1 Introduction

Let G be a simple connected graph with vertex set V(G) and edge set E(G). Its order is |V(G)|, and its size is |E(G)|. The maximum and the minimum degrees of G are denoted \(\delta (G)\) and \(\varDelta (G)\) (for short \(\delta\) and \(\varDelta\)), respectively.

A proper edge coloring of a graph G is an assignment of colors to the edges such that adjacent edges receive distinct colors. The chromatic index \(\chi ^{\prime }(G)\) is the minimum number of colors in a proper edge coloring of G. A strong edge-colouring (also called distance 2 edge-coloring) of G is a proper edge coloring of G, such that the edges of any path of length 3 use three different colors. We denote by \(\chi _{s}^{\prime }(G)\) the strong chromatic index of G which is the smallest integer k such that G can be strongly edge-colored with k colors. The concept of strong edge-coloring was introduced by Fouquet and Jolivet in [6, 7] and can be used to model the conflict-free channel assignment in radio networks [14, 15].

In 1985, Erdős and Nešetřil [5] proposed the following conjecture concerning the upper bound on \(\chi _{s}^{\prime }(G)\), and also provided an example to illustrate that this upper bound is sharp.

Conjecture 1

(Erdős and Nešetřil [5]) For every graph G,

$$\begin{aligned} \chi _{s}^{\prime }(G)\le \left\{ \begin{array}{lll} \frac{5}{4}\varDelta ^2, &{} &{} {if\,\varDelta \,is\,even,}\\ \frac{1}{4}(5\varDelta ^2-2\varDelta +1), &{} &{} {if\,\varDelta \,is\,odd.} \end{array} \right. \end{aligned}$$

Conjecture 1 was verified for graphs having \(\varDelta \le 3\) [1, 10]. For \(\varDelta =4\), an upper bound of 21 (recall that conjectured bound is 20 ) has been proved by Huang, Santana and Yu [11]. Moreover, Lv, Li and Yu [12] showed that Conjecture 1 is true for some special cases when for \(\varDelta =4\). For sufficiently large \(\varDelta\), using probabilistic techniques, Molloy and Reed in [13] proved that \(\chi _s'(G) \le 1.998\varDelta (G)^2\). This bound is improved to \(1.93\varDelta ^2\) by Bruhn and Joos [3], and very recently, is further improved to \(1.835\varDelta ^2\) by Bonamy, Perrett, and Postle [2]. Currently the best known upper bound on the strong chromatic index of a graph with maximum degree \(\varDelta\) is \(1.772\varDelta ^2\), it was proved by Hurley, de Joannis de Verclos and Kang [8] in 2020.

A graph that contains no induced subgraph isomorphic to \(K_{1,3}\) is said to be claw-free. Debski, Junosza-Szaniawski and Śleszyńska-Nowak [4] give the following upper bound for the strong chromatic index of claw-free graphs:

Theorem 1

(Debski, Junosza-Szaniawski and Śleszyńska-Nowak [4]) If G is a claw-free graph with maximum degree \(\varDelta\), then \(\chi _{s}^{\prime }(G)\le \frac{9}{8}\varDelta ^2+\varDelta\).

For claw-free subcubic graphs (claw-free graphs with \(\varDelta \le 3\)), we give the following somewhat stronger result.

Theorem 2

If G is a claw-free subcubic graph other than the prism graph, then \(\chi _{s}^{\prime }(G)\le 8\), where the prism graph is the graph isomorphic to the configuration depicted in Figure 1.

Fig. 1
figure 1

The prism graph \(G^\prime\) has \(\chi _{s}^{\prime }(G^\prime )=9\)

Remark 1

Recall that when \(\varDelta =3\), the upper bound in Conjecture 1 is 10, and the upper bound in Theorem 1 is \(\frac{105}{8}\), while our bound is 8.

