Which chips can sign a read

The table

The column that matters is the fourth one. Everything to the left of it is shopping; that column decides whether a copied tag can be detected at all.

"Signs each read" means the chip itself computes a message authentication code over a value that changes every read, using a key that never leaves the chip. Three of the nine rows do that. Two more cannot be judged. For the four that do not, a copy of the data a tag serves is indistinguishable from the tag, and no configuration changes that.

Read range, and why no honest table has a range column

Range is a property of the antenna, the reader and the field, not of the chip. The data sheets say so themselves, in the same breath as the numbers everyone quotes.

The standard sets the class. ISO/IEC 14443 is a proximity interface: the reader "generates a magnetic field with a minimum field strength of Hmin 1.5 A/m and a maximum field strength of Hmax 7.5 A/m".^[10] ISO/IEC 15693 is a vicinity interface, specified for a weaker field over a larger volume. We could not read a primary source for its field strength limits from here, so this page does not state them; the figures in circulation disagree with each other and we are not going to add to that.

Within a class, what you get depends on your inlay. The vendors are explicit:

• NTAG 213 / 215 / 216: "Operating distance up to 100 mm (depending on various parameters as e.g. field strength and antenna geometry)".^[1] The same data sheet adds that "communication to NTAG21x can be established only when the IC is connected to an antenna. Form and specification of the coil is out of scope of this document."^[1]
• MIFARE Ultralight EV1: "Operating distance up to 100 mm depending on antenna geometry and reader configuration".^[3]
• NTAG 424 DNA: "reading distances of up to 10 cm".^[5]
• ICODE SLIX2: "Operating distance: up to 1.5 m (depending on antenna geometry)", described elsewhere in the same document as a gate width.^[4]

So the only defensible statement is the structural one. A 13.56 MHz proximity chip on a 25 mm inlay read by a phone gives you a few centimetres, whatever the part number. An ISO/IEC 15693 chip on a large inlay read by a gate reader gives you up to about a metre. Those are facts about the interface and the antenna. Choosing a different proximity chip will not change your range, and any table that prints a range figure per chip is reporting the manufacturer's reference antenna rather than yours.

Price, and what we could not find

We set out to publish a price band per chip and could not do it honestly. Here is what is actually available, and what the numbers mean.

Bare die and wafer prices are quoted per project and are not published. Distributor pages for the individual integrated circuits could not be retrieved programmatically, so this page prints no price for any bare chip rather than a figure we did not read.

What is public is converted inlay pricing from resellers, which is the form most people actually buy. Three listings from one United States reseller, read on 2026-10-24, at the same format and from the same catalogue so the comparison is like for like:^[11]

Three caveats, all of which matter more than the numbers. The listings show a dollar sign and never state a currency code, so the currency is an inference from the seller's location rather than something the pages say. The three inlays are different diameters, so part of the difference is antenna and converting rather than silicon. And reseller pricing moves, so treat these as an order of magnitude with a date attached, not a quote.

The useful conclusion survives all of that. A signing chip costs roughly two to three times a non-signing one in converted inlay form at quantities a small project actually buys. If the difference between "a copy is detectable" and "a copy is not detectable" is worth less than about 45 cents a unit to your application, you did not need authentication; you needed a URL.

Two features that cannot be evaluated

This is the most useful thing on the page, and it is a negative result. Two major vendors ship anti-counterfeiting features whose specifications are not public, which means nobody outside the vendor can assess them.

ST's unique tap code

The ST25TN512 and ST25TN01K offer a feature ST calls Augmented NDEF, which inserts a "3-digit unique tap code" into the message without the host writing to memory. The data sheet describes it in four sentences, and the second one is the problem: "ANDEF_UTC is an ASCII value generated once every time the device is powered. The value is unique to each user tap of the tag, and predictable. The size of ANDEF_UTC is 3 bytes."^[8]

Predictable, in the vendor's own word. The data sheet names no key, no algorithm and no counter, and refers the reader to an application note that is not published: "Contact your STMicroelectronics sales office to get this document."^[8] The same applies to the TruST25 digital signature, whose algorithm, curve and length appear nowhere in the data sheet. A three-byte value that the vendor describes as predictable is not a message authentication code, and on the public evidence the tap code should be treated as a session marker rather than as authentication.

EM Microelectronic's web-based authentication

The EM4425 is interesting on paper: one die answering both at 13.56 MHz as an NFC Forum Type 5 tag and at UHF as an ISO/IEC 18000-63 tag, with "AES-128 crypto for NFC web based authentication", an "optional ACCESS counter increased at first reading", and an optional digital signature of 256, 384 or 512 bits.^[9] That describes something close to the NTAG 424 DNA's construction, on an interface with far longer range.

What is published is a short data sheet, which states on its final page that it is "an extract from a full datasheet" available "on request".^[9] It gives the counter no width, the authentication no construction, the key no diversification scheme and the signature no algorithm. So we cannot say whether the AES-128 feature computes a code over the counter in the way Secure Dynamic Messaging does, and therefore cannot place the part in the table's fourth column.

