TU Chemnitz — Patent Portfolio Analysis 2019-2024
The patent portfolio of Technische Universitaet Chemnitz — 68 DOCDB patent families from 107 applications across 8 patent offices, with a mechanical-engineering core, a MEMS and semiconductor cluster, and Fraunhofer-Gesellschaft as the strongest co-applicant.
Executive Summary
Technische Universitaet Chemnitz as a patent applicant, 2019-2024
Between 2019 and 2024, Technische Universitaet Chemnitz appears as applicant on 107 patent applications covering 68 DOCDB patent families. That works out at 1.57 applications per family, a ratio that describes a portfolio protected selectively rather than filed broadly across many jurisdictions. Filings reach 8 patent offices, but the distribution is steep: 61 of the 107 applications went to the DPMA, 17 to the EPO and 17 into the PCT route.
The portfolio is not uniform in kind. Of the 107 applications, 78 are national or regional patent applications, 17 are PCT applications, and 12 are German utility models covering 12 distinct families. Utility models are registered without substantive examination, which makes them a fast and inexpensive route to a protected right — a mix worth reading before any grant figure.
Technically, the portfolio sits in mechanical engineering. Machine tools is the largest WIPO technology field with 17 families, ahead of measurement (10), other special machines (9) and thermal processes and apparatus (9). A second cluster is formed by semiconductors (8 families) together with micro-structural and nano-technology (7 families); at IPC subclass level this shows as B81B and B81C, the two microsystems subclasses, with 6 and 5 families. Thermal energy is a third, distinct area: refrigeration (F25B, 6 families), heat exchange (F28D, 4) and ice production (F25C, 4).
Co-applications are present on 18 of the 68 families, involving 21 external partners once name variants are consolidated. Fraunhofer-Gesellschaft is the single strongest partner with 7 joint families spanning the entire period; every other partner contributes two families or one. Only one joint family is with another university (Brandenburgische Technische Universitaet Cottbus-Senftenberg) — the remaining partners are companies and named individuals.
The shape of this portfolio is that of a mid-sized technical university transferring mechanical-engineering research: 68 families in six years, a German-first filing pattern with 60 of 68 families reaching the DPMA, and a 29% PCT rate (20 families with at least one WO member) that marks the minority of inventions taken to the international stage. Filing volume moves between 6 and 17 families per year without a single direction over the six years.
Filing Trend 2019-2024
Annual patent families and applications with TU Chemnitz as applicant. Each family is counted once, in the year of its earliest in-scope filing.
Annual family counts move between 6 and 17 across the six years and do not follow one direction: 17 in 2019, 7 in 2020, back to 16 in 2021, 8 in 2022, 14 in 2023, 6 in 2024. At this scale a swing of five families is one research group filing or not filing in a given year, so the year-to-year movement should not be read as a trend. The six yearly values sum to exactly the 68 families in the portfolio.
Data completeness. Patent applications become visible in PATSTAT only after publication, normally around 18 months after filing. The 2024 value (and to a lesser degree 2023) is therefore incomplete in the Spring 2026 edition and will rise in later editions. No growth or decline claim in this report rests on the last available year.
Data table: Filing Trend 2019-2024
| Year | Patent Families | Applications | Co-Filed Families | Families with a Granted Member |
|---|---|---|---|---|
| 2019 | 17 | 29 | 4 | 11 |
| 2020 | 7 | 11 | 0 | 5 |
| 2021 | 16 | 17 | 4 | 12 |
| 2022 | 8 | 20 | 2 | 3 |
| 2023 | 14 | 18 | 6 | 9 |
| 2024* | 6 | 12 | 2 | 3 |
| Total | 68 | 107 | 18 | 43 |
* 2024: incomplete due to the approximately 18-month publication delay (PATSTAT Spring 2026, patstat-mtc.patstat_2026a). Grant counts for 2022 and later are additionally depressed by examination pendency.
Technology Profile
WIPO 35 technology fields, counted as distinct DOCDB families per field. The portfolio spreads across 28 of the 35 fields; the twelve largest are charted.