2 Preliminaries

We now introduce some notations used in the paper. For a graph G, a vertex is a k-vertex if it is of degree k. Similarly, we define i-cycle. A 3-cycle is called a triangle. For a vertex \(v\in V(G)\), let \(N_G(v)\) denote the set of all neighbors of v in G and let \(\varGamma _G(v)\) denote the set of all edges incident to v in G. Denote by \(SC(\varGamma _G(v))\) the set of colors used by edges in \(\varGamma _G(v)\). Given edges \(e_1\) and \(e_2\) in G, we say that \(e_1\) sees \(e_2\) if either \(e_1\) and \(e_2\) are adjacent, or there is another edge \(e_3\) adjacent to both \(e_1\) and \(e_2\). Denote by \([8]=\{1, 2,\ldots , 8\}\) the set of colors. A partial coloring of G is a strong edge-coloring of a proper subgraph of G. If \(e=uv\) is an uncolored edge in a partial coloring of G, let \(L_G(e)\) be the set of colors that is used on the edges incident to a vertex in \(N_G(u)\cup N_G(v)\), and let \(L^\prime _G(e)=[8]\backslash L_G(e)\). We write L(e) and \(L^\prime (e)\) for \(L_G(e)\) and \(L^\prime _G(e)\), respectively, if it is clear from the context. Let d(uv) be the distance between vertices u and v of G. For a cycle C, a vertex x and an edge e in G, let \(d(C,x)=min_{u\in C}d(u,x)\) be the distance of the cycle C and the vertex x of G, and let \(d(C,e)=min_{u\in C, v\in e}d(u,v)\) be the distance of the cycle C and the edge e of G. Let \(K_4^*\) be the graph obtained from \(K_4\) by subdividing one edge.

We will use the following well known result of Hall [9] in terms of systems of distinct representatives.

Theorem 3

(Hall [9]) Let \(A_1 ,\ldots , A_n\) be n subsets of a set U. A system of distinct representatives of \(\{A_1,\ldots , A_n\}\) exists if and only if for all \(k, 1\le k\le n\) and every choice of subcollection of size k, \(\{A_{i_1} ,\ldots , A_{i_k}\}\), we have \(|A_{i_1}\cup \cdots \cup A_{i_k}|\ge k\).

3 Proof of Theorem 2

Let G be a counterexample to Theorem 2 with \(|V(G)|+|E(G)|\) minimized. We will prove various statements about G. Clearly, G is connected.

Lemma 1

There is no 1-vertex in G.

Proof

Suppose that G contains 1-vertex v. Let u be the neighbor of v. By the minimality of G, \(G^*=G-v\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. G is claw-free so \(|L^\prime (uv)|\ge 3\). Thus, we color uv to obtain a strong edge-coloring of G, a contradiction. \(\square\)

Lemma 2

There are no adjacent 2-vertices in G.

Proof

Suppose that G contains adjacent 2-vertices u and v. Let \(N_G(u)=\{u^\prime ,v\}\) and \(N_G(v)=\{v^\prime ,u\}\). Note that \(u^\prime\) and \(v^\prime\) may be the same vertex. By the minimality of G, \(G^*=G-\{u,v\}\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. G is a claw-free graph so \(|L^\prime (uu^\prime )|\ge 3\), \(|L^\prime (vv^\prime )|\ge 3\) and \(|L^\prime (uv)|\ge 4\). Thus, we color \(uu^\prime\), \(vv^\prime\) and uv in turn to obtain a strong edge-coloring of G, a contradiction. \(\square\)

The following observation is true because G is claw-free.

Observation 1 Each 3-vertex is on at least one triangle in G.

Lemma 3

There is no 2-vertex in G.

Proof

Suppose that G contains 2-vertex v. Let \(N_G(v)=\{u,w\}\). By Lemma 2, u and w are 3-vertices. By Observation 1, u and w are on triangles.

Assume that \(uw\in E(G)\), by the minimality of G, \(G^*=G-\{v\}\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. Observe that \(|L^\prime (uv)|\ge 3\) and \(|L^\prime (wv)|\ge 3\). When we color uv, there are still two colors available for wv. Thus, we color uv and wv in turn to obtain a strong edge-coloring of G, a contradiction.

