr7N&md5=388cf716785a18d110eec401b8b810e8
  29. 29
    Koh, D.; Kang, J.; Kim, T.; Lee, J.; Noh, S.; Lee, H.; Kwon, J.; Lee, S.; Park, J.; Park, B.-G. Improved Robustness Against Magnetic Field in Spin-Orbit-Torque-Based Physical Unclonable Functions Through Write-back Operation. Adv. Electron. Mater. 2023, 9, 2201073,  DOI: 10.1002/aelm.202201073
    Google Scholar
    There is no corresponding record for this reference.
  30. 30
    Lee, S.; Kang, J.; Kim, J. M.; Kim, N.; Han, D.; Lee, T.; Ko, S.; Yang, J.; Lee, S.; Lee, S. Spintronic Physical Unclonable Functions Based on Field-free Spin-Orbit-Torque Switching. Adv. Mater. 2022, 34, 2203558,  DOI: 10.1002/adma.202203558
    Google Scholar
    There is no corresponding record for this reference.
  31. 31
    Shao, Y.; Davila, N.; Ebrahimi, F.; Katine, J. A.; Finocchio, G.; Khalili Amiri, P. Reconfigurable Physically Unclonable Functions Based on Nanoscale Voltage-controlled Magnetic Tunnel Junctions. Adv. Electron. Mater. 2023, 9, 2300195,  DOI: 10.1002/aelm.202300195
    Google Scholar
    There is no corresponding record for this reference.
  32. 32
    Meo, A.; Garzón, E.; De Rose, R.; Finocchio, G.; Lanuzza, M.; Carpentieri, M. Voltage-Controlled Magnetic Anisotropy Based Physical Unclonable Function. Appl. Phys. Lett. 2023, 123, 062405,  DOI: 10.1063/5.0166164
    Google Scholar
    There is no corresponding record for this reference.
  33. 33
    Chiu, Y.-C.; Khwa, W.-S.; Yang, C.-S.; Teng, S.-H.; Huang, H.-Y.; Chang, F.-C.; Wu, Y.; Chien, Y.-A.; Hsieh, F.-L.; Li, C.-Y. A CMOS-Integrated Spintronic Compute-In-Memory Macro for Secure AI Edge Devices. Nat. Electron. 2023, 6, 534– 543,  DOI: 10.1038/s41928-023-00994-0
    Google Scholar
    There is no corresponding record for this reference.
  34. 34
    Lee, J. K.; Lee, J.; Yoon, S. I.; Lee, M. H.; Lee, J. S.; Jang, Y.; Kim, D. Y.; Choe, S. B.; Park, J.; Kim, Y. K. Optical Verification of Physically Unclonable Function Devices Based on Spin-orbit torque Switching. Adv. Electron. Mater. 2023, 9, 2300056,  DOI: 10.1002/aelm.202300056
    Google Scholar
    There is no corresponding record for this reference.
  35. 35
    Finocchio, G.; Moriyama, T.; De Rose, R.; Siracusano, G.; Lanuzza, M.; Puliafito, V.; Chiappini, S.; Crupi, F.; Zeng, Z.; Ono, T. Spin-Orbit Torque Based Physical Unclonable Function. J. Appl. Phys. 2020, 128, 033904,  DOI: 10.1063/5.0013408
    Google Scholar
    There is no corresponding record for this reference.
  36. 36
    Zhang, L.; Fong, X.; Chang, C.-H.; Kong, Z. H.; Roy, K. Highly Reliable Memory-Based Physical Unclonable Function Using Spin-Transfer Torque MRAM. 2014. In IEEE International Symposium on Circuits and Systems (ISCAS); IEEE, 2014; pp 2169– 2172.
    Google Scholar
    There is no corresponding record for this reference.
  37. 37
    Failure Mechanism Based Stress Test Qualification for Integrated Circuits , AEC-Q100-Rev. H, Sep. 2014.
    Google Scholar
    There is no corresponding record for this reference.
  38. 38
    Wilde, F.; Gammel, B. M.; Pehl, M. Spatial Correlation Analysis on Physical Unclonable Functions. IEEE Trans. Inf. Forensics Secur. 2018, 13, 1468– 1480,  DOI: 10.1109/TIFS.2018.2791341
    Google Scholar
    There is no corresponding record for this reference.
  39. 39
    Rukhi, A.; Soto, J.; Nechvatal, J.; Smid, M.; Barker, E.; Leigh, S.; Levenson, M.; Vangel, M.; Banks, D.; Heckert, A.; Dray, J.; Vo, S. A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications. In NIST Special Publication 800–22:2020; National Institute of Standards and Technology (NIST): Gaithersburg, MD, USA, 2010; .
    Google Scholar
    There is no corresponding record for this reference.
  40. 40
    Hitaj, B.; Gasti, P.; Ateniese, G.; Perez-Cruz, F. PassGAN: A Deep Learning Approach for Password Guessing. 2019. In International Conference on Applied Cryptography and Network Security (ACNS); Springer, 2019; pp 217– 237.
    Google Scholar
    There is no corresponding record for this reference.