Choosing, in four questions

In the order that actually eliminates options.

1. Does a copy have to be detectable?If yes, you need a chip in the third column, and most of the table is gone. If no, buy the cheapest part that holds your URL and stop reading.
2. Does it have to work with no app?Secure Dynamic Messaging exists so that an unmodified phone browser can carry the proof. If you control the reader, you have more options and can use challenge-response rather than a one-way message.
3. Proximity or vicinity?A few centimetres with a phone, or up to about a metre with a gate reader. This decides the air interface, and the air interface decides which signing parts are even available. At ISO/IEC 15693 range the publicly specified signing options are thin.
4. How many reads, over how long?A 24-bit counter is 16.7 million reads, after which an NTAG 424 DNA stops serving the file. For a consumer item that is effectively unlimited. For a high-traffic access point it is a service interval.

One point that is not a question, because the answer is always the same: the tamper variants are worth considering only if the mechanical design can route a detection loop so that removing the tag must break it. The chip feature is the easy part. What it does and does not prove.

Sources

Every data sheet below was downloaded and read on 2026-10-24, and the figures in the table were taken from the sections named. Where a document contradicts itself or omits something, that is noted.

1. NXP, NTAG213/215/216 product data sheet, Rev. 3.2, 2 June 2015, document 265332. Sections 1.1 and 2 for memory, interface and distance; section 8.6 for the 24-bit counter; section 8.9 for the signature. Searching the document for CMAC, Secure Dynamic Messaging or Secure Unique NFC returns nothing.
2. NXP, NTAG 213 TT product data sheet, Rev. 1.1, 28 March 2017, document 398311. Sections 2.1 and 8.5 for memory, section 8.7 for the counter, section 8.9 for the lockable signature. NXP marks this revision an objective data sheet, meaning the specification was not final, and the plain document URL now returns a 404; only the query-string form served the file.
3. NXP, MIFARE Ultralight EV1, MF0ULX1 product data sheet, Rev. 3.3, 9 April 2019, document 234533. Table 1 for the two memory variants, section 8.7 and 10 for the reader-incremented counters, section 8.8 for the signature. The document names no curve, and does not mention the NFC Forum at all.
4. NXP, ICODE SLIX2 product data sheet, Rev. 4.2, 1 December 2021. Sections 2.1 and 2.2 for interface, distance and memory; section 9 confirms that "block 79 contains the 16-bit counter and cannot be used to store user data", which is why 2560 bits of memory yields 316 usable bytes; section 9.2.2 for the signature. The document does not mention AES or CMAC.
5. NXP, NTAG 424 DNA, NT4H2421Gx product data sheet, Rev. 3.0, 31 January 2019. Sections 1.1 and 2.1 for memory and distance; Table 5 for the three-file layout; sections 9.3.1 and 9.3.8 for the counter and the signature; section 10.10.1 for the originality signature. The signing mechanism in detail.
6. NXP, NTAG 424 DNA TagTamper, NT4H2421Tx product data sheet, Rev. 3.0, 31 January 2019. Same memory layout and signing mechanism, plus section 10.2 for the tamper status values.
7. NXP, MIFARE DESFire EV3 product page, and AN12753, MIFARE DESFire EV3 Quick start guide, Rev. 1.2, 30 September 2020. The product page is the source for Secure Dynamic Messaging support, "compatible with NTAG DNA", and the memory sizes, which it writes as "2kB, 4kB, 8kB and 16 kb" in one place and "2/4/8/16KB" in another. The full data sheet is not public, so the counter width is not stated anywhere we could read.
8. STMicroelectronics, ST25TN512 and ST25TN01K data sheet, DS13433 Rev. 3, December 2021. Features list and section 5.4.3 for the unique tap code, section 5.5 for TruST25. st.com was not reachable from this machine on any attempt, so the document was read on a third-party mirror whose footer confirms the revision and date. Re-verify against st.com before relying on it, and check for a later revision.
9. EM Microelectronic, EM4425 Em|Echo-V short data sheet, 4425-DS-02 version 4.0, 9 January 2020. Features list for the interfaces, the 2048-bit configurable memory, the AES-128 web-based authentication, the access counter and the optional signature sizes. The closing page states it is an extract and that the full data sheet is available on request.
10. Rohde & Schwarz, 1MA113, Measurements on RFID Components according to 14443, section 2 and Table 1, for the ISO/IEC 14443 field strength limits. This is a measurement application note quoting the standard, not the standard itself; ISO documents are paywalled and iso.org refused automated retrieval.
11. GoToTags store listings for the NTAG213 18 mm inlay, the ICODE SLIX2 38 mm inlay and the NTAG 424 DNA 25 mm inlay, read 2026-10-24. Listed prices only; no relationship with the seller, and no endorsement implied.