Mechanical engineering carries the portfolio. Machine tools is the largest single field with 17 families, a quarter of the 68, and three further mechanical fields follow directly behind: other special machines (9), thermal processes and apparatus (9) and mechanical elements (5). Alongside this sits a microsystems and semiconductor cluster — semiconductors (8) and micro-structural and nano-technology (7) — and a measurement and instruments strand led by measurement with 10 families.
Non-exclusive assignment. A patent family can be assigned to several WIPO technology fields at once. The 28 field values sum to 115, above the 68 distinct families in the portfolio. The chart therefore shows where the portfolio is present, not a division of it into shares — and for that reason it is drawn as bars, not as a pie or doughnut.
Data table: WIPO technology fields (all 28 with at least one family)
| Sector | Technology Field | Patent Families |
|---|---|---|
| Mechanical engineering | Machine tools | 17 |
| Instruments | Measurement | 10 |
| Mechanical engineering | Other special machines | 9 |
| Mechanical engineering | Thermal processes and apparatus | 9 |
| Electrical engineering | Semiconductors | 8 |
| Chemistry | Micro-structural and nano-technology | 7 |
| Electrical engineering | Electrical machinery, apparatus, energy | 6 |
| Mechanical engineering | Mechanical elements | 5 |
| Chemistry | Surface technology, coating | 5 |
| Electrical engineering | Audio-visual technology | 4 |
| Mechanical engineering | Textile and paper machines | 4 |
| Chemistry | Materials, metallurgy | 3 |
| Instruments | Medical technology | 3 |
| Other fields | Civil engineering | 3 |
| Mechanical engineering | Handling | 3 |
| Electrical engineering | Telecommunications | 2 |
| Electrical engineering | Computer technology | 2 |
| Mechanical engineering | Transport | 2 |
| Mechanical engineering | Engines, pumps, turbines | 2 |
| Chemistry | Macromolecular chemistry, polymers | 2 |
| Other fields | Other consumer goods | 2 |
| Chemistry | Chemical engineering | 1 |
| Chemistry | Basic materials chemistry | 1 |
| Chemistry | Organic fine chemistry | 1 |
| Instruments | Control | 1 |
| Instruments | Optics | 1 |
| Electrical engineering | Digital communication | 1 |
| Electrical engineering | Basic communication processes | 1 |
| Sum of field values | 28 fields | 115 |
| Distinct families | — | 68 |
IPC Analysis
IPC subclasses assigned to TU Chemnitz applications, counted as distinct DOCDB families per subclass. Titles are taken from IPC edition 2026.01.
At subclass level the manufacturing core resolves into joining and forming: B23K (soldering, welding, cladding) leads with 8 families, followed by B29C (shaping or joining of plastics) with 7 and B21D (sheet metal working) with 3. The microsystems pair B81B and B81C carries 6 and 5 families, and the thermal strand is visible across F25B (6), F28D (4) and F25C (4). Robotics appears with B25J (3) and acoustics with H04R (4).
Retired subclass kept in the count. H01L carries 5 families here, and in IPC edition 2026.01 that subclass was deprecated on 31 December 2025 — its subject matter was redistributed, with 818 recorded reclassification steps out of H01L in that edition alone. The assignments remain on the documents in PATSTAT Spring 2026, so the code is reported as it stands in the data rather than filtered out. It has no current title in IPC 2026.01, which is why the table below leaves that cell empty.
Non-exclusive assignment. A family carries several IPC symbols at once, so the subclass values sum above the 68 distinct families in the portfolio. Bars, not a pie or doughnut, for that reason.