Assume that \(uw\notin E(G)\). If \(|N_G(u)\cap N_G(w)|\ge 2\), then \(G\cong K_4^*\) and \(\chi _{s}^{\prime }(G)=7\). If \(|N_G(u)\cap N_G(w)|=1\), we use notations in Figure 2. By the minimality of G, \(G^*=G-\{v\}\) has a strong edge-coloring \(c^*\) with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. Observe that \(|L^\prime (uv)|\ge 1\) and \(|L^\prime (wv)|\ge 1\). If \(|L^\prime (uv)|\ge 2\), or \(|L^\prime (wv)|\ge 2\), or \(L^\prime (uv)\ne L^\prime (wv)\), we can assign distinct colors to uv and wv, and obtain a strong edge-coloring of G, a contradiction. Thus, we have \(|L^\prime (uv)|=1\), \(|L^\prime (wv)|=1\) and \(L^\prime (uv)=L^\prime (wv)\). Without loss of generality, let \(L^\prime (uv)=L^\prime (wv)=\{1\}\). Therefore, we may assume, without loss of generality, that \(c^*(uu_1)=2\), \(c^*(uu_2)=3\), \(c^*(u_1u_2)=4\), \(c^*(u_1u_3)=5\), \(c^*(u_2u_4)=6\), \(c^*(ww_1)=7\), \(c^*(ww_2)=8\), \(c^*(w_1w_2)=4\), \(c^*(w_1w_3)=5\) and \(c^*(w_2w_4)=6\). We claim that \(7,8\in SC(\varGamma _G(u_3))\), if not, we can recolor \(uu_1\) with 7 or 8, color uv with 2 and color vw with 1, and obtain a desired strong edge-coloring of G, a contradiction. Similar to the discussion above, we claim that \(7,8\in SC(\varGamma _G(u_4))\), \(2,3\in SC(\varGamma _G(w_3))\) and \(2,3\in SC(\varGamma _G(w_4))\). Thus, we can recolor \(uu_1\) and \(ww_1\) with 1, color uv with 2 and color wv with 7, and obtain a strong edge-coloring of G, a contradiction. \(\square\)

Fig. 2
figure 2

Illustrations of Lemma 3

Fig. 3
figure 3

Illustrations of Lemma 4

Lemma 4

Each 3-vertex is on exactly one triangle in G.

Proof

By Observation 1, each 3-vertex is on at least one triangle. Suppose that a 3-vertex v is on two triangles \(T_1=vv_1v_2\) and \(T_2=vv_3v_2\), we use notations in Figure 3. By the minimality of G, \(G^*=G-\{v,v_2\}\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. Observe that \(|L^\prime (vv_1)|\ge 4\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (vv_3)|\ge 4\), \(|L^\prime (v_2v_3)|\ge 4\) and \(|L^\prime (vv_2)|\ge 6\). Thus, we color \(vv_1\), \(v_1v_2\), \(vv_3\), \(v_2v_3\) and \(vv_2\) in turn to obtain a strong edge-coloring of G, a contradiction. \(\square\)

Lemma 5

Suppose that \(T_1=uvw\) and \(T_2=u_1v_1u_2\) are two different triangles in G such that \(vv_1\in E(G)\), then u (neither w) is not adjacent to \(u_1\) nor \(u_2\).

Proof

Without loss of generality, we assume that \(uu_1\in E(G)\). Since G is not the prism graph, \(u_2w\notin E(G)\), and we use notations in Figure 4. By the minimality of G, \(G^*=G-\{u,v,u_1,v_1\}\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. Observe that \(|L^\prime (u_1u_2)|\ge 5\), \(|L^\prime (v_1u_2)|\ge 5\), \(|L^\prime (uw)|\ge 5\), \(|L^\prime (vw)|\ge 5\), \(|L^\prime (uu_1)|\ge 6\), \(|L^\prime (vv_1)|\ge 6\), \(|L^\prime (u_1v_1)|\ge 7\) and \(|L^\prime (uv)|\ge 7\). Recall that \(u_2w\notin E(G)\), edge \(u_1u_2\) can not see edge vw in G. Note that \(|L^\prime (u_1u_2)|\ge 5\), \(|L^\prime (vw)|\ge 5\) and we have at most 8 different colors, so \(L^\prime (u_1u_2)\cap L^\prime (vw)\ne \emptyset\). We color \(u_1u_2\) and vw with the same color, and color \(u_2v_1\), uw, \(uu_1\), \(vv_1\), \(u_1v_1\) and uv in turn to obtain a strong edge-coloring of G, a contradiction. Thus, we obtain a strong edge-coloring of G, a contradiction. \(\square\)

Fig. 4
figure 4

Illustrations of Lemma 5

Lemma 6

Let \(v_1v_2v_3v_4v_5v_6\) be a 6-cycle without chord in G, and let \(c^\prime\) be a partial coloring of G such that all uncolored edges are \(v_1v_2\), \(v_2v_3\), \(v_3v_4\), \(v_4v_5\), \(v_5v_6\) and \(v_6v_1\), and edge \(v_1v_2\) can not see edge \(v_4v_5\), edge \(v_2v_3\) can not see edge \(v_5v_6\), edge \(v_3v_4\) can not see edge \(v_6v_1\). If \(|L^\prime (v_2v_3)|\ge 3\), \(|L^\prime (v_4v_5)|\ge 3\), \(|L^\prime (v_6v_1)|\ge 3\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (v_3v_4)|\ge 4\), and \(|L^\prime (v_5v_6)|\ge 4\), then the partial coloring \(c^\prime\) can be extended to a strong edge-coloring of G.