Cited By

Click to copy section linkSection link copied!

This article has not yet been cited by other publications.

Download PDF
Open PDF
Go to ACS Nano
Get e-Alerts
Get e-Alerts

ACS Nano

Cite this: ACS Nano 2024, 18, 20, 12853–12860
Click to copy citationCitation copied!
https://doi.org/10.1021/acsnano.4c00078
Published May 8, 2024

Copyright © 2024 The Authors. Published by American Chemical Society. This publication is licensed under

CC-BY-NC-ND 4.0 .

Article Views

1681

Altmetric

-

Citations

-
Learn about these metrics

Article Views are the COUNTER-compliant sum of full text article downloads since November 2008 (both PDF and HTML) across all institutions and individuals. These metrics are regularly updated to reflect usage leading up to the last few days.

Citations are the number of other articles citing this article, calculated by Crossref and updated daily. Find more information about Crossref citation counts.

The Altmetric Attention Score is a quantitative measure of the attention that a research article has received online. Clicking on the donut icon will load a page at altmetric.com with additional details about the score and the social media presence for the given article. Find more information on the Altmetric Attention Score and how the score is calculated.

  • Abstract

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 1

    Figure 1. Electrical and magnetic properties of the STT-MTJs. (a) Optical images of the STT MRAM (STT-MTJ) array (left and middle) and the device structure (right). The diameter of the circular pillar is 70 nm. STT switching characteristics: (b) current (I)–voltage (V) curve and (c) resistance (R)–I curve. (d) MR curves measured as a function of the out-of-plane magnetic field (Bz). Green (purple) symbols represent MR data obtained by sweeping the magnetic fields from +Bz (−Bz) to –Bz (+Bz). Red, blue, and gray arrows indicate the magnetization direction of the FL, RL, and PL, respectively. (e) A minor MR curve measured by sweeping Bz between ±0.27 T. Here, the magnetization direction of RL is fixed along the +z-direction.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 2

    Figure 2. Construction and characteristics of MRAM-PUFs. (a) MR curves of 64 MTJs after the randomization process. Red (blue) represents the clockwise (counterclockwise) switching polarity. Distributions of the switching polarity in 64 bit MRAM-PUFs (b) 1, (c) 2, and (d) 3. (e) Entropy of 31 MRAM-PUFs. PMF as a function of the (f) normalized inter-HD and (g) CC. The red curves indicate Gaussian fittings.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 3

    Figure 3. Temperature dependence of the MRAM-PUF. (a) Representative MR curves measured at different temperatures. (b) Temperature dependence of the TMR ratio. The error bar indicates the standard deviation. (c) MRAM-PUF consisting of 64 PUF unit devices at different temperatures of −40, 25, and 150 °C.

    High Resolution Image
    Download MS PowerPoint Slide

    Figure 4

    Figure 4. Resilience against machine learning attacks. (a) Representative MOKE images of 5 μm × 5 μm square-patterned devices after the randomization process. (b) Hysteresis loops of single square samples. The red (blue) curve is obtained from the dark (light) gray area, indicating that exchange bias forms along the positive (negative) z-direction. (c) Schematic illustration of the GAN. The GAN consists of a generator DNN and a discriminator DNN. The generator DNN is trained to generate fake CRPs, while the discriminator DNN is trained to distinguish the fake CRPs from the real ones. Here, we use 1200 CRPs for training, with the remaining 300 CRPs used for a comparison with CRPs generated by the GAN model. PMF as a function of the (d) inter-HD and (e) CC between the real CRPs and the fake CRPs produced by the GAN model.

    High Resolution Image
    Download MS PowerPoint Slide

  • References

    This article references 40 other publications.