Data table: Top IPC subclasses
| IPC Subclass | Title (IPC 2026.01) | Patent Families |
|---|---|---|
| B23K | Soldering or unsoldering; welding; cladding or plating … | 8 |
| B29C | Shaping or joining of plastics … | 7 |
| F25B | Refrigeration machines, plants or systems … | 6 |
| B81B | Microstructural devices or systems … | 6 |
| B81C | Processes or apparatus specially adapted for the manufacture … | 5 |
| H01L | Deprecated 31 December 2025 — no title in IPC 2026.01 | 5 |
| F25C | Producing, working or handling ice | 4 |
| F28D | Heat-exchange apparatus, not provided for in another subclass … | 4 |
| H04R | Loudspeakers, microphones, gramophone pick-ups … | 4 |
| B21D | Working or processing of sheet metal or metal tubes, rods … | 3 |
| F16B | Devices for fastening or securing constructional elements … | 3 |
| G01S | Radio direction-finding; radio navigation; determining distance … | 3 |
| H05B | Electric heating; electric light sources not otherwise provided for … | 3 |
| B25J | Manipulators; chambers provided with manipulation devices | 3 |
| B24C | Abrasive or related blasting with particulate material | 3 |
| G01N | Investigating or analysing materials by determining their … | 3 |
Titles shortened where the full IPC subclass title is long; the wording shown is the opening of the official title. Classification editions: IPC 2026.01, CPC 2026.08.
Geographic Strategy
Where TU Chemnitz files, and how the distribution differs between the first and the second half of the period.
The filing pattern is German-first: 60 of the 68 families reach the DPMA, and 61 of the 107 applications are German. Beyond that, EP and WO each account for 17 applications and the United States for 7; CN, AT, TW and CA appear once or twice each. Comparing the two halves of the period, DE families move from 33 to 27 while EP families move from 6 to 11 — on eight and eleven families the second figure is a small absolute change, but it is the only office pair that moves in opposite directions across the two halves.
Two counts of the PCT route, and why they differ. 17 applications in this portfolio were filed at WO. The PCT rate reported in the KPI strip is a different measurement: 20 of 68 families (29%) contain at least one WO member anywhere in the DOCDB family, including WO filings that are not themselves attributed to TU Chemnitz as applicant in the queried window. The family-level figure is the one used throughout this report; the application count is shown here because it is what the office table measures.
Data table: Top Filing Offices 2019-2024
| Office | Applications | Patent Families | Granted Applications |
|---|---|---|---|
| DE — DPMA | 61 | 60 | 34 |
| EP — EPO | 17 | 17 | 10 |
| WO — PCT | 17 | 17 | 3 |
| US — USPTO | 7 | 7 | 3 |
| CN | 2 | 2 | 1 |
| AT | 1 | 1 | 1 |
| TW | 1 | 1 | 0 |
| CA | 1 | 1 | 0 |
| Total | 107 | 68 distinct (per-office counts sum to 106) | 52 |
Data table: Geographic Shift by Period (families)
| Office | Families 2019-2021 | Families 2022-2024 | Families, whole period |
|---|---|---|---|
| DE — DPMA | 33 | 27 | 60 |
| EP — EPO | 6 | 11 | 17 |
| WO — PCT | 9 | 8 | 17 |
| US — USPTO | 4 | 3 | 7 |
| CN | 2 | 0 | 2 |
| AT | 1 | 0 | 1 |
| TW | 0 | 1 | 1 |
| CA | 1 | 0 | 1 |
Family counts per office sum to 106, above the 68 distinct families, because a family filed at several offices is counted once at each of them. 2024 filings are incomplete due to the approximately 18-month publication delay (PATSTAT Spring 2026). Grant counts for filings from 2022 onward are depressed by examination pendency and are not a measure of grant prospects; WO is the international phase of the PCT and is not itself a granting authority.
Collaborations
Organisations and individuals that appear as co-applicant on a TU Chemnitz application. Co-applicants are matched at application level, and every TU Chemnitz name variant is excluded from the partner list by the same rule that defines the portfolio.
18 of the 68 families (26%) carry at least one external co-applicant, spread over 21 partners once name variants are consolidated. Fraunhofer-Gesellschaft is the only partner above two families, with 7 joint families across the full period 2019-2024; five partners contribute two families each and fifteen contribute one. Only one joint family is with another university — Brandenburgische Technische Universitaet Cottbus-Senftenberg in 2023 — so the co-applicant network is essentially industry- and individual-facing rather than academic.