Proof

Notice that edge \(v_1v_2\) can not see edge \(v_4v_5\), edge \(v_2v_3\) can not see edge \(v_5v_6\), edge \(v_3v_4\) can not see edge \(v_6v_1\). If \(L^\prime (v_1v_2)\cap L^\prime (v_4v_5)\ne \emptyset\). We color \(v_1v_2\) and \(v_4v_5\) with same color in \(L^\prime (v_1v_2)\cap L^\prime (v_4v_5)\). Now \(|L^\prime (v_2v_3)|\ge 2\), \(|L^\prime (v_6v_1)|\ge 2\), \(|L^\prime (v_3v_4)|\ge 3\), and \(|L^\prime (v_5v_6)|\ge 3\), and we color \(v_6v_1\), \(v_2v_3\), \(v_3v_4\) and \(v_5v_6\) in this order to obtain a strong edge-coloring of G. Therefore, \(L^\prime (v_1v_2)\cap L^\prime (v_4v_5)=\emptyset\). Similarly, we have \(L^\prime (v_2v_3)\cap L^\prime (v_5v_6)=\emptyset\) and \(L^\prime (v_6v_1)\cap L^\prime (v_3v_4)=\emptyset\). Let \(A=\{v_1v_2\), \(v_2v_3\), \(v_3v_4\), \(v_4v_5\), \(v_5v_6,v_6v_1\}\), for any \(S\subseteq A\), we have \(|\cup _{e\in S} L^\prime (e)|\ge |S|\). By Theorem 3, we can assign distinct colors to \(v_1v_2\), \(v_2v_3\), \(v_3v_4\), \(v_4v_5\), \(v_5v_6\) and \(v_6v_1\). Thus, we obtain a strong edge-coloring of G. \(\square\)

Fig. 5
figure 5

Illustrations of Lemma 7

Lemma 7

Suppose that \(T_1=v_1v_2v_{12}\), \(T_2=v_3v_4v_{34}\) and \(T_3=v_5v_6v_{56}\) are three different triangles such that \(v_2v_3\in E(G)\) and \(v_4v_5\in E(G)\), then \(v_1\) (neither \(v_{12}\)) is not adjacent to \(v_6\) nor \(v_{56}\).

Proof

Without loss of generality, we assume that \(v_1v_6\in E(G)\). By Lemma 5, \(v_{12}v_{34}\notin E(G)\), \(v_{34}v_{56}\notin E(G)\) and \(v_{56}v_{12}\notin E(G)\), and we use notations in Figure 5. By the minimality of G, \(G^*=G-\{v_1,v_2,v_3,v_4,v_5,v_6\}\) has a strong edge-coloring with at most eight colors. The strong edge-coloring of \(G^*\) is a partial coloring of G. Observe that \(|L^\prime (v_1v_{12})|\ge 5\), \(|L^\prime (v_2v_{12})|\ge 5\), \(|L^\prime (v_3v_{34})|\ge 5\), \(|L^\prime (v_4v_{34})|\ge 5\), \(|L^\prime (v_5v_{56})|\ge 5\), \(|L^\prime (v_6v_{56})|\ge 5\), \(|L^\prime (v_2v_3)|\ge 6\), \(|L^\prime (v_4v_5)|\ge 6\), \(|L^\prime (v_6v_1)|\ge 6\), \(|L^\prime (v_1v_2)|\ge 7\), \(|L^\prime (v_3v_4)|\ge 7\), and \(|L^\prime (v_5v_6)|\ge 7\).