    1. 1
      Wang, Z.; Wu, H.; Burr, G. W.; Hwang, C. S.; Wang, K. L.; Xia, Q.; Yang, J. J. Resistive Switching Materials for Information Processing. Nat. Rev. Mater. 2020, 5, 173– 195,  DOI: 10.1038/s41578-019-0159-3
      1
      Resistive switching materials for information processing
      Wang, Zhongrui; Wu, Huaqiang; Burr, Geoffrey W.; Hwang, Cheol Seong; Wang, Kang L.; Xia, Qiangfei; Yang, J. Joshua
      Nature Reviews Materials (2020), 5 (3), 173-195CODEN: NRMADL; ISSN:2058-8437. (Nature Research)
      A review. The rapid increase in information in the big-data era calls for changes to information-processing paradigms, which, in turn, demand new circuit-building blocks to overcome the decreasing cost-effectiveness of transistor scaling and the intrinsic inefficiency of using transistors in non-von Neumann computing architectures. Accordingly, resistive switching materials (RSMs) based on different phys. principles have emerged for memories that could enable energy-efficient and area-efficient in-memory computing. In this Review, we survey the four phys. mechanisms that lead to such resistive switching: redox reactions, phase transitions, spin-polarized tunnelling and ferroelec. polarization. We discuss how these mechanisms equip RSMs with desirable properties for representation capability, switching speed and energy, reliability and device d. These properties are the key enablers of processing-in-memory platforms, with applications ranging from neuromorphic computing and general-purpose memcomputing to cybersecurity. Finally, we examine the device requirements for such systems based on RSMs and provide suggestions to address challenges in materials engineering, device optimization, system integration and algorithm design.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXotFKnug%253D%253D&md5=735b7fa0a0779ecb7d9d3186c3712466
    2. 2
      Chih, Y.-D.; Shih, Y.-C.; Lee, C.-F.; Chang, Y.-A.; Lee, P.-H.; Lin, H.-J.; Chen, Y.-L.; Lo, C.-P.; Shih, M.-C.; Shen, K.-H.; A 22nm 32Mb Embedded STT-MRAM with 10ns Read Speed, 1M Cycle Write Endurance, 10 years Retention at 150°C and High Immunity to Magnetic Field Interference. In 2020; IEEE International Solid-State Circuits Conference (ISSCC); IEEE, 2020; pp 222– 224.
      There is no corresponding record for this reference.
    3. 3
      Bhatti, S.; Sbiaa, R.; Hirohata, A.; Ohno, H.; Fukami, S.; Piramanayagam, S. N. Spintronics Based Random Access Memory: a Review. Mater. Today 2017, 20, 530– 548,  DOI: 10.1016/j.mattod.2017.07.007
      There is no corresponding record for this reference.
    4. 4
      Gallagher, W.-J.; Chien, E.; Chiang, T.-W.; Huang, J.-C.; Shih, M.-C.; Wang, C. Y.; Weng, C.-H.; Chen, S.; Bair, C.; Lee, G.; 22nm STT-MRAM for Reflow and Automotive Uses with High Yield, Reliability, and Magnetic Immunity and with Performance and Shielding Options. In 2019; IEEE International Electron Devices Meeting (IEDM); IEEE, 2019; pp 2.7.1– 2.7.4.
      There is no corresponding record for this reference.
    5. 5
      Choe, J. Recent Technology Insights on STT-MRAM: Structure, Materials, and Process Integration. In 2023; IEEE International Memory Workshop (IMW); IEEE, 2023; pp 1– 4.
      There is no corresponding record for this reference.
    6. 6
      Lee, K.; Chao, R.; Yamane, K.; Naik, V. B.; Yang, H.; Kwon, J.; Chung, N. L.; Jang, S. H.; Behin-Aein, B.; Lim, J. H.; 22-nm FD-SOI Embedded MRAM Technology for Low-Power Automotive-Grade-l MCU Applications. In 2018; IEEE International Electron Devices Meeting (IEDM); IEEE, 2018; pp 27.1.1– 27.1.4.
      There is no corresponding record for this reference.
    7. 7
      Yang, T.-H.; Li, K.-X.; Chiang, Y.-N.; Linn, W.-Y.; Lin, H.-T.; Chang, M.-F. A 28nm 32Kb Embedded 2T2MTJ STT-MRAM Macro with 1.3 ns Read-Access Time for Fast and Reliable Read Applications. 2018; IEEE International Solid-State Circuits Conference (ISSCC); IEEE, 2018; pp 482– 484.
      There is no corresponding record for this reference.
    8. 8
      Lee, K.; Bak, J. H.; Kim, Y. J.; Kim, C. K.; Antonyan, A.; Chang, D. H.; Hwang, S. H.; Lee, G. W.; Ji, N. Y.; Kim, W. J.; 1Gbit High Density Embedded STT-MRAM in 28nm FDSOI Technology. In 2019; IEEE International Electron Devices Meeting (IEDM); IEEE, 2019; pp 2.2.1– 2.2.4.
      There is no corresponding record for this reference.
    9. 9
      Ikegawa, S.; Mancoff, F. B.; Aggarwal, S. Commercialization of MRAM - Historical and Future Perspective. In 2021; IEEE International Interconnect Technology Conference (IITC); IEEE, 2021; pp 1– 3.
      There is no corresponding record for this reference.
    10. 10
      Song, Y.; Lee, J. H.; Han, S. H.; Shin, H. C.; Lee, K. H.; Suh, K.; Jeong, D. E.; Koh, G. H.; Oh, S. C.; Park, J. H.; Demonstration of Highly Manufacturable STT-MRAM Embedded in 28nm Logic. In 2018; IEEE International Electron Devices Meeting (IEDM); IEEE, 2018; pp 18.12.11– 18.12.14.
      There is no corresponding record for this reference.
    11. 11