Two harmonisation corrections applied. PATSTAT holds two harmonised name groups for Fraunhofer-Gesellschaft in this data set, one of them a spelling variant with the words run together. Read separately they suggest 7 plus 1 joint families; measured as one entity the correct figure is 7 distinct families, because the variant duplicates a family already counted. The same applies to Heinz-Dieter Teichert, who appears both as a natural person and as Tube Welding Tools; consolidated, that is 2 distinct families. Both figures below are the consolidated ones.
Data table: Co-applicants 2019-2024
| Co-Applicant (consolidated) | Type | Joint Families | Years |
|---|---|---|---|
| Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung | Research | 7 | 2019-2024 |
| 3D-plastic Hans Kintra GmbH | Company | 2 | 2023 |
| Dohlen & Krott Werkzeugbau GmbH | Company | 2 | 2023 |
| GBZ Mannheim GmbH & Co. KG | Company | 2 | 2022 |
| Heitkamp, Harald | Individual | 2 | 2023 |
| Teichert, Heinz-Dieter / Tube Welding Tools | Individual | 2 | 2023 |
| ANDRITZ AWEBA GmbH | Company | 1 | 2024 |
| Brandenburgische Technische Universitaet Cottbus-Senftenberg | University | 1 | 2023 |
| ElringKlinger AG | Company | 1 | 2024 |
| LIGenium GmbH | Company | 1 | 2020 |
| OptoGenTech GmbH | Company | 1 | 2021 |
| richter & hess Verpackungs-Service GmbH | Company | 1 | 2021-2022 |
| Solenis Technologies, L.P. (US) | Company | 1 | 2021 |
| Eight further named individuals (Haehnle, Hamers, Kassner, Seifert, Sommer, Spange, Taubert, Uhlig) | Individual | 1 each | 2021 |
Partner family counts sum above 18 because a co-filed family can carry more than one external applicant. 2024 co-filings are incomplete due to the approximately 18-month publication delay (PATSTAT Spring 2026).
Top Inventors
Named inventors on TU Chemnitz applications, counted as distinct DOCDB families per inventor. Names are normalised to upper case before grouping, so spelling case variants are counted once.
Inventive activity is distributed rather than concentrated. The most frequently named inventor, Thorsten Urbaneck, appears on 8 of the 68 families; the next two, Thomas Lampke and Markus Richter, on 5 each. No inventor reaches more than 12% of the portfolio, and beyond the first nine names every inventor in the top 25 stands at two families. This is consistent with the field spread seen in the technology profile: several research groups filing independently rather than one group accounting for most of the portfolio.
Data table: Top inventors by patent families
| Inventor | Country | Patent Families |
|---|---|---|
| Urbaneck, Thorsten | DE | 8 |
| Lampke, Thomas | DE | 5 |
| Richter, Markus | DE | 5 |
| Paczkowski, Gerd | DE | 4 |
| Nendel, Wolfgang | DE | 4 |
| Stoeckel, Chris | DE | 3 |
| Dittrich, Markus | DE | 3 |
| Kroll, Lothar | DE | 3 |
| Kanoun, Olfa | DE | 3 |
| Zhu, Hongxi | DE | 2 |
| Sauer, Konstantin | DE | 2 |
| Kroll, Martin | DE | 2 |
| Kraeusel, Verena | DE | 2 |
| Hofmann, Christian | DE | 2 |
| Froehlich, Alexander | DE | 2 |
| Haertel, Sebastian | DE | 2 |
| Thomas, Ulrike | DE | 2 |
| Naumann, Mario | DE | 2 |
| Gelbrich, Sandra | DE | 2 |
| Sommer, Michael | DE | 2 |
| Zucker, Tino | DE | 2 |
| Engelmann, Udo | DE | 2 |
| Zopp, Camilo | DE | 2 |
| Eichhorn, Sven | DE | 2 |
| Melzer, Marcel | DE | 2 |
Inventor names are recorded per application and are not harmonised in PATSTAT beyond the raw string; two people with the same written name would be counted as one. The chart shows the twelve largest; the table lists the top 25 returned by the query.