Note that \(L^\prime (v_1v_{12})=L^\prime (v_2v_{12})\), \(L^\prime (v_3v_{34})=L^\prime (v_4v_{34})\) and \(L^\prime (v_5v_{56})=L^\prime (v_6v_{56})\). Beacause \(L^\prime (v_1v_{12})\) and \(L^\prime (v_3v_{34})\) are 5-element subsets of 8-element set their intersection has at least 2 elememnts, denote them by \(\alpha _1\), \(\alpha _2\), that is \(\{\alpha _1,\alpha _2\}\subset L^\prime (v_1v_{12})\cap L^\prime (v_3v_{34})\). Similarly, we let \(\{\beta _1,\beta _2\}\subset L^\prime (v_1v_{12})\cap L^\prime (v_5v_{56})\) and \(\{\gamma _1,\gamma _2\}\subset L^\prime (v_3v_{34})\cap L^\prime (v_5v_{56})\). Recall that \(v_{12}v_{34}\notin E(G)\), \(v_{34}v_{56}\notin E(G)\) and \(v_{56}v_{12}\notin E(G)\), then edge \(v_1v_{12}\) can not see edges \(v_3v_{34}\), \(v_4v_{34}\), \(v_5v_{56}\), and edge \(v_2v_{12}\) can not see edges \(v_4v_{34}\), \(v_5v_{56}\), \(v_6v_{56}\), and edge \(v_3v_{34}\) can not see edges \(v_1v_{12}\), \(v_5v_{56}\), \(v_6v_{56}\), and edge \(v_4v_{34}\) can not see edges \(v_6v_{56}\), \(v_1v_{12}\), \(v_2v_{12}\).

If \(\{\alpha _1,\alpha _2\}\cap \{\beta _1,\beta _2\}\ne \emptyset\), we color \(v_1v_{12}\), \(v_3v_{34}\) and \(v_5v_{56}\) with same color in \(\{\alpha _1,\alpha _2\}\cap \{\beta _1,\beta _2\}\), and color \(v_2v_{12}\), \(v_4v_{34}\) and \(v_6v_{56}\) in this order. Now, we have \(|L^\prime (v_2v_3)|\ge 3\), \(|L^\prime (v_4v_5)|\ge 3\), \(|L^\prime (v_6v_1)|\ge 3\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (v_3v_4)|\ge 4\), and \(|L^\prime (v_5v_6)|\ge 4\). By Lemma 6, we obtain a strong edge-coloring of G, a contradiction.

If \(\{\beta _1,\beta _2\}\cap \{\gamma _1,\gamma _2\}\ne \emptyset\), we color \(v_1v_{12}\), \(v_3v_{34}\) and \(v_5v_{56}\) with same color in \(\{\beta _1,\beta _2\}\cap \{\gamma _1,\gamma _2\}\), and color \(v_2v_{12}\), \(v_4v_{34}\) and \(v_6v_{56}\) in this order. Now, we have \(|L^\prime (v_2v_3)|\ge 3\), \(|L^\prime (v_4v_5)|\ge 3\), \(|L^\prime (v_6v_1)|\ge 3\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (v_3v_4)|\ge 4\), and \(|L^\prime (v_5v_6)|\ge 4\). By Lemma 6, we obtain a strong edge-coloring of G, a contradiction.

If \(\{\alpha _1,\alpha _2\}\cap \{\gamma _1,\gamma _2\}\ne \emptyset\), we color \(v_1v_{12}\), \(v_3v_{34}\) and \(v_5v_{56}\) with same color in \(\{\alpha _1,\alpha _2\}\cap \{\gamma _1,\gamma _2\}\), and color \(v_2v_{12}\), \(v_4v_{34}\) and \(v_6v_{56}\) in this order. Now, we have \(|L^\prime (v_2v_3)|\ge 3\), \(|L^\prime (v_4v_5)|\ge 3\), \(|L^\prime (v_6v_1)|\ge 3\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (v_3v_4)|\ge 4\), and \(|L^\prime (v_5v_6)|\ge 4\). By Lemma 6, we obtain a strong edge-coloring of G, a contradiction.

Therefore, \(\{\alpha _1,\alpha _2\}\cap \{\beta _1,\beta _2\}=\emptyset\), \(\{\beta _1,\beta _2\}\cap \{\gamma _1,\gamma _2\}=\emptyset\) and \(\{\alpha _1,\alpha _2\}\cap \{\gamma _1,\gamma _2\}=\emptyset\). Thus, we color \(v_1v_{12}\) and \(v_3v_{34}\) with \(\alpha _1\), color \(v_2v_{12}\) and \(v_5v_{56}\) with \(\beta _1\), and color \(v_4v_{34}\) and \(v_6v_{56}\) with \(\gamma _1\). Now, we have \(|L^\prime (v_2v_3)|\ge 3\), \(|L^\prime (v_4v_5)|\ge 3\), \(|L^\prime (v_6v_1)|\ge 3\), \(|L^\prime (v_1v_2)|\ge 4\), \(|L^\prime (v_3v_4)|\ge 4\), and \(|L^\prime (v_5v_6)|\ge 4\). By Lemma 6, we obtain a strong edge-coloring of G, a contradiction. \(\square\)