      Ikegawa, S.; Mancoff, F. B.; Janesky, J.; Aggarwal, S. Magnetoresistive Random Access Memory: Present and Future. IEEE Trans. Electron Devices 2020, 67, 1407– 1419,  DOI: 10.1109/TED.2020.2965403
      11
      Magnetoresistive random access memory: present and future
      Ikegawa, Sumio; Mancoff, Frederick B.; Janesky, Jason; Aggarwal, Sanjeev
      IEEE Transactions on Electron Devices (2020), 67 (4), 1407-1419CODEN: IETDAI; ISSN:1557-9646. (Institute of Electrical and Electronics Engineers)
      A review. Magnetoresistive random access memory (MRAM) is regarded as a reliable persistent memory technol. because of its long data retention and robust endurance. Initial MRAM products utilized toggle mode writing of a balanced synthetic antiferromagnet (SAF) free layer to overcome problems with half-selected bits that challenged traditional Stoner-Wohlfarth switching. With the development of spin transfer torque (STT) switching in perpendicular magnetic tunnel junctions, the capability for scaling MRAM products increased markedly, enabling a 1-Gb device in 2019. Ongoing research will allow scaling to even higher capacities. Compared to traditional memories, STT-MRAM can save power, improve performance, and enhance system data integrity, which supports the growing computing demands for everything from data centers to Internet of Things (IoT) devices. This article provides a review of the technol. that enabled present toggle and STT-MRAM products, future STT-MRAM products, and new MRAM technologies beyond STT.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXhvVCgsbjO&md5=519031eae15190ce7ff1b89e1631508e
    12. 12
      Ryu, J.; Lee, S.; Lee, K.-J.; Park, B.-G. Current-induced Spin-Orbit Torques for Spintronic Applications. Adv. Mater. 2020, 32, 1907148,  DOI: 10.1002/adma.201907148
      12
      Current-Induced Spin-Orbit Torques for Spintronic Applications
      Ryu, Jeongchun; Lee, Soogil; Lee, Kyung-Jin; Park, Byong-Guk
      Advanced Materials (Weinheim, Germany) (2020), 32 (35), 1907148CODEN: ADVMEW; ISSN:0935-9648. (Wiley-VCH Verlag GmbH & Co. KGaA)
      A review. Control of magnetization in magnetic nanostructures is essential for development of spintronic devices because it governs fundamental device characteristics such as energy consumption, areal d., and operation speed. In this respect, spin-orbit torque (SOT), which originates from the spin-orbit interaction, has been widely investigated due to its efficient manipulation of the magnetization using in-plane current. SOT spearheads novel spintronic applications including high-speed magnetic memories, reconfigurable logics, and neuromorphic computing. Herein, recent advances in SOT research, highlighting the considerable benefits and challenges of SOT-based spintronic devices, are reviewed. First, the materials and structural engineering that enhances SOT efficiency are discussed. Then major exptl. results for field-free SOT switching of perpendicular magnetization are summarized, which includes the introduction of an internal effective magnetic field and the generation of a distinct spin current with out-of-plane spin polarization. Finally, advanced SOT functionalities are presented, focusing on the demonstration of reconfigurable and complementary operation in spin logic devices.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3cXksVKnu78%253D&md5=d40da93efcbd2999eee15efeeab76fe2
    13. 13
      Baek, S.-h. C.; Park, K.-W.; Kil, D.-S.; Jang, Y.; Park, J.; Lee, K.-J.; Park, B.-G. Complementary Logic Operation Based on Electric-Field Controlled Spin-Orbit Torques. Nat. Electron. 2018, 1, 398– 403,  DOI: 10.1038/s41928-018-0099-8
      There is no corresponding record for this reference.
    14. 14
      Jung, S.; Lee, H.; Myung, S.; Kim, H.; Yoon, S. K.; Kwon, S.-W.; Ju, Y.; Kim, M.; Yi, W.; Han, S. A Crossbar Array of Magnetoresistive Memory Devices for In-Memory Computing. Nature 2022, 601, 211– 216,  DOI: 10.1038/s41586-021-04196-6
      14
      A crossbar array of magnetoresistive memory devices for in-memory computing
      Jung, Seungchul; Lee, Hyungwoo; Myung, Sungmeen; Kim, Hyunsoo; Yoon, Seung Keun; Kwon, Soon-Wan; Ju, Yongmin; Kim, Minje; Yi, Wooseok; Han, Shinhee; Kwon, Baeseong; Seo, Boyoung; Lee, Kilho; Koh, Gwan-Hyeob; Lee, Kangho; Song, Yoonjong; Choi, Changkyu; Ham, Donhee; Kim, Sang Joon
      Nature (London, United Kingdom) (2022), 601 (7892), 211-216CODEN: NATUAS; ISSN:1476-4687. (Nature Portfolio)