Methodology
Search strategy, data source, and known limitations
As of 15 August 2026, this analysis is computed on EPO PATSTAT Global, Spring 2026
(patstat-mtc.patstat_2026a), with classification editions
IPC 2026.01 and CPC 2026.08 as recorded in
dpt_edition. Every figure in this report comes from a SQL query executed against
that edition; the full query set is published alongside the report in
queries.sql.
Search strategy, family counting, and data source
Applicant scope — how TU Chemnitz was identified
PATSTAT performs no entity resolution: one institution appears as many person records, one per spelling. The scope of this report was therefore built from applicant name variants, not from classification codes.
-
A broad discovery query listed every applicant whose
person_nameorhan_namecontains the string Chemnitz — deliberately broader than a name-prefix search, which would have missed variants such as Technische Universitaet Chemnitz, Koerperschaft des oeffentlichen Rechts and the inverted form UNIV CHEMNITZ TECH. - Inside the 2019-2024 window this returned 28 distinct Chemnitz-related applicant records. Of these, 10 name variants were kept as TU Chemnitz: seven spellings of Technische Universitaet Chemnitz with and without the legal-form suffix, two case and country variants, and UNIV CHEMNITZ TECH.
-
The kept set is matched by one rule, applied identically in every query and in the
co-applicant exclusion:
person_namebeginning with Technische Universit… Chemnitz, or equal to UNIV CHEMNITZ TECH. Because selection and self-exclusion use the same predicate, TU Chemnitz cannot appear in its own partner list.
Was han_name usable?
No — not as a grouping key on its own. The kept variants fall into
three different han_name groups:
TECHNISCHE UNIVERSITAT CHEMNITZ,
TECHNISCHE UNIV CHEMNITZ KORPERSCHAFT DES OFFENTLICHEN RE… and
UNIV CHEMNITZ TECH. Any single han_name would have captured a fraction
of the portfolio. Cross-checking han_name against person_name did,
however, show no foreign institution mapped into a TU Chemnitz group and no TU Chemnitz
record mapped out to a foreign group, so the harmonisation is fragmented here but not
misdirected. The same fragmentation was found and corrected on the partner side
(see Collaborations).
Institutions deliberately excluded
- ICM — Institut Chemnitzer Maschinen- und Anlagenbau e.V. (7 families in the window under its main spelling, plus 1 under a variant in which a lower-case L stands in for the letter I). ICM is an affiliated institute with its own legal personality as a registered association, filing under its own name. It is not counted as TU Chemnitz.
- Institut fuer Konstruktion und Verbundbauweisen e.V. an der TU Chemnitz — the same reasoning; it records no families inside the 2019-2024 window.
- Companies and individuals with a Chemnitz address — GEMAC Chemnitz, GPP Chemnitz, EDC Electronic Design Chemnitz, Chemieanlagenbau Chemnitz, NIWC, CLL Chemnitzer Laborleistungs, Chemnitz Power Labs, Maschinenbau-Service-Automatisierungstechnik Chemnitz and several partnerships. These share the city name only.
- A university of applied sciences in Chemnitz — none appears as an applicant in the discovery result for this window, so no such exclusion was needed.
Family counting
-
All portfolio figures use
COUNT(DISTINCT docdb_family_id). A DOCDB family groups the filings protecting one invention across countries, so an invention filed in DE, EP, WO and US counts once. - In the annual chart each family is assigned to the earliest filing year among its in-scope applications. The six yearly values therefore sum to exactly 68, the portfolio total, instead of counting a family again in every year it has a filing.
-
Co-applicants are matched at
appln_idlevel, not at family level. Matching on the family would pair the institution with its own name variants appearing on sibling applications and make it its own top partner. - The PCT rate is the share of families with at least one WO member anywhere in the DOCDB family (20 of 68), not the number of WO applications (17). The two are reported side by side in the Geographic Strategy section.