Proof of Theorem 2:

By Lemmas 13 and 4, G is 3-regular and each 3-vertex is on exactly one triangle in G. Let \(C=uvw\) be a triangle and \(N_G(u)=\{w, v, u_1\}\), \(N_G(v)=\{w,u,v_1\}\), \(N_G(w)=\{u,v,w_1\}\). Let \(C_1=u_1u_2u_3\) be the triangle where \(u_1\) is located, \(C_2=v_1v_2v_3\) be the triangle where \(v_1\) is located and \(C_3=w_1w_2w_3\) be the triangle where \(w_1\) is located. By Lemmas 4 and 5, \(V(C_1)\cap V(C_2)=\emptyset\), \(V(C_1)\cap V(C_3)=\emptyset\) and \(V(C_2)\cap V(C_3)=\emptyset\). We use notations in Figure 6. For C, \(C_1\) and \(C_3\), we have \(u_2w_2\notin E(G)\), \(u_2w_3\notin E(G)\) by Lemma 7. For C, \(C_2\) and \(C_3\), we have \(v_2w_2\notin E(G)\), \(v_2w_3\notin E(G)\) by Lemma 7. For C, \(C_1\) and \(C_2\), we have \(u_2v_2\notin E(G)\) by Lemma 7. Therefore, edge \(ww_1\) can not see edges \(u_1u_2\), \(v_1v_2\), and edge \(v_1v_2\) can not see edges \(u_1u_2\), \(ww_1\).

Precolor \(u_1u_2\), \(v_1v_2\) and \(ww_1\) with the same color. Color in a greedy way edges of \(G-\{u,v\}\) in the decresing order with the respect to the disctance from the cycle C, that is, if \(d(C,e_1)\ge d(C, e_2)\), then edge \(e_1\) precedes edge \(e_2\) (if egde \(e_1\) has two vertecies in distance k and \(k+1\) and edge \(e_2\) has two vertices both at distance k form C, we let \(e_1\) proceeds \(e_2\) in the ordering). For each \(e=xy\) in the order with \(d(C,x)\ge d(C,y)\ge 2\), y is adjacent to some vertex z with \(d(C,z)<d(C,y)\). So the three edges incident with z are after the edge e in the order. Clearly, at least two of the three edges at z are not precolored, thus e sees at least two uncolored edges in \(G-\{u,v\}\). On the other hand, by Lemma 4, each 3-vertex is on exactly one triangle in G. Therefore, e sees at most 7 different colors, and thus can be colored. For each \(e=xy\) in the order with \(d(C,x)\ge d(C,y)=1\), then \(e\in \{w_3w_1,w_2w_1,u_3u_1,v_3v_1\}\), and note that e sees at most 7 different colors, and thus can be colored.

Fig. 6
figure 6

Illustrations of Proof of Theorem 2

Now \(|L^\prime (uu_1)|\ge 3\), \(|L^\prime (vv_1)|\ge 3\), \(|L^\prime (uw)|\ge 4\), \(|L^\prime (wv)|\ge 4\) and \(|L^\prime (uv)|\ge 5\). Thus, we can color \(uu_1\), \(vv_1\), uw, wv and uv in this order to obtain a strong edge-coloring of G, which gives a contradition with the definition of G. This completes the proof of Theorem 2. \(\square\)

4 Final Remark

In this note, we have proved that if G is a claw-free subcubic graph other than the prism graph, then \(\chi _{s}^{\prime }(G)\le 8\). By the aid of computer experiment, for \(n\le 20\), we find the claw-free subcubic graph \(G^{\prime \prime }\) with \(\chi _{s}^{\prime }(G^{\prime \prime })=7\) (see Figure 7), and not find the claw-free subcubic graph G with \(\chi _{s}^{\prime }(G)=8\). Thus we leave the following problem for further study.

Fig. 7
figure 7

The claw-free subcubic graph \(G^{\prime \prime }\) with \(\chi _{s}^{\prime }(G^{\prime \prime })=7\)

Problem 1

If G is a claw-free subcubic graph other than the prism graph, then \(\chi _{s}^{\prime }(G)\le 7\).