      Implementations of artificial neural networks that borrow analog techniques could potentially offer low-power alternatives to fully digital approaches1-3. One notable example is in-memory computing based on crossbar arrays of non-volatile memories4-7 that execute, in an analog manner, multiply-accumulate operations prevalent in artificial neural networks. Various non-volatile memories-including resistive memory8-13, phase-change memory14,15 and flash memory16-19-have been used for such approaches. However, it remains challenging to develop a crossbar array of spin-transfer-torque magnetoresistive random-access memory (MRAM)20-22, despite the technol.'s practical advantages such as endurance and large-scale commercialization5. The difficulty stems from the low resistance of MRAM, which would result in large power consumption in a conventional crossbar array that uses current summation for analog multiply-accumulate operations. Here we report a 64 x 64 crossbar array based on MRAM cells that overcomes the low-resistance issue with an architecture that uses resistance summation for analog multiply-accumulate operations. The array is integrated with readout electronics in 28-nm complementary metal-oxide-semiconductor technol. Using this array, a two-layer perceptron is implemented to classify 10,000 Modified National Institute of Stds. and Technol. digits with an accuracy of 93.23 per cent (software baseline: 95.24 per cent). In an emulation of a deeper, eight-layer Visual Geometry Group-8 neural network with measured errors, the classification accuracy improves to 98.86 per cent (software baseline: 99.28 per cent). We also use the array to implement a single layer in a ten-layer neural network to realize face detection with an accuracy of 93.4 per cent.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhtVOlu7s%253D&md5=f42e9499f047cd7e8a03de243e651ae6
    15. 15
      Borders, W. A.; Pervaiz, A. Z.; Fukami, S.; Camsari, K. Y.; Ohno, H.; Datta, S. Integer Factorization Using Stochastic Magnetic Tunnel Junctions. Nature 2019, 573, 390– 393,  DOI: 10.1038/s41586-019-1557-9
      15
      Integer factorization using stochastic magnetic tunnel junctions
      Borders, William A.; Pervaiz, Ahmed Z.; Fukami, Shunsuke; Camsari, Kerem Y.; Ohno, Hideo; Datta, Supriyo
      Nature (London, United Kingdom) (2019), 573 (7774), 390-393CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
      Conventional computers operate deterministically using strings of zeros and ones called bits to represent information in binary code. Despite the evolution of conventional computers into sophisticated machines, there are many classes of problems that they cannot efficiently address, including inference, invertible logic, sampling and optimization, leading to considerable interest in alternative computing schemes. Quantum computing, which uses qubits to represent a superposition of 0 and 1, is expected to perform these tasks efficiently1-3. However, decoherence and the current requirement for cryogenic operation4, as well as the limited many-body interactions that can be implemented, pose considerable challenges. Probabilistic computing1,5-7 is another unconventional computation scheme that shares similar concepts with quantum computing but is not limited by the above challenges. The key role is played by a probabilistic bit (a p-bit)-a robust, classical entity fluctuating in time between 0 and 1, which interacts with other p-bits in the same system using principles inspired by neural networks8. Here we present a proof-of-concept expt. for probabilistic computing using spintronics technol., and demonstrate integer factorization, an illustrative example of the optimization class of problems addressed by adiabatic9 and gated2 quantum computing. Nanoscale magnetic tunnel junctions showing stochastic behavior are developed by modifying market-ready magnetoresistive random-access memory technol.10,11 and are used to implement three-terminal p-bits that operate at room temp. The p-bits are elec. connected to form a functional asynchronous network, to which a modified adiabatic quantum computing algorithm that implements three- and four-body interactions is applied. Factorization of integers up to 945 is demonstrated with this rudimentary asynchronous probabilistic computer using eight correlated p-bits, and the results show good agreement with theor. predictions, thus providing a potentially scalable hardware approach to the difficult problems of optimization and sampling.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXhvVWit7%252FE&md5=665cb85a1bf3ab1148cbe6495c94ccaf
    16. 16
      Herder, C.; Yu, M.-D.; Koushanfar, F.; Devadas, S. Physical Unclonable Functions and Applications: A Tutorial. Proc. IEEE 2014, 102, 1126– 1141,  DOI: 10.1109/JPROC.2014.2320516
      There is no corresponding record for this reference.
    17. 17
      Gao, Y.; Al-Sarawi, S. F.; Abbott, D. Physical Unclonable Functions. Nat. Electron. 2020, 3, 81– 91,  DOI: 10.1038/s41928-020-0372-5
      There is no corresponding record for this reference.
    18. 18
      McGrath, T.; Bagci, I. E.; Wang, Z. M.; Roedig, U.; Young, R. J. A PUF Taxonomy. Appl. Phys. Rev. 2019, 6, 011303,  DOI: 10.1063/1.5079407
      18
      A PUF taxonomy
      McGrath, Thomas; Bagci, Ibrahim E.; Wang, Zhiming M.; Roedig, Utz; Young, Robert J.
      Applied Physics Reviews (2019), 6 (1), 011303/1-011303/25CODEN: APRPG5; ISSN:1931-9401. (American Institute of Physics)