Classification scope
- The patent set is defined by applicant names, not by an IPC or CPC code list. There is therefore no code set to expand into subtrees and no predecessor codes to union in — the three completion steps that apply to a technology landscape do not apply here.
-
The IPC breakdown reports the symbols as they stand on the documents.
Deprecated codes are kept: no
is_activefilter is applied. This is what keeps H01L visible with 5 families even though that subclass was retired on 31 December 2025 in IPC edition 2026.01. -
IPC subclass titles are joined from
dpt_ipconsymbol_patstat, the join key that resolves the full symbol rather than a truncated form.
Data source
- EPO PATSTAT Global, Spring 2026 — dataset
patstat-mtc.patstat_2026a. - Classification views:
patstat-mtc.classification— IPC 2026.01, CPC 2026.08. - Platform: Google BigQuery, queried through the mtc.berlin PATSTAT MCP server.
- Plausibility cross-check: the DPMA register returns 72 German IP-right records with applicant Technische Universitaet Chemnitz filed 2019-2024, against 61 German applications measured here — a deviation of about 15%, within the tolerance normally accepted for a university profile. The two counts are not defined identically (register records versus PATSTAT applications), so the check establishes plausibility, not equality.
Scope Limitations
- Only filings where TU Chemnitz is recorded as applicant are counted. Inventions by TU Chemnitz staff filed solely under a company name, or assigned away before publication, are outside this portfolio.
- Affiliated institutes (ICM e.V., Institut fuer Konstruktion und Verbundbauweisen e.V.) are excluded by decision, not by oversight. Including ICM would add up to 8 families to the window.
-
Designs are not part of the data set; German utility models are — 12 of the
107 applications, covering 12 families, are utility models (
ipr_type = UM). They are registered without substantive examination, so any grant figure mixes examined and unexamined rights. - Filings from 2024 are incomplete because of the approximately 18-month publication delay, and grant counts for filings from 2022 onward are depressed by examination pendency. No trend claim in this report rests on the last available year.
- WIPO technology fields and IPC subclasses are multi-assigned: field values sum to 115 and subclass values likewise exceed 68. Neither is drawn as a pie or doughnut for that reason.
- Inventor names are ungrouped strings in PATSTAT; identical spellings of different people would merge, and a person filing under two spellings would split.
Glossary — Patent Terms Explained
- Patent Family (DOCDB)
- A group of patent applications protecting the same invention in different countries. Counting families rather than applications answers "how many inventions", not "how many filings".
- Applicant vs. Inventor
- The applicant owns the right applied for; the inventor is the natural person who made the invention. In PATSTAT the roles are separated by
applt_seq_nrandinvt_seq_nr. - IPC / CPC
- International and Cooperative Patent Classification — hierarchical schemes for classifying patents by technology. A subclass such as B23K is the four-character level.
- WIPO 35 Technology Fields
- A coarse mapping of IPC symbols onto 35 technology fields in five sectors, used for comparable technology profiles.
- PCT / WO application
- An international application under the Patent Cooperation Treaty. It reserves the right to file in member states later; it is not itself a granted patent.
- Utility model (Gebrauchsmuster)
- A German registered right, granted without substantive examination of novelty and inventive step, with a shorter maximum term than a patent.
- han_name
- PATSTAT’s harmonised applicant name. It reduces spelling variation but does not resolve entities, and one institution can carry several han_names.
Report Files
This report is published as a set of source files: meta.json with the report
metadata, cover.html and kpi-strip.html, the numbered section files
under sections/, and queries.sql containing every SQL query behind the
figures, with its PATSTAT and classification editions recorded in the header. The queries can be
re-run against the named edition to reproduce each number in this report.
Like what you see?
This report was built with a fully reproducible pipeline: EPO PATSTAT Global on BigQuery, a custom MCP server, and Claude AI for analysis and visualization.