      Authentication is an essential cryptog. primitive that confirms the identity of parties during communications. For security, it is important that these identities are complex, in order to make them difficult to clone or guess. In recent years, phys. unclonable functions (PUFs) have emerged, in which identities are embodied in structures, rather than stored in memory elements. PUFs provide "digital fingerprints," where information is usually read from the static entropy of a system, rather than having an identity artificially programmed in, preventing a malicious party from making a copy for nefarious use later on. Many concepts for the phys. source of the uniqueness of these PUFs have been developed for multiple different applications. While certain types of PUF have received a great deal of attention, other promising suggestions may be overlooked. To remedy this, we present a review that seeks to exhaustively catalog and provide a complete organisational scheme towards the suggested concepts for PUFs. Furthermore, by carefully considering the phys. mechanisms underpinning the operation of different PUFs, we are able to form relationships between PUF technologies that previously had not been linked and look toward novel forms of PUF using phys. principles that have yet to be exploited. (c) 2019 American Institute of Physics.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BC1MXjtVGrurc%253D&md5=215d6a7491cb431d49094c24fcb9bad9
    19. 19
      Suh, G. E.; Devadas, S. Physical Unclonable Functions for Device Authentication and Secret Key Generation. 2007; ACM/IEEE Dessign Automation Conference (DAC); IEEE, 2007; pp 9– 14.
      There is no corresponding record for this reference.
    20. 20
      Guajardo, J.; Kumar, S. S.; Schrijen, G.-J.; Tuyls, P. FPGA Intrinsic PUFs and Their Use for IP Protection. In 2007; International Workshop on Cryptographic Hardware and Embedded Systems; Springer, 2007; pp 63– 80.
      There is no corresponding record for this reference.
    21. 21
      Lee, J. W.; Lim, D.; Gassend, B.; Suh, G. E.; Van Dijk, M.; Devadas, S. A Technique to Build a Secret Key in Integrated Circuits for Identification and Authentication Applications. In 2004; Symposium on VLSI Circuits. Digest of Technical Papers; IEEE, 2004; pp 176– 179.
      There is no corresponding record for this reference.
    22. 22
      Zhang, J.-L.; Qu, G.; Lv, Y.-Q.; Zhou, Q. A Survey on Silicon PUFs and Recent Advances in Ring Oscillator PUFs. J. Comput. Sci. Technol. 2014, 29, 664– 678,  DOI: 10.1007/s11390-014-1458-1
      There is no corresponding record for this reference.
    23. 23
      Merli, D.; Schuster, D.; Stumpf, F.; Sigl, G. Side-Channel Analysis of PUFs and Fuzzy Extractors. 2011; International Conference on Trust and Trustworthy Computing (TRUST); Springer, 2011, pp 33– 47.
      There is no corresponding record for this reference.
    24. 24
      Gao, B.; Lin, B.; Pang, Y.; Xu, F.; Lu, Y.; Chiu, Y.-C.; Liu, Z.; Tang, J.; Chang, M.-F.; Qian, H. Concealable Physically Unclonable Function Chip with a Memristor Array. Sci. Adv. 2022, 8, eabn7753  DOI: 10.1126/sciadv.abn7753
      There is no corresponding record for this reference.
    25. 25
      Dodda, A.; Subbulakshmi Radhakrishnan, S.; Schranghamer, T. F.; Buzzell, D.; Sengupta, P.; Das, S. Graphene-Based Physically Unclonable Functions That Are Reconfigurable and Resilient to Machine Learning Attacks. Nat. Electron. 2021, 4, 364– 374,  DOI: 10.1038/s41928-021-00569-x
      25
      Graphene-based physically unclonable functions that are reconfigurable and resilient to machine learning attacks
      Dodda, Akhil; Subbulakshmi Radhakrishnan, Shiva; Schranghamer, Thomas F.; Buzzell, Drew; Sengupta, Parijat; Das, Saptarshi
      Nature Electronics (2021), 4 (5), 364-374CODEN: NEALB3; ISSN:2520-1131. (Nature Portfolio)
      Abstr.: Graphene has a range of properties that makes it suitable for building devices for the Internet of Things. However, the deployment of such devices will also likely require the development of suitable graphene-based hardware security primitives. Here we report a phys. unclonable function (PUF) that exploits disorders in the carrier transport of graphene field-effect transistors. The Dirac voltage, Dirac conductance and carrier mobility values of a large population of graphene field-effect transistors follow Gaussian random distributions, which allow the devices to be used as a PUF. The resulting PUF is resilient to machine learning attacks based on predictive regression models and generative adversarial neural networks. The PUF is also reconfigurable without any phys. intervention and/or integration of addnl. hardware components due to the memristive properties of graphene. Furthermore, we show that the PUF can operate with ultralow power and is scalable, stable over time and reliable against variations in temp. and supply voltage.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhtVGqtrjJ&md5=ffb58cd03cc82a2b42ee745d857a70b3
    26. 26
      Kim, J. H.; Jeon, S.; In, J. H.; Nam, S.; Jin, H. M.; Han, K. H.; Yang, G. G.; Choi, H. J.; Kim, K. M.; Shin, J. Nanoscale Physical Unclonable Function Labels Based on Block Copolymer Self-Assembly. Nat. Electron. 2022, 5, 433– 442,  DOI: 10.1038/s41928-022-00788-w
      26
      Nanoscale physical unclonable function labels based on block co-polymer self-assembly
      Kim, Jang Hwan; Jeon, Suwan; In, Jae Hyun; Nam, Seonho; Jin, Hyeong Min; Han, Kyu Hyo; Yang, Geon Gug; Choi, Hee Jae; Kim, Kyung Min; Shin, Jonghwa; Son, Seung-Woo; Kwon, Seok Joon; Kim, Bong Hoon; Kim, Sang Ouk
      Nature Electronics (2022), 5 (7), 433-442CODEN: NEALB3; ISSN:2520-1131. (Nature Portfolio)
      Hardware-based cryptog. that exploits phys. unclonable functions is required for the secure identification and authentication of devices in the Internet of Things. However, phys. unclonable functions are typically based on anticounterfeit identifiers created from randomized microscale patterns or non-predictable fluctuations of elec. response in semiconductor devices, and the validation of an encrypted signature relies on a single-purpose method such as microscopy or elec. measurement. Here we report nanoscale phys. unclonable function labels that exploit non-deterministic mol. self-assembly. The labels are created from the multilayer superpositions of metallic nanopatterns replicated from self-assembled block co-polymer nanotemplates. Due to the nanoscale dimensions and diverse material options of the system, phys. unclonable functions are intrinsically difficult to replicate, robust for authentication and resistant to external disturbance. Multiple, independently operating keys-which use elec. resistance, optical dichroism or Raman signals-can be generated from a single phys. unclonable function, offering millisecond-level validation speeds. We also show that our phys. unclonable function labels can be used on a range of different surfaces including dollar bills, human hair and microscopic bacteria.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB38XhvFGgt7bK&md5=e185c629460ee928fe6e0592cd9e3105
    27. 27
      John, R. A.; Shah, N.; Vishwanath, S. K.; Ng, S. E.; Febriansyah, B.; Jagadeeswararao, M.; Chang, C.-H.; Basu, A.; Mathews, N. Halide Perovskite Memristors as Flexible and Reconfigurable Physical Unclonable Functions. Nat. Commun. 2021, 12, 3681,  DOI: 10.1038/s41467-021-24057-0
      27
      Halide perovskite memristors as flexible and reconfigurable physical unclonable functions
      John, Rohit Abraham; Shah, Nimesh; Vishwanath, Sujaya Kumar; Ng, Si En; Febriansyah, Benny; Jagadeeswararao, Metikoti; Chang, Chip-Hong; Basu, Arindam; Mathews, Nripan
      Nature Communications (2021), 12 (1), 3681CODEN: NCAOBW; ISSN:2041-1723. (Nature Research)
      Phys. Unclonable Functions (PUFs) address the inherent limitations of conventional hardware security solns. in edge-computing devices. Despite impressive demonstrations with silicon circuits and crossbars of oxide memristors, realizing efficient roots of trust for resource-constrained hardware remains a significant challenge. Hybrid org. electronic materials with a rich reservoir of exotic switching physics offer an attractive, inexpensive alternative to design efficient cryptog. hardware, but have not been investigated till date. Here, we report a breakthrough security primitive exploiting the switching physics of one dimensional halide perovskite memristors as excellent sources of entropy for secure key generation and device authentication. Measurements of a prototypical 1 kb Pr pyridinium lead iodide (PrPyr[PbI3]) weak memristor PUF with a differential write-back strategy reveals near ideal uniformity, uniqueness and reliability without addnl. area and power overheads. Cycle-to-cycle write variability enables reconfigurability, while in-memory computing empowers a strong recurrent PUF construction to thwart machine learning attacks.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXhsFKnsr%252FP&md5=bc01c08ac4b95f9e11066ee066d3e02c
    28. 28
      Im, H.; Yoon, J.; Choi, J.; Kim, J.; Baek, S.; Park, D. H.; Park, W.; Kim, S. Chaotic Organic Crystal Phosphorescent Patterns for Physical Unclonable Functions. Adv. Mater. 2021, 33, 2102542,  DOI: 10.1002/adma.202102542
      28
      Chaotic Organic Crystal Phosphorescent Patterns for Physical Unclonable Functions
      Im, Healin; Yoon, Jinsik; Choi, Jinho; Kim, Jinsang; Baek, Seungho; Park, Dong Hyuk; Park, Wook; Kim, Sunkook
      Advanced Materials (Weinheim, Germany) (2021), 33 (44), 2102542CODEN: ADVMEW; ISSN:0935-9648. (Wiley-VCH Verlag GmbH & Co. KGaA)
      Since the 4th Industrial Revolution, Internet of Things based environments have been widely used in various fields ranging from mobile to medical devices. Simultaneously, information leakage and hacking risks have also increased significantly, and secure authentication and security systems are constantly required. Phys. unclonable functions (PUF) are in the spotlight as an alternative. Chaotic phosphorescent patterns are developed based on an org. crystal and at. seed heterostructure for security labels with PUFs. Phosphorescent org. crystal patterns are formed on MoS2. They seem similar on a macroscopic scale, whereas each org. crystal exhibits highly disorder features on the microscopic scale. In image anal., an encoding capacity as a single PUF domain achieves more than 1017 on a MoS2 small fragment with lengths of 25μm. Therefore, security labels with phosphorescent PUFs can offer superior randomness and no-cloning codes, possibly becoming a promising security strategy for authentication processes.
      >> More from SciFinder ®
      https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BB3MXitVGnsr7N&md5=388cf716785a18d110eec401b8b810e8
    29. 29
      Koh, D.; Kang, J.; Kim, T.; Lee, J.; Noh, S.; Lee, H.; Kwon, J.; Lee, S.; Park, J.; Park, B.-G. Improved Robustness Against Magnetic Field in Spin-Orbit-Torque-Based Physical Unclonable Functions Through Write-back Operation. Adv. Electron. Mater. 2023, 9, 2201073,  DOI: 10.1002/aelm.202201073
      There is no corresponding record for this reference.
    30. 30
      Lee, S.; Kang, J.; Kim, J. M.; Kim, N.; Han, D.; Lee, T.; Ko, S.; Yang, J.; Lee, S.; Lee, S. Spintronic Physical Unclonable Functions Based on Field-free Spin-Orbit-Torque Switching. Adv. Mater. 2022, 34, 2203558,  DOI: 10.1002/adma.202203558
      There is no corresponding record for this reference.
    31. 31
      Shao, Y.; Davila, N.; Ebrahimi, F.; Katine, J. A.; Finocchio, G.; Khalili Amiri, P. Reconfigurable Physically Unclonable Functions Based on Nanoscale Voltage-controlled Magnetic Tunnel Junctions. Adv. Electron. Mater. 2023, 9, 2300195,  DOI: 10.1002/aelm.202300195
      There is no corresponding record for this reference.
    32. 32
      Meo, A.; Garzón, E.; De Rose, R.; Finocchio, G.; Lanuzza, M.; Carpentieri, M. Voltage-Controlled Magnetic Anisotropy Based Physical Unclonable Function. Appl. Phys. Lett. 2023, 123, 062405,  DOI: 10.1063/5.0166164
      There is no corresponding record for this reference.
    33. 33
      Chiu, Y.-C.; Khwa, W.-S.; Yang, C.-S.; Teng, S.-H.; Huang, H.-Y.; Chang, F.-C.; Wu, Y.; Chien, Y.-A.; Hsieh, F.-L.; Li, C.-Y. A CMOS-Integrated Spintronic Compute-In-Memory Macro for Secure AI Edge Devices. Nat. Electron. 2023, 6, 534– 543,  DOI: 10.1038/s41928-023-00994-0
      There is no corresponding record for this reference.
    34. 34
      Lee, J. K.; Lee, J.; Yoon, S. I.; Lee, M. H.; Lee, J. S.; Jang, Y.; Kim, D. Y.; Choe, S. B.; Park, J.; Kim, Y. K. Optical Verification of Physically Unclonable Function Devices Based on Spin-orbit torque Switching. Adv. Electron. Mater. 2023, 9, 2300056,  DOI: 10.1002/aelm.202300056
      There is no corresponding record for this reference.
    35. 35
      Finocchio, G.; Moriyama, T.; De Rose, R.; Siracusano, G.; Lanuzza, M.; Puliafito, V.; Chiappini, S.; Crupi, F.; Zeng, Z.; Ono, T. Spin-Orbit Torque Based Physical Unclonable Function. J. Appl. Phys. 2020, 128, 033904,  DOI: 10.1063/5.0013408
      There is no corresponding record for this reference.
    36. 36
      Zhang, L.; Fong, X.; Chang, C.-H.; Kong, Z. H.; Roy, K. Highly Reliable Memory-Based Physical Unclonable Function Using Spin-Transfer Torque MRAM. 2014. In IEEE International Symposium on Circuits and Systems (ISCAS); IEEE, 2014; pp 2169– 2172.
      There is no corresponding record for this reference.
    37. 37
      Failure Mechanism Based Stress Test Qualification for Integrated Circuits , AEC-Q100-Rev. H, Sep. 2014.
      There is no corresponding record for this reference.
    38. 38
      Wilde, F.; Gammel, B. M.; Pehl, M. Spatial Correlation Analysis on Physical Unclonable Functions. IEEE Trans. Inf. Forensics Secur. 2018, 13, 1468– 1480,  DOI: 10.1109/TIFS.2018.2791341
      There is no corresponding record for this reference.
    39. 39
      Rukhi, A.; Soto, J.; Nechvatal, J.; Smid, M.; Barker, E.; Leigh, S.; Levenson, M.; Vangel, M.; Banks, D.; Heckert, A.; Dray, J.; Vo, S. A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications. In NIST Special Publication 800–22:2020; National Institute of Standards and Technology (NIST): Gaithersburg, MD, USA, 2010; .
      There is no corresponding record for this reference.
    40. 40
      Hitaj, B.; Gasti, P.; Ateniese, G.; Perez-Cruz, F. PassGAN: A Deep Learning Approach for Password Guessing. 2019. In International Conference on Applied Cryptography and Network Security (ACNS); Springer, 2019; pp 217– 237.
      There is no corresponding record for this reference.

  • Supporting Information

    Supporting Information

    The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.4c00078.

    • MR curves of MRAM-PUFs; MR measurements of an MRAM-PUF at various temperatures; generation of CRPs through optical measurements; evaluation of PUF performances of the bits generated by optical measurements; NIST randomness test results; GAN model; and performance comparisons of spintronics- or MRAM-based PUFs (PDF)


    Terms & Conditions

    Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.